Males - History, Political Science, and Philosophy

The Gender Gap in Political
Knowledge Among College
Students
Carol S. Botsch, Professor of Political
Science, USC Aiken
Robert E. Botsch, Professor of
Political Science, USC Aiken
Presented at the SWPSA Conference,
March 28, 2013, New Orleans
How we started this study
• Wanted to try teaching American Government
online in 1997
• Great skepticism from colleagues
• Developed very hard comprehensive 63 item pre
and post test—nearly all open-ended
• Also gathered attitudinal and demographic data,
including gender
• Noticed that women did considerably less well
• After 13 years of gathering data, now exploiting it
to look at many things, including the gender
gap—hence this little study!
Review of Literature on Gender
Gap in Political Knowledge
• General low levels of knowledge among Americans
and youth
• Negative implications for operation of a democratic
republic
• Gender gap well established in research
• Sources of gap
–
–
–
–
–
Socialization
Self-esteem and confidence
Education
Political Interest
And gender biased measurements that inflate gap
Methodological debate about
validity/reliability of measurements
• Gender based measurement bias
• Women more likely to know different things than
men
• Risk aversion theory
– Women more likely to choose “DK” on closed-ended
items—advice: discourage DK’s
– Leave blank open-ended items when only have partial
knowledge—advice: use closed-ended items
• Claim that up to 50% of gap is measurement
based
Others disagree or see measurement
problems as minor
• Delli Carpini and Keeter (1993) argue that
encouraging guessing leads to unreliability
• Cassese, Weber, and Khatib (2007) see gender
bias as a small factor and conclude that
conventional measures only modestly
overestimate the gender gap
• Our open-ended measures are conventional,
but we try to see evidence of female “risk
aversion” in our analysis
How the gap might be reduced
•
•
•
•
Taking American Government classes
Increasing political interest
Improving self-confidence and efficacy
Increasing incentives to answer all questions,
including those with partial knowledge and
guessing on closed ended items
Study Setting
• USC Aiken is public liberal arts institution with
a few MA programs
• 3400 students, mostly native, but 20 states
• 67% female
• 28% African-American, only a couple % other
• Sample representative of student body, in part
because American Government (or American
History) is required in General Education
across campus
Data—Two Studies
• 13 Year Survey of entering and exiting American
Government Students (1978-2010)
–
–
–
–
–
N = 3100 (or 6200 pre and posttests)
N = 2400 for this report—excluded online students
59 items, 1st Amend item 0-5, total score 0-63
Only 8 items can be considered close-ended—no DK
Scored as total # correct rather than each item as a
separate variable in data set
– Did go back and count # quest attempted for all 6200+
tests
• 2011 Study of Undergraduates
– N = 531; Resp Rate = 53%; Sample error: +/- 4.3%
– 8 item knowledge test, 7 open-ended
Quasi-Experimental Design
• Most of this paper relies on 13 year study
• Divided students into two panels
– Lo Incentive Panel: Posttest taken as pretest—
”just to see how much they knew” (n=1100)
– Hi Incentive Panel: Posttest counted as part of
final exam grade (n=1100)
• Two treatments
– LIP: the impact of completing Am Govt class
– HIP: Am Govt class + high incentives for
maximizing score on posttest
Findings
• Found gender gap no matter how measured
– Males answered more questions correctly
– Males attempted more questions
– Males got a higher percentage of questions
attempted correct—suggesting that they were not
just guessing based on little information
• Conclusion: most of any difference was real,
not due to female risk aversion
– Caveat: unless females were not answering
questions on which they had nearly full
information
Data showing Gender Gap—LIP
Pretest
# Correct
% of 63
Correct
# Attmptd
% Attmptd
Correct
Low Incentives Panel (LIP)
Males
Females
Diff
(n=383)
(n=843)
(p)
15.29
10.63
4.66
(0.000)
24.27%
16.88%
7.39% pt
(0.000)
31.09
24.69
6.40
(0.000)
43.23%
38.05%
5.18% pt
(0.000)
Taking American Government Class
Did Not Reduce Gap
• LIP panel did significantly better on the
posttest than the pretest
• But gaps on all measures of performance
remained—not significantly improved
from pretest
Data on Impact of Taking Am Gov
Low Incentives Panel (LIP)
Pretest
Males
(n=383)
Females
(n=843)
Diff (p)
# Cor
15.29
10.63
4.66 (0.000)
% Cor
24.27%
16.88%
7.39% pt (0.000)
# Att
31.09
24.69
6.40 (0.000)
% At Cor
43.23%
38.05%
5.18% pt (0.000)
Posttest
Females
(n=606)
24.64
Diff (p)
# Cor
Males
(n=285)
28.97
4.33 (0.000)
% Cor
45.98%
39.05%
6.93% pt (0.000)
# Att
43.11
38.57
4.54 0.000)
% At Cor
60.24%
56.79%
3.45% pt (0.003)
Raising Incentives Erased the Gap
# Cor
Males
(n=321)
14.50
High Incentives Panel (HIP)
Females
(n=770)
9.47
% Cor
23.01%
15.02%
7.99% pt (0.000)
# Att
32.15
24.82
7.33 (0.000)
% At Cor
39.46%
33.95%
5.51% pt (0.000)
Posttest
Females
(n=736)
43.31
Diff (p)
# Cor
Males
(n=299)
44.84
1.53 (0.071)
% Cor
71.17%
68.75%
2.42% pt (0.071)
# Att
54.12
53.15
0.97 (0.054)
% At Cor
75.81%
74.15%
1.66% pt (0.127)
Pretest
Diff (p)
5.03 (0.000)
Taking American Government Boosts
Political Interest and Reduces a
Gender Gap in Interest
Political
Interesta
Males
Females
Difference (p)
Pretest
2.60
2.26
0.34 (0.000)
Posttest
2.86
2.66
0.20 (0.000)
+0.26
+0.40
-0.14 (0.010)
Difference
Multivariate Approach: the relative
importance of gender and interest
GPA
Age
Race
Gender
Pol Intrst
Pol Efficy
News Paper
R2
Pretest:
Posttest:
# Correct
Betas
Both Panels
# Correct
Betas
Low Incentive
High Incentive
Panel
Panel
0.148**
0.081**
0.132**
0.210**(2)
0.252** (1)
0.200** (3)
0.065**
0.343**(1)
0.067
0.068
0.143**(3)
0.182**(2)
0.105**
0.049
0.348**(1)
0.095**
0.136**(2)
0.033
0.048
0.058
0.003
0.323
0.267
0.211
Focus on Political Interest
1. It is important in explaining
knowledge
2. Women have an interest gap
3. Why? We can do something about it
Possible strategy:
focus on boosting interest in
teaching American Government
Problems in Using Am Govt to boost
female political interest
• The short term reduction in the interest gap
washes out in the medium term (1-8
semesters)
• Data from 2011 study shows this
Political Interest
Males – Females (p)
Not taken Am Govt
0.53 (0.000)
Took Am Govt
0.41 (0.002)
Change in Difference
-0.12 (0.264)
Increasing Interest Does NOT Reduce
Know. Gap, & Females Lag at All Levels
Political
Pretest # Interest
Males
of
Correct Females
Answers Gap (m-f)
p
Males
Posttest
# of
Females
Correct Gap (m-f)
Answers
p
1
2
3
4
10.04
11.89
16.13
22.02
8.20
9.17
11.13
14.34
1.84
2.72
5.00
7.68
0.002
0.000
0.000
0.000
24.26
23.58
30.35
36.45
18.56
23.98
25.16
31.84
5.70
-0.40 (?)
5.19
5.01
0.020
0.796
0.000
0.033
Conclusions
• Gender Gap in Knowledge seems mostly real
• Taking Am Govt Gov has at best a slight impact in
reducing the gap—hi incentives can wipe out gap
• Political interest is an important stimulant for
political knowledge
• A political interest gap also exists
• Taking Am Govt produces a short term reduction
in interest gap, but washes out, and men have
higher knowledge at all interest levels
• Need special programs for women to boost
interest and knowledge—American Govt is not
enough