Hi all,
I did a multiple regression analysis with four IVs (autistic traits, ADHD traits, anxiety symptoms and depressive symptoms) and DV (sleep problem) and the overall regression was statistically significant, explaining 22% of the variance. However, only two of the IVs (anxiety and depressive symptoms) contributed significantly to the model (where I would have expected all four predictors to predict the DV according to the literature- more info below).
Some background info:
- Those diagnosed with autism and ADHD often show more sleep problems
- Those diagnosed with autism and ADHD experience more anxiety and depressive symptoms
- Having more anxiety and depressive symptoms cause more sleep problems
I somehow wanted to explain the reason of the non-significant results of the two IVs, so I conducted an additional Spearman’s correlation analysis. Results showed that all four IVs were significantly correlated with the DV, autistic and ADHD traits also significantly correlated with anxiety and depressive symptoms) (see image attached).
My questions are:
1. Can I interpret my non-significant results (autistic and ADHD traits) in regression using the results I obtained from the correlation analysis? If so, how?
[In other words, is it possible to use the results from correlation to explain/ interpret results from a regression analysis?]
2. Having said that, is it valid to say that having high levels of autistic and ADHD traits lead to more anxiety and depressive symptoms that then lead to sleep problems? Could I use the correlation results to explain that this could be a possible underlying mechanism?
I hope that’s not too complicated to understand and I would appreciate whatever help there is. Thanks in advance!