Query about Zero Inflated Poisson Regression

[Deepty Sarder]

I am conducting a study on demand of cinema where cinema goers and non-goers both are respondent. and 51% respondents are non-cinema goer. so the frequency of going to cinema hall within last 1 year is more of Zero. 49% respondents went to cinema hall and there responds were like 1,2,3,5 times. this is a cross sectional study. those are count data. so I felt ZIP regression is best because of excess of zero as it has the logit part for excess zero and poisson part for other count data. Now I need appropriate command of stata to conduct the study. I already got some command but I am skiptical. My inflated variables are not found statistically significant. I wonder is it a sign of chosing wrong model? Or i have to think about some other variables?

If I don't have any inflated variable in my mind, what to do? because i have few variables in mind as inflated variable. but i am confused. if i take few variables, will it create any problem? like losing degrees of freedom? because I have data of 185 respondents.

I would be grateful if you give me authentic guideline for my study.

Regards
Deepty Sarder

[Carlo Lazzaro]

Deepty:
you can take a look at -zip- entry in Stata .pdf model.
As per your description, your approach looks fine, as your dependent variable has zeros with different meanings.
Also, take a look at John Mullahy 's pivotal work on this topic: Mullahy, J. 1986. Specification and testing of some modified count data models. Journal of Econometrics 33: 341–365.
As usual, the lack of statistical significance should by no means be the most relevant aim of your research goal (and there are many reasons why statistical significance is out of reach).
As an aside, please note that your chances of getting (more) helpful replies is conditional on posting what you typed and what Stata gave you back (via CODE delimiters, please), as reminded by the FAQ.

[Phil Bromiley]

To add to Carlos helpful comment, you seem to be unclear about which variable must be inflated. Zero inflated Poisson is to deal with the dependent variable having too many zeros. So it's not clear what it would mean to say you have a few variables in mind as inflated variable. You also refer to inflated variables which in doesn't make sense to me.

It would also be worth thinking about running a conventional poisson.

[Deepty Sarder]

To add to Carlos helpful comment, you seem to be unclear about which variable must be inflated. Zero inflated Poisson is to deal with the dependent variable having too many zeros. So it's not clear what it would mean to say you have a few variables in mind as inflated variable. You also refer to inflated variables which in doesn't make sense to me.

It would also be worth thinking about running a conventional poisson.

Inflated variables are those variables that may be the reason of execess zeros . right? I have in mind few variables name. My study name is "factors influencing demand for cinema" so here frequency of going to cinema in last 1 year is dependent variable. i included cinema-goers and non-goers as sample. 51% respondents are non-cinema goer. In context of bangladesh, I think community attitude, family belief and marital status are probable inflated variables. I run the ZIP regression and i got conservative family belief is at satistical significant point. other two variables are not in significant level. what to do next? in poisson part of ZIP model i got 60% variables are at significant level (6 out of 10). what do you think now? am I in right path?

[Marcos Almeida]

Surely the presentation of data (a small fraction would do fine) plus command plus output would be helpful to clarify the query. That said, the “puzzling” aspect lack of low p values doesn’t necessarily mean the model is faulty.