Measurement instruments in behavioral medicine.

Measurements is the cornerstone of science. Measurement instruments (e.g., questionnaires, observational methods, and assessment of physiological parameters) are used to gather data: these data are used to test hypotheses (deductive approach) or to generate new hypotheses (inductive approach). In both approaches, a high-quality measurement instrument is an essential requirement for good science. This applies to behavioral medicine as well. It is therefore highly appropriate for the International Journal of Behavioral Medicine to publish the special series on ‘measurement instruments in behavioral medicine’.

One of the challenges facing behavioral medicine is the bewildering variety of measurement instruments. Several years ago, 31 questionnaires and 26 performance-based tests were identified to measure pain and activity limitations in osteoarthritis [1, 2]. And this is osteoarthritis only: in other areas an equally wide range of instruments is available for measuring functioning and quality of life.

This wide range of measurement instruments constitutes a huge impediment to scientific communication. Results of studies using different instruments cannot be fully compared. Are questionnaire X and questionnaire Y really measuring the same construct or do they measure different concepts? How does a 15-point improvement on questionnaire X compare to a nine-point improvement on questionnaire Y? In a way, the situation in measuring health resembles the pre-Napoleonic era, when cities and even villages had their own measures of length, volume, and weight: this created confusion and impediments to science and trade [3]. It is high time for the field of behavioral medicine to reach consensus on preferred measurement instruments. In the natural sciences, consensus on the International System of Units was reached a little bit more than 150 years ago. Behavioral medicine should aim for a similar consensus.

At least one step into the direction of reaching consensus on preferred measurement instruments is the translation of existing questionnaires into other languages, instead of developing fully new questionnaires. Six papers in the present special series concern translation of an existing questionnaire. Ng et al. [4] evaluated a Chinese translation of the Daily Spiritual Experience Scale. Spindler et al. [5] and Pedersen et al. [6] translated the questionnaire for assessing Type D personality into the Danish and Ukrainian language, respectively. Canavarro et al. [7] translated the World Health Organization Quality of Life scale into the Portuguese language, while Maïano et al. [8] translated the Body Image Avoidance Questionnaire into the French language. In all cases, the psychometric properties of the translated questionnaires were satisfactory: translating existing instruments is an excellent alternative to developing new instruments.

However, translation might introduce subtle differences in various language versions of the same questionnaire. Even if the measurement properties of the various versions of the questionnaire are good, there is no guarantee for full measurement equivalence across cultures. In a carefully designed study, Choi et al. [9] demonstrated differences in item scores of the Job Content Questionnaire between European cultures. To reduce cross-language differences in the future, the authors argue in favor of a stricter translation process and studies using only items which have been shown not to suffer from cross-cultural differences.

Three other papers in the special series concern the further validation of measurement instruments. Lee et al. [10] validated the Life Control Scale in a new population. Guo et al. [11] and Weck et al. [12] evaluated the factor structure of the Decisional Balance Scale and the Illness Attitude Scales, respectively. Cross-validation and evaluation of the factor structure of existing instruments are highly relevant objectives: in the context of reaching consensus on measurement instruments thorough information on measurement properties of measurement instruments is highly desirable.

The present special series is intended to contribute to information on properties of measurement instruments, in the context of reaching consensus on preferred measurements in behavioral medicine.