Two-sided test of equality for column proportions

#1
Hello!
I have a table of data that I have to show statistical significance for. The data compares the number of minority vs. non-minority persons in a program who were employed vs. unemployed during quarter 2 last year. I have to show that the differences between minority vs. non-minority pairs per row are either significant or not. I am having my work compared against another person who is doing the same work (if we get the same answers, they consider findings to significance testing confirmed, as this is a high-stakes report). Their output from SPSS shows significance in some pairs but not others, using a "two-sided test of equality for column proportions." The rule is that I must conduct the test using another tool besides SPSS to get that confirmation that our work is right. I attached the original data, the SPSS output I have to match, and a third tab to do this testing in. I don't know how to use R or SAS for this. My questions are:
1) Is this two-sided test of equality for column proportions the same thing as a z-test? If not, what IS this test?
2) With this specific dataset, how do I conduct the right test in excel? Or, should I use an online calculator of some sort to do it?
This has become time sensitive, and so any help is genuinely appreciated, thank you!
 

Attachments

obh

Active Member
#2
Hi Sciecnas,

I would say you need to check what differences are significant. (not have to show statistical significance...:))

I assume you will need to take a smaller significance level due to multiple comparisons.

Yes it is the same as z test since the test for two proportion use the normal approximation. (calculating mew and sigma based on the proportion)

You may use the following calculator, it will also generate the R code, so if you prefer R.
https://www.statskingdom.com/121proportion_normal2.html