[R-sig-eco] Help obtaining F statistics

[Dixon, Philip M [STAT]]

Luis,

The Analysis of deviance table you provided indicates that your data analysis has a serious issue, one much larger than calculating an F (or Chi-square, which is more common for GLM analyses) statistic.

The df for the sex:behavior interaction in your table is 7.  It should be 8, the product of sex df and behavior df.  I'm almost certain that one combination of sex and behavior does not occur in your data.  Appropriate analyses of data with missing cells is almost always very difficult.  You have to think carefully about the interpretation of each of the model effects (or at least, the p-values you got do not answer questions that I care about).  I recommend you find a local statistician to advise, because this is not an R issue.

Best wishes,
Philip Dixon


	[[alternative HTML version deleted]]