0

I am hoping to interpret the results of a logit regression model measuring hypertension as the dependent variable. I coded the model to calculate coefficients based on the marginal effect at the mean. I understand that all other continuous covariates are calculated at their mean value. But what value are other binary variable assigned when controlling for them? To illustrate my question more clearly, see the photo below. I've standardized age and BMI, so, for example, moving 1 SD in age is associated with an 11% greater probability of developing hypertension, for an individual with the mean BMI. But what is going on with my binary immigrant variable? Is it being held at 0 (therefore not an immigrant) when I evaluate the age coefficient? Thanks!

enter image description here

juliah0494
  • 43
  • 4
  • Would https://stats.stackexchange.com/questions/133623 answer your questions? – whuber Sep 29 '20 at 19:11
  • @whuber thank you for referring me to this! So it seems like the binaries are set to 0 when evaluating the marginal effect of a particular variable? – juliah0494 Sep 29 '20 at 19:19
  • I'm not sure that's a great way to think about it. The coefficients are estimated while *controlling for* all the other variables. Their meanings are best understood by analyzing the model. Neither the estimation procedure nor the model actually set the binaries to any particular value. This is very tractable, easily-interpreted model: it posits that the log odds of the outcome are a *linear* function of the explanatory variables. With linear functions, the marginal rates (aka partial derivatives) don't depend on what the values of the explanatory variables might happen to be. – whuber Sep 29 '20 at 19:22
  • @whuber ok thank you so much for clarifying! – juliah0494 Sep 29 '20 at 19:28

0 Answers0