OrientationObservationIn-depth interviewsDocument analysis and semiologyConversation and discourse analysisSecondary Data
SurveysExperimentsEthicsResearch outcomes
Conclusion
8.3.5.4 Interchangeability of indicators There is always a concern as to whether the indicators selected to represent a concept, be they single indicators or combined into an index, are appropriate. In essence, this concern can only be addressed through an analysis of the concept and the theoretcial 'fit' of the indicators to the concept.
However, there is an argument that one does nott need to be overly fussy about which indicators are selected. If one has a set of indicators, all of which seem to address the theoretical concept (and are thus theoretically valid), it doesn't much matter, for the analysis as a whole, which one(s) you select. This is known as the interchageability of indicators.
The idea is based on practical examples, rather than theoretical exposition and works as follows. If indicator 1 and indicator 2 are both appropriate indicators of the concept, then it doesn't matter which one you choose. Any single individual in the sample may have different responses to each indicator but the sample as a whole will demonstrate, overall, the same response whether it is indicator 1 or indicator 2. This has been demonstrated to be the case as far back as the the early 1960s, when the quantitative sociologist from Columbia, Paul Lazarsfeld, a proponent of quantitative scales, demonstrated the purely pragmatic process of scale-item selection (see Lazarsfeld, P. et al, 1972). He showed that more-or-less any question phrasing on a broad topic would result in so-called statistical reliability.
So, to reiterate, while individual people will be categorised in different ways in respect of a specified concept for each of the two indicators, the group as a whole will exhibit the same overall pattern for indicator 1 as for indicator 2. More important, the correlation of indicator 1 with a second concept (variable) for the group will be the same (more or less) as that of indicator 2 with the second variable.
For example, a study may be asking whether there is any correlation between gender and political party allegiance. There may be two indicators of political voting allegiance: I1, 'which party do you normally vote for?' and I2, 'who did you vote for at the last election?'. A sample of males and females might indicate that on I1, 30% overall voted for a specified right-wing party and on I2 that this figure was 28%. The results are different and so not everyone answered the same way for the two indicators. The main thing, though, is whether the correlation between gender and party allegiance works out the same for both indicators. The argument of those who support the interchangeability of indicators is that, in practice, the results will be very similar.
Activity 8.3.5.4 Operationalise the concepts that occur in your hypotheses in Activity 8.3.1.
