input of variables for excel regression

**marina1** · 11-14-2016, 01:32 PM

I am running a regression using annual salary as the dependent variable.
I want to enter four variables.

I understand that the variables are either entered using 1 or 0 or the numerical amounts.

One variable is gender (0= male, 1= female.
Time with company (entered as number of years)
Time in job (entered as number of years)

However, my fourth variable is "location" (city) of where the work is performed.

I need to use the location to help determine if that has an affect on the overall difference in salary between genders.

I do not know how I would enter the 4 different locations as numeric to determine if any particular location had a greater affect than the other three. I believe that the regression would only tell me if overall "location" has an affect, even if I use 1,2,3,4 to identify the different locations.

Am I wrong, or how would I reference each location to get which location has a greater affect on the r square?

Thanks for any help

**MrShorty** · 11-14-2016, 02:51 PM

Short, but possibly useless or misleading answer, to the title question: A quick description of setting up and performing multiple regression using the LINEST() function is here: https://support.office.com/en-us/art...a-fa7abf772b6d see example 3.

If you decide to pursue the multiple regression approach, you might be interested in http://simon.cs.vt.edu/SoSci/converted/MRegression/ which describes the statistics of basic multiple regression, including some statistical tests that can be performed.

I am not enough of a statistician to know, but I would be concerned about the applicability of multiple regression to the statistical test you say you want to perform. I don't know if there is a good statistical test to determine if the gender coefficient is dependent on location or independent of location. I would wonder if an ANOVA type analysis would be more appropriate for the test you want to perform. An introduction to ANOVA: http://www.statisticslectures.com/topics/introanova/ I would visualize an analysis that has your lists of mens salaries at each location, womens salaries at each location, your desired measure of difference between the two, and then perform an ANOVA on those to determine if location seems to impact mens, womens, both, and/or difference between men and women.