0
$\begingroup$

There are responses of my school's alumni on the skills they perceive they acquired during the studies and are required from their job now. 18 skills evaluated from 1 to 5 (ordinal values, right?). Also they are 18 classes of graduation (around 30 responses from each year). There are also such attributes of the respondents as a master's degree, current status, so on. There is no regression model I could construct up to this moment, but I'm at least interested if the difference in averages of the skills between different groups are significant. I try R over and over. I managed only to tabulate the means for the groups. But then I'm lost.

  • Are there tests I should run the data through first?
  • How to know what distribution my data follows?
  • And what are the ways to see if the means differ significantly across the groups?
  • Is it possible without a regression model at all?
$\endgroup$
  • $\begingroup$ Welcome to this site! What is the question? As stated, it looks like it is about how to use R to perform two- or multiple-group comparisons (as you title suggests). $\endgroup$ – chl Apr 5 '15 at 16:54
  • $\begingroup$ Thanks for editing my question. I made it more specific and added questions in the end. $\endgroup$ – sammax Apr 5 '15 at 17:05
  • $\begingroup$ Could you be more specific? I see at least five questions you included, all of them very broad, what makes quite unanswerable. Basically, your question could be summarized as "tell me everything one needs to know to analyze such data" what makes it too broad. $\endgroup$ – Tim Apr 6 '15 at 6:27
0
$\begingroup$

Ordinal responses on the Likert scale (1-5) are not quantitative, and you can't take, for example, a mean or average. There is no easily definable "distance" between a score of 4 and a score of 5, for example. You will have to look into some other types of statistics, depending on how deep you want to dig. Here is an interesting place to start, with some suggestions for ordinal or interval data:

http://www.ats.ucla.edu/stat/mult_pkg/whatstat/

Good luck!

$\endgroup$
  • $\begingroup$ I should add that you might want to look into frequency analysis - particularly contingency table analysis (tests of independence). Independent variable - Acquired/Required; Dependent variable(s) - ordinal values. $\endgroup$ – danno Apr 6 '15 at 12:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.