On Fri, 17 Nov 2023 12:39:04 -0800 (PST), anita nkomphela <
asxolisenkomphela@gmail.com> wrote:
Hi everyone,
how does one solve the ploblem of having negative cronbach's alpha
on the PSPP. Is it because my data is wrong or what, becuase when i
try to do the item deleting it just makes it worse. please help
Reliability procedures will assume that the correlation matrix
of the items is (almost) all positive r's. If they aren't, then you
don't have much of a latent factor. Or you are looking at the
wrong data/ looking wrong at the data.
Negative r's imply that you need to reverse the scores for items
whose implicit direction is opposite of whatever you are calling the
factor.
- for var with values 1-5, you can reverse the scores by using, like,
COMPUTE r_var = 6- var.
After computing the new variables, be sure that you name the
right ones for the analysis. (Look at the correlation matrix to
be sure.) (I've made the dumb mistake of creating a new var
and then forgetting to change the analysis.)
--
Rich Ulrich
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)