STATS MAKE ME CRY

Hi,

Thanks for offering such a great forum. I have a question about MANOVA and Levene's test. I ran the MANOVA with 2 groups (N=101 and N=95) and with 3 variables - positive affect; negative affect; and psychological well-being (PWB). All assumptions were fine except for Levene's. The sig levels were >.05 for PA but for NA it was .000 and for PWB it was .031. What are your thoughts about whether this will still be okay to continue (fingers are crossed!) or do I need to adjust anything. Alpha was set at .05.

Kind regards,
Julie

Sorry for double posting! The computer just sat there, and sat there, and nothing happened so I posted it again...

Consider that both PA and PWB both are sig. at 0.01. For the NA which is sig. at 0.000 the corrective remidies are not needed (Black and Hair, 6th ed. P 462) For more details check an example of Levene's Test of Equality of Error Variancesa
F df1 df2 Sig.
ER 13.767 2 796 .000
MD 1.046 2 796 .352
IS 1.107 2 796 .331
BD 2.530 2 796 .080
.
Levene's test of equality of error variances is applied again to check the assumption of homoscedasticity for each dependent variable. Three dependent variables morale decline, insecurity and brain drain are insignificant at 0.01level of significance, confirming the homoscedasticity. In the case of economic recession, the significance level is 0.000, indicating the possible existence of hetroscedasticity for this variable. However the sample size is very large in each group and the presence of homoscedasticity for the other three dependent variables, the corrective remedies are not needed. (Black and Hair, 6th ed. P 462)

Thank you both for your replies - it gives me a couple of directions to follow up on.