Basic Stats - Emoji Assessment

Hi All, I really need some stats help! I have no background in stats at all but I am a work study student helping out with something called "Therapy Dogs" where students can play with dogs during high stress periods. The group doing this project has participants rate their moods using emojis before and after they play with the dogs. My manager has asked me to input the data. At a glance, it seems clear that most participants see a positive shift in their mood but I would like to set up the spreadsheet with a formula that will do a calculation that is more meaningful. To start, I grouped the data into LOW MOOD SPECTRUM EMOJIS (LMS), NEUTRAL MOOD SPECTRUM EMOJIS (NMS), and HIGH MOOD SPECTRUM EMOJIS (HMS) and did SUMs for PRE and POST.

Here are the SUMS for 1 session grouped by spectrum:

Pre LMS = 10
Pre NMS = 10
Pre HMS = 4

Post LMS = 1
Post NMS = 3
Post HMS = 21

My question is, is there something more I should do, calculation wise, to make this data more meaningful? Once I input 40 sessions for example, I'd like to be able to represent this in a graph. What would be the best calculation to do and if you have any advice about how to set it up I would love to hear it!

1st you should do paired t-test in excel to show that dog therapy is good to reduce stress
2nd plot separate graphs for pre lms ,nms,hms and post lms,hms,nms
and compare them.
hoping you are doing well !


TS Contributor
I don't think a t-test is appropriate here.

Do you have a unique linker/ID for each participant that matches their pre and post scores?
Does each participant have only 1 pre and 1 post?
Then instead of finding the sum of LMS,NMS AND HMS you can change it to scale 0-10
where 0-3-->LMS
then run a paired t-test before and after the dog therapy
hope it will work.


TS Contributor
@abhay9318 are you set on forcing this into a t-test framework? The dependent variable is not (at a minimum) interval scale; the recode you suggest still doesn't make this more appropriate for a t-test, in my mind.
These are ordered levels of a variable with a repeated measurement after some treatment. There also appears to be no control group (i.e. subjects who did nothing but gave a second measurement after the same time period).