An integrative approach to measurement: focus groups as a survey pretest

Level 1 of their figure, the background concept, is not a part of the interactions involved in measurement (Adcock and Collier 2001: 530).

Focus groups are often used to simultaneously capture multiple individual opinions on one topic. This focus on the individual unit of analysis is most common when focus groups are used for marketing purposes, although it is also used in social science research, even though one of its unique strengths rests in its social nature (Cyr 2017).

Focus groups are not the only method to yield emic data. Individual interviews and ethnography also generate data that are emic in nature (Stewart et al. 2009: 592). Either of these methods could generate data to contextualize survey questions, but neither is as efficient. With three or four focus groups, a researcher can speak with dozens of individuals and attain the data saturation necessary to establish differences in language and domains of observation.

The lack of generalizability of focus groups is a characteristic that must be acknowledged when they are used in a research design.

Each focus group consisted of six to eight individuals. Two focus groups took place in Lima; one took place in Callao; and the other two took place in the Southern cities of Ayacucho and Ilave. For more information, see Cyr (2017).

The relational nature of the identity was due to participants’ tendency to compare APRA to other parties. Apristas were more disciplined and loyal than other party militants. While not conveyed here, the relational nature comes through in the broader analysis of the focus groups (Cyr 2017).

One might suggest that the meaning behind all concepts is acquired socially (see e.g., Vygotsky 1978). In using this term, I do not take such an absolute view. Instead, I adopt the term “socially constructed” to distinguish the more abstract concepts that we study in the social sciences, including race, culture, power, and corruption, from other more concrete terms, such as primary school education or election. In these latter examples, while the referent for the concept may change (e.g., primary school education may be better translated as “basic” or “grade” school in certain cultures), the real-world phenomenon itself is, broadly speaking, generally straightforward to convey. With a word such as corruption, by contrast, there is extensive discussion about what the term denotes and how people actually perceive it in practice (see, e.g., Andersson and Heywood 2009, 749–750; Chadda 2004; Brown and Cloke 2004).

In this sense, socially-constructed concepts are subject to the problems of cross-cultural comparability highlighted in the previous section. Still, problems of measurement validity with socially-constructed concepts may arise even in the absence of cross-cultural comparison.

This is an important, but not the only, kind of interaction that may be germane to the analysis of data from focus groups. Interactions occur among participants, as emphasized here, but they also occur between the participants and the moderator. This second type of interaction is potentially imbued with power asymmetries and can, if not properly mitigated, affect the interactions that occur among participants. See Ayrton (2018) for a review.

An exception is Posner (2005), which uses both survey research and focus groups to measure ethnic identity.

The larger research project addressed multiple political parties in each focus group.