Discussion:
Negative Cronbach's alpha
(too old to reply)
Morgan Kagande
2021-03-02 13:40:00 UTC
Permalink
can someone explain if reliability is tested using one variable at a time of it has to relate to the independent variable.
My alpha value continue to be negative even if i have tried putting uniform responses to every respondents
Does the alpha value also get to be affected by bigger sample sizes
May you kindly help
i can share my data with you if you will not mind for you to just have a look and explain the problem.
Rich Ulrich
2021-03-02 20:55:06 UTC
Permalink
On Tue, 2 Mar 2021 05:40:00 -0800 (PST), Morgan Kagande
Post by Morgan Kagande
can someone explain if reliability is tested using one variable at a time of it has to relate to the independent variable.
You may have to re-phrase this question.

Your SUBJECT line is, "Negative Cronbach's alpha."

Cronbach's alpha is a coefficient of "internal reliability," and
it is a function (essentially) of the average correlation. It is
larger for a larger number of items in the scale with the same
average r. It can be negative when you have negative correlations.

Apha shows the expected reliability (reproducibility) of a scale
score which is computed as the sum of items listed.

Where do negative alphas come from? -- SPSS Reliability does
not know about "reversed scored" items; you need to be sure
that all variables which you pass to the Reliability program
are scored so that low numbers always represent "low" on
whatever you label the scale. (To fix: Scores may be "reflected"
by subtraction -- X in the range of 1-4 becomes X_rev in the
range of 4-1 after you subtract from 5). The resulting matrix
of correlations will be (almost entirely) positive values.

So, you might pass to Reliability a list like, Var1, Var2_rev,
Var3_rev, Var4 to Var10.


There is no "independent variable" in the Reliabilty analysis, so
that question is a puzzle.

And, "one variable at a time"?
The SPSS program can show statistics on individual variables,
like "alpha when item is deleted" or multiple R with other
variables. The MR isn't useful if the N of cases is not larger
than the number of variables.
Post by Morgan Kagande
My alpha value continue to be negative even if i have tried putting uniform responses to every respondents
? I don't understand, at all. I can't figure a word to finish the
sentence which makes it make any sense, either.
Post by Morgan Kagande
Does the alpha value also get to be affected by bigger sample sizes
No.
--
Rich Ulrich
Loading...