Tipultech logo

2nd Hypothesis

Sarah Ogilvie

In order to determine whether rates of rejection of influence were predicted by implicit knowledge, explicit beliefs, or neither, a standard multiple regression analysis was conducted. Even though the two factors obtained in the previous analysis did not separate measures of explicit beliefs from implicit knowledge measures, they were selected as possible predictors of P(S) scores based on the following reasoning. As the ability of the implicit knowledge measures to actually measure implicit knowledge was undetermined, it was acknowledged that the pattern of measures loading onto the two factors may have better represented a possible separation between implicit knowledge and explicit beliefs. Hence, the two factors obtained in the principal components analysis were entered as independent variables into the regression.

In employing this regression method it was recognised that the outcome may have been affected by the smaller than recommend number of cases that were included (Tabachnick & Fidell, 1996). The results of the regression analysis are represented in Table 9.

Table 9

Standard Multiple Regression of Two Factors Underlying Measures of Implicit Knowledge and Explicit Beliefs on P(S) scores (N = 40)

Variables B β R R² Change F Change Sig. F Change
Factor 1 2.62E-02 0.17 0.17 0.03 0.03 1.13 0.30
Factor 2 9.16E-03 0.06 0.19 0.04 0.01 0.14 0.72

It is evident (Table 9) that neither of the factors contributed significantly to prediction of P(S): together they explained only 4% of the variability in P(S) scores. This result may be explained by the relatively low variance obtained in the P(S) scores (M = 0.58, SD = 0.15). If not, however, it may be the case that the constructs represented by the two factors do not predict rejection of influence.

In order to determine whether this finding was the result of the small number of cases included in the principal components analysis, cases from another experiment that employed similar measures were combined (N = 23) (Foddy, 1998). However, as the measures used in Foddy's experiment did not include the SentComp, a second principal components analysis was required. As the ASI and the WSQ subset were found to be highly correlated (r = 0.63, p < 0.01), these were combined to represent one variable.

In evaluating this combined data set for the assumptions of principal component and regression analyses satisfactory results were obtained. A principal components extraction with varimax rotation was then conducted. In specifying a two-factor solution, the resultant eigenvalues were 2.54 for the first factor (accounting for 28% of the variance), and for the second factor 1.63 (accounting for 18% of the variance). Together the two factors explained a total of 46% of the variance, a similar result to that obtained in the smaller sample. Table 10 represents the factor loadings of each index of stereotyping for this combined sample.

Table 10

Factor Loadings for Principal Components Extraction and Varimax Rotation of Stereotyping Measures (N=63).

Measures of Stereotyping Factor One Factor Two
OthsVwsF 0.66 0.37
Consistent pairs (WrdTask) 0.56 0.21
Inconsistent pairs (WrdTask) 0.78 -2.47E-03
OthsVwsM 0.33 0.44
Correct response rate (WrdTask) 0.37 0.44
YrVwsM 4.43E-02 0.71
YrVwsF 7.64E-02 0.71
ASIWSQ 0.02 0.68

Note. Measures loading onto Factor One are represented above the upper line, and measures loading onto Factor Two are below the lower line. A third factor is apparent between the two lines.

In this combined sample, the pattern of factor loadings was similar to that obtained in the smaller sample (Table 10). However, in this case, the OthsVwsM and the correct response rate measure formed their own factor. Factor One was internally consistent, and well defined by its variables (implicit knowledge measures), as was the case with Factor Two. However, this factor comprised measures of explicit beliefs. These factors were then entered as predictors of P(S) scores in a standard multiple regression.

On performing this analysis none of the factors were found to predict the P(S) scores: a result in accordance with the smaller sample results. Compared with the smaller sample, the amount of variance in the P(S) scores explained by the three factors increased from 4% to 5%. Together, the results from the two regression analyses suggest that the measures of implicit knowledge and of explicit beliefs, are not predictive of P(S) scores.

In light of the hypotheses, if it can be assumed that the measures employed in this study are tapping two types of stereotypic processing, measured by explicit beliefs measures and variously by implicit knowledge measures, and that these types of processing do not predict rejection of influence (the measure of competence-related behaviour), then it may be the case that stereotypic processing does not affect competence-related behaviour, and by inference does not affect expectations for competence. However, it may also be the case that expectations for competence mediate the relationship between stereotypic processing and competence-related behaviour.

In order to asses this potential mediation effect, correlations were assessed between measures of stereotyping, P(S) and expectations for competence as measured by participants' direct reportage of their own ability relative to their partner (selfother). It must be noted that although this selfother rating was obtained after performance in the pattern recognition task, high correlations have previously been obtained between ratings before and after this task (r >0.8, Foddy & Smithson, 1996). Thus for the purposes of this paper this measure will be treated as one representative of expectations for competence.

In assessing the relationships between stereotyping, P(S), and selfother (Table 7), the only significant correlation to evident was between P(S) and selfother (r = 0.46, p < 0.001). As a result it was speculated that expectations for competence (selfother) may mediate the relationship between stereotypic processing and competence-related behaviour (rejection of influence).

To test this mediation relationship the two factors of stereotyping measures obtained using this study's sample (N=40), were used to predict selfother in a standard regression analysis. The results are presented in Table 11.

Table 11

Standard Multiple Regression of Two Factors Underlying Measures of Implicit Knowledge and Explicit Beliefs on Selfother Ratings (N = 40)

Variables B β R R² Change F Change Sig. F Change
Factor 1 1.55 0.23 0.23 0.05 0.05 3.32 0.07
Factor 2 1.11 0.16 0.28 0.08 0.03 1.72 0.19

It is apparent (Table 11) that together these two stereotyping factors explained 8% of the variance in selfother ratings, more variance than P(S) scores (rejection of influence) were explained by (4%). This result suggests that expectations for competence mediates the relationship between stereotyping and competence-related behaviour.

<< >>

Online Premature Ejaculation CBT Treatment
Only 99USD$