Year 12s: what will be the first way you research your future options? >>

Which statistical test is most suitable?

Hi all.

I have been asked to propose an experiment related to prenatal alcohol assumption and working memory. The experiment i've designed has three groups: a control group; a group containing children whose mother drank within the recommended limits whilst pregnant; and a group containing children diagnosed with FAS.

I planned on giving each participant four tests: two testing the phonological loop capabilities and two testing the visio-spatial sketchpad. From there, I want to firstly compare the scores from tests measuring the same component to determine whether there is a statistical link - I presume this is testing for validity? Then I want to combine each participants' scores into two - one score representing phonological capabilities and one viso-spatial - and then compare these results across groups to determine if there's a statistical difference in results.

My hypothesis is that as alcohol consumption increases, there will be a significant reduction in the power of working memory components.

Could anybody possible explain the statistical tests I would need to do to achieve this?

Thank you! :smile:

Reply 1

iammichealjackson

Original post by Anoth3rhobo

Firstly you haven't defined what your control group is? Surely the mothers who drank within the recommended limits would be your control (or those who haven't drank at all). Given that in real life you would have actual time/money constraints, it isn't a good idea to have three groups.

You are right to test for the validity of your working memory measures. IF you have studied factor analysis before, I would conduct a confirmatory factor analysis and try and compare a "nested model" which has the two WM components you listed versus a model which has just one working memory component. Alternatively, if that is all gobeldegook to you, you could use cronbach's alpha to test how reliable the scale is, however this doesn't really tell you much about whether all your tests measure the same thing. You could just analyse the correlations between each test and decide if they all correlate well with each other, and you get a good score on the cronbach alpha, that you can make a aggregate score from all the measures. You would do this by converting each individual score on the tests to a standardized score (http://en.wikipedia.org/wiki/Standard_score) and then find the mean from each test for each individual.

Next, if you decide to only use two groups, you could simply use a t-test to examine whether there is any difference between the two groups (based on the p-value) and examine the magnitude (using cohen's d).

It is likely that some extraneous variables will make interpretting the results difficult, however. You COULD control for any extraneous variables using an ANCOVA design, which is imply an ANOVA, where you can control for covariates. ANOVAs are like t-tests in that they let you test for a mean difference between two groups, but in the ANCOVA you can control for any continuous variable. Some extraneous variables which might be relevant are: birth weight and SES (using income as a measure). In this case you could use the ANCOVA to test for the main effect of alcohol consumption on WM whilst controlling for these other factors.