Significance between samples

F

frank1987

New Member
#1
#1
I am not an expert user when it comes to SPSS so maybe this is more than I can handle. Here is the situation:

I have two products I am comparing, Product A and Product B. I use the typical 7-point scale of Strongly Dislike (1) to Strongly Like (7) with Neither Like Nor Dislike (4). I recoded the three dislike opinions as one value, the three like opinions as one value, and the neither as one value - 1,7, and 4 respectively. I run a crosstab on with my recoded variable against the product variable and get the combined results I am looking for.

Here is what I don't know how to do. Lets say 40% of the people "liked" Product A and 42% of the people like Product B. Is there something in SPSS that would allow me to take those two percentages and calculate their significance? I looked at the One-Way ANOVA under compare means, but that groups them all together. I need it broken down between my recoded values and the two products.

Is this possible? If it is possible and you are willing to help, I would be really grateful.
 
I

Injektilo

New Member
#3
#3
You have a couple of choices:
1. Run an independent sample t-test, with the test variable being the rating and the group variable being the product. However, the test variable needs to be the full 7 point scale, not the one you recoded. Technically, you probably should use the Mann-Whitney test, but it will give you the same answer.
2. Run a chi-square test (when you run a cross-tabs, click on "statistics" and select chi-square). Here, you should use the recoded rating variable. Note that the chi-square will only tell you that there's a difference going on somewhere, you'll need to check the data to see where that difference could be. If you're only focusing on the top 3 box of the scale, I'd recode further into a binomial variable (either rated in the top 3 box, or not), and run the chi-square test on that.

If your results are actually 40% for product A and 42% for product B, I can tell you right away that the results are not statistically significant, unless if your sample size is huge (as in, the tens of thousands).
 
You must log in or register to reply here.