Testing logit model

I'm finishing my calculations, but after I performed analysis with logit model, I got stuck. Stata software calcualted coeficients that look like in the attached file. And it gave Pseudo R2 equal to 1. If I understand correctly, the parameters in this model are not suitable, since calculations look very unrealistic?
Also, I tried to check each parameter separately, and only 3 variables were meaningful. Other 3 parameters gave "Prob > chi2" more then 0,30, which is more than p value, so those parameters should be rejected if I understand corectly?
If I include two parameters in the model, one of them gets p-value more than criteria, which means that some parameters should be equal to 0. So the model demands that I should use only one parameter. Could you advise, are there any methods to improve my model, so that I could use more than one variable? I'm not a statistician, so don't know much about such methods.



Less is more. Stay pure. Stay poor.
I would be highly concerned with the output, since it is saturated with ".",

Are there model convergence issues, something seems highly wrong with your output. Also, OR's are invertible, meaning you can divide 1 buy your OR and get what the OR would be if you changed the reference group. For you, 1/0.00145 would be about 700. That is a huge OR. I would wonder if the predictor perfectly predictions the outcome? If so, you may have a convergence issues in regards to the maximum likelihood and may need to run logistic regression with Firth correction or run exact logistic regression to get model to converge. I would elicit help from a statistician, to make sure you aren't eventually misreporting results or poorly specifying your model..

Side note, people typically don't use pseudo-R^2 values in logistic regression. Yeah, the model outputs it, but it has little meaning in logistic regression. People usually use accuracy and its components, sensitivity and specificity, etc.
Ok. Then I have few more questions. I'm quite familiar with linear regressions, but logit regressions are new to me.
1. For linear regressions, many tests should be performed to check the data, variables, etc. Are there any tests that must be performed for logit regressions?
2. I just read about support vector machines method. One working paper indicated, that SVM method produces better results than logit model. Do you know if this is true? Would it help to use SVM instead of logit?
3. Regarding my data for logit model, goodness of fit test produces "Prob > chi2" equal to "1". As I understand this means, that model isn't good.
4. Classification test shows, that 100% of data are correctly classified.
5. Correlation matrix between variables show, that 3 variables have significant correlation: independent variables "Size", "Corr" and "Svert" have more than 0.9 correlation.
6. My purpose of research is to test the impact of independent variables on the dependent variable. Since I don't need to calculate the value of dependent variable, I use bivariate dependent variable with values of 1 and 0. My model shows, that there's something wrong with it. So I would say that the model is misspecified. In this case I tried to test the model with less variables. My idea is that, since the purpose of research is to test the impact of independent variables on the dependent variable, can I run logit regression only for one independent variable at the time? If results show, that goodness of fit is high, "Prob > chi2" is low, then I would say that independent variable has high impact on dependent variable. I would run six logit regressions for each independent variable, and would estimate, which ones have impact, and which ones don't.