The following is a distribution of a sample of 100 taken from a large population. The distribution is measured on a "rating" scale, which my research shows lies somewhere between an ordinal scale and an interval scale.
Values Frequencies
5 25
4 25
3 20
2 15
1 15
(1) Assuming that there is NO explicit hypothesis, besides descriptive statistics, what inferential tests (if any) would you perform?
For example, being that the distribution is obviously skewed, would you still compute a t-estimate of the mean?
Estimate of the mean = Mean of the sample TIMES +/- t* TIMES Standard error of the mean WHERE t* is the t-value for df=n-1 and a confidence interval of, say, 95%.
(2) What if you summarize the table as follows:
Weighted values Frequencies
4.5 50
2.1 50
What inferential tests (if any) would you then perform?
Thanks!
Tony
([log in to unmask])
|