Non-significant variables in a logistic regression model

[nyla]

If some variables of a logistic regression models are non significant, should they be considered for a risk index calculation?
Should the logistic model include only relevant variables?

Thanks for the attention.

 

[Office_Shredder]

If you have enough data to statistically determine the variables are non significant, then you can include them and you should get a 0 coefficient (probably not literally). Regression only gets interesting when your data is small for the number of predictors you want to include, in which case throwing in extra predictors you know are bad is going to cause issues.