Conceptual and Methodological Issues in Adapting Tests

Abstract

Suppose that a Dutch psychologist decides to produce a Dutch-language version of an American intelligence test and that a subtest of general knowledge contains the item “Who is the president of the United States?” Such an item may well have good psychometric properties in an American sample. The Dutch psychologist may decide to use a verbatim translation, “Wie is de president van de Verenigde Staten?” The Dutch question is as clear as the English, and a back translation (Brislin, 1980, 1986; Werner & Campbell, 1970) will produce the English original. Moreover, the president of the United States may be better known to American citizens, but he is certainly not unknown in Europe. When using the test, the difficulty level of the item will presumably be lower in the United States than in The Netherlands. This difference is a reflection of the difference in knowledge of the populations: There are relatively more Americans than Dutch who know the president’s name. Assuming that the subtest is unidimensional, it is quite likely that the item measures the underlying construct in an appropriate way in both countries. The conclusion may seem inescapable: The item is useful because the two language versions are similar and measures the same under-lying construct in the two countries.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 162 citations exist upstream.