Don't know the detail of your model, it seems that the following should be helpful.
Before you do any logistic regression, the following are two basic things to check:
1)missing value pattern of your dependent and independent variables since logistic regression use listwise deletion of observation values if any of your variables have missing values
2)perform univariate analysis between your dependent and independent variables to
a) have a general idea which independent variables should be in the final model
b) check if some of your independent variable have "complete separation" for dependent variable (an example: assume "death" is your outcome, you find all death are male)
Consider X as a single predictor of Y. There is a situation
in these data called separability. Note that a cutpoint on
X of 5.5 separates the Ys. Try a logistic regression of
Y on X, and you get Bs and Standard errors that blow up.
Likewise, consider Z as a single predictor. Here, Z=5
straddles the boundary of Y being 0 or 1. A similar thing
happens with Z as a single predictor.
X and Z are highly but not perfectly correlated. When you
do a logistic regression of Y on both X and Z, Logistic Regression prints
the kind of messages that you report.
So, take a closer look at your data, and look for situations
of separability or near-separability (complete or near-complete separation) and/or high correlation
in your predictors.
Actually, the last example I presented have given you answer.
Anyway,
1)as regards to "complete/near-complete separation", run univariate logistic regression between your dependent variable and each of your independent variable, your will see very large "S.E" (blowing up standard error) in "Variables in the Equation" in SPSS output if there is "complete/near-complete separation"
2)as regards to high correlation (multicolliniarity), for continuous variable, just run correlation matrix of all your independent variables to have a general idea of their correlation
扫码加好友，拉您进群
全部版块
我的主页
收藏
