0

I try to make a binary logistic regression model from 85871 samples. I used enter method. There are seven explanatory variables and all of them have significant in Wald Test. But, in Hosmer Lemeshow Test also significant, so my model doesn't fit the data right? What can I do in order to make my model fit? Is there another goodness of fit test?

vanny septiana
  • 1
  • 1
  • 1
    With your large sample size, you can get low p-values for the explanatory variables even if your model's explanatory power is low. – mkt Mar 19 '18 at 09:43
  • You need to read up on logistic regression, many post here which could help. A few: https://stats.stackexchange.com/questions/18750/hosmer-lemeshow-vs-aic-for-logistic-regression https://stats.stackexchange.com/questions/53078/how-to-visualize-bayesian-goodness-of-fit-for-logistic-regression For us to be able to help more, we need more information! Please read http://www.statisticalanalysisconsulting.com/how-to-ask-a-statistics-question/ and write a better question! – kjetil b halvorsen Mar 19 '18 at 11:00
  • Do you intend to use your model for prediction? What is your goal? Goodness of fit is pointless in a prediction setting. – Chamberlain Mbah Mar 19 '18 at 14:18

0 Answers0