09.12.2012 Views

I__. - International Military Testing Association

I__. - International Military Testing Association

I__. - International Military Testing Association

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Overall, the results replicate the findings from our previous research. As in the earlier experiment,<br />

the scores of soldiers given the “fake good” instructions were significantly higher than soldiers’ in the<br />

honest condition. This shows that the “fake good’ instructions were effective in producing positive<br />

response distortion.<br />

Also, scores resulting from the honest and coached-with warning conditions were not significantly<br />

different from each other. Thus, response distortion on the ASP was reduced (but not necessarily<br />

eliminated) in the group given the ‘coached-with warning” instructions. The combination of the warning<br />

of fake detection items and instructions not to appear “too perfect” may be responsible for this<br />

suppression of positive response distortion.<br />

In our extension of the research, we also examined the effect of coaching when no warning about<br />

fake detection items is given. As shown in Table 2, soldiers in this condition had significantly higher<br />

scores than those given the “honest” instructions. However, the scores of those in the coached-without<br />

warning group did not differ significantly from those in the fake good condition. Thus, the aeneral<br />

(faking) strategy (i.e., describing oneself in a way that insures being selected by the Army) and the more<br />

soecific (coached) strategy (trying to present oneself as mature, responsible, well-adjusted, hardworking,<br />

well organized, and easy to get along with) were equally effective in producing response<br />

distortion. Finally, in comparison with the “coached” instructions, the addition of a warning about faking<br />

detection items resulted in significantly lower scores on 4 out of the 6 scales. This demonstrates that the<br />

warning was at least partially effective in reducing response distortion.<br />

Grouo Differences in Correlations Amona ASP Scale Scores<br />

Correlations of the Validity scale with the other ASP scales were examined in each of the four<br />

conditions. As expected, the lowest correlations with the Validity scale were found when examinees<br />

were responding honestly & = .20 to .37, all p-z.05). The highest correlations with the Validity scale<br />

were found when subjects were coached or told to fake in the socially desirable direction (r = .30 to .71,<br />

all g< .05). The correlations with the Validity scale within the coached-with warning group @ = .11 to<br />

SO) were generally higher than the correlations found within the honest group, but smaller than the<br />

correlations for the two other groups. This indicates that the coached-wi?h warning group distorted their<br />

responses in a positive direction, but not as much as the faking or coached groups.<br />

Utilitv of the Validitv Scale<br />

for Detectina Resoonse Distortion<br />

The purpose of the ABLE Validity scale is to identify individuals who have distorted their responses in<br />

a socially desirable direction. We examined how effective this scale would be in correctly classifying<br />

persons who were coached or instructed to distort their ASP responses.<br />

Table 3 shows how well the Validity scale discriminates among the groups, for each possible cut<br />

score that might be used to classify distorted responses. For example, with a cut score of 27, no one in<br />

the honest group would be incorrectly classified as faking (i.e., deliberately distorting responses in a<br />

socially desirable direction). However, this cut score would correctly classify 22% of those given the<br />

fake good instructions, 15% of those coached, and 3% of those coached-with warnings. Thus, all<br />

individuals in the fake good or coached conditions who were at or above the cut score would be<br />

correctly classified as fakers. Moreover, this would be done without misclassifying anyone in the honest<br />

group (since no one in this group had a Validity score above 26). The results also show that response<br />

distonion among those given the coached-with warning instructions is most difficult to detect. This is<br />

consistent with the finding that Validity scores between the honest and coached-with warning groups do<br />

not differ significantly.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!