urgent stats help

[ceire]

I need some help desperately. I would like to find out whether there is any difference between three conditions using a t test (I dont know if I need an independent or paired t test), and have tried this in SPSS but cannot seem to make sense of it.
Basically I have 6 participants who watched three kinds of animation sequence (1, 2, 3) and had to describe what they saw in their first and second language. I want to discern whether there was any difference between the number of words they used in each animation sequence. (I have also calculated the proportion action and mental state words used for each sequence). Basically I think I need someone to help me know how I enter the data I have into columns, and then to help me with how to ask for the relevant t test. If necessary I can pay for this help, and send the data I am looking at, and go from there via Skype? Really struggling with what is supposed to be a simple statistical analysis.

 

[GretaGarbo]

If you can start the spss program you should be able to see an excel looking window.

Enter the 6 participants scores for the first animation in the first column.
Do the same in the next columns for animation 2 and 3.

Save the data with a name (like: “blahblah.sav)

click on “analyse/compare means/paried samples t-test”

chose variables and click on OK.

If necessary I can pay

If you are in the US, send a cake to Jake. (He is “in the middle there somewhere”!) If you are down under send beer to Bugman!

 

[EdGr]

You know, often when a consultee seems to have a "simple" question I find it quickly gets more complicated when you look into it. First of all you have 3 conditions. Most statisticians would start with a one way repeated measures ANOVA and only do paired t-tests (with a correction for 3 comparisons) if it was significant. But you also mention "first and second language" -- that sounds suspiciously to me like a second independent variable, which would raise the design to a 2-way ANOVA. There is also the question of whether your data are normally distributed -- an assumption of t-tests, although with 6 subjects you may not have much choice (rank tests lose power badly with small N's).

Hopefully wherever you work there is some statistical support or mentoring -- this sounds like an academic study. Isn't there a statistics department in your institution?

 

[GretaGarbo]

You know, often when a consultee seems to have a "simple" question I find it quickly gets more complicated when you look into it

True.

Hopefully wherever you work there is some statistical support or mentoring -- this sounds like an academic study. Isn't there a statistics department in your institution?

But I don't have the impression that it is the intention from this site to suggest to a user to go and get advice somewhere else.

But this does not seem to be so urgent for Ceire. So I guess we can drop this. I was just curious if he would send any cakes to Jake.

 

[EdGr]

Hi GretaGarbo: I agree that our purpose is not to send someone elsewhere. But I think that sometimes that is the most appropriate answer! I'm talking about cases in which a questioner seems to be struggling to set up a very basic analysis (t-test) and yet most of us would do something an order of complexity higher. Unless the person asking is a quick study or willing to stretch themselves quite a bit, a site like this cannot replace one-on-one consulting or a good statistics class.