Response to Another Student Discussion- Significant Results

Deborah Anthony

Top of Form

An F statistic is generated by running an ANOVA or MANOVA.  The goal of the F statistic is to help researchers determine if the means between two populations is significantly different.  F statistics are similar to T-tests or statistics, which can indicate whether a single variable is statistically significant.  If a researcher has found a significant F with her MANOVA, she can interpret it as meaning that the results did not occur by chance and retain the null hypothesis that the means are equal.  If the results are an F statistic that so large that it falls way outside of the critical value of the distribution, the test data can be thrown out – meaning that you can’t reject the null hypothesis (Warner, 2013).   

In the event of a significant F statistic the next step would be to assess Levene’s Test of Equality of Error Variances to determine if the p-values contained therein are greater than .05.  If the p-values are greater than .05 (which we want them to be), they will provide some evidence that the equal variance assumption is satisfied on the univariate level.   We then use separate ANOVA tests to look at the individual dependent variables.  If there is a significant result in the p-value of the ANOVA tests, the researcher can then look at the means in order to describe the differences they’ve discovered (NOTE: the p-values displayed in the ANOVA tests is the Bonferoni adjusted level -- .05 is divided by the number of dependent variables) (Foltz, 2014). 

Deborah

References