Abstract:
The purposes of this study were to (a) outline a three-stage
methodology combining functional/conceptual
equivalence, item equivalence, and equivalence in construct
operationalization to investigate the cross-cultural
validity of psychological test instruments, and (b) to
examine the cross-cultural equivalence of the Physical
Estimation and Attraction Scale (Sonstroem, 1978) for
English-speaking and Thai adolescent boys.
Functional/conceptual equivalence or translation accuracy
was assessed in the first stage using four well-known
translation methods: pragmatic translation, a modified
Delphi technique, back translation, and a bilingual method.
Based on these analyses the Thai version of the PEAS was
judged to have adequate functional/conceptual equivalence.
In the second stage the item equivalence of PEAS items
across cultures was analyzed via item response theory. The
subjects consisted of 499 boys aged 14-19 years attending
Oregon public schools and 1009 boys aged 14-19 years in
Thailand public schools. Employing a two-parameter logistic
model, IRT difficulty and discrimination parameters were
estimated using the PC-BILOG program for the 54 attraction
and 33 estimation items in each PEAS version. Statistical
comparison of IRT parameters across cultures for each PEAS
item separately revealed that twelve attraction and three
estimation items had acceptable item equivalence, six
attraction items and four estimation items contained
translation inaccuracies, while thirty six attraction and
26 estimation items were judged to have differences in
cross-cultural meaning. Stage three of the model assessed
the equivalence in construct operationalization of the
translated instrument (i.e., the equivalence in the meaning
of the underlying latent trait). The presence of ill-conditioned
interitem correlation matrices in both the
English and Thai data sets prohibited such an analysis in
the present study.
