Which statistical test shall I use?

#1
Hi,

I want to do a statistical analysis on my work but I am not sure which test to do.


My sample size is 53 individuals. The survey is looking at sex differences in attitudes to monogamy. I have developed 7 scenarios, where participants have to answer whether the individuals in the scenarios were acceptable and whether they were forgivable. Acceptable and forgivable are measured through a multiple choice scale (A-F). I would like to do a statistical test to see whether there is a significant difference in the choices in which each sex answered, for each scenario.

Additionally, it would be helpful (but not necessary) to do a statistical test within the genders, to statistically show a choice is significant.

What test shall I use / do you have any tips to help?
Thank you so much!
 

Karabiner

TS Contributor
#2
Acceptable and forgivable are measured through a multiple choice scale (A-F).
The dependent variable seems a bit unclear. Do you have something like an "acceptable-forgivable scale"
with 6 levels, and level 1 means least acceptable-forgivable and level 6 means maximum acceptable-forgivable?
Or, is it not possible to rank the answering options A to F with regard to some construct, and instead
there are 6 categories?
Additionally, it would be helpful (but not necessary) to do a statistical test within the genders, to statistically show a choice is significant.
What do you mean by this?

With kind regards

Karabiner
 
#3
Hi Karabiner,

Thank you for your reply!
So currently - it is a two part answer, the first being the measurement of acceptability, A = totally acceptable, B = somewhat acceptable, C = not sure, D = somewhat unacceptable, E = totally unacceptable, F = Doesn't need accepting - and then the second measurement being forgivable using the same scale for forgiveness. This, i guess, could be made into a scale from 1-6.
I want to measure, if the answers from Males differs to the answers from Females relating to the same scenario.

I meant by this "Additionally, it would be helpful (but not necessary) to do a statistical test within the genders, to statistically show a choice is significant." that I would like to measure if there is a statistical difference within the genders relating to different scenarios. (I know I could do this simply through percentages, but I was wondering if it was necessary to calculate the statistical significance of the differences).
 

Karabiner

TS Contributor
#4
The test depends on sample size, and whether the rating scale here is considered interval scaled or just ordinal scaled.
If sample size is small (n < 20 or so), or the answers are considered ordinal, then you could use Mann-Whitney
U-tests. If sample size is larger, and you consider the measurement as interval scaled, then you can use
the t-test for independent groups. Or, if you want to test males vs. females jointly for all 6 scenarios, you can
use Hotellings T².

With the second, you mean whether responses within the group of males (or females, respectively) vary between
scenarios? You can do this using tests for dependent measures. For a small sample or an ordinal scale, the global
test would be the Friedman test. If this is statistically significant, then pairwise comparisons can be done using the
sign test. If the measurements are considered interval, then repeated measures analysis of variance (global test),
followed by dependent samples t-test (pairwise).

With kind regards

Karabiner