Psychometrics tutorial - Internal consistency

Resource type : this resource contains a tutorial or tutorial notes.

This tutorial involves learning how to examine internal consistency using SPSS 's reliability function.

General steps

To examine the internal consistency of a set of related items:

Recode if necessary so that all items are scored in the same direction
Compute the Cronbach's alpha (α) and examine the reliability analysis results (especially Cronbach's alpha if item removed) to determine the internal consistency and whether any items should be dropped
Recompute the reliability analysis after dropping any unwanted items
Note that the example involves only 10 items to measure one factor. If you have a multi-factorial measure, then examine the internal consistency for each set of items which represent a factor.

Exercise 1: Openness
edit
View the accompanying screencast: [1]

Data file: data_15_1.sav
Allen and Bennett 15.2 ^{[1]} , pp. 205-213

Rating scale

1
Very inaccurate
2
Moderately inaccurate
3
Neither inaccurate or accurate
4
Moderately accurate
5
Very accurate

Items

1
Believe in the importance of art.
2
Tend to vote for conservative political candidates
3
Have a vivid imagination
4
Avoid philosophical discussions
~~5~~
~~Tend to vote for liberal political candidates.~~
6
Do not like art.
7
Carry the conversation to a higher level.
8
Am not interested in abstract ideas.
9
Enjoy hearing new ideas.
10
Do not enjoy museums.

Syntax

* Openness reliability analysis.
* Recoding.
compute Open2r=6-Open2.
compute Open4r=6-Open4.
compute Open6r=6-Open6.
compute Open8r=6-Open8.
compute Open10r=6-Open10.
EXECUTE.
variable labels
Open2r 'O2. Tend to vote for conservative political candidates.'
Open4r 'O4. Avoid philosophical discussions.'
Open6r 'O6. Do not like art.'
Open8r 'O8. Am not interested in abstract ideas.'
Open10r 'O10. Do not enjoy going to art museums.'.
* Reliability.
RELIABILITY
/VARIABLES=Open1 Open2r Open3 Open4r Open5 Open6r Open7 Open8r Open9 Open10r
/SCALE('ALL VARIABLES') ALL
/MODEL=ALPHA
/STATISTICS=CORR
/SUMMARY=TOTAL.
RELIABILITY
/VARIABLES=Open1 Open2r Open3 Open4r Open6r Open7 Open8r Open9 Open10r
/SCALE('ALL VARIABLES') ALL
/MODEL=ALPHA
/STATISTICS=CORR
/SUMMARY=TOTAL.

Solution

10 items; reverse-code 2, 4, 6, 8, 10
α = .77, N = 60
drop #5, which gives α = .84