Statistical difference between two proportions over time

#1
Hi everyone,

I face a problem in which I have two groups: a treated one and a control group.
The treatment group is composed of people who for instance passed ahead of a theater in the last 3 days. The control group is composed of people who did not pass ahead of a theater in the last 3 days. In addition, we must imagine that theaters did not exist 3 days ago (for instance all the theaters were bakeries until 3 days ago).
I want to know if the theater treatment had a significant impact on the proportions of theater visitors in each group.

1) I think there might be a problem of localization bias because not everyone lives in the same area. If someone lives in the countryside he/she will be less likely to go as frequently to the theater as someone who lives in the city. Am I right?

I have two test ideas:

2) Just take randomly N people and put those who saw a theater in the treatment group (the other ones will be part of the control group). Then measure the difference in proportions between the two groups with Z test.

3) My other idea was to create my audience a week before the study. I measure the proportion of people who passed ahead a bakery and actually went to the theater, I do the same for the control group. Then I measure the proportion of people who in the last 3 days passed ahead a theater and actually went to the theater, I do the same for the control group. At the end I compare the evolution of the two proportions between the two groups.

If someone could help me to solve this problem I would be grateful,

Thank you very much for your help
 

obh

Active Member
#2
Hi Hitsuji,

I don't understand exactly what you described...so my answer may not be to the point.

If you want to compare proportions and it isn't just simple 2 groups, you may run a logistic regression

In this case, you can use predictors like the following:
x1 - categorical: treated/not treated
x2 - distance from the hospitol,
etc

Y - is binary DV