Subgroup Statistical Testing

#1
I am looking at the my company's termination data and comparing it to the data that was taken on the termination survey (30% response rate) and am trying to determine how to decide if the subset (survey data) is statistically a good representation for the overall terminated population? Is it statistically significant

I thought of using paired T-Test which seems to follow my test mapping chart, but in reading, it says not the right thing, so considering Z-test…

Any other recommendations on what type of test to use for this, or would it be better to pull out the subgroup from the population and measure them that way?

There are ten integer variables for both the true terminations and terminations taking the survey.

My hypothesis is that there is no difference between my termination survey data compared to the overall terminated population. Moreover, I will be doing this in Excel...

Thanks
Diesel70
 

Karabiner

TS Contributor
#2
am trying to determine how to decide if the subset (survey data) is statistically a good representation for the overall terminated population? Is it statistically significant
"Statistical significance" deals with question whether the difference between respondents and nonrespondents is exactely 0.00000000000. Of course, one will not assume that the subgroup is perfectly representative for the total group. Therefore, with a large enough sample size the 0.0000 assumption will certainly be rejected. This could be the case even if differences are small. On the other hand, a non-significant result could just be due to small sample size (low power). This would not help you answering the question whether the subgroup is sufficiently representative for the total group.

Maybe you just compare respondents and non-respondents and report the differences in terms of effect size measures or something like that.

With kind regards

Karabiner
 
Last edited: