**This Chapter is Under Construction**
VII. Adaptation
Webpage last modified: 2008-Jun-20
Introduction
This module will deal with adaptation in either source questionnaires or in other language versions of questionnaires. Some disciplines use adaptation as a synonym for good translation. This is not how adaptation is used here. Under adaptation we understand the deliberate modification of such features as content, format or visual presentation of a question in order to better fit the needs of a new population, location, language, or mode, or any combination of these (see also Survey Instrument Design guidelines).
In terms of question design strategies, adaptation, after replication, is the second most popular strategy for using existing questions. Whenever existing questions are modified (adapted) they need to be treated and tested as new questions.
Translation always involves some kinds of adaptation, but adaptation does not necessarily involve translation. In this module we will discuss adaptation required because questions are to be fielded in multiple languages. The identification of adaptation needs can be made in the source questionnaire if the questionnaire development procedures have identified such needs. Oftentimes the need to adapt may only become apparent in the course of preparing translations. The concrete form of adaptation chosen for a given location is usually decided at the local level. Some adaptations may in fact be motivated by the wording of the source, others by the restricted potential of the target language. In many cases, however, adaptations match social realities more often than they address just language system requirements alone.
Forms of adaptation
This module on adaptation will treat the following adaptation needs:
Language-driven adaptation: Strictly language-driven changes are fairly predictable instances of adaptation. For example, English twenty-eight is "eight and twenty" in German, whereas "a week" may be seven days or eight days, depending on the language involved .At the same time, lexical and structural differences across languages can pose considerable problems for comparability. Thus achieving a good rendering of a source question that accommodates language-driven change and maintains required measurement properties is often a major challenge. English has multiple forms of negation beyond the negative particle "not", including negative prefixes and suffixes and negatively poled words such as seldom and rarely. Other language may not have the same possibilities in every case.
Socio-cultural, system-driven adaptation: Measurement systems are a good example of this kind of adaptation (yards, pounds, Fahrenheit vs. metres, kilos, Centigrade), as are "functionally equivalent" institutions (parliamentary elections, primary school, Value Added Tax vs. presidential elections, grade school, and purchase tax). Depending on the purpose of a question, adaptations might be simple or complex. Some, such as indications of distance, could be directly calibrated if necessary or roughly matched if that were sufficient.
Adaptation to maintain or reduce level of difficulty: Educational tests are biased if it is easier for one population to have access to the knowledge tested or perform the task required than it is for another population of equal ability. Knowledge questions are thus sometimes adapted to maintain the same level of difficulty across different populations. Language-based memory and vocabulary tests also need to accommodate differing average lengths of words and the relative frequency and difficulty of words chosen across languages. Depending on the test, other aspects, such as ease of pronunciation or visual complexity, might bias re-call repetition or interpretation. In social science disciplines, reducing respondent burden is more the issue; adjustments are thus often made to the level of vocabulary used in a translation for populations with expected low levels of education.
Adaptation to ensure local coverage of a concept: Health research has become increasingly cognizant of the fact that translated questions may not ask for the local information needed to ascertain the presence of a given medical condition (Rogler, 1999; Cheng, 2001; Bolton, 2001; Andary et al, 2003). The 2000 version of the Diagnostic and Statistical Manual of Mental Diseases (DSM-IV-TR), for example, includes localized indicators for depression not present in earlier versions (Cheung, 2004). Similar needs of local or localized questions to improve construct or concept coverage could be identified for many areas — for political or social commitment, religious identification, supervision of others’ work, or environmental perceptions and behaviours.
Adaptation to ensure questions are understood as intended: Vision assessment questions are sometimes formulated along the lines of Do you have difficulty reading a newspaper, even with spectacles? Such questions assume a scenario in which respondents are literate, that is, can "read" in the skills sense, have access to newspapers and, if their vision is impaired, also have access to corrective aids. Someone who is illiterate, for example, might understand the questions as one about whether they know how to read. If newspapers or access to eye care are not readily accessible, unintended readings of the question could become salient for respondents. In order to acquire the information intended (how well in rough terms a respondent can see), the question would need to be adapted. Adaptation in this context might in fact call for an entirely new question for the indicator.
Adaptation related to cultural norms of communication and disclosure: Speech communities differ in how they frame and conduct communication. Depending on cultural expectations regarding politeness, more or less overt expressions of politeness may be required (polite imperatives, apologies for asking a question, etc.). In similar fashion, populations unfamiliar with the survey question and answer "game" may need and expect more explanation and more directions than would survey-savvy populations. Sensitive questions are best asked in ways that allow respondents covert disclosure of the answers.
Adapting design features: Changes in the design of an instrument can be motivated by many factors including a number mentioned above. The direction languages are read or written in, familiarity with certain visual representations (thermometers, Kunin faces) and an array of culturally anchored conventions related to visual presentation, including colour symbolism, may call for design adaptation (cf. Tanzer, 2005 on diagram processing). Lexicon (a language’s vocabulary) and grammar may also motivate a change in design. For example, the English mid-scale response category neither agree nor disagree is rendered in Hebrew ISSP questionnaires as "in the middle". A word-for-word equivalent of the English in Hebrew would produce "no agree no no agree". Since this means as little in Hebrew as it does in English, a label considered to satisfy the middle category function is used instead. This is an adaptation.
Documentation of adaptation: The module will also deal with recording adaptations made and the rationale behind these changes. It will outline basic information needed to record such changes and suggests forms to use. As information about adaptations and the rationale behind them are increasingly recorded and shared across research communities, a clearer picture of regularities in adaptation needs can be gained. This will result in a more refined typology within and across disciplines which can, in turn, further inform instrument revision and adaptation practices.
Previous chapter | Next chapter | Home
© 2008 The authors of the Guidelines hold the copyright. Please contact us if you wish to\n publish any of this material in any form.