Test for equality of means / proportions

[Meike Wenzl]

Hello everyone,

I have conducted a test for equality of means but have propblems interpreting it.

I am comparing two groups I created before with the command "regroup" of my DV. Now I want to compare the two means of my IV "ROA change" but I do not quite understand which t-test is the correct one for me and how to interpret it. I have two groups; one downsizes, the other does not. I also read this document here http://www.stata.com/manuals13/rttest.pdf but I do not quite understand any of the interpretations. I think my example is close to the example 3; however there is no explanation. May the Ha: diff ! = 0 (means are statistically different from each other) be the right one? This would be at a level of 23.48%, correct?

I appreciate any explanation and help! I hope I included all the necessary information.

[Carlo Lazzaro]

Meike:
- I would run an unpaired -ttest- with unequal variance;
- if you're interested in a two-tail hypothesis testing, the H you mention is the right one;
- however, for no one of the three He there' s evidence that the difference of the means of the two groups reaches statistical significance at the usual arbitrary 0.05 p-value. However, the absence of evidence is not evidence of absence (I.e., a difference may well exist, but you samples are too limited to allow you to detect it).

[Meike Wenzl]

Carlo:

thanks so much for your input. You've helped me so much with my analysis for my thesis. I can never really thank you enought!

[Carlo Lazzaro]

Meike:
my pleasure.
All the best for your dissertation.