1

Possible Duplicate:
ANCOVA and its disturbing assumptions

I am testing whether prenatal smoking affects intelligence, while controlling for social economic status. I want to use an ANCOVA (Intelligence - DV, Prenatal smoking - IV and SES -covariate) but I carried out a preliminary homogenity of regression slopes but the results are significant for interaction effect which suggests I should not use an Ancova. I don't think I can use a regression either as I only have one IV. Any ideas?

Community
  • 1
Marie
  • 11
  • 1
  • 2
    You may want to read my answer here: [ancova-and-its-disturbing-assumptions](http://stats.stackexchange.com/questions/44928//44959#44959). If you have any further questions, edit your Q to clarify them. I don't see that you have any problem; there is nothing wrong with a multiple regression model w/ just 1 IV, 1 covariate, & 1 interaction. – gung - Reinstate Monica Jan 13 '13 at 03:48
  • I imagine that parental smoking depends on social economic status. If so, you should be very careful about interpreting your ANCOVA. Look at this: http://www.ncbi.nlm.nih.gov/pubmed/11261398 – Stephan Kolassa Jan 13 '13 at 14:40
  • Thanks gung and Stephan. Gung I wanted to use a hierarchical multiple regression so that I could control for ses but my understanding was that I would need at least two IVs for this? – Marie Jan 13 '13 at 17:18

0 Answers0