Reliability of TEPS Section Scores and the Total Score

The purpose of the current study is to investigate the reliability and stability of the section and total (composite) scores of TEPS from the classical test theory perspectives. For reliability analyses, multiple sets of data were collected not only from the operational TEPS administrations before and after the revision but also from the four pilot tests administered during the TEPS revision process. Cronbach’s (1951) alpha coefficients were computed for four different section scores of the tests while Feldt and Brennan’s (1989) composite score reliability coefficients were computed for the total scores of these tests. These coefficients are examined and compared across different test forms of the original and revised TEPS. Coefficients of equivalence and stability and correlation coefficients between forms before and after the revision were also examined to see how stable TEPS scores were. The results showed that TEPS section scores and the total score were reliable and that the changes introduced by the revision did not reduce the stability. The total score had high reliability above 0.9 indicating that TEPS can be used as a dependable indicator of Korean English language learners’ English proficiency to inform language-related, decision-making.