ANOVA of Race vs Racial Acceptance and Religiosity

Previously, I looked for a correlation between religiosity and racial acceptance, as determined by the 2012 Outlook on Life survey.  My alternative hypothesis was that high religiosity would correlate with high racial acceptance; however, visualization of the data in a scatterplot showed no correlation, so I accepted the null hypothesis.

This week, I decided to take the question further:  we were assigned to run an ANOVA on our data, so I chose to run two:  Race vs Racial Acceptance, and Race vs Religiosity.  In each case, race is the categorical explanatory variable; racial acceptance or religiosity is the quantitative response variable.  Here's what I found:

For  Race vs Racial Acceptance, ANOVA found a significant relationship: F(4, 2264) = 59.49, p<0.001.  Because race has more than two levels (black, white, latino, other, and mixed) and the ANOVA was significant, I ran post hoc analysis using the Tukey HSD test.  The pairwise comparisons that were significant were:  black-white, black-latino, white-latino, white-other, white-mixed.  The means for each group were:

So this means that whites showed significantly lower racial acceptance than did other ethnicities.  This was an unfortunate surprise.

For Race vs Religiosity, ANOVA again found significant relationship:  F(4, 2289) = 29.35, p<0.001.  Again for the reasons listed above I ran the Tukey HSD test. The pairwise comparisons that were significant were:  black-white, black-latino, and black-other.  The means for each group were:

So, as you can see, blacks showed significantly higher religiosity as compared to the other races.