I am doing a study on locus of control and coping ability moderated by social support. For the subscale of locus of control - internal belief, it was found that there were no significant prediction on coping ability when using the linear regression statistical analysis. However, when the internal belief subscale was categorized into high and low levels and tested with a 2 way anova on coping ability, it can be seen that there was a significant main effect. So, how should i report this contrasting find in my research study? It is confusing and I am stuck in the discussion section because of this contrast in results obtained. Please help and advice.
These are actually two slightly different questions:
Regression: Is coping ability linearly related to locus of control-internal belief?
ANOVA: Are people who are high on internal belief different on coping ability than people who are low on internal belief?
Before reporting anything, I would examine plots of the two variables against each other. I'd also take a careful look at whether the assumptions of the models were violated.
Then I might use splines in the regression.