چکیده:
The aim of the current study was twofold: (1) to explore the internal
structure of the general English (GE) section of the university entrance
examination for Ph.D applicants (PhD UEE)into the English programs
at state universities in Iran, and (2) to examine the factor structure
invariance of the test across two proficiency levels. Multigroup
confirmatory factor analysis was used to analyze the responses of a
random sample of participants (N=1009) who took the test in 2014.
First, four models (unitary, uncorrelated, correlated and higher
ordered) were estimated and compared to find the model that best
represented the data. Then, the factor structure invariance of the test
across the two proficiency levels was explored using multigroup
confirmatory factor analysis. The higher-order and correlated three
factor models showed the best fit to the data. The results also showed
that the structure of the test remained invariant across both
proficiency levels. The results supported the multi-componential view
of language proficiency. It was found that there is no relationship
between levels of language proficiency and the structure of the test.
However, the results called into question the score-reporting policy for
the PhD UEE and led to the conclusion that a single total score does
not reflect the structure of the test.
خلاصه ماشینی:
com Abstract The aim of the current study was twofold: (1) to explore the internal structure of the general English (GE) section of the university entrance examination for Ph. D applicants (PhD UEE)into the English programs at state universities in Iran, and (2) to examine the factor structure invariance of the test across two proficiency levels.
To the best knowledge of the authors, few studies have explored the factor structure (invariance) of the UEE tests in general and construct validity of the PhD UEE in particular.
Studying the factor structure invariance of the PhD UEE can provide evidence on fairness and consequently, in Messick’s (1989) terms, generalizability aspect of the construct validity of the test.
There are still studies which have found that the structure of language proficiency remained the same across different proficiency levels ( Shin, 2005; Stricker & Rock 2008).
Stricker and Rock (2008) studied the factor structure invariance of TOEFL across test takers with different native languages and different amounts of exposure to English.
Wang (2006) investigated the factor structure invariance of Certificate of Proficiency in English (ECPE), and the Michigan English Language Assessment Battery (MELAB) tests across different genders.
Factor structure studies on high-stakes proficiency tests have mostly compared the fit of a series of factor models such as a unitary factor model, a correlated factor model, and a higher order factor model (e.
, language proficiency Furthermore, to examine the factor structure invariance of the test across two proficiency groups, a multigroup CFA was conducted on the baseline model obtained in the previous stage of analysis.