parametric test??? what test?

#1
hi im doing a study to see whether childrens vocabulary develops quicker when children are cared for at hom or in day care. i would like to get 50 children who attend day care and 50 children who stay at home and then measure the amount of words they can produce. my hypothesis is that children cared for at home will have a higher vocabulary due to joint attention between parent and child whereas in a day care setting this occurs less frequently due to having a ratio of 1:5 carer to children.
i am trying to figure out what statitical test to use on this data and can not figure out whether my data meets parametric assumption i know that a parametric test assumes a normal distribition and homogenity of variances but i just dont get how i would know this. therefore i dont know which test to use.
ANY help would be greatly appreciated
 

JohnM

TS Contributor
#2
You would probably use a two-sample t-test. If the underlying assumptions are grossly violated, then use the nonparametric analogue of the t-test which is the Mann-Whitney U-test.

To test whether or not your data violates the assumption of normality, use a normal probability plot and also use the Anderson-Darling test.

To check whether the "homogeneous variances" assumption is violated, you can use an F-test.