Repeating a study, in exactly the same format, to check whether the same results are obtained a second or subsequent time (usually by a different researcher). One variation is simultaneous replication of the same study, either in the same context, or in contexts carefully selected for their known variation on one key variable. There can also be replication via secondary analysis of a major data-set, to re-test hypotheses, and to assess the impact of the particular program or software employed in the original analysis.
Arguably, and for a variety of reasons, not nearly enough resources in the discipline are devoted to replicating results: the effort, cost, and time involved in conducting social surveys usually preclude systematic replication; funding agencies are less likely to pay for investigations which seek only to confirm earlier findings; and there is little kudos for researchers (and, one suspects, not much career advancement) to be had in reproducing existing studies. (In psychology, by comparison, the dominant experimental methods lend themselves to the widespread use of replication studies.) However, although whole surveys are all too infrequently replicated, individual questions and batteries of questions often are—especially where these have been assembled into scales (for example to measure attitudes or personality traits).
Strict replication shades gradually into the idea of a ‘re-study’. Re-studies are also in a sense replications, but are usually conducted at time intervals such that observable differences from earlier findings can reasonably be attributed to real change in the subjects or processes under investigation, rather than to either chance or measurement error in the original study. Prominent and informative re-studies include the two community studies of the English town of Banbury, conducted by Margaret Stacey and her colleagues in 1948–51 and 1966–8 (compare Tradition and Change, 1960, and Power, Persistence and Change, 1975), and the studies by Robert Redfield and Oscar Lewis of Tepoztlan in Mexico (in 1926 and 1943 respectively). Redfield depicts the community in question as an ideal-typical type of folk society (smooth functioning, well-integrated, contented, well-adjusted) whereas Lewis paints a picture dominated by individualism, fear, lack of co-operation, envy, schism, and mistrust. Each party accused the other of methodological failings (for example asking the wrong questions of the wrong informants). The celebrated disjuncture between the two accounts was eventually resolved, when a further investigation revealed that the society had in all probability changed dramatically in the intervening years, due to increases in population pressure and a programme of radical land redistribution (see P. Coy, ‘A Watershed in Mexican Rural History’, Journal of Latin American Studies, 1971). See also reliability.