Plot predicted probability for dummies independent variable

[Yara Issa]

Hi,

I am using logistic regression analysis to test the effect of marital, employment and education pathways on the probability of remaining childless at age 45 for men in the UK.
My model is as follow:


xtlogit cl ib4.Educ2 ib3.early_job ib0.never_married ib3.earlymarriage ib3.ethnicity ib4.fatherclass ib2.health ib7.bc , or
bc: dummy variable for 5 birth cohort.

after running the model, I have saved the estimates and would like to plot the predicted probability for the main independent variables (i.e., education, marital, employment). These include dummies and categories. for example:
Education has 4 categories
age at early job 3 categories
never married, dummy

Would it be possible to plot their predicted probability?

Thanks

[Rich Goldstein]

you can get predicted probabilities for the overall model with the -predict- command but I don't know that is what you really mean; you should probably read

help margins
help marginsplot

[Yara Issa]

thank you. could you please help me with another question related to Mark command in Stata. I have some missing values in some of the variables that i want to use for a logistic regression. I want to produce a descriptive statistics tables with full sample size. how can I use the mark, markout command to have this done please?

[Rich Goldstein]

if you mean by "descriptive statistics" what is produced by the -summarize- command, then there is no reason to use -mark- or -markout-; just use -summarize-; if you mean something else, please clarify

[Yara Issa]

thanks for getting back to me. Yes, I am trying to describe the sample I have. for example their highest education level by cohorts, marital statues by cohort, etc. however, there are some variable that has missing data so in these variable I would have less sample size compared to the other with full size. I have been advised to use the Mark command to have a table with full sample size across all variables, this is achieved by using the Marks command. this is only used in descriptive but in the analysis the missing data will b dropped.

[Rich Goldstein]

sorry, but I am either confused or I don't believe what you are saying; why don't you produce your descriptive statistics and, if for some reason that isn't what you want, then describe how it differs from what you want and maybe someone can help you; however, note that no matter what command you use (other than, say, multiple imputation followed by -mi estimate-) you cannot get statistics for observations that have missing values