Just checking - do you mean multivariate regression that is multiple responses or multiple regression (albeit with categorical variables), that is multiple predictors ?

In my case it is multiple regression - one outcome and multiple predictor variables. Both the outcome variable and predictor variables are dichotomous.

Hello Srdjan,

Selecting independent variables (IVs) by first insisting that each show a significant relationship with the DV is not an ideal method for building a model. Doing so will guarantee that you miss out on inclusion of suppressor variables, for example. Nor does it guarantee that you will wind up with the best ensemble of IVs, regardless of whether your purpose is prediction or explanation.

In any event, if your question is how to present the results, I see several options:

1. One is the "confusion matrix." That is, a 2 x 2 table in which one dimension is the estimated/predicted value or category on the outcome variable (DV) and the other is the true category on the DV. Percents of correct classifications for each level of the true category, as well as overall, would be helpful.

2. From the logistic regression, the odds ratio for each IV [sometimes noted as "exp(B)"] should be furnished, as this offers one way to evaluate the relative impact of an IV (in comparison to the others in the model), at least for your sample. Reporting significant vs. non-significant IVs in the final model would also be useful.

3. You could do something like this, for the significant IVs (hypothetical example):

Highest likelihood for a case being in the target category on the DV:

IV#1 = "Female" and IV#2 = "Married" and IV#3 = "No prior cardiac episode/s"

Lowest likelihood for a case being in the target category on the DV:

IV#1 = "Male" and IV#2 = "Not married" and IV#3 = "Prior cardiac episode/s"

Where DV might be: Survival for next 5 years among senior citizens (Yes/No),

IV#1 = Sex; IV#2 = Marital status; IV#3 = History of heart attack.

Note that classification tress take this type of approach a step further, but that's a different kind of analysis than what you have chosen.

Good luck with your work.

David Morse Thank you very much for you helpful answer.

You might also take a look at something like the attached. Best wishes, David Booth

Following this Picture

David Eugene Booth Rebaz Yahya Thank you for your helpful answer. It is very much appreciated.

Have a look at the Out of sample predictions chapter in Article Manual supplement for MLwiN Version 2.14

In Stata this is known as marginal analysis: https://www.stata.com/features/overview/marginal-analysis/

In R https://cran.r-project.org/web/packages/margins/vignettes/Introduction.html

Thank you prof. Kelvyn Jones for your informative answer.

Article Introduction to Multivariate Regression Analysis

You can use R programming language and this link will be useful for you

https://cran.r-project.org/doc/contrib/Faraway-PRA.pdf

Best regards

Thank you Ahlam H. Tolba . Actually, I've already performed all the analyses and just needed the suggestion for graphical presentation of results (type of graph, etc.) other than table.

Actually, I am here to learn as well. Thanks to the massive contributions from scholars