The impact of measurement invariance and the provision for partial invariance in confirmatory factor analytic models on factor intercorrelations, latent mean differences, and estimates of relations with external variables is investigated for measures of two sets of widely assessed constructs: Big Five personality and the six Holland interests (RIASEC). In comparing models that include provisions for partial invariance with models that do not, the results indicate quite small differences in parameter estimates involving the relations between factors, one relatively large standardized mean difference in factors between the subgroups compared and relatively small differences in the regression coefficients when the factors are used to predict external variables. The results provide support for the use of partially invariant models, but there does not seem to be a great deal of difference between structural coefficients when the measurement model does or does not include separate estimates of subgroup parameters that differ across subgroups. Future research should include simulations in which the impact of various factors related to invariance is estimated.

There has been much speculation regarding the influence of cultural norms on the acceptance and use of personnel selection testing. This study examined the Responses dummy coded as 1 5 process is same for all countries, 2 5 different processes used across countries, 3 5 mixture of custom and standardised processes across countries d Response scale is 1 5 never, 2 5 very rarely, 3 5 somewhat rarely, 4 5 somewhat frequently, 5 5 very frequently 450 RYAN ET AL.

No abstract

In this study, 249 users of the TOEFL ITP® assessment series (e.g., admissions officers, English‐language teachers, academic staff) in Japan, Mexico, and Indonesia were surveyed about their uses and opinions of TOEFL ITP scores, followed by in‐depth interviews with 21 of these users. Overall, the most common use of the test was as an exit requirement from English‐language programs to demonstrate proficiency in English listening and reading. The majority of participants saw TOEFL ITP scores as very useful indicators of students' English‐language proficiency. Interviews helped clarify the user needs met by this assessment and how test scores were actually applied. Study participants indicated that they need a relatively inexpensive and practical English‐language assessment that also provides them with enough information to make decisions about test takers' proficiency in all relevant skill areas. Some of the ways interviewees talked about using TOEFL ITP scores were consistent with the recommendations of Educational Testing Service (ETS), while other uses (e.g., for workplace applications) were more questionable, as they might imply potentially higher stakes than were intended for this test. The results of the study highlight areas where TOEFL ITP users might need additional informational support with regard to score interpretation and use.

No abstract