Yes, you have repeated measures/within subjects data. The complexity element is that your IVs have different values for each person-post combination. I want to do a terminology shift. In the ANOVA context, IVs, the 'by' variables are expected to be nominal. The 'with' variables are expected to be ordinal/continuous and are usually called 'covariates'. You had people make ratings on 1-5 scales. So: ordinal, continuous, thus covariates. Because you have three person-post specific covariates, you have to use the Mixed command. The alternative, GLM, assumes that the covariates are constant for each person. Mixed is complex; it will do a lot of things. I assume you are at a university. I urge you to look around for somebody to work with.
By the way, can we assume that the presentation order of the posts was randomized rather than the same for everybody?
Try out this. Change names to fit your data.
Data are in long format (use varstocases if needed).
mixed y by post with xa xb xc /fixed post xa xb xc/print solution r/
repeated post | subject(rec) covtype(csr).
Look carefully at the covtype. CSR may not be the best choice. Read up on Mixed!!
Ryan, Bruce, or Rich may well be able to offer more knowledgeable syntax.
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of Wiebke.90
Sent: Monday, May 12, 2014 7:58 AM
To:
[hidden email]
Subject: Which test should I use?
Hi everyone!
Very thankful for this board as I am in big trouble finding out which analysis I should use. Let me shortly explain my study:
I showed 120 people each 8 facebook posts (screenshots) and had them rank it on a 5 point Likert scale on three IV, namely "Level of entertainment", "Level of information", "Level of emotional arousal", and one DV "Willingsness to Share".
I am pretty sure its a repeated measures within subjects design. However I am too confused to know which test to use. I have thought about categorizing each IV as high/low (mean under 3 is low, and above 3 is high), and then perform a paired samples t-test for each IV variable. So I have two columns in SPSS, one with the share scores for the low and one for the high and then I compare the means? Is that possible to do this for each? I also have control variables, and I would have no idea how to include them in a paired t-test.
Secondly, I would also like to know which one influences the DV the most, and also interactions. The only test that we use regularly is a normal multiple regression analysis. Is there any way for me to use it? Would make it so much easier...
Thank you soo much in advance!
Cheers,
Wiebke
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Which-test-should-I-use-tp5725975.htmlSent from the SPSSX Discussion mailing list archive at Nabble.com.
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
I have thought about categorizing each IV as high/low (mean under 3 is low, and above 3 is high), and then perform a paired samples t-test for each IV variable. So I have two columns in SPSS, one with the share scores for the low and one for the high and then I compare the means? Is that possible to do this for each? I also have control variables, and I would have no idea how to include them in a paired t-test.
