2. Results
2.1. Preliminary analyses
Spearman's rho correlations determined relations between sensory sensitivity and response bias to each expression. Response bias was not correlated with sensory sensitivity for either the positive or negative expression, ps > .10. However, mothers’ sensory sensitivity to the positive infant expression was positively correlated with their sensory sensitivity to the negative expression, r = .24, p < .05, as well as their response bias to the positive expression being positively correlated with their response bias to the negative expression, r = .25, p < .04. Additionally, correlational analyses and univariate ANOVAs determined that neither sensory sensitivity or response bias were related to maternal education, age, household income, scores on the ITQ, or in the distribution of parity, infant sex, and employment status ps > .10, except for depression which was positively related to response bias for the positive infant expression, r = .27, p < .03, with more depressed mothers having a more conservative response bias (i.e., tendency to respond with same, as opposed to different). 3
Univariate ANOVAs and Chi-square analyses were conducted to determine whether mothers in the three illusion of control groups or mothers assigned to the two gender labels differed on socio-demographic or questionnaire measures. No differences were found between illusion of control groups on maternal education, age, household income, scores on the ITQ and CES-D, or in the distribution of parity, infant sex, and employment status, ps > .10. Regarding gender label groups, mothers differed in depression, F(1, 67) = 4.23, p < .05, View the MathML sourceηp2=.06, and age, F(1, 67) = 5.76, p < .02, View the MathML sourceηp2=.08, as well as in ratings of infant temperament, F(1, 67) = 4.91, p < .03, View the MathML sourceηp2=.07. Consequently, maternal depression and age, as well as infant temperament were entered as covariates in the analyses testing manipulation effects. Because we manipulated infant gender, mothers’ own infants’ sex was also entered as a covariate. Lastly, on the manipulation check, all mothers indicated correctly the gender they had been told.
2.2. Sensory sensitivity and response bias: effects of gender labeling and maternal self-efficacy
To determine whether infant gender labeling affected mothers’ sensory sensitivity to the positive and negative infant expressions, either directly or as a moderator of illusion of control, a 2(gender label) × 3(illusion of control) MANCOVA was conducted with sensory sensitivity to each expression entered as the dependent variables. Initial results revealed a significant multivariate effect for gender label, F(2, 58) = 3.89, p < .03, View the MathML sourceηp2=.12, and the gender label × illusion of control interaction, F(2, 59) = 4.79, p < .02, View the MathML sourceηp2=.14. Associated ANCOVAs revealed a significant main effect for gender label for the positive expression, F(1, 59) = 6.08, p < .02, View the MathML sourceηp2=.09. Mothers who received the female label were more sensitive to the positive infant expression compared with mothers who received the male label (see Table 1). A significant gender label × illusion of control interaction predicted mothers’ sensory sensitivity to the negative expression, F(2, 59) = 4.59, p < .02, View the MathML sourceηp2=.14 (see Fig. 2). Simple effects tests indicated this was due to mothers with high illusion of control being less sensitive when receiving the male label (M = 3.55, S.D. = 1.57) than when receiving the female label (M = 2.25, S.D. = .74). Sensitivity of mothers with low and moderate illusory control did not differ based on the male and female labels: For low illusion of control, M = 3.15, S.D. = 1.09 and M = 2.91, S.D. = .94, respectively; for moderate illusion of control, M = 2.40, S.D. = .87 and M = 2.97, S.D. = .89, respectively. The multivariate main effect for illusion of control was not significant, p > .10.
Table 1.
Infant gender label × maternal sensory sensitivity and response bias: means and standard deviations
Infant gender label
Male (n = 33) Female (n = 36)
M S.D. M S.D.
Maternal sensory sensitivity
Positive expression 3.10a 1.12 2.43b .77
Negative expression 3.04 1.25 2.74 .90
Maternal response bias
Positive expression .74a .40 .57b .25
Negative expression .71 .34 .63 .31
Note. For maternal sensory sensitivity, lower scores indicate greater sensitivity. For maternal response bias, scores less than .75 indicate a less conservative response bias (i.e., a tendency to favor the different response choice) and scores more than .75 indicate a more conservative response bias (i.e., a tendency to favor the same response choice). Means in the same row with different superscripts differ significantly, ps < .05.
Table options
Mean maternal sensory sensitivity to the negative infant expression during the ...
Fig. 2.
Mean maternal sensory sensitivity to the negative infant expression during the signal detection task as function of maternal illusion of control and infant gender label. Lower scores indicate greater sensory sensitivity.
Figure options
To determine whether infant gender labeling affected mothers’ response bias to the positive and negative infant expressions either directly or moderated by illusion of control, a 2 (gender label) × 3 (illusion of control) MANCOVA was conducted with response bias to each expression entered as the dependent variables. Although the multivariate main effect for gender label was marginal, F(2, 58) = 3.00, p < .06, View the MathML sourceηp2=.09, the associated ANCOVA for gender label for the positive expression, F(1, 59) = 6.03, p < .02, View the MathML sourceηp2=.09, was significant. With a value of .75 indicating no bias, mothers who received the female label had a response bias reflecting a tendency toward pressing the different response choice (less conservative) compared with the same response choice. In contrast, those who received the male label exhibited no bias (see Table 1). The multivariate main effect for illusion of control and the gender label × illusion of control interaction were not significant, ps > .10.
