Transcript Exercise 6 - People Server at UNCW

(EXERCISE 6)
THE IMPACT OF
ATTITUDES TOWARD GAY MARRIAGE
ON 2004 PRESIDENTIAL VOTE CHOICE
CONTROLLING FOR PARTY IDENTIFICATION
Roger C. Lowery
PLS 401, Senior Seminar
Department of Public & International Affairs
UNC Wilmington
28 March 2016
1
Univariate Hypothesis
• Theory:
– Immediately prior to Election Day 2004, the
nationwide trial-heat margin between Bush and
Kerry was too close to call.
• H1: Neither Bush nor Kerry was a consensus choice
among pre-election voters in 2004.
28 March 2016
2
Table 1:
V002
Presidential vote choice
Text of this Question or Item
Whom did you vote for in the presidential election?
(Nonvoters are recorded as NA).
% Valid
50.1
48.5
1.5
100.0
28 March 2016
% All
38.6
37.4
1.1
22.8
100.0
N
412
399
12
243
1,066
Value
1
2
3
9
Label
Bush
Kerry
Other
NA
Total
3
Univariate Findings
• H1 (neither Bush nor Kerry was a consensus choice in 2004) is
supported by the sample data in Table 1 because:
1. The pattern predicted by H1 is observed in the sample data.
There is very little difference (less than 2%) between Bush and Kerry
support in the sample.
2. The differences in Bush/Kerry support that are observed in
the sample are too small to be statistically significant.
The random-sampling error margin for the sample results in Table 1
(3.0 %)* is larger than the vote-choice margin between Bush and
Kerry supporters (1.6 %).
* http://www.custominsight.com/articles/random-sample-calculator.asp
28 March 2016
4
Bivariate Hypothesis
• Theory:
– Bush supported a constitutional ban on gay
marriage and Kerry opposed.*
*http://www.cbsnews.com/stories/2004/07/13/eveningnews/main629360.shtml
• H2: Voters who opposed gay marriage were
more likely to support Bush in 2004 than
voters who supported gay marriage.
28 March 2016
5
Table 2:
2004 Presidential Vote Choice by Attitude toward Gay Marriage
V125 Attitude toward Gay Marriage
(Should same-sex couples be allowed to marry?)
Cells contain:
-Column percent
-Weighted N
1
Yes
2
No but permit
unions
3
No
ROW
TOTAL
1: Bush
25.4
63
61.0
13
63.5
301
50.7
376
2: Kerry
74.6
184
39.0
8
36.5
173
49.3
365
COL TOTAL
100.0
247
100.0
22
100.0
473
100.0
742
V002RCL:
2004
Presidential
Vote Choice
Summary Statistics
Tau-b =
28 March 2016
-.35
Chi-square
probability =
0.00
6
Bivariate Findings
• H2 (voters who opposed gay marriage were more likely
to support Bush in 2004 than voters who supported
gay marriage) is supported by the sample data in Table
2 because:
1. The pattern predicted by H2 is observed in the sample
data. Taub = 0.35, which indicates that gay-marriage
attitudes were a strong predictor of vote choice.
2. This sample finding is statistically significant. The chisquared probability of random-sampling error is less
than 0.05 (χ2 = 0.00).
28 March 2016
7
Multivariate Hypothesis
• Theory:
– Because some (but not all) gay-rights supporters have
gravitated to the Democratic Party and some (but not
all) gay-rights opponents have moved to the
Republican Party; therefore, there is less conflict
within each party than between the two parties on
the issues of gay rights.
• H3: the impact of attitudes toward gay marriage on 2004
presidential vote choice will be weaker within partisans
than in the total population. [Party identification will be
a confounding variable.]
28 March 2016
8
Table 3a
2004 Presidential Vote Choice by Attitude toward Gay Marriage
Controlling for Party Identification
(Democrats)
V125RCL Attitude toward Gay Marriage
(Should same-sex couples be allowed to marry?)
Cells contain:
-Column percent
-Weighted N
V002RCL: 1: Bush
2004
2: Kerry
Presidenti
al Vote
Choice COL TOTAL
1
Yes
2
No or civil
unions only
ROW
TOTAL
3.7
6
15.7
27
9.7
33
96.3
164
84.3
146
90.3
311
100.0
171
100.0
174
100.0
344
Chi-square
probability =
0.00
Summary Statistics
Tau-b =
28 March 2016
-.20
9
Table 3b
2004 Presidential Vote Choice by Attitude toward Gay Marriage
Controlling for Party Identification
(Independents)
V125RCL Attitude toward Gay Marriage
(Should same-sex couples be allowed to marry?)
Cells contain:
-Column percent
-Weighted N
V002RCL: 1: Bush
2004
2: Kerry
Presidenti
al Vote
Choice COL TOTAL
1
Yes
2
No or civil
unions only
ROW
TOTAL
36.6
4
44.4
13
42.3
17
63.4
7
55.6
16
57.7
23
100.0
11
100.0
30
100.0
40
Chi-square
probability =
(p= 0.65)
Summary Statistics
Tau-b =
28 March 2016
-.07
10
Table 3c
2004 Presidential Vote Choice by Attitude toward Gay Marriage
Controlling for Party Identification
(Republicans)
V125RCL Attitude toward Gay Marriage
(Should same-sex couples be allowed to marry?)
Cells contain:
-Column percent
-Weighted N
V002RCL: 1: Bush
2004
2: Kerry
Presidenti
al Vote
Choice COL TOTAL
1
Yes
2
No or civil
unions only
ROW
TOTAL
83.0
51
93.6
269
91.7
320
17.0
11
6.4
18
8.3
29
100.0
62
100.0
287
100.0
349
Chi-square
probability =
0.01
Summary Statistics
Tau-b =
28 March 2016
-.15
11
Multivariate Findings
• H3 (the impact of attitudes toward gay marriage on 2004
presidential vote choice will be weaker within partisans
than in the total population) is supported by the sample
data. Party identification is a confounding variable in this
analysis.
1. The strength of the bivariate relationship did weaken as
predicted in the partisan subgroups. [The taub for
Democrats (0.20) and Republicans (0.15) was less than in
the total sample (0.35).
2. The impact of gay marriage on vote choice (although
weakened) was still statistically significant within
Democratic (χ2 = 0.00) and Republican (χ2 = 0.01)
subgroups.
28 March 2016
12
Substantive Implications
• The Democratic Party is more internally divided on the issue
of gay marriage than is the Republican Party.
• However, party identification out-weighed the impact of gaymarriage attitudes in presidential vote choice in 2004.
– even if the electorate had been limited to only Democratic
identifiers who opposed gay marriage, then Kerry would
still have easily defeated Bush.
– even if the electorate had been limited to only Republican
identifiers who supported gay marriage, then Bush would
still have easily defeated Kerry.
– There were relatively few single-issue gay-rights voters in
2004 who voted against their party’s candidate.
28 March 2016
13
Methodological Implications
• Why is gay marriage is more of a “wedge issue” for the
Democratic Party than the Republican Party?
• What important demographic groups are most likely to oppose
their party’s stand on gay marriage; and, therefore, more likely
to defect?
• Do other gay-rights issues (adoption, employment, hate crimes)
align with or cross-cut the gay-marriage issue cleavage?
• Do other group cleavages (age, gender, race, ethnicity, religion,
etc.) align with or cross-cut the gay-rights issue cleavage?
• Do other public morality issues (public-school prayer, sex
education in public schools, abortion, torture, etc.) align with or
cross-cut the gay-rights issue cleavage?
28 March 2016
14