Pooling data for multiple regression

[funkymunky]

Hi All, :wave:

I have a few issues with some collected data.

I have been collecting attitude/perception questionaires using likert scales for a new technology application.

I want to perform some regression analysis on these, however the sample groups are relatively small (15-20) and I have 5 groups - since there are several independant variables, i need a large sample size in order to perform the analysis. Can anyone tell me the required tests i need to complete in order to allow for the pooling of data?

Will appreciate any help

 

[JohnM]

Can you give us a better sense of what the groups are and what independent variables need to be looked at? It will make it easier to answer your question....

 

[funkymunky]

ok here goes,

I have 4 variables in total {a, b, c, d} each is a higher order construct calculated by several response items {a1, a2, ...b1, b2, ...c1, c2, ...d1, d2...}

My hypotheses include testing several unidirectional relationships, namely:
a -> b
a -> c
b -> c
b, c -> d

The sample groups are taken from different sites, all using the same piece of kit to achieve a given task. These include school classes from sweden, finland, greece and the UK.

Do i need to conduct reliability analysis such as alpha, and factor analysis on each individual group? or am i able to treat these as a single population, before doing these analysis'?

ideally i would like to use the the sample in its entirety so would like to know what conditions are requred for me to treat these samples as a single population.

best regards