I had to standardize the scores of a variable because the items composing it were measured on different Likert-type scales (5, 6, 7 points). After this, I ran a correlation between this variable and other variables, with which theoretically it is supposed to correlate negatively. This yielded though a positive correlation. Correct me if I'm wrong, but is this result due to the score standardization? Thank you!
-
$\begingroup$ When you say you 'standardized the scores of a variable' do you mean you standardized the components before adding them, or you standardized the result after adding them? $\endgroup$ – Glen_b Jan 5 '14 at 23:06
-
$\begingroup$ The scores were standardized before adding them. $\endgroup$ – andree Jan 5 '14 at 23:18
No, it can't be due to score standardization, unless the standardization somehow involved multiplying by a negative number. I can't see why standardization would involve that, but you never know!
More usually, the standardization would be something like:
$S = 7/5*A + 6/5*B + C$
as opposed to
$S = A + B + C$
these two S will correlate very highly with each other and in the same direction with some other variable.
-
$\begingroup$ Indeed, it could not have been standardization, thank you! Actually, I just realized it's a coding problem.. which is an entirely different problem. $\endgroup$ – andree Jan 5 '14 at 23:20