0

Brief summary of my study: I am studying two types of organizations, A and B. For both organizations A and B, I measured how centralization affected innovation. There were significant differences between the results in organization A compared to Organization B.

I have 303 total responses--212 from Organization A and 91 from Organization B. The Pearson correlations of centralization on innovation were -0.04 and 0.33, respectively.

So I have two populations, the centralization score (independent variable), and innovation score (dependent variable).

Here are my questions:

  1. It was suggested that I do a t-test. I was wondering if I should do the t-test on the independent variable, the dependent variable, or both? Also, if the independent t-test on the innovation score says there is no significant difference between populations, does that mean these Pearson correlations are bogus?
  2. Similarly, do I find the p-value from the all 303 responses, or for each of the two types of organizations?
  3. Finally, what other tests should I run?
gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
Ed Clark
  • 1
  • To be clear: Are you trying to see whether the differences between the two organizations in the effect of centralization on innovation? Why not use regression? – Marquis de Carabas Apr 28 '16 at 20:51
  • Yes, as you see in my results, there were differences in how centralization affected innovation. As a stats newbie, I am not sure how regression helps me. I can take organization type 1 and run a simple regression on how C affects I. Then I can do the same with organization type 2. Is that what you recommend? How does that bolster my findings? – Ed Clark Apr 29 '16 at 00:34

1 Answers1

1

If you want to check if correlation is the same in the two kinds of organizations, you can can do a test on equality of correlations.

You may also want to check if slopes of regression lines in both kinds of organizations are equal. Then, you can see this previous question in the site.

A t-test is useful to compare means. In your case it could be used to check if both organizations have the same centralization mean (or innovation mean), but that doesn't seem to be what you are looking for.

Community
  • 1
Pere
  • 5,875
  • 1
  • 13
  • 29