Interval and dichotomus variable

[agalvez]

Dear mates

I have a model with the following structure:

- IV. One scale consist of 5 constructs (interval variables, likert scale, multi-item)
- IV. One scale consist of 3 constructs (interval variables, likert scale, single-item).
- IV. Dichotomus variable (membership of a group=1, non-membership=0)

- DV: the contribution (in number) of each surveyed user to a specific cause (for example, "10", "50", "22", "3512"...).

Aim:

I would like to study the determinant factors of that contribution, so that I build paths from each IV to DV.

Questions:

a) it is possible to build a model with different IV?

b) When some indicators of a construct have a loading lower than 0.707 I delete it. After deleting problematic items, if a construct has only two valid indicators, I delete the whole construct because the minimum number of indicators if three. Is this right?

c) I use a validated scale that consist of three single-item constructs. In this case the item loading doesn't matter since it is "1". Is this right?

Thank you in advance

[jmbecker]

a) Yes.

b) You should not always delete items if their loading is below 0.7. Just delete them if you Cronbachs Alpha, Composite Reliability and AVE are below their thresholds. If they are above the thresholds, it is better to retain the item.

c) Yes.

[agalvez]

Just as I expexted :D

Thank you very much, jmbecker