I am comparing the effect of several genes on blood parameters such as hemoglobin. I have done this in a multiple regression model using one of the genes as the reference category e.g CC CT TT, in this case if CC is reference category and Hb is the outcome variable, how would you interpret the results if the beta coefficient for CT was -0.45? Is it correct to say that in comparison to CC those with genotype CT had a reduction of -0.45g/dl?
Yes (if the response was measured in g/dl and no transformation was done).
The coefficients are the estimated to the reference.
The intercept is the mean of the reference group.
If you log-transform the results, the coefficients are the log fold-changes relative to the reference, the intercept is the mean log response of the reference group (it might be reasonable to transform some of the responses).
Yes (if the response was measured in g/dl and no transformation was done).
The coefficients are the estimated to the reference.
The intercept is the mean of the reference group.
If you log-transform the results, the coefficients are the log fold-changes relative to the reference, the intercept is the mean log response of the reference group (it might be reasonable to transform some of the responses).
I agree with Jochen if there exist no confounding and no interactions among expressed genes.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics