Internal Consistency Reliability in SPSS
Background
In psychology studies, we often measure constructs using scales—a series of several questions that all
ask about aspects of the same construct. For example, to measure extraversion, you might ask people
to rate the following scale items from 1 (strongly disagree) to 5 (strongly agree):
1. I see myself as talkative
2. I see myself as quiet
3. I see myself as outgoing
Whenever you use a scale in a scientific study, you’ll need to report its internal consistency reliability (α)
in your Method section.
Reversing Items
Before computing the reliability for a scale, you’ll need to make sure your items are all properly
reversed, if necessary. “Reversing” items means that you need to make sure all items are scored such
that higher numbers represent higher levels of the construct. For example, in the sample extraversion
scale above, higher responses to item #2 would indicate that people saw themselves as LESS extraverted
(because they’re agreeing that they’re quiet). So, before computing reliability for the extraversion scale,
you will need to reverse item #2. (If you’ve already reversed the scoring for item #2 in your survey when
you collected the data, you do not need to reverse it again!)
To reverse an item, the formula is:
(Reversed Score) = (Scale Max) + (Scale Min) – (Original Score)
So, imagine that you have the three extraversion items in your dataset:
We need to reverse item e2. It was rated on a scale from 1 to 5. So, using the above formula, we’ll use
the following syntax:
compute e2r = 5+1-e2.
execute.
Whenever you reverse a variable, always create a new variable (e.g., e2r). Don’t simply overwrite the
old variable. You never want to overwrite your original data. Always add new variables instead.
