Date of Award
Doctor of Philosophy
Educational Leadership, Research and Technology
Dr. Brooks Applegate
When factorial invariance is established across translated forms of an instrument, the meaning of the construct crosses language/cultures. If factorial invariance is not established, score discrepancies may represent true language group differences or faulty translation. This study seeks to disentangle this by determining whether cultural/linguistic variance can be decomposed separately from construct variance intended in the measuring instrument.
Translated forms of the God Mediated Control factor of the Belief in Personal Control Scale (BPCS) (Berrenberg, 1987) was analyzed across multiple samples for measurement and structural invariance among American native English speakers, Arab native Arabic speakers, and Arab bilingual Arabic/English speakers. Moreover, the linguistic proficiency factor of the bidimentional acculturation scale (BAS) (Marin and Gamba, 1996) was included in some models as a possible invariance mediator.
Multiple Groups Confirmatory Factor Analysis (MGCFA) and Multiple Indicator Multiple Causes (MIMIC) Models showed weak factorial invariance between American native English speakers and Arab native Arabic speakers and between the Arab native Arabic speakers and Arab bilingual speakers when they responded to the Arabic version of the BPCS. Structural invariance was established between Arab native bilingual speakers across the two BPCS forms. Lastly, Arab native speakers and the Arab bilingual speakers responding to the English version showed strong factorial invariance but not structural invariance. This finding was further examined in a MGCFA/MIMIC model to determine if the BAS might mediate the level of structural invariance. The analysis showed that there was no effect for acculturation on structural invariance of Arab bilingual speakers when they completed the English form of the BPCS.
The results of this study indicate evidence of linguistic/cultural differences in translated instruments when administered to mono- and bilingual speakers demonstrating that conventional translation methods fail to create two interchangeable instruments. However, when a bilingual group takes both language versions two weeks apart, results show that structural invariance exists, indicating that these participants are internalizing the construct equivalently crosstranslated forms. Finally, modeling acculturation as a linguistic/cultural mediator did not alter the level of invariance in these groups. Results are discussed in terms of language and culture and the meaning of constructs following translation.
Ayyad, Fatma, "Invariance in the Factor Structure of Translated Instruments: Multiple Group Confirmatory Factor Analysis and MIMIC Models" (2011). Dissertations. 322.