Dh. Rost et K. Hoberg, CHANGING THE POSITION OF ITEMS IN PERSONA LITY QUESTIONNAIRES - METHODOLOGICAL MALPRACTICE OR TOLERABLE PRACTICE, Diagnostica, 43(2), 1997, pp. 97-112
The starting point of the study calls attention to changes in the vali
dities and in the psychometric properties that may occur when items ge
t extracted out of multidimensional personality questionnaires and are
presented as blocks of homogenous content. To investigate the effects
of item blocking, two different versions of a questionnaire containin
g the same items were presented to two independent samples of pupils a
ged 13 to 18 years (standard version: n(1) = 295, experimental version
: n(2) = 305). - In the standard version of the questionnaire, 18 targ
et items comprising three scales to assess the self concept of scholas
tic performance and ability were randomly shuffled amongst 107 distrac
ter items. The target items consisted of six different basic items, id
entically formulated for each subject (mathematics, German, geography)
, the only difference being the naming of the subject. Six target item
s to assess the general academical self concept were also mixed among
the distracter items. - In the experimental version of the questionnai
re, the three equivalently formulated self concept items for the three
subjects were placed one under another and accentuated with a backgro
und hatch to make six blocks. On either page of the experimental versi
on of the questionnaire one block was presented, fitted into the distr
acter items whose order was maintained. - The mode of presentation (st
andard vs. experimental) showed few effects. The construct validity (f
actor structure of items, intercorrelation of the scales) did not chan
ge with the blocking of the items. While the standard deviations of th
e scales was not influenced by item blocking, in two out of three scal
es small statistically significant increases of means and homogeneitie
s (Cronbach's alpha) could be observed although these were of little p
ractical relevance. It seems premature to declare the extraction of su
bscales out of multidimensional questionnaires and their presentation
as blocks to be generally inadmissable and to call it malpractice.