THE PRESENT STUDY
5.2 S urvey D esign
The survey design determ ines the m anner in which individuals or other units are compared and analysed. It is the basis for making interpretations from the data. Standard textbooks in social research m ethods usually contain a variety of research designs from which a researcher interested in studying some social phenom enon may choose. Since th ere is no general and inherent superiority in one research design over the other, research designs differ on the basis of several other factors.
Surveys may differ in their objectives, cost, time, and scope. Whatever the units of analysis, d ata are collected for purposes of describing each in d iv id u al un it. The m any d escrip tio n s are th e n aggregated and m an ip u lated in o rd er to describe the whole sam ple stu d ied and, by extension, the population represented by th at sample. The choice of a p articu lar research design, however, m ust be d irected by the central research question the researcher wishes to address (Moser & Kalton, 1979; Shipman, 1988; Babbie, 1990; Fowler, Jr., 1993).
The present study is concerned with the relationships am ongst activity, quality of life determ inants, and life satisfaction of the elderly. This central concern led to the adoption of the correlational (or cross-sectional) survey design. This is a quasi-experim ental approach to social scientific inquiry which compares two or more units at one point in time (Labobitz & Hagedom, 1981; Miller, 1991; Oppenheim, 1992). Apart from the study's central concern, two im p o rtan t factors, tim e co n strain ts and lim ited resources, influenced the choice of the research design. A tight budget
and time lim itations typical of dissertation research m eant th at repeated observations or a longitudinal study, even if desirable, were impossible.
The design adopted in this study, the correlational (or cross-sectional) survey method, collects data at one point in time from a sample selected to describe some larger population at that time. Such a survey can be used not only for purposes of description but also for the d eterm in atio n of relatio n sh ip s betw een variables at the tim e of the study (Labobitz & Hagedorn, 1981; Miller, 1991). This survey m ethod is not w ithout its problems. One m ajor weakness associated with it is to do with the inability to control the effects of extraneous factors that may account for variations between different groups.
Labovitz & Hagedom (1981) and Fowler, Jr. (1993) note th at surveys, as com pared with experiments, are often characterised by a high degree of representativeness but a low degree of control over extraneous factors. Because surveys do not have experim ental and control groups, o th er factors besides the independent variables may have produced the changes in the dependent variable. In contrast to surveys, a case study has a low level of representativeness. One can hardly differentiate cause from effect, an d inferring from the intensive study of one or a few cases involves a high and generally unknown am ount of risk. The m ajor advantage of a case study lies in the richness of its descriptive examples th at results from the intensive study of one or a few units. The results of these studies often suggest perceptive hypotheses th at subsequently should be tested under experimental and survey situations.
Whilst it is difficult to counter the weaknesses of case study methods, the problem s of surveys can be rem edied 'through statistical control techniques, such as partialling, partial correlation and standardisation and through the combination of several data collection m ethods' (Labovitz & Hagedom, 1981: 49) .
Although a majority of surveys utilise a single data collection method, it is not uncommon for a combination of methods to be used. Babbie (1990: 63) points out th at ’the best studies are often those that combine more than one design', since each design provides a different perspective on the subject u n d e r th e study. The social science research m ethod of m ultiple inform ation-gathering techniques was adopted in this study. The use of d iffe re n t m ethods in com bination, in which th e stre n g th of one com pensates for the weakness of another, and the potential for linking them, was initially christened as methodological triangulation by Campbell & Fiske (1959), and its usefulness was stressed by Miller (1991), who calls for attem pts to strengthen the \ alidity of empirical evidence in the social sciences by reliance on more than one approach. When a hypothesis can survive the confrontation of a series of complem entary m ethods of testing, it contains a degree of validity unattainable by a hypothesis tested with the re s tric te d fram ew ork of a single m ethod. The basic fea tu re of m ethodological triangulation is the combination of two or m ore different research m ethods in the stu d \ of the same em pirical units. Such an approach allows one m ethod to complement the o th er in term s of the interrelationships of problems, theories and m ethods (Denzin & Lincoln, 1994). Flick (1992) argues fu rth er th at the value of m ethodological triangulation is apparent, not least because the weakness of one style may be balanced against the strengths of another style.
In applying methodological triangulation, this study is based on the use of secondary sources, a stru c tu re d survey, and u n stru c tu re d in -d ep th interviews with the aged and social workers, with the aim of establishing a m ore reliable and clear p ictu re of the quality of life of th e elderly. Secondary sources consist of existing documents th at were collected by the researcher, such as governm ental records or newspapers. The structured q u estio n n aire was applied by the research er an d six w ell-educated interviewers, and the unstructured in-depth interviews were conducted by the researcher himself.