Hi Cheuk Fai Martin Chan

I think there is no harm in setting up a single regression model as long as its assumptions are satisfied. By incorporating interaction terms between the street dummy variable and other independent variables of interest in the regression model, you will be able to test street-specific differences in those variables. This is more elegant than conducting multiple separate chi-square tests.

But if you are not interested in street-specific differences and only want to include them to make a robust model, you can use a variable selection method to eliminate unnecessary interactions in the model. I prefer the use of an information criterion or a shrinkage method. For details, please see: Article Variable selection - A review and recommendations for the pr...

Another factor to consider is that you have repeated measurement data, i.e., multiple characteristics for the same shop which might be correlated. To take care of these correlations, you may consider a mixed regression model. For examples, see https://ourcodingclub.github.io/2017/03/15/mixed-models.html

@ Cheuk Fai Martin Chan ,

It is not clear about the numerical data you have on the two variables whose relationship is your thrust.

As I understand, you can combine the two sets of data without any test and go for analysis since your objective is to know the relationship between two variables.

You can also apply another alternative. You can study the relationship between the two variables separately for the two data sets and then apply meta analysis of the two sets of findings in order to obtain the net findings.

To compare multivariate data from 2 sources I used Mantel Test...it may be applicable...check...

>>it is programmed in statistical package GENSTAT.....

MANTEL test. Assesses the association between similarity matrices References

Mantel, N. (1967). The detection of disease clustering and a generalized regression approach. Cancer Research, 27, 209-220.

>>>Manly, B.F.J. (1991). Randomization and Monte Carlo Methods in Biology. Chapman & Hall, London.

>>>Dhanoa, M. S., Lister, S. J. (1995). Mantel test of distance matrices calculated from Near Infrared Spectral data of forage tree species Proceedings of Tenth International Workshop on Statistical Modelling, Innsbruck, Austria, 10-14 July 1995,