Chi Square Test For Independence with Likert Scale

G

Gaëlle

Guest
#1
Hello,

I would like to do a Chi square test for independence with a Likert scale to see if there are any associations in gender and education level with regards to ethical consumption.

Each question of my likert scale has 5 categories (definitely wouldn't buy it, wouldn't buy it, might buy it, would buy it and definitely would buy it). I would like to know if there is a way for me to reduce the question categories so that there are only two of them for the qustion variable: wouldn't buy it (which would take into account the responses from definitely wouldn't buy and wouldn't buy it) and would buy (which would take into account the responses from would buy it and definitely would buy it).
I would also like to have only two categories with education level, namely higher education and below higher education. At the moment, I have 8 categories.

By doing this, I'll have two variables with only two categories for each Chi square test and therefore, it might be easier to analyse.

Is there a way to put categories together when doing a Chi square test on SPSS 18?

Thank you for your help.

Gaëlle

PS: Sorry if there are any mistakes, English is not my mother tongue.
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
I wouldn't recommend collapsing the data unless you have a good reason, beyond ease. You will lose some of the exactness. I typically collapse groupings of cells if there are low number of values or if collapsing them will better answer the question. The amount of work to examine all groups then individual groupings isn't much more once you have the concepts down. Don't currently use SPSS, but guessing you can create a new variable based on Bolean statements (e.g. New_Response = no, if response = 1-3; New_Response = no if response is > " " and = 4-5)