Search

This paper examines reading achievement when the maternal/paternal language has become a de facto second language. The performance of a cohort of Greek-Australian high school students (N = 270) on a diagnostic Greek reading test was significantly below that of pupils in second to fourth grades in Greece. The mean item difficulty for Greek-Australian high school students was 0.35 compared with 0.51 for second grade, 0.69 for third grade and 0.80 for fourth grade pupils in Greece. The pattern of responding indicated that the Rasch model fitted the data. The effects of background factors were also examined. Students whose mother or father spoke Greek had statistically significant higher levels of ability than those who spoke Greek and English or English alone. It was also found that the length of the key words (correct responses) had a large effect on the difficulty of the questions; the longer the key word, the more difficult the question was. The implications of the results for assessment of reading in and the acquisition of an ethnic or second language are discussed. (Contains 3 tables and 4 figures.)

Heated discussions about the comparability of standards between examination subjects have kept Qualification Authorities, Testing Services, independent researchers and academics around the world busy for many years. As a result, many countries have adopted statistical techniques which aspire to make aggregated scores based on different subjects comparable. This paper presents eight international case studies and highlights the criticisms against the statistical comparability methods. The side-effects (problems) they may have sparked in various countries are discussed in length. The author maintains the view that the Gordian knot in this case is not statistical: it is one of perceived unfairness. The Qualification Authorities and the Testing Services should step forward and explain to the stakeholders what can be done and what cannot be done using statistical techniques. Once the stakeholders accept the limitations of any statistical technique, it will be easier for them to strike a balance between the desire for comparability and the need to democratically allow students to choose their own combination of subjects to study. The psychometricians should not be forced to apply statistical solutions to complex problems caused by political decisions; they are more likely to fail than to succeed. (Contains 1 figure.)