Development and validity of a Dutch version of the Remote Associates Task: An item-response theory approach

Abstract

The Remote Associates Test (RAT) developed by Mednick and Mednick (1967) is known as a valid measure of creative convergent thinking. We developed a 30-item version of the RAT in Dutch with high internal consistency (Cronbach's alpha = 0.85) and applied both Classical Test Theory and Item Response Theory (IRT) to provide measures of item difficulty and discriminability, construct validity, and reliability. IRT was further used to construct a shorter version of the RAT, which comprises of 22 items but still shows good reliability and validity—as revealed by its relation to Raven's Advanced Progressive Matrices test, another insight-problem test, and Guilford's Alternative Uses Test.

Introduction

Most researchers agree that creativity is the ability to generate behavior and behavioral outcomes that are unique, useful, and productive (Sternberg & Lubart, 1996). Therefore, creativity is considered as a performance or ability, manifested in original, valuable, and socially accepted ideas, products, or works of art. The creativity level of an individual can be assessed by means of performance measures derived from creative thinking tasks. Guilford (1967), who can be considered the founder of modern creativity research, drew a distinction between convergent and divergent thinking. Convergent thinking aims for a single, highly constrained solution to a problem, whereas divergent thinking involves the generation of multiple answers to an often loosely defined problem.

Influenced by Guilford's suggestions to distinguish convergent and divergent thinking, many creativity measures have been developed, such as Guilford's Alternative Uses Test, considered to assess divergent thinking, and Mednick's Remote Associates Test (RAT; Mednick & Mednick, 1967), considered to assess convergent thinking. The latter was designed in accordance with Mednick's (1962) associative theory of creativity. According to this theory, the creative thinking process consists in using associative elements to create new combinations which either meet specified requirements or are in some way useful.

The RAT aimed at measuring creative thought without requiring knowledge specific to any particular field. Two college-level versions of the test were developed, each consisting of 30 items (Mednick, 1968, Mednick and Mednick, 1967). Each item consists of three words that can be associated in a number of ways, such as by forming a compound word or a semantic association. “Creative thought” is required to find a valid solution because the first and most obvious solution is often not correct, so that more remote connections need to be retrieved in order to relate the three words to each other. Even though this arguably introduced an aspect of divergent thinking, the basic structure of the RAT (finding a highly constrained, single solution) fits rather well with Guilford's (1967) concept of convergent thinking. Notwithstanding Guilford's distinction, in most studies of problem solving and creative thinking the RAT has been used as a test of general creativity (e.g., Ansburg, 2000, Beeman and Bowden, 2000, Bowers et al., 1990, Dallob and Dominowski, 1993, Dorfman et al., 1996, Schooler and Melcher, 1995, Shames, 1994, Smith and Blankenship, 1989). The RAT has also been employed in a wide range of research including studying psychopathologies (e.g., Fodor, 1999), success and failure experiences (e.g., Vohs & Heatherton, 2001), and affect (e.g., Mikulincer & Sheffi, 2000).

Performance on the RAT is known to correlate with performance on classic insight problems (e.g., Ansburg, 2000, Dallob and Dominowski, 1993, Schooler and Melcher, 1995), suggesting that at least some items in the RAT reflect insight. The materials used in the test involve verbal associative habits that could reasonably be assumed to be familiar to almost all individuals brought up in the United States, especially in the English speaking part of the US culture. However, it has been noted that the RAT is rather difficult for non-native speakers of English (e.g., Estrada, Isen, & Young, 1994). Several non-English versions have therefore been developed: Hebrew, Japanese, and Jamaican (Baba, 1982, Hamilton, 1982, Nevo and Levin, 1978), but to our knowledge there is no Dutch version of this test available. Therefore, the aim of the current study was to develop a Dutch version of the RAT: a short, reliable, and valid measurement instrument to measure creative convergent thinking in the Dutch language. To do so we first developed and administered 30 Dutch RAT-like items. Next, we used Item Response Theory (IRT) to evaluate the psychometric properties of this 30-item test, and to shorten the test with the least possible loss of psychometric quality and information. To validate this short version, we related the RAT measures to measures from two other tasks that are assumed to assess aspects of convergent thinking: the Raven's Advanced Progressive Matrices test (Raven, 1965), which is also considered to provide an estimate of fluid intelligence, and an insight-problem test. Finally, we contrasted RAT measures with estimates of divergent-thinking performance derived from Guilford's Alternative Uses Test.