W
EAVER
19
The Color of the Campaign
favoring Turner in treatment 1A and slightly less favoring Turner in treatment 1B (Table 5); the
differences are not statistically significant. Ergo, the reactions of men in the two treatments are
consistent with the control group. While less than 20 percent of women identified with the white
Turner, at least half identified with the black Turners (Table 5a). Women “switched” from
supporting Sheldon in the control group to completely reversing and supporting the disfavored
Turner. The differences in favorability across treatments among women are statistically
significant except for the vote (p = .002 for “identify” and p = .061 for the feeling
thermometer).
Therefore, an important qualification of the bivariate results is in order: black
candidates entering the race made women more likely to vote for them.
Table 5: Candidate Evaluation by Subject Group, Men Only, Pilot 2
I
DENTIFY
F
EELING
T
HERMOMETER
V
OTE
S
UBJECT
G
ROUP
Sheldon Turner
Non-
response
Favors
Sheldon
Same
Favors
Turner Sheldon
Turner
Non-
response
Control:
White Turner
52.2% 39.1%
8.7% 47.8%
26.1%
26.1% 47.8%
39.1%
13.0%
Treatment 1a:
Light-Skinned Turner
40.0 52.2
8.0 25.0
41.7
33.3 40.0
44.0
16.0
Treatment 1b:
Dark-Skinned Turner
47.1 41.2
11.8 35.3
23.5
41.2 35.3
35.3
29.4
Identify:
χ
2 = 1.061 p = .900
Feeling Thermometer:
χ
2 = 3.741 p = .442
Vote:
χ
2 = 2.143 p = .709
41
Includes respondents who reported that they were Independent, None, and Other.
42
The crosstab is insignificant when we include men and women but it becomes significant when we look only at
differences across treatments for women.
Convention