| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||
COMMENTARY |
Center for the Study of Aging and Human Development, Duke University Medical Center; and Geriatric Research, Education and Clinical Center, Veterans Administration Medical Center, Durham, North Carolina.
You need a brief measure, say, of stress, or social resources, or functional status. You find one that not only seems to address the conceptual issues about which you are concerned, but for which information on validity and reliability have also been published. Moreover, it has been used frequently, and its name is well recognized. You are happy—you know you have done the right thing. But have you? Perhaps, and up to a certain extent; but not necessarily.
The article by Burholt and colleagues (this issue) is an important reminder that an assessment developed in one location, or with a particular sample, may perform differently in another location and with a different sample. They wished to measure social resources, a complex concept, using a minimum number of items. They selected the Social Resources measure developed in the United States for the Older Americans Resources and Services questionnaire because of its theoretical basis, statistically and clinically sound selection of items, and established reliability ("gold standard" validity had not been assessed because the "gold standard" had already been incorporated). They then did what few people do—they gathered data to check whether this measure was understood equally well in the six developed countries of Europe in which they were interested. They identified some serious problems. Using comparable sampling frames in each country and similar modes of administration (face-to-face-interviews, except for one site where mailing was preferred), they found major differences in participation rate. There appeared to be no obvious explanation for why the participation rate was so low in some countries but much higher in others. All questions but one were well accepted, but the question concerned with the availability and likely duration of help when sick was problematic and had a high nonresponse rate. In a country such as Sweden, where substantial formal resources are available (reflecting the preference of the population), this item appeared to lose the meaning it had in the United States, where support may be needed from (and expected of) family and friends in preference to formal resources. Although the factor structure of the Older Americans Resources and Services social resources section still held overall, it held with distinctly less strength than in the original.
It is obvious that translation is needed when using a measure with people who are not completely familiar with the language in which the measure was originally developed. Scholars have paid considerable attention to developing translations that correspond in meaning with the intent of the original (an example of sound procedure is given at http://www.mapi-research.fr/i_02_meth.htm). However, although critical, translation does not ensure that the measure is as relevant in one location as another, or, as Burholt and colleagues show, that the items have the same meaning and assess the same theoretical concepts among different population groups or in different political settings.
There is a lesson here. Researchers cannot assume that because a measure has been carefully developed, has demonstrated appropriate reliability and validity, and is properly translated that it will work as well in a setting other than the one in which it was developed. When the setting is markedly different (e.g., with respect to race/ethnicity, socioeconomic status, religious practice, legal and political situation, economic development), researchers must give careful consideration to determining measure equivalence. In this regard, using different types of conceptual and statistical approaches, investigators have paid attention to equivalence in areas such as assessments of quality of life (e.g., Crane, Gibbons, Narasimhalu, Lai, & Cella, 2007
), cognitive measures (e.g., Mungas, Reed, Crane, Haan, & Gonzalez, 2004), activities of daily living (e.g., Jitapunkul, Kamolratanakul, & Ebrahim, 1994), and cross-national studies designed to compare residents of developed and less developed countries (e.g., Ganguli et al., 1996). Although expensive and time consuming, such steps are necessary if findings and comparison across studies are to be meaningful.
Footnotes
Decision Editor:Kenneth F. Ferraro, PhD
Received for publication August 6, 2007. Accepted for publication August 14, 2007.
References
| ||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
