Dear list,
I have patients nested within physicians in a two-level models.
Following the section 4.6 of Tom Snijders and Roel Bosker's book, I have found the following regression coefficients in a random intercept model :
- within physicians regression coefficient for variable X1 : -0.17 (SE of 0.06), p <.05
- between physicians coefficient for variable X1 : -0.08 (0.18) non significiant
- contextual effect : 0.09 (0.19) ns
so since contextual effect = inter - intra, and since here contextual effect is non significant, it means that 'inter' and 'intra' coefficients are NOT different (p. 59 of the book)
Does it still make sense to say that within physician coefficient is significant while between coefficient is not? if it does not make sense, what model should I specify instead?
Thank you in advance for you help.