topic Re: Logit odds missing valuer in Statistical Procedures

Logit odds missing valuer

sameer112217 — Sat, 26 Nov 2016 10:40:18 GMT

logit(p)== b0 + b1x1.

If x1 value is missing. what will be the value of the odds? Missing or some value?

Can anyone answer?

Re: Logit odds missing valuer

Reeza — Sat, 26 Nov 2016 11:04:04 GMT

Missing.

Re: Logit odds missing valuer

Ksharp — Sun, 27 Nov 2016 04:14:05 GMT


You mean a single x1 value is missing 
or all the x1 value are missing ?

Re: Logit odds missing valuer

sameer112217 — Sun, 27 Nov 2016 06:11:33 GMT

Single x1 with rest of values

Re: Logit odds missing valuer

Ksharp — Sun, 27 Nov 2016 12:52:43 GMT

If one single obs (x1) is not missing. SAS will remove that obs from fitting MODEL.
Therefore, ODDS should have a value .

Re: Logit odds missing valuer

Ksharp — Sun, 27 Nov 2016 12:54:47 GMT

Sorry. 
If one single obs(x1) is MISSING ......

Re: Logit odds missing valuer

Reeza — Sun, 27 Nov 2016 12:54:56 GMT

@Ksharp you've assumed the OP is fitting a model, given the formula it could be a simple data step.

Either way the answer is missing. It would also be much faster to check this out than to ask.

Re: Logit odds missing valuer

sameer112217 — Sun, 27 Nov 2016 19:47:16 GMT

Hello Ksharp and Reeza!

Thanks for the explanation. Just to make it clear I am explaining this

The logistic regression equation is: (just an example

logit(p) = -4.48 + 0.11 x AGE + 1.16 x SMOKING

I am asking for one below

logit(p) = -4.48 + 0.11 x AGE (Just one equation)

and if age value is missing ...what would be outcome? 0? or missing value? or some value

Sameer

Re: Logit odds missing valuer

Rick_SAS — Sun, 27 Nov 2016 20:26:07 GMT

The predicted value is missing for every observation in which any explanatory variable in the model is missing. The regression model cannot compute a predicted value (or a residual) when there is a missing value on the right-hand side of the model.

Re: Logit odds missing valuer

sameer112217 — Sun, 27 Nov 2016 20:50:43 GMT

Thank you everyone for the help