106655 GPO
National Endowment for the Arts
Reading At Risk:
A Survey of Literary Reading in America
Research Division Report #46
Research Division Report #46
June 2004
National Endowment for the Arts
1100 Pennsylvania Avenue, NW
Washington, DC 20506-0001
Telephone: 202-682-5400
Produced by Tom Bradshaw and Bonnie Nichols
Research Division, National Endowment for the Arts
Mark Bauerlein, Director, Office of Research and Analysis
Editorial and Publication Assistance by Don Ball
Prepared for the Research Division, National Endowment for the Arts
by Kelly Hill, Hill Strategies Research, Hamilton, Ontario, Canada
Designed by New Vision Communications, Alexandria, Virginia
www.newviscom.com
Front Cover Photo: ©Dennis Marsico/CORBIS
Printed in the United States of America
Library of Congress Cataloging-in-Publication Data
Information not available at time of publication.
202-682-5496 Voice/TTY
(a device for individuals who are deaf or hearing impaired)
Individuals who do not use conventional print materials may contact the
Arts Endowment’s Office for AccessAbility at 202-682-5532 to obtain
this publication in an alternate format.
This publication is available online at www.arts.gov, the Web site of the
National Endowment for the Arts.
Table of Contents
Preface ....................................................................................................................................Page vii
Executive Summary ..............................................................................................................Page ix
1. Introduction..........................................................................................................................Page 1
2. Literature Participation in 2002 ........................................................................................Page 3
3. Factors in Literary Reading ..............................................................................................Page 9
4. Trends in Literature Participation, 1982 – 2002 ..........................................................Page 21
5. Summary and Conclusions ............................................................................................Page 29
Appendices..............................................................................................................................Page 33
List of Tables
Table 1. U.S. Adult Participation in Literary Activities
in the 12-Month Period Ending August 2002 ............................................................Page 4
Table 2. U.S. Adult Participation in Cultural, Sports, and Leisure Activities
in the 12-Month Period Ending August 2002 ............................................................Page 5
Table 3. Performing Arts Attendance by U.S. Adults, 2002 ............................................Page 6
Table 4. Art Museum Attendance by U.S. Adults, 2002 ..................................................Page 6
Table 5. Volunteer and Charity Work by U.S. Adults, 2002............................................Page 6
Table 6. Sporting Events Attendance by U.S. Adults, 2002..............................................Page 7
Table 7. U.S. Adults Playing Leisure Sports, 2002 ............................................................Page 7
Table 8. Demographic Characteristics of U.S. Adults Reading Literature
in the 12-Month Period Ending August 2002 ............................................................Page 9
Table 9. Literary Reading by Race and Ethnicity, U.S. Adults,
in the 12-Month Period Ending August 2002 ..........................................................Page 11
National Endowment for the Arts
iii
Reading at Risk
Table 10. Literary Reading by Region, U.S. Adults, 2002 ..............................................Page 14
Table 11. Literary Reading by Occupation Group, U.S. Adults, 2002 ........................Page 14
Table 12. Literary Reading by Parents' Education Levels, U.S. Adults, 2002 ............Page 14
Table 13. Average Number of Hours Per Day Watching TV, U.S. Adults, 2002......Page 15
Table 14. Demographic Characterizes of U.S. Adults That Read Novels or
Short Stories, Poetry, or Plays in the 12-Month Period Ending August 2002. ....Page 17
Table 15. Demographic Characteristics of U.S. Adults Listening to
Live or Recorded Readings of Novels or Poetry, 2002 ............................................Page 18
Table 16. Demographic Characteristics of U.S. Adults
Doing Creative Writing, 2002 ......................................................................................Page 19
Table 17. Literary Reading, U.S. Adults, 1982, 1992, and 2002 ....................................Page 21
Table 18. Reading or Listening to Poetry, U.S. Adults, 1982, 1992, and 2002 ............Page 22
Table 19. Creative Writing, U.S. Adults, 1982, 1992, and 2002 ....................................Page 22
Table 20. Creative Writing Classes or Lessons, U.S. Adults, 1982, 1992, and 2002....Page 22
Table 21. Literary Reading by Gender, 1982, 1992, and 2002 ........................................Page 23
Table 22. Literary Reading by Ethnicity and Race, 1982, 1992, and 2002....................Page 24
Table 23. Literary Reading by Education, 1982, 1992, and 2002 ..................................Page 25
Table 24. Literary Reading by Age, 1982, 1992, and 2002 ..............................................Page 27
Table 25. Literary Reading by Age Cohort, 1982 and 2002............................................Page 27
Appendix Table 1. Summary of Multiple Logistic Regression Predicting
Whether Adults have Read Literature ......................................................................Page 42
Appendix Table 2. Summary of Multiple Logistic Regression Predicting
Whether Adults have Read 12 or More Books ......................................................Page 44
Appendix Table 3. Correlation Analysis of Predictor Variables in the Literature
Reading and Frequent/Avid Book Reading, Logistic Regression Models ..........Page 46
National Endowment for the Arts
iv
Reading at Risk
List of Figures
Figure 1. Participation in Literary Activities, 2002 ............................................................Page 3
Figure 2. Literary Reading Rates by Age, 2002 ................................................................Page 10
Figure 3. Literary Reading Rates by Education, 2002 ....................................................Page 12
Figure 4. Literary Reading Rates by Region, 2002 ..........................................................Page 13
Figure 5. Literary Reading Rates by Gender, 1982, 1992, and 2002 ..............................Page 23
Figure 6. Literary Reading Rates by Ethnicity and Race, 1982, 1992, and 2002 ........Page 24
Figure 7. Literary Reading Rates by Education, 1982, 1992, and 2002 ........................Page 25
Figure 8. Literary Reading Rates by Age, 1982, 1992, and 2002 ....................................Page 26
National Endowment for the Arts
v
Reading at Risk
Preface
R eading at Risk is not a report that the National
While oral culture has
Endowment for the Arts is happy to issue. This
a rich immediacy that is
comprehensive survey of American literary
not to be dismissed, and
reading presents a detailed but bleak assessment of the
electronic media offer the
decline of reading’s role in the nation’s culture. For the
considerable advantages of
first time in modern history, less than half of the adult
diversity and access, print
population now reads literature, and these trends reflect a
culture affords
larger decline in other sorts of reading. Anyone who loves
irreplaceable forms of
literature or values the cultural, intellectual, and political
focused attention and
nce Jacobs
a
importance of active and engaged literacy in American
contemplation that make
society will respond to this report with grave concern.
complex communications
Reading at Risk is not a collection of anecdotes,
and insights possible. To
Photo by V
theories, or opinions. It is a descriptive survey of national
lose such intellectual
trends in adult literary reading. Based on an enormous
capability – and the many sorts of human continuity it
sample size of more than 17,000 adults, it covers most
allows – would constitute a vast cultural impoverishment.
major demographic groups – providing statistical
More than reading is at stake. As this report
measurements by age, gender, education, income, region,
unambiguously demonstrates, readers play a more active
race, and ethnicity. Conducted by the U.S. Bureau of the
and involved role in their communities. The decline in
Census and spanning 20 years of polling, the Survey of
reading, therefore, parallels a larger retreat from
Public Participation in the Arts, the data source for
participation in civic and cultural life. The long-term
Reading at Risk, is as reliable and objective as any such
implications of this study not only affect literature but all
survey can be. While not every measurement of reading
the arts – as well as social activities such as volunteerism,
was built into the study, the report provides so much data
philanthropy, and even political engagement.
in such detail that it constitutes a comprehensive factual
What is to be done? There is surely no single
basis for any informed discussion of current American
solution to the present dilemma, just as there is no single
reading habits.
cause. Each concerned group – writers, teachers,
The key results of the survey are condensed in the
publishers, journalists, librarians, and legislators – will
“Executive Summary,” which follows, but the report can
legitimately view the situation from a different
be further summarized in a single sentence: literary
perspective, and each will offer its own
reading in America is not only declining rapidly among
recommendations. The important thing now is to
all groups, but the rate of decline has accelerated,
understand that America can no longer take active and
especially among the young. The concerned citizen in
engaged literacy for granted.
search of good news about American literary culture will
Reading is not a timeless, universal capability.
study the pages of this report in vain.
Advanced literacy is a specific intellectual skill and social
Although the news in the report is dire, I doubt that
habit that depends on a great many educational, cultural,
any careful observer of contemporary American society
and economic factors. As more Americans lose this
will be greatly surprised – except perhaps by the sheer
capability, our nation becomes less informed, active, and
magnitude of decline. Reading at Risk merely documents
independent-minded. These are not qualities that a free,
and quantifies a huge cultural transformation that most
innovative, or productive society can afford to lose.
Americans have already noted – our society’s massive
shift toward electronic media for entertainment and
information.
Reading a book requires a degree of active attention
and engagement. Indeed, reading itself is a progressive
skill that depends on years of education and practice. By
Dana Gioia
contrast, most electronic media such as television,
Chairman, National Endowment for the Arts
recordings, and radio make fewer demands on their
audiences, and indeed often require no more than passive
participation. Even interactive electronic media, such as
video games and the Internet, foster shorter attention
spans and accelerated gratification.
National Endowment for the Arts
vii
Reading at Risk
Executive Summary
R eading at Risk presents the results from the
10 Key Findings
literature segment of a large-scale survey, the
Survey of Public Participation in the Arts,
1. The percentage of adult Americans reading
conducted by the Census Bureau in 2002 at the request
literature has dropped dramatically over the past 20
of the National Endowment for the Arts. This survey
years.
investigated the percentage and number of adults, age
18 and over, who attended artistic performances,
visited museums, watched broadcasts of arts
programs, or read literature. The survey sample
Decline in Literary Reading
numbered more than 17,000 individuals, which makes
it one of the most comprehensive polls of art and
1982
1992
2002
literature consumption ever conducted.
% of U.S. Adult Population Reading Literature
56.9
54.0
46.7
Reading at Risk extrapolates and interprets data on
Number of Literary Readers (in millions)
96
100
96
literary reading and compares them with results from
similar surveys carried out in 1982 and 1992. The
survey asked respondents if, during the previous
■ Less than half of the adult American population
twelve months, they had read any novels, short stories,
now reads literature.
plays, or poetry in their leisure time (not for work or
■ The 10 percentage point decline in literary reading
school). The report establishes trends in the number
represents a loss of 20 million potential readers.
of adults reading, listening to, and writing literature
■ Only the strong growth in overall U.S. population
by demographic categories of age, race, region,
of nearly 40 million adults from 1982 to 2002
income, and education. This report also compares
allowed the actual number of readers to remain
participation in literary activities with other leisure
flat at 96 million.
activities, such as watching movies and exercising.
Reading at Risk provides an invaluable snapshot of
2. The decline in literary reading parallels a decline
the role of literature in the lives of Americans. It
in total book reading.
comes at a critical time, when electronic media are
becoming the dominant influence in young people’s
worlds. Reading at Risk adds new and distressing
information to the discussion. It contains solid
Trends in Book and Literary Reading
evidence of the declining importance of literature to
our populace. Literature reading is fading as a
Percentage of U.S. Adult Population
Change, 1992 to 2002
Percentage Point (pp)
meaningful activity, especially among younger people.
If one believes that active and engaged readers lead
1992
2002
Difference
Rate of Decline
Read Any Book
60.9
56.6
-4.3 pp
-7%
richer intellectual lives than non-readers and that a
Read Literature
54.0
46.7
-7.3 pp
-14%
well-read citizenry is essential to a vibrant democracy,
the decline of literary reading calls for serious action.
■
Total book reading is declining significantly,
although not at the rate of literary reading.
■
The percentage of the U.S. adult population
reading any books has declined by -7 percent over
the past decade.
National Endowment for the Arts
ix
Reading at Risk
3. The rate of decline in literary reading is accelerating.
Rates of Decline in Literary Reading
■ The ten-year rate of decline has
1982
1992
2002
accelerated from -5 percent to -14
% of U.S. Adult Population
56.9
54.0
46.7
Percentage Point (pp) Decline
–
-2.9 pp
-7.3 pp
percent since 1992.
Rate of Decline
–
-5%
-14%
4. Women read more literature than men do, but literary reading by both groups is declining at significant rates.
Literary Reading by Gender
■ Only slightly more than one-third
of adult American males now read
Percentage by Group
Percentage Point (pp) Change
literature.
1982
1992
2002
1992-2002
1982-2002
Men
49.1
47.4
37.6
-9.8 pp
-11.5 pp
■ Reading among women is also
Women
63.0
60.3
55.1
-5.2 pp
-7.9 pp
declining significantly, but at a
slower rate.
5. Literary reading is declining among whites, African Americans, and Hispanics.
Literary Reading by Race/Ethnicity
Percentage by Group
Percentage Point (pp) Change
1982
1992
2002
1992-2002
1982-2002
White
59.8
58.0
51.4
-6.6 pp
-8.4 pp
African American
42.3
45.6
37.1
-8.5 pp
-5.2 pp
Hispanic
36.4
34.0
26.5
-7.5 pp
-9.9 pp
Other
50.2
42.7
43.7
+1.0 pp
-6.5 pp
National Endowment for the Arts
x
Reading at Risk
6. Literary reading is declining among all education levels.
Literary Reading by Education
■ The higher the education level, the
Percentage by Group
Percentage Point (pp) Change
higher the reading rate, but reading
1982
1992
2002
1992-2002
1982-2002
among every group has declined over
Grade School
21.2
17.3
14.0
-3.3 pp
-7.2 pp
the past 20 years.
Some High School
38.8
34.5
23.3
-11.1 pp
-15.4 pp
High School Graduate 54.2
49.0
37.7
-11.3 pp
-16.5 pp
Some College
72.9
65.0
52.9
-12.1 pp
-20.0 pp
College Graduate/
82.1
74.6
66.7
-7.9 pp
-15.4 pp
Graduate School
7. Literary reading is declining among all age groups.
Literary Reading by Age
Percentage by Group
Percentage Point (pp) Change
1982
1992
2002
1992-2002
1982-2002
18-24
59.8
53.3
42.8
-10.5 pp
-17.0 pp
25-34
62.1
54.6
47.7
-6.9 pp
-14.4 pp
35-44
59.7
58.9
46.6
-12.3 pp
-13.1 pp
45-54
54.9
56.9
51.6
-5.3 pp
-3.3 pp
55-64
52.8
52.9
48.9
-4.0 pp
-3.9 pp
65-74
47.2
50.8
45.3
-5.5 pp
-1.9 pp
75 & Older
40.9
40.4
36.7
-3.7 pp
-4.2 pp
8. The steepest decline in literary reading is in the youngest age groups.
■
Literary Reading by Young Adults
Over the past 20 years, young adults (18-34)
have declined from being those most likely to
read literature to those least likely (with
Percentage of Group
Rate of Decline
the exception of those age 65 and above).
Age Group
1982
1992
2002
1982-2002
18-24
59.8
53.3
42.8
-28 %
25-34
62.1
54.6
47.7
-23
■ The rate of decline for the youngest adults
All Ages
56.9
54.0
46.7
-18
(18-24) is 55 percent greater than that of the
total adult population (-28 percent vs. -18
percent).
National Endowment for the Arts
xi
Reading at Risk
9. The decline in literary reading foreshadows an
erosion in cultural and civic participation.
Participation in Cultural and Social Activities
■ Literary reading strongly correlates to
Percentage of U.S. Adult Population
other forms of active civic participation.
Literary Readers Non-Literary Readers
Perform Volunteer and Charity Work
43.0
17.0
■ Literary readers are more likely than
Visit Art Museums
44.0
12.0
non-literary readers to perform volunteer
Attend Performing Arts Events
49.0
17.0
and charity work, visit art museums,
Attend Sporting Events
45.0
27.0
attend performing arts events, and attend
sporting events.
10. The decline in reading correlates with increased
participation in a variety of electronic media,
including the Internet, video games, and portable
digital devices.
■ Literature now competes with an enormous array
of electronic media. While no single activity is
responsible for the decline of reading, the
cumulative presence and availability of these
alternatives have increasingly drawn Americans
away from reading.
■ Non-readers watch more television than do
readers.
■ In 1990, book buying constituted 5.7 percent of
total recreation spending, while spending on audio,
video, computers, and software was 6 percent. By
2002, electronic spending had soared to 24 percent,
while spending on books declined slightly to 5.6
percent.
■ A 1999 study showed that the average American
child lives in a household with 2.9 televisions, 1.8
VCRs, 3.1 radios, 2.1 CD players, 1.4 video game
players, and 1 computer.
National Endowment for the Arts
xii
Reading at Risk
Conclusion
Reading at Risk presents a distressing but objective
overview of national trends. The accelerating declines
in literary reading among all demographic groups of
American adults indicate an imminent cultural crisis.
The trends among younger adults warrant special
concern, suggesting that – unless some effective
solution is found – literary culture, and literacy in
general, will continue to worsen. Indeed, at the
current rate of loss, literary reading as a leisure activity
will virtually disappear in half a century.
Reading at Risk is testimony that a cultural legacy is
disappearing, especially among younger people.
Twenty years ago, just after the NEA 1982 survey, the
landmark study A Nation at Risk warned that “a rising
tide of mediocrity” had overtaken the school system
and threatened a generation of students. The report
sparked a massive reform effort whose consequences
are still evolving today. Reading at Risk reveals an
equally dire situation, a culture at risk. The National
Endowment for the Arts calls upon public agencies,
cultural organizations, the press, and educators to take
stock of the sliding literary condition of our country.
It is time to inspire a nationwide renaissance of
literary reading and bring the transformative power of
literature into the lives of all citizens.
National Endowment for the Arts
xiii
Reading at Risk
Chapter 1: Introduction
W ho reads literature in America?
(e.g., gender, ethnicity, race, education, and age),
socioeconomic (e.g., family income, employment
Novels, short stories, poetry, and plays have
status, and occupation), and geographic (e.g., regions
their own intrinsic value, as do all of the arts. Beyond
and states) factors.
the benefits of literature to the individual lie the
benefits to a culture as a whole. Who reads literary
Data Collection
works, what kinds, how often, and how much – all
The 2002 Survey of Public Participation in the Arts
measure the well being of a culture. Do we value
(SPPA) was conducted as a supplement to the U.S.
reading? Encourage it in our homes, schools, and the
Census Bureau’s Current Population Study (CPS),
wider community?
which scientifically selects households to represent the
The question of who reads is one of particular
socio-economic characteristics of the U.S. population.
social importance. Reading is both a reflection of
The Census Bureau also conducted arts participation
disparate education levels and a way of bridging the
surveys in 1982 and 1992.
differences among them. Reading is obviously related
In the 2002 SPPA, 17,135 adults, age 18 and older,
to the literacy of a nation, which in turn is related to
were interviewed, primarily by telephone.
the quality of life of its citizens. If literacy is the
Demographic data including gender, income,
baseline for participation in social life, then reading –
educational level, age, and race/ethnicity were used in
and reading of literary work in particular – is essential
weighting the survey data so that the results match
to a sound and healthy understanding of, and
characteristics of the total U.S. adult population. The
participation in, a democratic society.
Census Bureau determined the weight for each survey
Levels of public participation in literature also
record.
matter to the economic vitality of the publishing
Interviewed in August 2002, survey respondents
industry – from publishers large and small,
were asked about their arts participation activities in
distributors, and bookstores to individual poets,
the prior 12 months. The response rate to the survey
dramatists, and fiction writers. Books are big business
was 70 percent, similar to the rate in the second half of
in the United States. In 2000, the book industry
the 1992 survey (68 percent). Appendices A and B
published 122,000 new titles and sold a total of 2.5
provide background information and data collection
billion books, a number that has tripled over the past
procedures for both the SPPA and CPS.
25 years.1 Given these figures, some have expressed
the opinion that there is currently an “explosive
Literature vs. Books
growth in America’s book culture.”2
The SPPA investigates arts participation in a variety
Research into the habits of readers is important to
of art forms, including attendance at live concerts,
provide information about the state of literature and
plays, and dance performances; visits to art museums
literacy, to examine the market for books, and to assess
and historic sites; and participation through
the relationship of literature to other areas of arts and
broadcasts, recorded media, and the Internet. This
culture. Although this monograph cannot explore all
monograph focuses on the survey’s questions related to
of these issues, it examines a number of demographic,
reading literature, listening to readings or recordings
lifestyle, and other factors that may influence
of literature, and creative writing activities.
literature participation. In particular, the Survey of
The 2002 SPPA asked respondents if, during the
Public Participation in the Arts allows us to examine
past 12 months, they had read any novels or short
whether literary reading varies based on demographic
stories, plays, or poetry. A positive response to any of
1 See Gayle Feldman, Best and Worst of Times: The Changing Business of Trade Books, 1975-2002, Columbia University’s National Arts Journalism Program, (2003).
2 J. Peder Zane, “America the Literate,” newsobserver.com, Dec. 15, 2002.
National Endowment for the Arts
1
Reading at Risk
those three categories is counted as reading literature,
The rest of this monograph is organized as follows:
including popular genres such as mysteries, as well as
contemporary and classic literary fiction. No
• Chapter 2 summarizes the participation in literary
distinctions were drawn on the quality of literary
activities in 2002;
works.
• Chapter 3 analyzes factors affecting literature
In addition to the three questions pertaining to
participation in 2002;
literature, the SPPA asked respondents if they had
• Chapter 4 examines trends in literature participation
read any books, and, if so, how many. The distinction
between 1982 and 2002; and
between reading literary works and reading any books
• Chapter 5 provides a summary and conclusion for
is important to the analysis presented in this report.
the monograph.
Books can be of any type and cover a vast array of
subjects, literary and non-literary alike, and for the
purposes of the survey, the respondents need to have
read as a leisure time activity, not for work or school.
Literature, of course, can be found in sources
other than books. Poetry, drama, and fiction can be
read in magazines and literary journals, even on
subway and bus placards. The Poetry Society of
America, for example, provides poetry to transit
authorities throughout the country, including Los
Angeles, New York, Philadelphia, Chicago, and
Dallas.
In several sections of this monograph, literary
readers will be classified in terms of how many books
of any type that they have read. This measure of
book-reading frequency is used to investigate broad
characteristics of literary readers. For example, Table
5 (shown in Chapter 2) shows the percentage of adults
who volunteered or did charity work. About 17
percent of those who did not read literature did
charity work; more than 43 percent of literary readers
did. Almost half of literary readers who also read 50
or more books in 2002 (i.e., “avid readers”) did charity
work. Therefore, the number of books read is used to
investigate how literary readers spend their time, and
how their time usage differed from those who did not
read literature.
Throughout this monograph, information from
the SPPA will be supplemented with information on
literature from a range of sources, including research
reports, essays, newspaper articles, and books.
National Endowment for the Arts
2
Reading at Risk
Chapter 2: Literature Participation in 2002
A mericans participate in a wide range of literary percent) indicated that they read at least one book in
activities, including book reading, listening to
2002 (not counting books required for work or
recorded literary readings or attending live
school).3 This amounts to about 117 million people 18
readings, and pursuing their own creative writing.
years of age or older. The SPPA also asked about the
The 2002 Survey of Public Participation in the Arts
types of literature read.
(SPPA) asked people 18 years of age or older a series
The most common types of literature read were
of questions about their literary activities. This chapter
novels or short stories (45 percent of adults or 93
highlights findings from the SPPA on general levels of
million people).4 Poetry was read by 12 percent of
participation in literary activities; numbers of books
adults, or 25 million people.5 A smaller number of
read; and literature participation compared with other
adults (4 percent or 7 million people) indicated that
cultural and leisure activities. The SPPA shows the
they read a play during the previous year.6
levels of “crossover” participation, that is, the number
Grouping novel, short story, poetry, and play
of literary readers who also participate in a range of
readers into a single category of literary readers shows
additional cultural and leisure activities. Finally, levels
that just under one-half of all adults (47 percent) read
of literature participation in the U.S. are compared
literature in 2002. This represents 96 million people
with literature and reading surveys in other countries.
and forms the group that is the principal focus of this
Unless otherwise noted, all information in this chapter
monograph. The definition of literature used in this
refers to SPPA respondents’ activities during the one-
report encompasses any type of fiction, poetry, and
year period between August 2001 and August 2002.
plays that the SPPA respondents felt should be
For simplicity, this time period will be referred to
included and not just what literary critics might
throughout this report as 2002.
consider literature. Respondents were not asked what
Figure 1 and Table 1 summarize responses to a
genre of works they read, nor were they asked
range of literature participation questions. These
whether they read literary non-fiction such as
graphics show that over one-half of respondents (57
criticism, commentary, and essays. By this definition,
Figure 1: Participation in Literary Activities, 2002
Americans 18 years of age or older
60%
56.6%
Read any book
Read novel or short story
50
Read poetry
45.1%
Read plays
40
Listened to readings of novels
Listened to poetry readings
Did personal creative writing
30
Took creative writing classes or lessons
Source: 2002 Survey of Public Participation
20
in the Arts
12.1%
10
9.3%
5.9%
7.1%
3.6%
1.0%
0
3The wording of the question asked is: “With the exception of books required for work or school, did you read any books during the last 12 months?”
4“During the last 12 months, did you read any novels or short stories?”
5“During the last 12 months, did you read any poetry?”
6“During the last 12 months, did you read any plays?”
National Endowment for the Arts
3
Reading at Risk
Table 1
U.S. Adult Participation in Literary Activities in the 12-Month Period Ending August 2002
% of population
Millions of people
if the proportion of those reading
Read any book
56.6 %
117
literary works is subtracted from
Read literature
46.7
96
the total number of people who
Including:
Read a novel or short story
45.1
93
read at least one book of any kind,
Read poetry
12.1
25
about ten percent of the population
Read a play
3.6
7
read only non-literary books.
Listened to live or recorded readings of novels or books
9.3
19
The SPPA results also reveal
Read or listened to poetry
14.3
30
Including:
that just less than one in ten people
Read poetry
12.1
25
(9 percent) listened to live or
Listened to live or recorded readings of poetry
5.9
12
recorded readings of novels or
Did personal creative writing
7.0
14
books in the previous year.7 This
-published
1.0
2
-unpublished
6.1
13
amounts to 19 million people. A
Took creative writing classes or lessons
smaller number (6 percent or 12
-in past year
1.0
2
-ever
13.3
27
million people) listened to live or
Used Internet to learn about, read or discuss topics related to literature
9.2
19
recorded readings of poetry.8
Seven percent of adults (or 15
million people) indicated that they did some creative
Number of Books Read
writing during 2002.9 For most, this activity appears
The 2002 SPPA asked all respondents how many
to be done more for personal fulfillment than for
books they read in 2002.14 Respondents who answered
income or public enjoyment, as only 1 percent (or 2
affirmatively indicated that they read approximately
million people) had a work published.10
six books. Extrapolated for the population as a whole,
Respondents were also asked whether they took
this means Americans read about 2.1 billion books in
creative writing classes or lessons, and 13 percent of
the survey year.15
adults (or 27 million people) indicated that they had
The general book-reading habits of literary readers
taken a class at some point in their lives.11 Most
vary widely. Readers of literary works can be divided
respondents indicated that the classes were taken
into four categories: “light” readers (1-5 books during
while they were in elementary or high school. Only a
the year, both literary and non-literary), “moderate”
small percentage (1 percent or 2 million people)
readers (6-11 books per year, both literary and non-
indicated that they took creative writing classes or
literary), “frequent” readers (at least one book every
lessons during the survey year.12
month, i.e., 12-49 books per year, both literary and
Given the increasing importance of the Internet as
non-literary), and “avid” readers (about one book every
a tool for the arts, the 2002 SPPA asked respondents
week, i.e., 50 or more books per year, both literary and
whether they use the Internet to learn about, read, or
non-literary). The percentage of people in each
discuss topics related to literature. About 9 percent of
category is as follows: light readers 21 percent,
adults (or 19 million people) said they had.13
moderate readers 9 percent, frequent readers 12
percent, and avid readers 4 percent. In other words,
about one in six people reads 12 or more books in a
7 “During the last 12 months, did you listen to a reading of novels or books, either live or recorded?”
8 “During the last 12 months, did you listen to a reading of poetry, either live or recorded?”
9 “With the exception of work or school, did you do any creative writing such as stories, poems or plays during the last 12 months?”
10 “Were any of your writings published?”
11 “Have you ever taken lessons or classes in creative writing?”
12 “Did you take any of these lessons or classes in the past year?”
13 “Do you use the Internet to learn about, read or discuss topics related to literature - novels, poetry or plays?”
14 The number of books read includes all books read (outside of work or school), not just novels, short stories, poetry, and plays.
15 This median number of books read (6) is the point where half of all readers read more books and half read less. The mean (commonly called “average”) number of books read
was 18.5. (The mean is much higher than the median because of the impact of those readers who indicated that they read a very large number of books.) The total books read estimate
of 2.1 billion is a product of the number of readers (117 million) times the mean number of books read (18).
National Endowment for the Arts
4
Reading at Risk
year (just fewer than 17 percent are in the “frequent”
attending sporting or performing arts events or
or “avid” reader categories).
visiting art museums or galleries. In 2002, about the
same percentage of people read literature as watched
Comparison of Literature Participation with Other
an average of three or more hours of TV per day.
Cultural and Leisure Activities
Table 2 also shows that, with about 93 million
How does literary reading compare to other cultural,
readers, novels and short stories have a significant
sports, and leisure activities? Literary reading is
audience in the U.S. Poetry, with 25 million readers, is
clearly an important component of Americans’ leisure
about as popular as attendance at performances of
activities. The data in Table 2 demonstrate that the
jazz, classical music, or non-musical plays. About as
proportion of people reading literature is higher than
many people – 7 million – read plays as attend live
participation in most cultural, sports, and leisure
opera or ballet.
activities. In fact, of the activities included in the 2002
SPPA, only TV watching, moviegoing, and exercising
Participation of Literary Readers in Other Cultural
attract significantly more people than reading literary
and Leisure Activities
works. Literary reading is much more popular than
Readers are highly social people, frequently engaged
in the arts, sports, and community life. Analysis of
Table 2
the 2002 SPPA data in Tables 3 through 7 show
U.S. Adults Participation in Cultural, Sports, and Leisure Activities
in the 12-Month Period Ending August 2002
that people who read literature are active,
attending a variety of arts events, volunteering in
% of
Millions
their communities, and participating in sports. In
population of people
Watch at least one hour of TV per day (on average)
95.7 %
197
fact, literary readers are much more likely to
Go out to movies
60.0
124
participate than those who do not read. For
Jog, lift weights, walk or other exercise program
55.1
113
example, literary readers are nearly three times as
Work with indoor plants or do any gardening (for pleasure)
47.3
97
likely to attend a performing arts event, almost
Read literature
46.7
96
Including:
four times as likely to visit an art museum, over
Read novel or short story
45.1
93
two-and-a-half times as likely to do volunteer or
Read poetry
12.1
25
charity work, over one-and-a-half times as likely
Read play
3.6
7
to attend sporting events, and over one-and-a-half
Watch 3 or more hours of TV per day (on average)
46.2
95
Make repairs or improvements on home
42.4
87
times as likely to participate in sports activities. In
Go to amusement/theme park or carnival
41.7
86
fact, people who read larger numbers of books
Attend amateur or professional sports (excluding youth sports)
35.0
72
tend to have the highest levels of participation in
Visit art or craft festival or fair
33.4
69
other activities, especially arts activities.
Attend any of 7 live performing arts activities
31.8
65
(except school performances)
The first row of Table 3 shows that, in 2002,
Including:
32 percent of people attended a performing arts
Musical stage play or operetta
17.1
35
event (including jazz, classical music, opera,
Non-musical stage play
12.3
25
musical and non-musical plays, ballet, and other
Classical music
11.6
24
Jazz
10.8
22
dance performances). The second and third rows
Other dance
6.3
13
of Table 3 show that the performing arts
Ballet
3.9
8
attendance rates of literary readers and those who
Opera
3.2
7
did not read are strikingly different: 49 percent of
Visit historic park or monument / Tour historic buildings or neighborhoods 31.6
65
Do outdoor activities, such as camping, hiking or canoeing
30.9
64
literary readers attended a performing arts event,
Participate in any sports activity
30.4
63
compared to 17 percent of those who did not read
Do volunteer or charity work
29.0
60
literature during the year. Literary readers are
Attend art museum or gallery
26.5
55
National Endowment for the Arts
5
Reading at Risk
Literary readers are also engaged citizens. Table 5
Table 3
shows that 29 percent of all survey respondents did
Performing Arts Attendance by U.S. Adults, 2002
volunteer or charity work in 2002. As with the
Number of Attenders
findings noted above for performing arts and art
% Attending
(millions)
museum attendance, there is a striking difference
All adults
31.8 %
65
between the rates of volunteer or charity work of
Did not read literature
16.7
18
literary readers (43 percent) and non-readers (17
Read literature
49.0
47
*Light book readers (1-5 books)
41.2
18
percent). Frequent readers are most likely to have
*Moderate book readers (6-11 books)
51.9
10
*Frequent book readers (12-49 books)
57.0
14
Table 5
*Avid book readers (50 books or more)
58.2
5
Volunteer and Charity Work by U.S. Adults, 2002
*Note: Books can be of any type, literary and non-literary
% Volunteering Number of Volunteers
(millions)
nearly three times as likely to attend performing arts
All adults
29.0 %
60
events.
Did not read literature
16.6
18
The last four rows of Table 3 show that
Read literature
43.3
42
*Light book readers (1-5 books)
36.5
16
performing arts attendance rates are highest for those
*Moderate book readers (6-11 books)
45.9
8
who have read the most books. Moreover, the 47
* Frequent book readers (12-49 books)
51.1
13
million literary readers who attended a performing
*Avid book readers (50 books or more)
49.2
5
arts event in 2002 account for nearly three-quarters (72
percent) of all attendees.
*Note: Books can be of any type, literary and non-literary
Table 4 shows that 26 percent of people visited an
art museum in 2002. Literary readers are much more
volunteered or done charity work in 2002.
likely to visit art museums (44 percent) than non-
Readers of literary works attend sporting events
readers are (12 percent). Art museum attendance rates
and participate in sporting activities themselves at rates
higher than non-readers. The first row in Table 6
are highest for those who have read the most books.
shows that 35 percent of all adults attended an amateur
The 42 million literary readers who visited an art
or professional sporting event (excluding youth sports)
museum in 2002 comprise over three-quarters (77
in 2002. Sports attendance was much more common
percent) of the total art museum visiting public.
among literary readers (45 percent) than non-readers
(27 percent). Sports attendance does decrease
Table 4
somewhat for avid readers, but still remains above the
Art Museum Attendance by U.S. Adults, 2002
sports attendance rate of those who do not read literary
works.
% Attending Number of Attenders
In terms of active sports participation, Table 7
(millions)
All adults
26.5 %
55
shows that 30 percent of all adults participated in a
Did not read literature
11.6
13
sports activity in 2002. Again, literary readers are more
Read literature
43.5
42
likely to participate in sports (38 percent) than non-
*Light book readers (1-5 books)
34.9
15
readers (24 percent). Sports participation is somewhat
*Moderate book readers (6-11 books)
47.2
9
*Frequent book readers (12-49 books)
53.1
13
lower for readers of 50 books or more than for less-
*Avid book readers (50 books or more)
50.9
5
frequent readers. Still, the sports participation
rate of avid readers (31 percent) is higher than the
*Note: Books can be of any type, literary and non-literary
participation of people who do not read literary works
(24 percent).
National Endowment for the Arts
6
Reading at Risk
Table 6
(15 years of age or older).17 Like the Canadian
Sporting Events Attendance by U.S. Adults, 2002
figure reported above, this amount includes reading
any type of book outside of work or school, not just
% Attending
Number of Attenders
novels, short stories, plays or poetry. The most
(millions)
All adults
35.0 %
72
comparable U.S. figure (57 percent) is quite high
Did not read literature
26.7
29
compared to the overall European average, and is
Read literature
44.5
43
similar to the reading rate in a number of
*Light book readers (1-5 books)
43.2
19
European countries (Luxembourg 56 percent,
*Moderate book readers (6-11 books)
46.9
9
*Frequent book readers (12-49 books)
47.5
12
Denmark 55 percent, and the Netherlands 53
*Avid book readers (50 books or more)
37.7
3
percent). The highest European reading rates are in
Sweden (72 percent), Finland (66 percent), and the
United Kingdom (63 percent), and the lowest rates
*Note: Books can be of any type, literary and non-literary
are in Belgium (23 percent) and Portugal (15
percent).
Table 7
U.S. Adults Playing Leisure Sports, 2002
The European study also defined “strong readers”
as those who read eight or more books during the
% Participating
Number of Participants
year. Overall, about 37 percent of Europeans fall
(millions)
into this category, with a high of 52 percent in the
All adults
30.4 %
63
U.K. to a low of 15 percent in Portugal. A
Did not read literature
23.9
26
recalculation of the American figures shows that
Read literature
37.9
36
*Light book readers (1-5 books)
36.1
16
about 24 percent of Americans read eight or more
*Moderate book readers (6-11 books)
41.7
8
books in 2002. The percentage of “strong readers”
*Frequent book readers (12-49 books)
40.8
10
in the U.S. falls in the bottom third of the 15
*Avid book readers (50 books or more)
31.1
3
European countries surveyed.
*Note: Books can be of any type, literary and non-literary
Summary of Literature
Participation in 2002
International Comparisons
The results from the 2002 Survey of Public
A survey similar to the SPPA was conducted in
Participation in the Arts show that literary reading is a
Canada in 1998. In response to a question about their
popular but declining leisure activity, reaching about
reading habits, two-thirds of Canadians (15 years of
one-half of the adult population (47 percent), or 96
age or older) indicated that they read a book during
million people.
the survey year.16 The most comparable figure from
Among readers of literary works, the most
the U.S. survey would be the overall book reading rate
popular category is fiction, with 45 percent of the
(outside of work and school) among Americans 18
population reading novels and/or short stories. About
years of age or older. This figure is 57 percent,
12 percent read poetry, while about 4 percent read
markedly lower than the Canadian percentage of 67
plays.
percent.
In terms of the frequency of reading, during the
A recent study reported that the overall book
survey year, for literary readers:
reading rate in 15 European countries was 45 percent
16 Calculations by Hill Strategies Research Inc. from Statistics Canada’s General Social Survey 1998. This figure is based on the population 15 or older. The question asked was: “During
the past 12 months, as a leisure activity (not for paid work or studies) did you read a book?”
17 The percentages reported here reflect the percentage of respondents (15 or older) indicating that they read a book for reasons other than work or study in the survey year (2000/01).
The figures are taken from Key Figures on Cultural Participation in the European Union, a presentation by Michail Skaliotis to the International Symposium on Culture Statistics,
Montreal, October 2002. Available at www.readingeurope.org/observatory.nsf/InternationalSurveyA?OpenPage.
National Endowment for the Arts
7
Reading at Risk
• About one in five (21 percent) read 1 to 5 books;
• About one in 11 (9 percent) read 6 to 11 books;
• Almost one in eight (12 percent) read 12 to 49 books;
and
• Approximately one in 25 (4 percent) read 50 books
or more.
The SPPA results reveal that almost one in ten people
(9 percent) listened to live or recorded readings of
novels or books and 6 percent listened to poetry
readings during the survey year. About 7 percent
wrote creative works of their own, and 9 percent used
the Internet to learn about, read, or discuss topics
related to literature.
An examination of literary readers’ participation
in other cultural and leisure activities clearly shows
that literary readers are active and social. The high
degree of crossover attendance indicates that book
marketing directed at arts attenders may be effective.
This also means that arts organizations could target
literary readers in order to increase attendance. The
high level of volunteer participation by literary readers
suggests that organizations trying to develop their
volunteer base should target literary readers.
While no direct comparisons are available for
literary reading, the SPPA and other research indicate
that the U.S. falls behind Canada and several
European nations in overall reading rates.
National Endowment for the Arts
8
Reading at Risk
Chapter 3: Factors in Literary Reading
A number of demographic, lifestyle, and other
factors influence literature participation.
Table 8
Knowledge of these factors is important both
Demographic Characteristics of U.S. Adults Reading
to understand current readership and to develop
Literature in the 12-Month Period Ending August 2002
readers in different demographic groups. On one
hand, statistics demonstrate that literary reading varies
Demographic
% reading literature
Number reading literature
significantly by gender, ethnicity, race, education, and
characteristics
(millions of adults)
other variables. On the other hand, literary reading
Overall population
46.7 %
96
varies somewhat less by age and family income.
Gender
In addition to the literary reading inquiries
Female
55.1
59
presented in this chapter, two statistical models were
Male
37.6
37
created in order to identify the factors most closely
associated with literary reading and frequent literary
Ethnicity and race
Hispanic
26.5
6
reading. The statistical models help establish an
White
51.4
77
association between certain demographic
African American
37.1
9
characteristics and literary reading rates. In particular,
Other
43.7
4
the models give estimates of the likelihood of someone
Age
in a particular group (e.g., men) reading literature,
18 to 24
42.8
11
compared to others (e.g., women), accounting for the
25 to 34
47.7
18
fact that men and women do not necessarily have the
35 to 44
46.6
21
same levels of education, income, or other
45 to 54
51.6
20
55 to 64
48.9
13
characteristics that can lead to higher rates of literary
65 to 74
45.3
8
reading. The major findings of the statistical models
75 or older
36.7
6
are highlighted throughout the chapter, and Appendix
C to this monograph provides more information about
Education
Grade school
14.0
2
the models.
Some high school
23.4
5
High school graduate
37.7
24
Demographic and Geographic Factors in Literary
Some college
52.9
30
Reading
College graduate
63.1
23
Graduate school
74.3
13
Table 8 summarizes the estimates of literary reading
rates of Americans in various demographic groups
Family income
based on the 2002 Survey of Public Participation in the
$9,999 or less
32.1
5
Arts. Overall, 47 percent of adults, about 96 million
$10,000 to $19,999
37.5
8
$20,000 to $29,999
37.5
9
people, read literature during the survey year. This
$30,000 to $39,999
44.1
11
reading rate provides a benchmark measure for all
$40,000 to $49,999
47.9
8
other estimates. That is, any group with a literary
$50,000 to $74,999
52.3
18
reading rate lower than 47 percent can be considered
$75,000 or more
60.8
28
Income not reported
39.8
9
underrepresented among literary readers, and any
group with a rate over 47 percent can be thought of as
overrepresented.
More women read literature than men. The
survey found that, in 2002, 55 percent of women read
literature vs. 38 percent of men. This gap is supported
National Endowment for the Arts
9
Reading at Risk
by the statistical model, which found that men were
rates. In fact, the percentage of literary readers in
less than half as likely as women to read literature,
every age group is only a percentage point or two
once differences in education, income, and other
different from each age group’s share of the overall
demographic and economic factors were accounted
population. The report of the 2001 Consumer Research
for.
Study on Book Purchasing confirms the SPPA findings
Literary reading is also unequal among different
by stating that “books appeal to consumers of all
ethnic and racial groups.18 White Americans have the
ages.”20
highest literature participation rate (51 percent), much
In the statistical model of literary reading, the only
higher than that of African Americans (37 percent)
age group that was found to be a significant factor in
and Hispanic Americans (26 percent). The statistical
literary reading is the youngest age group (18- 24).
model of literary reading shows that, compared to
Accounting for differences in education, income, and
Hispanic Americans, white Americans are almost
other socioeconomic factors, people in this age group
twice as likely to read literature, even keeping other
were about 15 percent less likely than others to read
factors constant, such as the higher overall levels of
literature.
education among white Americans. African
Despite the relative consistency in reading among
Americans are about 30 percent more likely than
the different age groups, the fact that the youngest age
Hispanic Americans to read literature.19
group has the lowest literary reading rate raises
A percent distribution of literary
readers by ethnicity and race shows that 80
Figure 2: Literary Reading Rates by Age, 2002
percent are white, 9 percent are African
Americans 18 years of age or older
American, and 6 percent are Hispanic
60%
American. A more in-depth focus on
National Average
ethnicity and race is provided later in this
46.7%
51.6%
chapter.
50%
48.9%
47.7%
46.6%
45.3%
Age
42.8%
40%
In contrast to the distinct differences in
36.7%
literary reading by gender and ethnic or
racial background, Figure 2 shows that
30%
literary reading is reasonably consistent
between people in different age ranges,
20%
with the exception of a very low reading
rate among those 75 or older. The next
10%
lowest literary reading rate is among
young adults aged 18-24 years (43 percent),
while the highest rate is for those between
0%
18-24
25-34
35-44
45-54
55-64
65-74
75 and
45-54 years (52 percent). All of the age
older
groups (below 75) fall within this
Source: 2002 Survey of Public Participation in the Arts
reasonably narrow range of participation
18 The ethnic and racial groups examined include Hispanic Americans (of any race), non-Hispanic white Americans (referred to as simply “white Americans” in this monograph), non-
Hispanic African Americans (referred to as simply “African Americans”), and Other ethnic or racial groups.
19 The SPPA literature questions did not specify the language of books read.
20 2001 Consumer Research Study on Book Purchasing, Book Industry Study Group, p. 63.
National Endowment for the Arts
10
Reading at Risk
questions about the future of literary reading in
Ethnicity and Race
America. Is this an indication of a decline in reading,
Literary reading rates differ among ethnic and racial
possibly due to competition from TV, videogames,
groups in America, due in part to differences in
movies, and the Internet? Or is it simply a “natural”
culture, language, socioeconomics, education, and
function of the fact that younger people are
other factors outside the scope of the SPPA. able 9
T
temporarily preoccupied with other types of reading
gives a breakdown of literary reading rates by gender,
(textbooks and coursework) or other leisure activities?
age, education, and income for people in each of these
Will those in the youngest age group read more as
ethnic and racial groups. Some interesting differences
they age? Chapter 4 looks at trends in literary reading
emerge from the data in Table 9. For instance, an
since 1982, including an analysis of trends in literary
examination of literary reading rates by gender shows
reading for specific age groups.
that women have much higher literary reading rates
Table 9
than men in all ethnic and racial groups.
Literary Reading by Race and Ethnicity, U.S. Adults
In addition, the data show that the lowest
in the 12-Month Period Ending August 2002
literary reading rate is among male
Hispanic Americans (18 percent),
Demographic characteristics
Hispanic Americans
African Americans*
White Americans*
(% reading literature) (% reading literature)
(% reading literature)
followed by male African Americans (30
Overall literature reading rate
26.5 %
37.1 %
51.3 %
percent), and female Hispanic Americans
(34 percent). Male white Americans and
Gender
female African Americans read literature
Male
18.4
29.8
41.4
Female
34.1
42.9
60.7
in similar proportions (41 percent and 43
percent, respectively). Female white
Age
Americans have by far the highest
18 to 24
21.6
34.8
48.6
literary reading rate – 61 percent.
25 to 34
31.9
43.8
51.7
35 to 44
24.5
43.7
51.3
There are also interesting age-related
45 to 54
28.7
35.9
56.8
differences in literary reading habits
55 to 64
32.2
32.8
53.1
among people in different ethnic and
65 to 74
14.3
27.0
50.5
racial groups. For white Americans,
75 or older
17.7
16.7
40.3
literary reading is fairly evenly
Education
distributed by age. This is less true for
Grade school
11.7
6.5
19.2
African Americans, for whom literary
Some high school
19.9
22.2
14.0
reading is most common among younger
High school graduate
28.7
29.3
40.4
Some college
34.0
45.8
55.6
age groups (25 to 34 and 35 to 44). This is
College graduate
52.2
58.6
66.1
undoubtedly related to higher education
Graduate school
57.0
59.9
76.1
levels among younger African
Americans. There is no consistent pattern
Family income
$9,999 or less
19.5
26.9
38.6
of literary reading by Hispanic
$10,000 to $19,999
26.8
33.5
41.6
Americans. The highest reading rates are
$20,000 to $29,999
21.9
33.4
42.4
for those in the 25 to 34 and 55 to 64 age
$30,000 to $39,999
25.4
38.4
48.9
groups.
$40,000 to $49,999
23.4
45.7
50.3
$50,000 to $74,999
37.1
45.6
54.9
The data in Table 9 show that literary
$75,000 or more
39.6
51.8
62.8
reading is strongly related to education
Income not reported
22.7
30.2
44.5
levels in all ethnic and racial groups.
Those with college education have much
*Not including Hispanics
National Endowment for the Arts
11
Reading at Risk
Figure 3: Literary Reading Rates by Education, 2002
The statistical model shows that, compared
to high school graduates:
Americans 18 years of age or older
80%
74.3%
• those with a grade school education are
almost 60 percent less likely to read
70%
63.1%
literature;
• those with some high school education
60%
National Average
46.7%
52.9%
(but no diploma) are about a third less likely
50%
(35 percent) to read literature;
• those with some college education (but no
37.7%
40%
degree) are about 35 percent more likely to
read literature;
30%
23.4%
• those with a college degree are about 75
percent more likely to read literature; and
20%
14.0%
• those with a graduate school degree are
240 percent more likely to read literature.
10%
0%
Research into consumers’ book purchases
Grade
Some
High
Some
College
Graduate
School
High
School
College
Graduate
School
confirms the SPPA findings that highly
School
Graduate
educated people are overrepresented among
Source: 2002 Survey of Public Participation in the Arts
book readers. Data from the 2001 Consumer
Research Study on Book Purchasing indicate
higher literary reading rates than those with high
that, while 27 percent of all adults have at least a
school education or less. The literary reading rates in
college degree, 39 percent of books for adults are
all three ethnic and racial groups increase as family
bought by those with at least a college degree.21
income increases. A close examination of the figures
shows that the differences in literary reading between
Family Income
the higher and lower income groups are larger for
The literary reading rate does not vary as strongly
Hispanic and African Americans than for white
based on family income as it does for education,
Americans.
although literature participation does increase fairly
Education
steadily for each increase in family income (see the
figures in Table 8). About one-third of the lowest
Both the literary reading rates and the statistical
income group (those with family income under
model of literary reading show that the most
$10,000) read literature during the survey year
important factor is educational attainment. Figure 3
compared to 61 percent of the highest income group
shows that only 14 percent of those with grade school
(those with family incomes of $75,000 or more). The
education read novels, short stories, poetry, or plays in
statistical model shows that having a low family
2002. Those with a graduate school education are
income (or not reporting the family income level) is
more than five times (74 percent) more likely to read
associated with low literary reading rates. Those with
literary works. In between these two extremes, literary
low family incomes and those who did not report their
reading increases consistently at every educational
income level are about 17 percent less likely to read
level, with literature participation rates for those with
literature as those in other income levels (after
various levels of educational attainment.
adjusting for other factors, such as the fact that those
21 2001 Consumer Research Study on Book Purchasing, p. 35.
National Endowment for the Arts
12
Reading at Risk
data in Table 10 show that, of nine sub-
Figure 4: Literary Reading Rates by Region, 2002
regional groupings of states, literary
Americans 18 years of age or older
reading is most common in the two sub-
60%
regions of the West (Mountain – 53
National Average
percent and Pacific – 50 percent) and
51.2%
46.7%
49.7%
50%
least common in the three sub-regions of
46.7%
the South (South Atlantic – 43 percent,
42.1%
West South Central – 41 percent, and
40%
East South Central – 41 percent). The
sub-regions of the Northeast (New
30%
England and Mid-Atlantic – both 50
percent) and the Midwest (West North
Central – 50 percent and East North
20%
Central – 45 percent) rank in the middle
of regional participation in literature.23
10%
In addition to these regional
differences, the SPPA found several
0%
other differences. People who live in the
West
Northeast
Midwest
South
suburbs are more likely to be readers
Source: 2002 Survey of Public Participation in the Arts
than either those who live in the city or
the country. The literary reading rate for
with lower family incomes often have lower education
rural residents is 41 percent, compared with 47 percent
levels).
for central-city dwellers and 49 percent for those who
Higher income people are overrepresented among
reside in the suburbs of metropolitan areas.
those who buy books. While 23 percent of people have
household incomes of $75,000 or more, 33 percent of
Employment and Occupation
books are bought by those with household incomes in
Employed people are most likely to have read
this range.22
literature in 2002. Indeed, 49 percent of employed
Geographic Areas
people did so, compared to 41 percent of those
unemployed. Of those people who are not in the labor
Figure 4 shows that literary reading is most popular in
force (including retired and disabled Americans as
the Western states (51 percent), followed by the
well as those not looking for a job), some 43 percent
Northeast (50 percent), Midwest (47 percent) and South
read literary works in the survey year. People in
(42 percent). (See Table 10 for a list of the states
managerial, professional, and technical occupations are
included in each of these regions.) After accounting for
more likely to read literature than those in other
regional differences in education, income, etc., the
occupation groups. Table 11 shows a breakdown of
statistical model shows that, compared with residents of
literature participation in four broad occupation
the Northeast, residents of the West are 14 percent
groups.
more likely to read literature, while residents of the
Parents Education Level
Midwest and South are less likely to read literature (13
percent and 20 percent less likely, respectively).
Survey respondents were also asked about their
Table 10 lays out the literary reading rates by
parents’ education levels. Table 12 shows that reading
varies strongly based on parents’ education levels,
region and in sub-regional groupings of states. The
22 lbid, p. 35.
23 Because of the small number of survey respondents in some states, literary reading rates are not reported on a state-by-state basis.
National Endowment for the Arts
13
Reading at Risk
Table 10
Table 11
Literary Reading by Region, U.S. Adults, 2002
Literary Reading by Occupation Group, U.S. Adults, 2002
Region
Sub-region
States included in sub-region
Occupation groups
% reading literature
(% reading literature)
(% reading literature)
Managerial, professional, technical and support occupations
58.8 %
West (51.2%)
Mountain (53.4%)
Arizona
Sales occupations
40.5
Colorado
Service occupations
39.9
Idaho
Production, craft, repair, operator occupations
27.7
Montana
Nevada
New Mexico
undoubtedly due to the fact that better educated
Utah
Wyoming
parents are more likely to read to their children and
Pacific (50.4%)
Alaska
encourage children to read when they are young.24 An
California
earlier SPPA survey asked respondents about the
Hawaii
encouragement to read that they received from their
Oregon
parents. A summary of the results from the 1982
Washington
survey indicated that “the relationship between
Northeast (49.7%)
New England (50.0%)
Connecticut
Maine
parental encouragement to read and adult literary
Massachusetts
reading was quite strong, stronger even than the
New Hampshire
relationship between the respondent’s education level
Rhode Island
and literary reading.”25 The statistical model of
Vermont
literary reading based on the 2002 SPPA survey shows
Mid-Atlantic (49.7%)
New Jersey
that respondents whose mothers had attended college
New York
Pennsylvania
are nearly 50 percent more likely than others to read
Midwest (46.7%)
West North Central (49.9%) Iowa
literature in 2002.
Kansas
Minnesota
Missouri
Table 12
Nebraska
Literary Reading by Parents' Education Levels, U.S. Adults, 2002
North Dakota
South Dakota
Education levels of
Father’s education
Mother’s education
East North Central (45.5%) Illinois
father or mother
(% reading literature)
(% reading literature)
Indiana
Grade school
37.0 %
34.2 %
Michigan
Some high school
44.6
43.6
Ohio
High school graduate
49.2
49.8
Wisconsin
Some college
60.1
63.4
South (42.1%)
South Atlantic (43.3 %)
Delaware
College graduate
63.1
65.2
District of Columbia
Graduate school
73.4
76.0
Florida
Georgia
Maryland
North Carolina
Impact of Television
South Carolina
Virginia
Does TV watching affect literary reading? This is a
West Virginia
key question posed by many who feel that today’s
West South Central (40.9%) Arkansas
frequent TV watchers may not read as much as
Louisiana
previous generations have. The 2002 SPPA data show
Oklahoma
Texas
24 Statistics on the topic of parental involvement in reading can be found through the
East South Central (40.9%) Alabama
National Household Education Survey. A summary is available on the National
Institute for Literacy’s website, www.nifl.gov.
Kentucky
25 Nicholas Zill and Marianne Winglee, Who Reads Literature?, National Endowment
Mississippi
for the Arts Research Report #22, 1990. Questions about parental encouragement
Tennessee
were not asked in the 2002 SPPA.
National Endowment for the Arts
14
Reading at Risk
that literary readers watch an average of 2.7 hours of
disappointed when Oprah Winfrey cancelled the book
TV per day, while people who do not read literary
club related to her talk show.28 The effects of mass
works watch an average of 3.1 hours of TV per day.26
media, particularly television, movies, and the Internet,
Not surprisingly, a statistical model created to analyze
merit further scrutiny and research.
frequent readers found that watching four hours or
more of TV per day had a negative impact on the
Literacy
chances of someone reading 12 books or more per year.
Although 46.7 percent of the adult population read
Watching no TV had a positive impact on the
literature in 2002, a comparable percentage of adults
may not have been capable of reading and
Table 13
understanding most novels, short stories, poetry, or
Average Number of Hours Per Day Watching TV, U.S. Adults, 2002
plays. A 1995 report from the National Center for
Education Statistics (NCES) shows that 45 percent of
Average Hours Per Day
adults read at “prose literacy levels” one and two.29
All adults
2.9
Rather than reporting a single illiteracy rate, NCES
Did not read literature
3.1
classifies adults into five levels of literacy. People
Read literature
2.7
*Light book readers (1-5 books)
2.8
scoring at levels one and two probably do not have the
*Moderate book readers (6-11 books)
2.6
skills necessary to read many types of literature.
*Frequent book readers (12-49 books)
2.4
*Avid book readers (50 books or more)
2.6
Participation of Literary Readers in Other Cultural
and Leisure Activities
*Note: Books can be of any type, literary and non-literary
Previous research into performing arts and art
probability of someone reading 12 books or more.
museum attendance has shown that attendance at one
Overall, however, frequent readers watch only slightly
type of arts activity is strongly related to attendance at
less TV per day than infrequent readers.
other types of activities.30 The 2002 SPPA data show
The SPPA results cannot show whether people who
that literary reading is also strongly related to
never read literary works would do so if they watched
participation in other arts activities. For example, 77
less TV, or whether they would use this extra time in
percent of art museum visitors read literature in 2002,
other ways. A 2001 Gallup survey of 512 people
and 72 percent of people who attend performing arts
showed that regular computer users spent 1.5 hours
events also read literary works. Those who participate
per day using the Internet and 1.1 hours reading
in any of the arts are likely to participate in more than
books. However, those who did not regularly use a
one form. Some people appear to be predisposed to
computer also spent 1.1 hours per day reading a
partake in a wide range of arts activities, while others
book.27
choose not to participate in any artistic activities at all.
In some cases, TV watching may have a positive
Arts participation, once begun, tends to become a
impact on literary reading. Authors regularly appear
habit.
on TV to promote their books, and some TV book
In the statistical model of literary reading,
clubs have been extremely popular. In fact, in the
attendance at art museums and performing arts events
spring of 2002, most book publishers were very
is shown to be a significant factor in predicting literary
26 These figures were calculated from responses to the question “Approximately how many hours of television do you watch on an average day?”
27 Gallup Poll Tuesday Briefing, “Does reading still stack up?” September 3, 2002.
28 Oprah’s book club has been reinstated with a focus on classic literary works rather than contemporary books. A recent selection, for example, is East of Eden by John Steinbeck.
29 NCES Fast Facts, National Center for Education Statistics, at www.nces.ed.gov/fastfacts.
30 See, for example, two reports by Hill Strategies, Performing Arts Attendance in Canada and the Provinces and Museum and Art Gallery Attendance in Canada, 2003, available from
www.hillstrategies.com. For more information about crossovers between disciplines, see 2002 www.hillstrategies.com. For more information about crossovers between disciplines, see
2002 Survey of Public Participation in the Arts, Report #45, National Endowment for the Arts, April 2004. Also see Timothy Fisher and Stephen Preece, “Evaluating Performing Arts
Audience Overlap” in International Journal of Arts Management, vol. 4, no. 3, Spring 2002.
National Endowment for the Arts
15
Reading at Risk
reading, even controlling for education, ethnicity, race,
• white Americans are 63 percent more likely than
and other factors. Respondents who attended both an
people from other ethnic and racial groups to be
art museum and a performing arts event are nearly six
frequent readers;
times as likely to read literature as those who
• those under 45 years of age are 36 percent less likely
participated in neither of these arts activities. Those
to be frequent readers;
who visited an art museum or attended a performing
• those with at least some college education are 28
arts event are over twice as likely to read literature as
percent more likely than those without any college
those who participated in neither of these arts
education to be frequent readers;
activities.
• those respondents whose mothers had at least some
college education are 19 percent more likely to be
Demographic Characteristics of Frequent Readers
frequent readers than are those whose mothers had
Of those who read any literary works in 2002, 17
no college education;
percent are classified as frequent readers (12 or more
• those who are not in the labor force are 17 percent
books in the survey year). Women are much more
more likely than others to be frequent readers;
likely than men to be frequent readers: 22 percent of
• those who visit an art gallery and attend a
women, compared to 11 percent of men. Frequent
performing arts events in a year are more than 250
readers among racial and ethnic groups break down
percent more likely to be frequent readers than
into these percentages: 20 percent of white Americans,
those who did only one or neither of these arts
9 percent of African Americans, and 5 percent of
activities;
Hispanic Americans. Those between 45 and 74 years
• those who did charity work are 26 percent more
of age tend to read the most books in a year. About
likely to read 12 or more books per year than those
one in five people in this age group is a frequent
who did not;
reader, compared to only one in ten people under 25.
• those who did not watch TV in a typical day are 48
Frequent readers also tend to have high levels of
percent more likely to be frequent readers than are
education and family income. The West, Midwest
those who watched one to three hours of TV per
and Northeast regions of the country have similar
day; and
proportions of frequent readers (just under one in five
• those who watched four or more hours of TV in a
residents of each of these regions), while the South has
typical day are 12 percent less likely to be frequent
the lowest proportion of frequent readers (about one
readers than are those who watched one to three
in seven residents).
hours of TV per day.
Art museum and performing arts attenders are
much more likely than non-attenders to be frequent
Readers of Novels, Short Stories,
readers. In fact, 33 percent of art museum visitors
Poetry, and Plays
(compared to 11 percent of non-visitors) read 12 or
The 2002 SPPA included questions regarding three
more books during the year. Similarly, 30 percent of
types of literature: novels or short stories (a single
performing arts attenders (compared to 10 percent of
question), poetry, and plays. Novels or short stories
non-attenders) read 12 or more books in 2002.
were by far the most popular types of literature, with
The regression model (see Appendix C) shows the
45 percent of the survey respondents indicating that
demographic patterns of frequent book readers who
they had read novels or short stories in the survey year.
reported reading literature. This model demonstrates
In comparison, 12 percent read poetry and 4 percent
that:
read plays.
• men are 37 percent less likely than women to be
Many demographic patterns are consistent among
frequent readers (12 or more books of any kind per
readers of the different types of literature examined in
year);
the SPPA. For all three types of literature:
National Endowment for the Arts
16
Reading at Risk
• women read more than men;
Table 14
• white Americans are more likely than any
Demographic Characteristics of U.S. Adults That Read Novels or
other racial or ethnic group to read literature;
Short Stories, Poetry, or Plays in the 12-Month Period Ending August 2002
• reading varies slightly across age groups, with
% reading novels or short stories % reading poetry % reading plays
the exception of a drop for those 75 or older
Overall reading rate
45.1 %
12.1 %
3.6 %
and a declining rate for 18-24 year olds;
• reading rates increase strongly with education
Sex
Male
36.1
7.8
3.0
levels, with college graduate rates more than
Female
53.4
16.0
4.1
50 percent higher than the overall reading
rate for poetry or plays and more than one-
Ethnicity and race
third higher for novels or short stories;
Hispanic
24.8
6.7
3.1
White
50.0
13.1
3.8
• reading rates also increase with income levels,
African American
34.7
11.8
2.8
with people with family incomes under
Other
42.3
9.8
3.1
$40,000 reading at rates below the overall
average, and those with family incomes of
Age
18 – 24
40.7
13.9
5.9
$40,000 or more exceeding average reading
25 – 34
45.8
10.9
3.5
rates;
35 – 44
45.1
10.9
3.4
• the West and Northeast regions of the
45 – 54
50.2
14.2
3.7
country have the highest reading rates.
55 – 64
47.9
12.4
2.5
65 – 74
43.8
11.9
2.5
75 or older
35.4
9.4
3.2
There are some differences in the reading rates
of different types of literature for non-white
Education
ethnic and racial groups. The gap between
Grade school
13.4
2.5
1.1
Some high school
22.7
5.1
1.5
white Americans – the group with the highest
High school graduate
36.5
6.8
1.6
reading rates for all types of literature – and
Some college
51.1
14.3
3.9
African or Hispanic Americans varies between
College graduate
60.7
18.0
6.4
the different types of literature. For example,
Graduate school
72.3
26.1
8.2
there is a large gap between the novel or short
Family income
story reading rates of white Americans (50
$9,999 or less
29.6
10.0
3.4
percent), African Americans (35 percent), and
$10,000 to $19,999
36.2
10.4
3.1
Hispanic Americans (25 percent.) In contrast,
$20,000 to $29,999
36.2
9.2
2.5
$30,000 to $39,999
42.5
11.0
2.8
African Americans and white Americans have
$40,000 to $49,999
46.4
12.7
4.1
similar rates for poetry (12 percent and 13
$50,000 to $74,999
50.7
12.9
3.8
percent). Hispanic Americans are only
$75,000 or more
59.4
16.0
4.9
somewhat less likely than white Americans to
Income not reported
38.0
9.5
2.8
read plays (3.1 percent and 3.8 percent).
Region
West
49.5
15.4
4.8
Listening to Live or Recorded
Northeast
48.3
11.9
3.9
Readings of Novels or Poetry
Midwest
45.5
10.8
2.8
South
40.5
10.8
3.2
In addition to reading books, many people enjoy
listening to literary readings, either in person or
National Endowment for the Arts
17
Reading at Risk
through recordings. The 2002 SPPA asked
Table 15
respondents whether they listened to live or
Demographic Characteristics of U.S. Adults Listening to Live or
recorded readings of novels31 or poetry.32 Just
Recorded Readings of Novels or Poetry, 2002
over 9 percent said that they listened to live or
recorded readings of novels or other books, and
% listening to novels or other books
% listening to poetry
about 6 percent indicated that they listened to
Overall rate
9.3 %
5.9 %
poetry readings during the survey year.
Sex
Table 15 provides a detailed demographic
Male
7.9
4.8
breakdown of those who listened to novels or
Female
10.6
6.8
poetry in 2002. Women are more likely than
Ethnicity and race
men to listen to novels or poetry. While white
Hispanic
5.6
3.4
Americans are most likely to listen to book
White
10.3
5.9
readings, African Americans are most likely to
African American
7.4
8.6
listen to poetry readings. This may be due,
Other
6.6
5.1
in part, to the popularity of dub and slam poetry
Age
readings in the U.S.33 Although readings of
18 – 24
5.6
6.2
both novels and poetry are most commonly
25 – 34
7.7
5.8
attended by those between 45 and 54, there is
35 – 44
10.5
5.8
45 – 54
12.1
7.1
decidedly younger audience listening to poetry
55 – 64
10.9
5.9
readings. The second-most likely age group to
65 – 74
9.3
4.8
attend poetry readings is 18-24 year olds, while
75 or older
6.2
3.8
55-64 year olds are most likely to listen to a
Education
reading of a novel or other literary work.
Grade school
2.6
1.3
There is a strong connection between
Some high school
3.6
2.9
education and listening to both novels and
High school graduate
5.4
2.8
Some college
9.7
7.3
poetry; college graduates are more than 50
College graduate
15.3
9.1
percent more likely than the average person to
Graduate school
20.9
12.4
listen to novels or poetry. The rates of listening
to novel or poetry readings also increase with
Family income
$9,999 or less
6.1
4.6
income. As was the case with literary reading,
$10,000 to $19,999
4.9
4.5
the West has the highest rates of listening to
$20,000 to $29,999
5.9
4.7
novels or poetry of all regions of the country.
$30,000 to $39,999
8.5
6.4
$40,000 to $49,999
8.3
6.7
Personal Creative Writing
$50,000 to $74,999
11.8
6.1
$75,000 or more
14.0
7.2
The third major set of questions dealt with
Income not reported
7.7
4.9
personal creative writing,34 and about one in 14
people (7 percent) said that they wrote creative
Region
West
11.7
7.3
works during the survey year. As is the case
Northeast
8.8
6.6
with all facets of literary reading and writing,
Midwest
9.7
5.6
South
7.8
4.7
31 “Did you listen to a reading of novels or books, either live or recorded
during the last 12 months?”
32 “Did you listen to a reading of poetry, either live or recorded during
the last 12 months?”
33 Dub poetry is an oral presentation of poetic works usually combined
with music (often drumming).
34 “With the exception of work or school, did you do any creative writing
such as stories, poems or plays during the last 12 months?”
National Endowment for the Arts
18
Reading at Risk
women are more likely than men to pursue creative
Table 16
writing. Similarly, creative writing is most popular
Demographic Characteristics of U.S. Adults
among those with high education levels and among
Doing Creative Writing, 2002
residents of the western region of the country.
% doing creative writing
However, an analysis of creative writers shows
Overall rate
7.0 %
some differences between the demographic
characteristics of creative writers and literary readers
Sex
Male
5.8
(see Table 16). Interestingly, creative writing does not
Female
8.2
increase consistently by income levels. Low- and
middle-income people are about as likely to write
Ethnicity and race
creative works as those with high incomes. Other
Hispanic
4.0
White
7.6
comparisons of the data show that African Americans
African American
7.4
are as likely as white Americans to do creative
Other
5.3
writing. In addition, creative writing is most common
among those under 25.
Age
18 – 24
12.7
25 – 34
7.9
Summary of Factors in Literature
35 – 44
6.7
Participation
45 – 54
6.8
With nearly one-half of Americans participating,
55 – 64
5.0
65 – 74
4.1
literary reading is clearly one of the nation’s favorite
75 or older
3.7
pastimes. But the pleasures of reading literature are
not shared by all. Educational attainment is the most
Education
Grade school
1.0
important factor in literature participation. Highly
Some high school
2.8
educated people read literature much more than those
High school graduate
4.1
with lower levels of education. This chapter has also
Some college
9.1
shown that reading and listening to literature are
College graduate
10.6
Graduate school
12.7
consistently higher for women, urban residents, and,
generally, those with higher family incomes. White
Family income
Americans have the highest participation rates of any
$9,999 or less
7.8
ethnic or racial group in almost all literature-related
$10,000 to $19,999
5.5
$20,000 to $29,999
5.6
activities, except for the high participation rate of
$30,000 to $39,999
7.6
African Americans listening to poetry and doing
$40,000 to $49,999
7.4
creative writing. Literature participation is fairly
$50,000 to $74,999
7.0
consistent between those in different age ranges, with
$75,000 or more
8.5
Income not reported
5.9
the exception of the youngest and oldest people.
Literary activities are most popular in the West,
Region
followed by the Northeast, Midwest, and South.
West
8.3
It is not clear from the SPPA data how much
Northeast
7.3
Midwest
7.5
influence TV watching has on literary reading. People
South
5.9
who read literary works, even frequent book readers,
watch slightly less TV each day than those who do not
read at all. A statistical model created to analyze
National Endowment for the Arts
19
Reading at Risk
frequent book readers found that watching four hours
or more of TV per day had a negative impact on the
chances of someone reading 12 books or more per
year, while watching no TV had a positive impact on
the probability of someone reading 12 books or more.
The SPPA results cannot show whether people would
read more if they watched less TV, or whether they
would use this extra time in other ways.
The 2002 SPPA data show that one of the most
important factors in literary reading is participation in
other arts activities. Art museum and performing arts
attendance are significant factors in literature
participation, even adjusting for education, ethnicity,
race and other factors. The book industry might target
arts participants in their marketing in the future.
National Endowment for the Arts
20
Reading at Risk
Chapter 4: Trends in Literature Participation, 1982-2002
O ver the past 20 years, the U.S. has experienced society. Results from statistical tests (chi square) are
significant demographic change. The
also presented in this chapter in order to examine
population has grown considerably since 1982
whether the changes in literary reading between 1982,
and is becoming increasingly racially and ethnically
1992, and 2002 are statistically significant.37
diverse. Recent waves of immigration have come from
a wide range of countries. The Census Bureau noted
Changes in Literature Participation
in a recent report that “the minority population grew
Literary reading in America has not grown since 1982.
11 times more rapidly as the white non-Hispanic
In fact, it is the strong growth in the population that
population between 1980 and 2000.”35 The changing
has allowed the overall number of people reading
demographics of the United States ensure continued
literature to remain stable at about 96 million between
change in literary and artistic circles and in public
1982 and 2002.38 Along with these figures, Table 17
participation in literature and the arts.
shows that there has been a substantial decrease in the
Myriad other social and economic transformations
percentage of people reading literature, from 57
have also taken place since 1982. For example,
percent in 1982 to 47 percent in 2002, a decline of 10
technological change has increased in intensity over
percentage points.
the past 20 years. In 1982, personal computers were a
relatively new phenomenon, and the Internet was a
Table 17
small network. Now, computers and the Internet are
Literary Reading, U.S. Adults, 1982, 1992, and 2002
readily available in many homes and in most
workplaces. Videogames have also proliferated since
1982
1992
2002 Change 1982 to 2002
the early 1980s, a time when Atari sets were fairly
Number reading literature (in millions)
95.6
100.3
96.2
0.6
American population 18 or older (in millions) 168.0
185.8
205.9
37.9
new. These changes, along with the growth in
% reading literature
56.9 % 54.0% 46.7%
–10.2
network, cable, and satellite television and the advent
of video rentals, have had a significant impact on
people’s time use over the past 20 years. As early as the
Changes in Poetry Reading and Listening
1980s, fears were expressed that the U.S. was
During the 1990s, the growth in popularity of live
becoming a “nation of watchers” rather than a “nation
readings, poetry slams, and other forms led some to
of readers.”36 In the late 1990s, the Center for the
speculate about a revitalization of poetry in
Book in the Library of Congress chose “Building a
America.39 If such revitalization is occurring, it is
Nation of Readers” as its National Reading Promotion
not apparent in the figures from the 1982, 1992, and
Campaign theme. By the time of the 2002 SPPA,
2002 Surveys of Public Participation in the Arts. A
public participation in literature faced very real
comparison of the poetry readers and listeners in
challenges born out by the Survey’s findings.
these years (see Table 18) shows that, despite
This chapter examines changes in literary reading
significant population growth, the number of people
in the 1982, 1992, and 2002 Surveys of Public
reading poetry or listening to a poetry reading
Participation in the Arts, and relates these changes to
decreased from about 34 million in 1982 to just under
literary, social, and cultural trends in American
30 million in 2002. This amounts to a loss of more
35 Demographic Trends in the 20th Century, Census 2000 Special Report issued November 2002.
36 See, for example, President Reagan’s Radio Address on Education, September 8, 1984, available at http://www.reagan.utexas.edu/resource/speeches/1984/90884a.htm
37
.
The chi-square statistic can be used to test whether the literary reading rates of Americans in various groups (e.g., men, women, Hispanic Americans, white Americans, etc.) are
independent of the year the sampling was done. Within each group, a statistically significant chi-square statistic would indicate that the literary reading rates were significantly
different between the years tested.
38 In 1982, only one question was asked regarding literary reading: “During the last 12 months, did you read novels, short stories, poetry or plays?” In 1992 and 2002, three separate
questions were asked: “During the last 12 months, did you read any plays?”; “During the last 12 months, did you read any poetry?”; and “During the last 12 months, did you read any
novels or short stories?” Respondents who said “yes” to any of these three questions in 1992 or 2002 are included in Table 17. In all three years, a statistically representative sample of
Americans was surveyed. In 1982 and 2002, there were about 17,000 survey respondents, compared to about 13,000 in 1992.
It should be noted that the estimate of 96 million literary
readers differs from the 93.3 million estimate that appears on page 2, Table 1, of Research Report #45, 2002 Survey of Public Participation in the Arts, because it is based on the revised
estimate of the 1982 adult population, 168 million, and not the 164 million originally provided by the Census Bureau.
39 See, for example, “10 Years After, Poetry Still Matters” in The Chronicle of Higher Education, September 27, 2002.
National Endowment for the Arts
21
Reading at Risk
than 4 million people, or 13 percent of the 1982
Changes in Creative Writing
audience for poetry.40 The percentage of people
Classes or Lessons
reading poetry or listening to poetry decreased
Table 20 shows that the number of people who
substantially, from about 20 percent of adults in 1982
indicated that they had ever taken a creative writing
and 1992 to 14 percent in 2002.
class or lesson decreased from 30 million in 1982 to 27
million in 2002.42 The percentage of
Table 18
people taking creative writing classes or
Reading or Listening to Poetry, U.S. Adults, 1982, 1992, and 2002
lessons at some point in their lives also
decreased, from 18 percent of adults in
1982
1992
2002 Change 1982 to 2002
1982 to 13 percent in 2002.
Number reading or listening to poetry (in millions) 33.8
38
29.5
-4.3
% reading or listening to poetry
19.8 % 20.5 % 14.3 %
-5.5
Changes in Personal Creative Writing
Table 20
Contrary to the overall decline in literary
Creative Writing Classes or Lessons, U.S. Adults, 1982, 1992, and 2002
reading, the number of people doing
Change
creative writing – of any genre, not
1982
1992
2002
1982 to 2002
exclusively literary works – increased
Number ever taking creative writing classes
29.5
29
27.3
-2.2
substantially between 1982 and 2002. In
or lessons (in millions)
1982, about 11 million people did some
% ever taking creative writing classes or lessons
18.0%
15.6% 13.3%
-4.7
form of creative writing. By 2002, this
number had risen to almost 15 million
people (18 or older), an increase of about 30 percent.41
Factors in the Changes in Literature Participation
The percentage of people doing creative writing
Due to higher overall levels of education in America
increased by a much smaller amount, from just fewer
over the past 20 years and the correlation between
than 7 percent of adults in 1982 to just greater than 7
literature participation and education, one might
percent in 2002. Table 19 shows the trends in creative
think that there would have been an increase in the
writing activity.
popularity of literature since 1982. However, an
analysis of the demographic
characteristics of literary readers in 1982,
Table 19
1992, and 2002 shows a widespread
Creative Writing, U.S. Adults, 1982, 1992, and 2002
decline in the literary reading rates of
people from a range of demographic
1982
1992
2002
Change 1982 to 2002 *
backgrounds. In fact, literary reading
Number doing creative writing (in millions)
11.5
13.7
14.4
2.9
rates decreased for men, women, all
% doing creative writing
7.0 %
7.4 %
7.1 %
0.1
ethnic and racial groups, all education
* The amounts in the change column are calculated from non-rounded figures. Because of rounding, the amounts in this column
may not equal the difference between the figures in the other columns.
groups, and all age groups.
40 In 1982, only one question was asked regarding poetry reading and listening: “During the last 12 months, did you read, or listen to a reading of, poetry?” In 2002, two separate
questions were asked: “During the last 12 months, did you read any poetry?” and “During the last 12 months, did you listen to a reading of poetry, either live or recorded?”
Respondents who said “yes” to either of these questions in 2002 are included in Table 18.
41 The creative writing question asked did not vary much between 1982 and 2002. In 1982, the wording was: “During the last 12 months, did you work on any creative writings such as
stories, poems, plays and the like? Exclude any writing done as part of a course requirement.” In 1992 and 2002, the wording was: “With the exception of work or school, did you do
any creative writing such as stories, poems or plays during the last 12 months?”
42 The question regarding creative writing lessons or classes did not change between 1982 and 2002. In all three years, the wording was: “Have you ever taken lessons or classes in
creative writing?”
National Endowment for the Arts
22
Literature Participation in America
The results of the statistical test (chi square) confirm
percent to 37 percent, a drop of 12 percentage points.43
the significance of the widespread decline in literary
In contrast, the female literary reading rate decreased
reading. The statistical test shows that, between 1982
from 63 percent to 55 percent, a drop of 8 percentage
and 2002, there were statistically significant decreases
points. Figure 5 also illustrates the overall decline in
in literary reading for the following demographic
literary reading, from 56 percent in 1982 to 47 percent
groups:
in 2002.
• men and women;
Gender
• Hispanic Americans, white Americans, African
The gap in the literary reading rates between women
Americans and other ethnic groups;
and men increased between 1982 and 2002. In 1982,
• people in all categories of educational attainment;
the literary reading rate among women (63 percent)
and
• the three age groups under 45 (18-24, 25-34, and 35-
Table 21
44).
Literary Reading by Gender, 1982, 1992, and 2002
(Millions of U.S. Adults)
There are some differences in the rates of decline
1982
1992
2002
Change
% change
among different demographic groups. For example,
Men
39.5
42.2
36.9
-2.6
-6.6%
the decrease in literary reading was more pronounced
Women
56.8
58.5
59.1
2.3
4.0%
among men than among women. Figure 5 illustrates a
decrease in the male literary reading rate from 49
was 14 percentage points higher than the
rate among men (49 percent). The
Figure 5: Literary Reading Rates by Gender, 1982 , 1992, and 2002
difference increased to 17 percentage
Americans 18 years of age or older
points in 2002, with 55 percent of
80%
women and 38 percent of men reading
1982 1992 2002
literature. Changes in the number of
70%
people reading literature depend on two
main factors: changes in the literary
60%
63.0% 60.3%
reading rate and changes in population.
56.4% 54.2%
55.1%
Because of the overall 21 percent
50%
49.1%
46.7%
47.4%
increase in the population (18 years of
40%
age and older), the trends in the number
37.6%
of people reading literature were not as
30%
strongly downward as the changes in the
literary reading rates.
20%
Different demographic groups saw
varying rates of population growth and
10%
differing changes in literary reading
rates between 1982 and 2002.
0%
Overall
Men
Women
Consequently, changes in the number of
Source: 2002 Survey of Public Participation in the Arts
people reading literature varied between
demographic groups. Table 21 shows
43 This drop in percentage points should not be confused with the “percentage decrease.” The drop in percentage points is simply the difference between the literary reading rates in
1982 and 2002. The percentage change relates the difference in rates to the initial reading rate. The percentage change in the proportion of men reading literature was 24 percent
between 1982 and 2002, while the percentage change was 13 percent for women.
National Endowment for the Arts
23
Reading at Risk
that the number of male literary readers decreased
percent of white Americans and 36 percent of Hispanic
slightly, from 39 million in 1982 to 37 million in 2002.
Americans read literature, for a difference of about 23
The number of American women reading literature
percentage points. In 2002, the difference was 25
increased slightly, from 57 million to 59 million
percentage points.
between 1982 and 2002.
The gap between the literary reading rates of
white and African Americans decreased slightly, from
Race and Ethnicity
17 percentage points in 1982 to 14 percentage points in
Among all four ethnic and racial groups examined, the
2002. The gap between the literary reading rates of
literary reading rate decreased most strongly among
white Americans and people from other ethnic groups
Hispanic Americans. At 36 percent, the literary
also decreased slightly between 1982 and 2002. In 1982,
reading rate of Hispanic Americans was the lowest of
the gap was nearly 10 percentage points; in 2002, the
all ethnic and racial groups in 1982. This rate
Table 22
decreased by 10 percentage points to 26 percent
Literary Reading by Ethnicity and Race, 1982, 1992, and 2002
in 2002. Figure 6 shows that there were
(Millions of U.S. Adults)
significant, but lower, rates of decline in literary
reading among white Americans (-8.4 points),
1982
1992
2002
Change
% change
Hispanic
African Americans (-5.2), and people of Other
3.4
5.2
6
2.6
74.7 %
White
83.3
83.5
77
-6.2
-7.5
ethnic and racial origins (-6.5).
African American
7.6
9.5
8.8
1.2
15.4
The gap between the literary reading rates
Other
2
2.4
4.1
2.2
111.5
of white and Hispanic Americans increased
slightly between 1982 and 2002. In 1982, 60
gap was 6.5 percentage points.
At the same time as the literary
Figure 6: Literary Reading Rates by Ethnicity And Race, 1982, 1992, and 2002
reading rate decreased very strongly
Americans 18 years of age or older
among Hispanic Americans, there was
60%
also a sharp increase in the Hispanic
1982 1992 2002
American population. Given this
59.8%
58.0%
population increase, the number of
50%
Hispanic Americans reading literature
51.4%
50.2%
increased by nearly 3 million between
40%
45.6%
1982 and 2002. This increase, which
42.3%
43.7%
42.7%
amounts to nearly 75 percent of the
37.1%
30%
36.4%
number of Hispanic American literary
34.0%
readers in 1982, is the largest increase
26.5%
in the number of readers from any
20%
ethnic or racial group.
There was also significant growth
10%
in the African American population
and in the population from other
ethnic groups. Because of these
0%
Hispanic
White
African American
Other
important demographic shifts, the
Source: 2002 Survey of Public Participation in the Arts
number of readers from African
National Endowment for the Arts
24
Reading at Risk
American and other backgrounds increased between
decreased significantly for people with all levels of
1982 and 2002. In contrast, the number of white
educational attainment. In fact, the literary reading
Americans reading literature fell by more than 6
rate decreased by 15 percentage points or more for
million between 1982 and 2002. In summary, because
those in all except the lowest education group (grade
of the changing demographics of the U.S., there was
school only).
an increase in the number of literary readers from
all ethnic and racial groups except white
Table 23
Literary Reading by Education, 1982, 1992, and 2002
Americans.
(Millions of U.S. Adults)
These changes in the number of literary
readers are highlighted in Table 22. White
1982
1992
2002
Change
% change
Grade school
4.2
2.4
1.6
-2.5
-60.9
Americans represented 80 percent of literary
%
Some high school
9.9
7.6
4.7
-5.2
-52.6
readers in 2002, down from 87 percent in 1982.
High school graduate
33.1
32.6
24
-9.1
-27.5
African Americans constituted 9 percent of literary
Some college
26.1
27.2
30
4
15.2
readers in 2002, a slight increase from 8 percent in
College graduate / Graduate school
23.5
31.1
35.6
12.2
52.0
1982. Hispanic Americans comprised 6 percent of
literary readers in 2002, up from 4 percent in 1982.
The gap between the literary reading rates of
Finally, Americans from other ethnic and racial groups
college graduates and high school graduates remained
represented 4 percent of literary readers in 2002, an
large but stable between 1982 and 2002. In 1982, the
increase from 2 percent in 1982.
difference between the reading rates of college
graduates (82 percent) and high school graduates (54
Education
percent) was about 28 percentage points. By 2002, after
Figure 7 illustrates that the literary reading rate
a significant drop in the literary reading rates of both
groups, the gap was 29 percentage
points.
Figure 7: Literary Reading Rates by Education, 1982, 1992, and 2002
Despite the sharp decreases in literary
Americans 18 years of age or older
reading at all education levels, rising
100%
1982 1992 2002
levels of education in American society
led to an increase in the number of
literary readers who had some college
80%
education or a college degree. Table 23
82.1%
shows that the number of readers with
72.9%
74.6%
a college degree or graduate education
60%
66.7%
increased by about 12 million. The
65.0%
number of literary readers with some
54.2%
52.9%
college education increased by about 4
40%
49.0%
million. There were decreases in the
38.8%
37.7%
number of literary readers at the three
34.5%
other education levels. In particular, the
20%
number of literary readers with a high
21.2%
23.4%
17.3%
school education decreased by 9 million.
14.0%
0%
Grade School
Some
High School
Some
College Graduate/
High School
Graduate
College
Graduate School
Source: 2002 Survey of Public Participation in the Arts
National Endowment for the Arts
25
Reading at Risk
Age Groups
literary reading rate of people under 45 did not rank
Although literary reading rates decreased between
as highly among the seven age groups. In 2002, the 18
1982 and 2002, the only statistically significant
to 24 age group had a literary reading rate (43 percent)
decrease was for those 18 to 44 years of age. While all
that placed it sixth among the seven age groups. The
three age groups under age 45 dropped in percentage,
48 percent literary reading rate of those between 25
for those adults 18 to 24 years of age, the literary
and 34 ranked third among all age groups, while the
reading rate decreased from nearly 60 percent in 1982
47 percent rate for people between 35 and 44 ranked
to 43 percent in 2002 – a drop of 17 percentage points.
fourth. People between 45 and 54 had the highest
Young adults are reading much less than they
literary reading rate in 2002 (52 percent). For people
used to. Making literary reading appeal to teenagers
between 55 and 64, the literary reading rate of 49
also appears to be a significant problem. Long-term
percent placed this group second among the seven age
reading assessments, summarized by the National
groups. The literary reading rates of those between 65
Institute for Literacy, show that:
and 74 (45 percent) and people 75 and older (37
percent) ranked these groups fifth and seventh
• a smaller percentage of 13- and 17-year-olds read for
respectively.
fun daily in 1999 than in 1984;
In 1982, reading literature peaked between ages 18
• a smaller percentage of 17-year-olds saw adults
and 34, gradually falling off as readers aged. In 2002,
reading in their homes in 1999 than in 1984; and
fewer younger people are reading literature; the peak
• a greater percentage of 17–year-olds were watching
age group is 45-54, with sloping off rates at both ends
three or more hours of television each day in 1999
of the spectrum.
than in 1978.44
Population growth and aging led to increases in
the number of people reading literature in all age
For those in the 25 to 34 age
group, literary reading decreased
Figure 8: Literary Reading Rates by Age, 1982, 1992, and 2002
Americans 18 years of age or older
by more than 14 percentage
points, from 62 percent in 1982 to
80%
1982 1992 2002
47 percent in 2002. The literary
reading rate among those 35 to 44
70%
years of age decreased by 13
percentage points, from 60 percent
60%
62.1%
in 1982 to 47 percent in 2002.
59.8%
59.7%
58.9%
50%
A close examination of Figure
54.9%
56.9%
53.3%
54.6%
52.8%
52.9%
51.6%
50.8%
8 also reveals a shift in the
48.9%
40%
47.7%
47.2%
46.6%
45.3%
popularity of literary reading
42.8%
40.9%
40.4%
between 1982 and 2002, especially
30%
36.7%
among the younger age groups. In
1982, the 25 to 34 age group had
20%
the highest literary reading rate
(62 percent), followed by the 18 to
10%
24 and the 35 to 44 age groups
0%
(each 60 percent). In 2002, the
18-24
25-34
35-44
45-54
55-64
65-74
75 and older
Source: 2002 Survey of Public Participation in the Arts
44 See the National Institute for Literacy’s online Reading Facts available at www.nifl.gov.
National Endowment for the Arts
26
Reading at Risk
groups 35 or older. The increase in the number of
Although this seems to show that literary reading is
literary readers was largest for those between the ages
not being replaced by moviegoing, it should be noted
of 45 and 54 (this age group encompasses the largest
that it is possible that those who went to the movies in
share of the baby boomer cohort). There were more
2002 may have gone to more movies per year (than
than 7 million more literary readers in this age
moviegoers in 1982), leaving them with less time to
category in 2002 than in 1982. This is about double the
read literature.
increase in the number of literary readers between the
Watching movies at home is also very popular,
ages of 35 and 44. For those 55 and older, the increase
with many people setting up full “home theaters” to
in the number of literary readers was fairly small.
watch movies acquired through satellite or cable
There were significant decreases in the number of
connections or the rental of videotape and digital
literary readers in the 18 to 24 and 25 to 34 age groups.
video disks. In 1982, videocassette recorders were just
Not only have the percentages fallen for young adults,
starting to achieve widespread home popularity.
but the actual numbers of young readers have fallen,
Between 1985 and 2000, annual consumer spending on
despite overall population grown. See Table 24 for
television, radios, and sound equipment increased by
full details.
68 percent, from $371 per household in 1985 to $622 in
An important question about literary reading
2000. In comparison, annual spending on reading
concerns trends in age cohorts over time. This section
increased by only 4 percent, from $141 per household
compares the literary reading rates by age in 1982 with
in 1985 to $146 in 2000.45
the age groups to which those people would belong in
2002. As an example, the literary reading rate of 25 to
Table 24
Literary Reading by Age, 1982, 1992, and 2002
34 year olds in 1982 was 62 percent. In 2002, this
(Millions of U.S. Adults)
group of people would have been between 45 and 54.
The literary reading rate of this group in 2002 was
1982
1992
2002
Change
% change
under 52 percent, a decrease of more than 10
18 to 24
17.8
12.9
11.4
-6.3
-35.7 %
percentage points. Table 25 shows that there has been
25 to 34
24.9
23.1
17.6
-7.3
-29.2
35 to 44
17
23.4
20.5
3.5
20.7
a substantial decrease in literary reading between 1982
45 to 54
12.7
15.8
20.1
7.4
58.4
and 2002 for every age cohort. In fact, all of the age
55 to 64
12
11.3
12.6
0.6
5.0
cohorts have seen decreases of about 10 percentage
65 to 74
7.8
9.3
8
0.2
2.5
points or more since 1982.
75 and older
4.1
5
5.7
1.6
39.3
Other Leisure Activities
A comparison of the 1982, 1992, and 2002 Surveys of
Table 25
Public Participation in the Arts shows that time spent
Literary Reading by Age Cohort, 1982 and 2002
watching TV has remained stable for those 18 and
(Millions of U.S. Adults)
older. In 1982 and 1992, the average amount of TV
watched per day was 3.0 hours. This decreased slightly
1982
2002
Change
18 - 24 in 1982 / 35 - 44 in 2002
59.8 %
46.6 %
-13.2
to 2.9 hours per day in 2002. These figures suggest
25 - 34 in 1982 / 45 - 54 in 2002
62.1
51.6
-10.5
that TV watching may not be an important factor in
35 - 44 in 1982 / 55 - 64 in 2002
59.7
48.9
-10.8
the overall decrease in literary reading.
45 - 54 in 1982 / 65 - 74 in 2002
54.9
45.3
-9.6
Movies are a popular leisure activity in the United
55 - 64 in 1982 / 75 and older in 2002
52.8
36.7
-16.1
States, but the percentage of people going to at least
one movie in a year decreased slightly over the past 20
years, from 63 percent in 1982 to 60 percent in 2002.
45 U.S. Census Bureau, Statistical Abstract of the United States: 2002, Table 1212, p. 748.
National Endowment for the Arts
27
Reading at Risk
Summary of Trends in Literature
Participation
The statistics presented in this chapter clearly
demonstrate that there has been a widespread decline
in literary reading over the past 20 years. There were
significant decreases in the literary reading rates of
Americans from a wide range of demographic groups,
including:
• men and women;
• people from all ethnic and racial groups;
• people with all levels of educational attainment; and
• those under age 45.
Poetry suffered between 1982 and 2002 with fewer
people reading poetry or listening to live poetry
readings or recordings. Creative writing is one of the
few literary activities in which a higher percentage of
people participated in 2002 than in 1982.
Although nearly half of Americans read literature
in 2002, literary participation is clearly less popular
than it used to be, possibly due to competition for
entertainment time and money from a range of other
options, including videogames, movies, and the
Internet. Consumer spending data show a significant
increase in spending on television, radios, and sound
equipment between 1985 and 2000. Because of the
overall increase in the population between 1982 and
2002, the trends in the number of people reading
literature were not as strongly downward as the
changes in the literary reading rates.
For all who are interested in reading habits in
America – from the book industry, literacy
organizations, and arts funders to governments and
businesses that depend on a well-educated, literate
workforce – the trends presented in this chapter point
to the importance of attracting and exciting readers
from a broad range of demographic groups.
National Endowment for the Arts
28
Reading at Risk
Chapter 5: Summary and Conclusions
I n a recent essay, “What use is literature?,”
exception to this is the high participation rate of non-
Myron Magnet stated that “data are
Hispanic African Americans listening to poetry.
meaningless until we can articulate a story that
The pleasure of literature participation appears to
makes sense out of them, and literature makes sense
be shared fairly equally among those in different age
out of the data of human experience.”46
ranges, although the youngest and oldest age groups
have the lowest participation rates.
Who Reads Literature?
Literary activities are most popular in the West,
Data from the 2002 Survey of Public Participation in
followed by the Northeast, Midwest and South.
the Arts (SPPA) demonstrate that many people enjoy
It is not clear from the SPPA data how much
literature. Novels, short stories, poetry, and plays
influence TV watching has on literary reading. Not
attract almost one-half of those 18 or older (47 percent
surprisingly, a statistical model created to analyze
or about 96 million people). Each part of the literary
frequent readers found that watching four hours or
puzzle examined in this monograph – novels, short
more of TV per day had a negative impact on the
stories, poetry, and plays – attracts a significant
chances of someone reading 12 books or more per
number of people. Poetry (read by 25 million adults) is
year.47 Watching no TV had a positive impact on the
about as popular as attendance at jazz performances
probability of someone reading 12 books or more.
or at classical music events. About as many people
Literary readers watch slightly less TV each day than
read plays (7 million) as attend live opera or ballet.
non-readers, and frequent readers watch only slightly
Novels and short stories have an audience (93 million)
less TV per day than infrequent readers. The SPPA
that is larger than almost any other cultural or leisure
results cannot show whether non-readers would read
pursuit. A number of people have a particularly
more if they watched less TV, or whether they would
strong attachment to books; about one in six literary
use this extra time in other ways.
readers (17 percent) read 12 or more books in 2002.
Literacy rates and levels (see Literacy on page 15)
Americans participate in literature in a variety of
have a significant impact on literature participation.
other ways. Almost one in ten (9 percent) listened to
More research into the leisure time habits of well-
live or recorded readings of novels or books, and 6
educated non-readers would be required in order to
percent listened to poetry readings during the survey
determine ways of raising literacy in the U.S.,
year. About 7 percent wrote creative works of their
although it is evident that people have more arts and
own, and 9 percent used the Internet to learn about,
entertainment options competing for their leisure time
read, or discuss topics related to literature. Most
than ever before.
literary readers are active in a wide range of other
cultural and leisure pursuits.
Participation in Other Arts Activities
The 2002 SPPA data show that one of the most
Factors in Literature Participation
important factors in literary reading is participation in
Those with high levels of education are much more
other arts activities. Art museum and performing arts
likely to read literature than those with lower levels of
attendance are significant indicators for literature
educational attainment. Reading and listening to
participation, even adjusting for education, ethnicity,
literature are consistently higher for women, urban
race, and other factors.
residents, and, generally, people with higher family
incomes. Non-Hispanic white Americans have the
highest participation rates of any ethnic or racial
group in almost all literature-related activities. One
46 In City Journal, Summer 2003, www.city-journal.org.
47 The details of the statistical models created for this report are included in Appendix C.
National Endowment for the Arts
29
Reading at Risk
Trends in Literature Participation
surfing for reading literary works.
The percentage of U.S. adults reading literature
Demographic factors may have also contributed to
dropped from 56.4 percent in 1982 to 46.7 percent in
the overall decline in literary reading rates. Over the
2002 – a decline of almost 10 percentage points. This
20-year span of this analysis, Hispanics doubled their
may indicate a downward trend over the past two
share of the total U.S. population, rising from 6.4
decades, but it is important to note that the SPPA is
percent of the total in 1980 to 12.5 percent in 2003.49
not conducted on a yearly basis. This monograph
At the same time, the 2002 SPPA reports that the
looks at the surveys held in 1982, 1992, and 2002 – ten-
literary reading rate for Hispanics was only half that
year snapshots. No information is available for non-
of non-Hispanic whites. So, the dramatic population
SPPA years, and it is possible that the 2002 drop is a
growth of Hispanics may have contributed to lower
short, one-year change.
literary reading rates for the adult population as a
If the 2002 data represent a declining trend, it is
whole.
tempting to suggest that fewer people are reading
But this assumption has shortcomings. If
literature and now prefer visual and audio
Hispanic populations were a significant reason for
entertainment. Again, the data – both from SPPA
falling rates of literary reading, presumably the SPPA
and other sources – do not readily quantify this
would have reported heavier decreases in literary
explanation.
reading by adults with lower levels of education (given
As discussed in Chapter 3, television does not
that education levels are lower for Hispanics).
seem to be the culprit. In 2002, those who do read and
Instead, the survey reports that literary reading fell
those who do not read literature watched about the
among all education levels. Still, population growth
same amount of TV per day – three hours’ worth.
for this ethnic group is considerable. For this reason
The Internet, however, could have played a role.
alone, arts agencies and policy makers may want to
During the time period when the literature
target Hispanics for programs to raise literary reading
participation rates declined, home Internet use soared.
rates.
According to a 2000 Census Bureau report, 42 percent
These are only two factors of many that could
of households used the Internet at home – up
have contributed to the drop in the share of adults
dramatically from 26 percent in 1998, one of the
reading literature in 2002. Newspaper and magazine
earliest years of the Bureau’s tracking.48 By contrast,
articles about post-September 11 developments and
literary reading rates reported in 1982 and 1992 were
the war in Afghanistan may have hindered literary
virtually identical in a period before the Internet was
reading during the survey year.
widely available. It was not until 2002 that the
Additional data and research are needed to
reported percentage of adults reading literature
explain why literary reading fell in 2002 and if indeed
dropped considerably.
there is a pattern of decline over time. More research
Also, home Internet users have a similar profile to
is needed to complete the portrait of the reader of
literary readers. They are likely to be well educated
literary works. Some questions for a research agenda
(bachelor’s degree or higher) and belong to the age
and a national conversation on literature participation
group whose reading rates show the greatest
might include:
percentage drop. Home Internet use was the greatest
(50.2 percent) for people ages 25-44. However, this
• How does literature, particularly serious literary
pattern of falling literary reading rates timed with
work, compete with the Internet, popular
rising home Internet use may only be coincidental.
entertainment, and other increased demands on
The SPPA did not ask if people substituted Internet
leisure time?
48 U.S. Department of Commerce, U.S. Census Bureau, “Home Computers and Internet Use in the United States: August 2000.” Current Population Report, P23-207, September, 2001.
49 U.S. Department of Commerce, U.S. Census Bureau, “The Hispanic Population.” Census 2000 Brief, May 2001.
National Endowment for the Arts
30
Reading at Risk
• How do parents, communities, schools, and the
education system respond to illiteracy and aliteracy?
• Have changes in the ways publishers choose and
market books had any effect on literature
participation?
• If education levels are the surest predictor of
literature participation, what can be done to increase
the reading level by less educated adults?
• What factors are at work in the decline in reading
literary works among people ages 18 to 45? Are we
losing a generation of readers?
The National Endowment for the Arts and other
organizations involved with literature can look
forward to future arts participation surveys (the next
SPPA is scheduled for 2007) to further investigate who
reads literature.
National Endowment for the Arts
31
Reading at Risk
Appendix A
Survey of Public Participation in the Arts
and sections. For example, the arts education category
Background and Methodology
includes new questions about the location of arts
In 1982, the National Endowment for the Arts (NEA)
education and if survey respondents had taken arts
initiated the Survey of Public Participation in the Arts
education lessons in the last year. The survey dropped
(SPPA) to examine the public’s involvement in a
questions about craftwork lessons. Some questions
variety of arts and art forms. The survey examines
concerning live attendance, recreation lifestyle, music
general participation in the performing arts, visual
preferences, arts creation, and media participation
arts, historic site visits, music, and literature.
were also modified.
1982 and 1985 Surveys of Public Participation
1997 Survey of Public Participation
in the Arts
in the Arts
The U.S. Census Bureau administered the first SPPA
From June to October of 1997, the NEA conducted a
in conjunction with the 1982 National Crime Survey,
fourth edition of the SPPA. This survey was not
which surveyed over 72,000 individuals. More than
administered through the U.S. Census Bureau’s
17,000 respondents over 18 years of age were
National Victimization Survey. Instead, the NEA
interviewed for the SPPA.
retained Westat Corporation of Rockville, Maryland, a
The survey concentrated primarily on live arts
private research firm, to conduct a stand-alone
participation, consisting of ten questions. The survey
telephone survey. Using a random-digit dial telephone
asked the live arts questions of all survey respondents.
survey plan, more than 12,000 individuals over 18
On a rotating basis, it asked questions pertaining to
years of age were interviewed. However, due to the
arts education, non-arts leisure activities, arts facilities,
considerable differences in survey methodologies, the
music preferences, arts creation and other
1997 telephone survey produced results that are not
participation, media engagement, and barriers to
comparable to previous or subsequent surveys. The
attending live performances categories. As a result, the
1997 survey can only be analyzed as a stand-alone,
sample sizes for the latter questions were smaller than
point-in-time survey, and not in an SPPA historical
for the live arts category. Approximately 25 percent of
analysis.
the interviews were completed by phone.
2002 Survey of Public Participation
In 1985, the NEA sponsored a second SPPA
in the Arts
through the U.S. Census Bureau as part of the newly
named National Crime Victimization Survey. It was
In 2002, the NEA sponsored its fifth and most recent
nearly identical to the 1982 survey.
SPPA. For the first time, the 2002 SPPA was
implemented as a supplement to the U.S. Census
1992 Survey of Public Participation
Bureau’s Current Population Survey (CPS), a research
in the Arts
instrument issued monthly to approximately 60,000
The third SPPA, conducted in 1992, surveyed 12,736
households throughout the nation for use by the U.S.
individuals over 18 years of age and followed the same
Bureau of Labor Statistics. The U.S. Census Bureau
format as previous surveys. However, questions about
weighted each survey record so that the survey sample
arts education, non-arts leisure activities, etc. were
would reflect the nation’s estimated demographic and
asked throughout the year, thereby providing a larger
geographic profile.
sample size for these questions than in the 1982 survey.
A total of 17,135 respondents in the August 2002
Additionally, the NEA refined the survey instrument
CPS participated in the 2002 SPPA. Respondents were
by adding, omitting, and modifying various questions
asked all of the 2002 SPPA questions, resulting in an
National Endowment for the Arts
33
Reading at Risk
overall sample size applicable to the entire survey.
In general, the 2002 survey instrument followed
that used in the 1992 SPPA, only incorporating slight
modifications. For instance, the media engagement
category added a video (DVD) discs option to
questions about watching arts programs via television.
The music preferences category added and omitted
various music categories, and two new sections were
added to the end of the survey. The first new section
asked about trips away from home to attend arts
events or art museums, and the second new section
asked the individual’s frequency of reading about,
discussing, or accessing information about the arts via
the Internet.
National Endowment for the Arts
34
Reading at Risk
Appendix B
Data Collection
The supplement contained questions about the
household member’s participation in various artistic
Basic CPS
activities from August 1, 2001 to August 1, 2002. It
The basic Current Population Survey (CPS) collects
asked about the type of artistic activity, the frequency
primary labor force data about the civilian
of participation, training and exposure, music
noninstitutional population. Interviewers ask
preferences, length of travel for trips to artistic events,
questions concerning labor force participation about
school-age socializations, and computer usage related
each member 15 years and older in every sample
to artistic information. Interviews were conducted
household.
during the period of August 18-24, 2002.
The 2002 monthly CPS sample was selected from
the 1990 Decennial Census files with coverage in all 50
states and the District of Columbia. The sample is
continually updated to account for new residential
construction. To obtain the sample, the United States
was divided into 2,007 geographic areas. In most
states, a geographic area consisted of a county or
several contiguous counties. In some areas of New
England and Hawaii, minor civil divisions are used
instead of counties. These 2,007 geographic areas
were then grouped into 754 strata, and one geographic
area was selected from each stratum.
About 60,000 occupied households are eligible for
interview every month out of the 754 strata.
Interviewers are unable to obtain interviews at about
4,500 of these units. This occurs when the occupants
are not found at home after repeated calls or are
unavailable for some other reason.
Each month’s sample is composed of eight
rotation groups that rotate on a schedule of four
months in sample, eight months out, and four months
back again.
August 2002, Survey of Public Participation in
the Arts (SPPA) Supplement
In addition to the basic CPS, interviewers asked
supplementary questions on public participation in the
arts. Households in the outgoing rotation groups are
eligible sample households for the supplement;
therefore only 25 percent of the basic CPS households
were sampled. The SPPA supplement attempted to
obtain self-responses from household members ages 18
and over. Proxy responses were allowed if attempts
for a self-response were unsuccessful.
National Endowment for the Arts
35
Reading at Risk
Appendix C
The main source of information for this research
characteristics that cannot be quantified. Ethnic
monograph is the 2002 Survey of Public Participation
background is a typical categorical variable where
in the Arts (SPPA), conducted in August 2002 by the
there are a number of categories with no intrinsic
Census Bureau on behalf of the National Endowment
order. Gender is a particular type of categorical
for the Arts. More than 17,000 people 18 years of age
variable known as a dichotomy since it can equal only
or older responded to the SPPA’s questions about
one of two possible values. Similarly, whether a person
participation in the arts, including questions about the
has read literature could be described as a
number and types of books they read. Respondents
dichotomous variable (i.e. person has read
were asked to respond to the survey questions based
literature/person has not read literature). In the
on their activities during a 12-month period leading
analysis reported here, whether a person has read
up to August 2002. The response rate to the SPPA was
literature is also the dependent variable
70 percent.
(conventionally known as ‘Y’) since it is the outcome
This appendix provides an overview of two
to be predicted based on a set of independent
statistical models that were created in preparing this
variables.
monograph:
Mathematical and Statistical Considerations
1. a model of factors that can be used to predictÄ
The following mathematical and statistical discussion
literary reading in 2002; andÄ
has been derived from Logistic Regression Using the
2. a model of factors that can be used to predictÄ
SAS System: Theory and Application by Paul D. Allison
frequent reading in 2002Ä
(1991). Further review can be obtained from Pampel
(i.e., reading 12 books or more).Ä
(2000) and Wright (1995) as listed in the References
section at the end of this appendix.
Variables and statistical considerations common to the
Ordinary least squares (OLS) regression or linear
two models will be discussed first.
regression is the most common statistical technique
employed to predict the value of a dependent variable
Variables
based on a given set of independent variables. In OLS
To predict whether or not a person will read
regression, independent variables can be continuous,
literature, some basic background knowledge is
categorical, or interval. However, if OLS is used to
required. These basic facts are known as independent
predict the value of a dichotomous dependent variable
variables (conventionally known as x variables) and
such as ‘literary reading,’ violations of the statistical
often include items such as a person’s age, gender,
assumptions of ordinary linear regression and a
ethnic background, education, etc. These examples
conceptual problem occur. As a result, OLS regression
represent several different types of independent
is not suitable to an analysis of literary reading. A
variable; identification of each variable type is
demonstration of the conceptual problem can be
pertinent since it determines the statistical technique(s)
obtained by setting Y equal to ‘1’ (read literature) or ‘0’
available. Age could be treated as a continuous
(did not read literature) and then attempting to
variable since it is quantifiable and can be any
predict the expected value of Y using a set of
conceivable value within a specific range (e.g. 45.6667
dependent variables. Since 0 and 1 are the only
years old), but is more commonly identified as an
possible values for Y, the expected value of Y is
interval variable where the difference between each
equivalent to the probability that Y is equal to 1 (i.e.
value is measurable and constant (e.g. 44, 45, 46 years
the probability that a given person read literature).
old). Ethnic background and gender are examples of
categorical variables and measure qualitative
National Endowment for the Arts
36
Reading at Risk
This can be seen in the following equations where
Equation 2
E(Yi) is the expected value of Y for any given i person:
O = p/1-p = probability of event/probability of no
event
E(Yi) = 1 x Probability (Yi=1) + 0 x Probability (Yi=0),
and be rearranging the terms,
and substituting pi = Probability (Yi=1)
p = O/1+O
we obtain E(Yi) = pi.
or more explicitly in the following table:
Furthermore, application of the statistical assumptions
Probability
Odds
that the error in the expected value of Y does not vary
0.1
0.11
with x and that Y is a linear function of x, we obtain in
0.2
0.25
its simplest form what is called the linear probability
0.3
0.43
0.4
0.67
model:
0.5
1.00
0.6
1.50
Equation 1
0.7
2.33
pi = α + βxi.
0.8
4.00
0.9
9.00
where α is the Y intercept and β is the regression
coefficient or slope. This equation says that the
It is clear from the table that if the probability of an
probability that Y=1 is a linear function of x. If x
event occurring is less than 0.5 that the odds are less
happens to be a continuous variable that has no upper
than 1. Furthermore, from equation 2 it can be seen
or lower bound it is theoretically possible to obtain
that while there is no upper boundary on odds, the
values of Y which are less than 0 or greater than 1.
lower boundary is still 0. However, by taking the
Since linear functions are by nature unbounded, and
logarithm of the odds the lower boundary can be
probabilities by definition cannot exceed 1 or fall
removed. Starting with the linear probability equation
below 0, a nonsensical situation with little predictive
discussed above (Equation 1), followed by changing
value arises.
probabilities to odds (Equation 2) and then taking the
Logistic regression (also called logit analysis) has
logarithm of these odds, we obtain the logit model as
been developed to address statistical and conceptual
shown in the equation below (from Allison, 1991):
issues that arise from the use of OLS with
dichotomous dependent variables. Logistic regression
Equation 3
can also be used with quantitative and/or categorical
log (pi/1-pi) = α + βxi1 + βxi2 + ... + βkxik
independent variables making it an optimal method
for social sciences data analyses.
where k refers to the number of independent
One of the key concepts of logistic regression is
variables, i = 1,...., n individuals, and pi is the
the relationship between probability and odds. Odds
probability that Y=1. The term ‘log (pi/1-pi)’ is also
are commonly used by gamblers and refer to the ratio
known as the logit or log-odds. Natural logarithms
between the expected number of times an event will
tend to be used by most researchers although base-10
occur to the expected number of times an event will
logarithms may be used as well.
not occur. For example, if the odds are 5 to 1 then we
expect the event to occur 5 times as many times as not
and if the odds are 1/5 then we would expect the
reverse. The relationship between probability and
odds can be shown as:
National Endowment for the Arts
37
Reading at Risk
Multiple Logistic Regression Model to Predict
inflation in the variance of each coefficient compared
Literary Reading in Adults
which what it would be if the variable was
Prior to the development of the model, the
uncorrelated with any other variable in the model
unweighted 2002 SPPA dataset was examined for
(Allison, 1991).
trends and correlated variables. The dependent
Table 1 summarizes the results of the final
variable was “literary reading in adults” and this was
multiple logistic regression to determine statistically
scored as ‘1’ if the respondent had reported reading
significant (p<0.05) demographic predictors of literary
literature in the previous 12 months and ‘0’ if they did
reading in adults. Table 3 shows the partial correlation
not. The frequency distribution of literary readers and
coefficients of all significant independent variables for
non-readers was investigated for survey responses and
the literary reading and frequent/avid reader logistic
lead to the identification of trends and a subset of
regression models. The partial correlation coefficient
independent variables most likely to explain whether
or r gives a measure of the degree of correlation
people read literature or not. These variables were
(positive or negative) between each independent
then subjected to a multiple correlation analysis to
variable and the dependent variable while excluding
determine if any were correlated with each other (also
the effects of all the other variables. The ranking for
called collinear or multicollinear analysis). The
each variable indicates the relative order of the
independent variables were also correlated with the
correlation of each independent variable to the
dependent variable ‘literary reading’ to further
dependent variable.
determine which were likely to play the largest role in
the prediction of literary reading. Collinearity analyses
Coding of Independent Variables
were performed primarily to eliminate correlated
Specific levels of the independent variables within the
independent variables in the regression model since
broad groups of education, gender, ethnicity and race,
highly correlated independent variables may result in
region, age, annual family income, mother’s education,
inflated standard errors and unstable regression
and attendance at an art museum and/or performing
coefficients. For example, the education levels of the
arts in the previous 12 months were determined to be
mothers and fathers of respondents were found to be
statistically significant (p<0.05) predictors of literary
highly correlated (r=0.724,
reading in adults. The variables and their
p<.001). At the same time,
the mother’s education was found to be more highly
corresponding reference variables are shown in Table
correlated with literary reading in respondents
1. Only variables that were statistically significant
compared to the father’s education. As a result, the
were included in the model and as a result some levels
father’s education variable was eliminated from the
were grouped together.
final model. Each potential independent variable was
All levels of respondent education were found to
examined in this manner prior to construction of the
be statistically significant predictors of literary reading
regression model.
and, as shown in Table 1, were broken into several
Following preliminary exploration of the 2002
dummy variables corresponding to grade school only,
SPPA data, the maximum-likelihood estimation
some high school, some college, college graduate, and
method in the PROC LOGISTIC program of the
graduate school. Each survey respondent was assigned
SAS version 8.02 statistical software package was used
a ‘1’ for the highest education level achieved and a ‘0’
to develop the logistic regression model.
for all other education levels. High school graduation
Multicollinearity of independent variables was
was used as the reference variable for odds ratio
checked for the variables in the final model using the
interpretation.
PROC REG program with the TOL and VIF options.
Gender was coded ‘1’ for male and ‘0’ for female
The TOL and VIF options indicate the level of
and by default female was the reference variable (for
National Endowment for the Arts
38
Reading at Risk
dichotomous variables the reference variable is the one
interval for each statistically significant predictor of
coded as ‘0’). Ethnicity and race were initially divided
literary reading in adults. The regression coefficient
into ‘Hispanic,’ ‘non-Hispanic white,’ ‘non-Hispanic
for each variable measures the effect of that particular
African American,’ and ‘other’ ethnic groups. In the
variable alone on the log-odds that a person read
final analysis, cases belonging to the ‘other’ ethnic
literature, given that all other variables are equal.
category were eliminated since this category was not
Examination of the predictor variable ‘Graduate
found to be statistically significant. ‘Hispanic’ was set
school’ reveals that the coefficient for the regression
as the reference variable. The remaining respondents
model is 0.8725 which is the logged odds that a person
were coded as ‘1’ for the ethnicity and race category to
with this education level read literature (see equation
which they belonged and ‘0’ for all others.
3) with all other factors held constant. By taking the
Respondents were also divided into Northeast,
inverse of the natural log of this coefficient (e0.8725) we
Midwest, South, and West depending on the region of
obtain the odds ratio point estimate of 2.393. This
residence. Northeast was set as the reference variable
means that, everything else being equal, the odds of a
and respondents were coded as ‘1’ depending on
person who went to graduate school reading literature
residence and coded as ‘0’ for all other regions.
is more than twice that of a person whose education
Since the only age group found to be a significant
ended with high school graduation. (High school
predictor of literary reading was the 18-24 group, the
graduation is the reference variable for the odds ratios
variable age was set so that correspondents were given
for all education variables in this model.). Put another
a score of ‘1’ if they were aged 18-24 and a score of ‘0’
way, the odds of a person who graduated from high
if they were aged 25 or over.
school (and didn’t continue to college) reading
Similarly, income was broken into two categories;
literature is 0.4178 or 42 percent of the odds of a
respondents were coded as ‘1’ if their income was less
person who went to graduate school (i.e. 1/2.393)
than $10,000 or if they did not report their income,
reading literature. The interpretation of the odds
otherwise they were coded as ‘0’.
ratios for dichotomous variables such as ‘gender’ is
The respondent’s mother’s education variable was
identical. In this case, the chances of a male reading
scored as ‘1’ if the respondent’s mother attended at
literature (net of the effects of all other predictor
least some college and as ‘0’ if she did not.
variables) are less than half (0.431) of the chances of a
The arts attendance variable was broken into
female reading literature. For each odds ratio point
three dummy variables. Attendance at either an art
estimate the corresponding 95 percent confidence limit
museum or a performing arts event in the previous
was calculated using the standard error for the
year resulted in a score of ‘1’ for the first variable.
regression coefficient and outlines the degree of
Attendance at both an art museum and a performing
sampling variability in that coefficient.
arts event in the previous year was scored as ‘1’ for the
At the bottom of Table 1 the proportion of
second variable. Otherwise respondents were coded as
correctly identified cases, false positives and false
‘0’. The variable ‘attended neither’ was used as the
negatives at the 0.500 probability level is given for the
reference variable for odds ratio interpretation.
regression model. These statistics give the proportion
of observed cases which would be classified correctly
Interpretation of Regression Coefficients, Standard Error
using the set of independent variables identified in the
and Odds Ratios
given model when the calculated probability of being
Table 1 illustrates the regression coefficients, the
a literature reader is 0.500 or higher. These statistics
standard error associated with each coefficient, and
also estimate the proportion which would be classified
the odds ratios including the 95 percent confidence
as literary readers when they were not (false positives),
National Endowment for the Arts
39
Reading at Risk
and the proportion classified as non-literary readers
adult read 12 or more books in the previous 12
when in fact they did read literature (false negatives).
months. As before, only variables that were
statistically significant were included in the model.
Multiple Logistic Regression Model to Predict
The results are summarized in Table 2.
Reading of 12 or More Books Per Year in 2002
Respondents with at least some college (coded as
‘1’) were predicted to be nearly 28 percent more likely
(“Frequent” and “avid” readers)
to be a frequent/avid reader (everything else being
The goal of the second statistical model is to predict
equal) than those with a high school diploma or less
whether literature-reading adults were likely to read
(coded as ‘0’).
12 or more books based on the same set of
Gender of the correspondent was coded as ‘1’ if
demographic predictor variables as in the previous
male and ‘0’ if female. Analysis of the odds ratios
analysis. Prior to the construction of the logistic
suggested that males are 0.633 times as likely as
regression model based on the unweighted 2002 SPPA
females to read 12 or more books in a year.
dataset, all data from respondents who did not read
Analysis of the ethnicity and race variables
literature in the previous year were eliminated. The
indicated that within this category, only the non-
dependent variable was ‘frequent/avid reader’ and the
Hispanic white group was a statistically significant
remaining respondents were given a score of ‘1’ if they
predictor of frequent/avid reading. As a result,
reported reading 12 or more books in the previous
respondents were scored as ‘1’ if they were non-
year and ‘0’ if they had not.
Hispanic white Americans and ‘0’ if they were
The maximum likelihood estimation method
Hispanic Americans, non-Hispanic African
within the PROC LOGISTIC program was used to
Americans, or from another ethnic or racial group.
construct the logistic regression model as previously
The logistic regression analysis suggested that non-
described. Multicollinearity of independent variables
Hispanic white people were 1.6 times as likely to fall
within the final model was analyzed as in the previous
into the frequent/avid reader group compared with all
model.
the other ethnic and racial groups.
Table 2 provides a summary of the final multiple
Television viewing habits were broken into three
logistical model containing the statistically significant
dummy variables: 0 hours of television watching per
(p<0.05) predictors of readers of 12 or more books per
day, 1-3 hours per day, and 4 or more hours per day.
year. The regression coefficients, standard errors, p
The group corresponding to 1-3 hours of television
per day was set as the reference variable and
values and reference variable interpretation are as
respondents were scored as ‘1’ for the television
before. The odds ratio point estimates (with 95 percent
viewing class to which they belonged and ‘0’ for all
confidence intervals) give the odds of being a
others. Examination of the odds ratios reveals that
frequent/avid reader for each particular variable, all
people who watch no television are 1.475 times as
other variables being equal. The multiple correlation
likely to read 12 or more books in a year than those
analysis of the predictor variables is shown in Table 3.
who watch 1-3 hours per day. Alternatively, those who
reported watching 4 or more hours of television per
Coding of Independent Variables and
day would be expected to be only 88 percent as likely
Interpretation of Odds Ratios
to read 12 or more books compared with those in the
Specific levels of the independent variables (education,
1-3 hour reference group.
gender, ethnicity and race, television viewing habits,
The variable ‘age’ was divided into two groups
age, employment, mother’s education, participation in
representing respondents less than 45 years old (scored
charity/volunteer work, and attendance at art
as ‘1’) and those aged 45 or older (scored as ‘0’). The
museums and performing arts events) were shown to
logistic regression model predicts that respondents
be significant (p<0.05) predictors of whether or not an
National Endowment for the Arts
40
Reading at Risk
under the age of 45 were only 63 percent as likely to
References
be frequent/avid readers compared with those aged 45
Pampel, F.C. 2000. Logistic Regression: A Primer. Sage
or greater.
University Papers Series on Quantitative Applications in
Respondents who were not in the labor force were
the Social Sciences, 07-132. Sage Publications, Inc.,
given a score of ‘1’ while those who were in the labor
Thousand Oaks, CA, 84 pp.
force (unemployed or employed) were scored as ‘0’.
The odds ratio analysis suggested that respondents
Allison, P.D. 1991. Logistic Regression Using the SAS®
who were not in the labor force were more likely (17
System: Theory and Application. SAS Institute Inc.,
percent) to be frequent/avid readers than those who
Cary, NC: pp 5-78.
self-identified as being in the labor force.
The respondent’s mother’s education variable was
scored as ‘1’ if the mother attended at least some
Wright, R.E. 1995. Logistic Regression. In Reading and
college and as ‘0’ if she did not. Respondents whose
Understanding Multivariate Statistics, edited by L. G.
mothers attended college were nearly 20 percent more
Grimm and P.R. Yarnold, American Psychological
likely to be frequent/avid readers than those whose
Association, Washington, DC, pp 217-244.
mothers had a high school diploma or less.
The variable ‘charity work’ was scored as ‘1’ if
respondents reported participation in volunteer or
charity work in the previous year and as ‘0’ if they did
not. Participation in charity work is associated with a
26 percent increase in the odds of being a
frequent/avid reader compared to those who did not
perform volunteer or charity work.
Attendance at both an art museum and a
performing arts event was a significant predictor of
reading 12 or more books in a year. Respondents were
scored as ‘1’ if they attended both an art museum and
performing art event, and ‘0’ if they only attended one
or didn’t attend either in the previous year.
Respondents who attended both were 2.6 times as
likely to be a frequent/avid reader compared with
those who attended one or neither activity.
National Endowment for the Arts
41
Reading at Risk
Appendix Table 1
Summary of Multiple Logistic Regression Predicting Whether Adults have Read Literature (n=16,183).
Regression
Standard
p
Variable
Coefficient
Error
value
Intercept
-0.8747
0.0843
<0.0001
Education
Grade school
-0.8454
0.1029
<0.0001
Some high school
-0.434
0.0707
<0.0001
Some college
0.3279
0.0456
<0.0001
College graduate
0.5652
0.0574
<0.0001
Graduate school
0.8725
0.078
<0.0001
Sex
Male
-0.8423
0.0369
<0.0001
Ethnicity & Race
Non-Hispanic white
0.6186
0.0711
<0.0001
Non-Hispanic African American
0.2701
0.0913
0.0081
Region
Midwest
-0.1434
0.053
0.0068
South
-0.2264
0.0516
<0.0001
West
0.1296
0.0557
0.02
Age
Young (under 25)
-0.1599
0.0635
0.0118
Income
Low or not reported
-0.1813
0.0494
0.0002
Mother's Education
At least some college
0.3871
0.0482
<0.0001
Arts Attendance
Art museum OR performing arts
0.9654
0.0437
<0.0001
Art museum AND performing arts
1.7851
0.058
<0.0001
At the 0.5000 probability level, the proportion of cases correctly classified with the model above is 71.3%; the false positive rate is 26.8%; and the false negative rate is 30.2%.
National Endowment for the Arts
42
Reading at Risk
Odds Ratios
Reference
Point
Lower 95%
Upper 95%
Variable
Estimate
Con. Limit
Con. Limit
High school graduate
0.429
0.351
0.525
as above
0.648
0.564
0.744
as above
1.388
1.269
1.518
as above
1.76
1.572
1.97
as above
2.393
2.054
2.788
Female
0.431
0.401
0.463
Hispanic
1.856
1.615
2.134
as above
1.31
1.095
1.567
Northeast
0.866
0.781
0.961
as above
0.797
0.721
0.882
as above
1.138
1.021
1.27
Age 25+
0.852
0.753
0.965
Reported income > $9,999
0.834
0.757
0.919
High school or less
1.473
1.34
1.619
Attended neither
2.626
2.41
2.861
Attended neither
5.96
5.32
6.677
National Endowment for the Arts
43
Reading at Risk
Appendix Table 2
Summary of Multiple Logistic Regression Predicting Whether Adults have Read 12 or More Books (n=7,555)
Regression
Standard
p
Variable
Coefficient
Error
value
Intercept
-0.9113
0.0915
<0.0001
Education
At least some college
0.2432
0.0577
<0.0001
Sex
Male
-0.4575
0.0517
<0.0001
Ethnicity & Race
Non-Hispanic white
0.4898
0.0725
<0.0001
TV Viewing (hours/day)
none
0.3889
0.114
0.0006
4 or more hours
-0.1273
0.061
0.0369
Age
<45 years old
-0.4538
0.052
<0.0001
Employment
Not in Labor Force
0.1599
0.0569
0.005
Mother's Education
At least some college
0.177
0.0554
<0.0014
Charity Work
Some volunteer/charity work
0.2337
0.0501
<0.0001
Arts Attendance
Art museum AND performing arts
0.3738
0.0542
<0.0001
At the 0.5000 probability level, the proportion of cases correctly classified with the model above is 62.3%; the false positive rate is 42.7%; and the false negative rate is 36.4%.
National Endowment for the Arts
44
Reading at Risk
Odds Ratios
Reference
Point
Lower 95%
Upper 95%
Variable
Estimate
Con. Limit
Con. Limit
High school or less
1.275
1.139
1.428
Female
0.633
0.572
0.7
African American, Hispanic
1.632
1.416
1.881
and other
1-3 hours
1.475
1.18
1.845
as above
0.88
0.781
0.992
Age 45+
0.635
0.574
0.703
In Labor Force
1.05
1.312
(Unemployed or employed)
1.173
High school or less
1.194
1.071
1.33
No volunteer/charity work
1.263
1.145
1.394
Attended only one or
1.453
2.41
2.861
didn’t attend either
National Endowment for the Arts
45
Reading at Risk
Appendix Table 3
Correlation Analysis of Predictor Variables in the Literature Reading and Frequent/Avid Book Reading, Logistic Regression Models
Independent Variable
y*
Rank
Education
Grade school
-0.159
6
Some High School
-0.155
8
Some College
0.072
14
College Graduate
0.167
5
Graduate School
0.168
4
Sex
Male
-0.179
3
Ethnicity
Non-Hispanic White
0.156
7
& Race
Non-Hispanic African American
-0.077
12
Residence
Midwest
-0.008
16
South
-0.084
11
West
0.073
13
Age
Young
-0.034
15
Income
Low/Not Reported
-0.093
10
Mother's
At least some college
0.207
2
Education
Arts
Art museum OR
0.155
9
Attendance
Performing Arts
Art museum AND
0.327
1
Performing Arts
* indicates the partial coefficient ('r) and is a measure of the degree of correlation between the independent variable and the dependent variable excluding the effects of all the other variables. All pa
reading correlation (this variable was statistically significant in the regression model, see Appendix Table 1).
National Endowment for the Arts
46
Reading at Risk
Independent Variable
y*
Rank
Education
At least some college
0.084
6
Charity Work
Volunteer or
0.103
3
Charity Work
Sex
Male
-0.101
5
Ethnicity
Non-Hispanic White
0.104
4
& Race
Television
None
0.047
9
Viewing
4+ hours/day
-0.047
10
Age
<45
-0.113
2
Employment
Not in labor force
0.048
7
Mother's
At least some college
0.056
8
Education
Arts
Art museum AND
0.129
1
Attendance
Performing Arts
artial correlation coefficients are statistically significant at the p<0.001 level with the exception of the variable "Midwest" in the literature
National Endowment for the Arts
47
Reading at Risk
