# How do the correlations between Big 5 personality change in studies examining “faking bad”?

I have conducted several studies that get participants to complete the IPIP personality test while role play going for a job. A common finding is that the average correlation (i.e., $\bar{r}$ between big 5 personality scales increases the more you ask people to fake. For example, the rough pattern in the data is:

• Get participants to complete a personality test honesty ($\bar{r} = .25$)
• Get participants to complete do a job applicant role-play ($\bar{r} = .40$)
• Get participants to complete a job applicant role-play and explicitly tell then to fake it ($\bar{r} = .60$)

Thus, increasing the amount of faking, increases the average intercorrelation. I'm interested in what happens when participants "fake bad" (e.g., specifically try to present a negative impression such as depressed, disturbed, or socially undesirable).

• What have previous studies found to be the effect of "faking bad" on the average scale intercorrelations of Big 5 personality?
• Have any researchers offered theoretical explanations for any observed effect of faking bad on such changes average scale intercorrelations?
-