Good day everyone! I have searched and read a lot of topics about this one and yet there are still varying answers left and right. I need a specific one so I hope someone can help me with this.
I adapted these questionnaires from other authors and I noticed they have different ranges.
My study is all about finding the relationship of employees' ergonomic assessment of their workspace and the employees' individual work performance.
The ergonomic assessment questionnaire (EAQ) is Likert scale 1-5:
1-Poor, 2-Satisfactory, 3-Good, 4-Very Good, 5-Excellent
The individual work performance questionnaire (IWPQ) is Likert scale 0-4:
0-Seldom, 1-Sometimes, 2-Quite Often, 3-Often, 4- Always
0-Never, 1-Seldom, 2-Sometimes, 3- Quite Often, 4- Often
I know the IWPQ has different scales so I have to solve for the Cronbach alpha first to measure their internal consistency. Assuming that the alpha is in the accepted & above range and now I have to test the relationship of EAQ and IWPQ, will it be hard for me to compute since one is 0-4 and the other is 1-5? or are the results going to be inaccurate?
PS: I am still in undergraduate, doing this for my work as a student research assistant. I nor my school does not have SPSS or any other statistical software. I am just planning to solve this through Excel or manually. So please no advice on how to do this or that through SPSS and the likes thank you
The "LIckert Scale" is just a way to try to quantify subjective answers. The names applied are arbitrary but follow a general trend. As long as both are 5-units (0-4 or 1-5) you can just adjust the numerical value of either one for a consistent set of values. You can also use your own titles for each category, if you like. Remember most of all - statistics are just guidelines! They are neither absolute nor definitive. Other than giving a PROBABILITY (itself not guaranteed!) of some correlation being "non-random" you can't use them to state that some hypothesis is even "true" - only that it might be.
In a nutshell, can just change the numbers of one of the scales so that all are 1 to 5 or 0 to 4. The absolute numbers are meaningless, since you are dealing with interval scales. You can even compute correlations without such an adjustment. However, the recoding makes it easier for you to compare means, otherwise you have to mentally add or subtract one.