Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Using Cronbach's alpha to generate sub scales.

From   Stephen McKay <[email protected]>
To   [email protected]
Subject   st: Using Cronbach's alpha to generate sub scales.
Date   Tue, 14 Oct 2003 19:34:00 +0100

Hi Francis

> Date: Sun, 12 Oct 2003 17:58:46 -0500
> From: Francis Owusu <[email protected]>
> Subject: st: Using Cronbach's alpha to generate sub scales. 
> Hi all,
> I need help using Stata to calculate Cronbach's alpha. I have data on
> different measures of performance and I want to group them into
> subgroups for further analysis. 

This sounds like a job for -factor- analysis.  Worth a try before you 
set up the subgroups

> I have 2 problems:
> 1. Several of the questions were in negative, however in many cases,
> the negatives were not reversed in the calculation even though I did 
not include "asis" command. Any suggestions?

The reversing in the calculation is empirically driven and not 
automatic - the algorithm did not 'think' they should be reversed.  In 
version 8 there is a reverse() option; otherwise do the reversing 
yourself using -recode- or -replace- (or -gen- a new var and use that).
> 2. All of the survey questions are not on the same scale, so I
> include 
> "std" command to standardize the items before summing. However, the
> new 
> variable generated for the new scale had many negative scores.  I
> know that 
> the scale should not be in negative. What would you advise?

The std command 'centres' variables on zero, so the std-ised variables 
will then contain negative values (perhaps half, if the var is normally 
distributed!) and so will the resulting scale.

Advice - personally I wouldn't worry about the negative values, and if 
I did I'd do the standardisation/summation using normal data derivation 

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index