Page 1 of 1

Which analysis to use?

Posted: Fri Jul 03, 2015 7:57 am
by lawrence90
Hi, I am writing my master thesis and want to analyse the effect of the presence of a sticker on mobile phone placement and usage of students.
For DV: placement (in sight or out of sight) and sticker presence (yes/no) I used a binary logistic regression.
For DV: Usage I do have some problems finding the right test.
I would like to test Usage in 2 ways: between the no sticker and sticker group. and within with these 2 questions from the survey with the normally as a baseline.

Furthermore I do have some moderators which possibly could have an influence on the effect of the sticker on usage.

Is there someone who can help me or explain me which test to use?

How often did you check your mobile phone while studying today?
 Every minute
 Every 5 minutes (12x an hour)
 Every 10 minutes (6x an hour)
 Every 15 minutes (4x an hour)
 Every 20 minutes (3x an hour)
 Every 30 minutes (2x an hour)
 Less than once in an hour
 Did not check my phone

How often do you normally check your phone while studying?
 Every minute
 Every 5 minutes (12x an hour)
 Every 10 minutes (6x an hour)
 Every 15 minutes (4x an hour)
 Every 20 minutes (3x an hour)
 Every 30 minutes (2x an hour)
 Less than once in an hour
 Do not check my phone while studying

Re: Which analysis to use?

Posted: Mon Jul 06, 2015 8:31 am
by GerineL
If you ignore the fact that your variable is actually an ordinal variable, you can use repeated measures design.

Re: Which analysis to use?

Posted: Wed Jul 22, 2015 10:42 am
by AlexTht
Hello everyone!

I am a student doing her master thesis. I have never studied or used statistics, but have been advised to do for my thesis.
Let me describe the subject briefly and what I am trying to do : I am looking into the determinants of the EU's compliance with international trade law.
My dependant variable is compliance (with international law).
My independent variables are retaliation, reputation costs, level of democracy, third party presence, nature of EU's
cometence, export dependency.My data comprises 50 WTO disputes where the EU is a defendant.
I am trying to see which of them significantly impact the EU's behaviour in WTO disputes. One of my lecturers told me to use multivariate regression to do so.
However, I verified the assumptions and my data does not appear to be normally distributed or have variance homogeneity.
So I looked into nonparametric tests, thinking a correlation test could still be useful. However, my data does not present monotonic relationship
between the variables, so I can't use Spearman's or Kendall's tests either. I have looked at a lot of things for the past days, but have no idea
what to do from now on.
If somebody knows about this, it'd really kind to help