the Position Generator

social capital measurement with network members' occupations and prestiges

the instrument
The Position Generator (Lin & Dumin, 1986; Lin, Fu, and Hsung, 2001) is currently the most often used survey instrument for the measurement of individual social capital. It does not include the mapping of an ego-centered network (as with the use of Name Generators), which may save a lot of interview time. Instead, it is based on the idea that a list of `positions' in the form of occupation titles can 1) be used as cues to report network members, and 2) can be used to give indications of the social resource collections that are embedded in an individual's social network by means of occupational prestige indices. The Position Generator is specifically designed for the measurement of access to social resources useful in instrumental actions. Its operationalisation is directly based on the notion that occupational prestige is a good indicator of social resources. More specifically, it is assumed that network members with higher occupational prestiges control more (financial and cultural) resources, and have more access to and influence on third parties (political resources).

strong and weak points
The current popularity of the Position Generator in social capital studies can easily be explained by its methodological advantages. First, the instrument is constructed from a firm theoretical basis, which is universal enough to enable similar applications across various populations and cultures. Second, the construction of indicators from its data has largely been standardised (see Van der Gaag & Snijders, 2003; Van der Gaag, Snijders & Flap, 2008). Both aspects allow for easy comparisons between studies. However, where Name Generators can lead to lengthy interviews, and Resource Generator instruments can be troublesome in their construction because of interculturally disputable answers to the question "what are useful and usable social resources?", Position Generators also have disadvantages. The most obvious of these is that they require an extra interpretation step in terms of the instrumentality of accessing social prestige: not every social capital question can be translated into such terms, especially when the use of social resources in expressive actions is studied (Van der Gaag, Snijders & Flap (2008)).

There are several practical issues to consider when constructing a Position Generator instrument: which and how many occupations should be selected and how these should be worded (see Erickson, 2004; Van der Gaag, 2005: ch.6). Many practical issues respondents may encounter when answering Position Generator questions are discussed in (van der Gaag, Appelhof and Webber (2012). A brief summary is that respondents like these questions, sometimes find them odd to answer, wording of the names of the occupation items may pose a problem for them (when too generic, too old-fashioned or too unfamiliar), and they may overlook a part of their network. Because they generally know little about their network members' occupations, or rarely consider them in social interactions, some sets of network members fitting Position Generator items seem to get overlooked. Instead of developing a new Position Generator, you may want to check whether you can use existing, name generated data, which can sometimes be converted into responses 'as if' resulting from Position Generator items, provided they contain sufficient numbers of Name Generators, and name interpretation data about network members' occupations.

Which resource indices should be used in the construction of Position Generator measures, and which measures should be constructed is also an issue. In most studies so far, indicators calculated from Position Generator data have been the highest accessed prestige, the range in accessed prestige, the number of accessed occupations, the average accessed prestige or the total accessed prestige. Here is some SPSS syntax ready to adjust for your set of dichotomous Position Generator items to directly calculate this set of measures. These and other options to model indicators from Position Generator data are discussed in van der Gaag, Snijders & Flap (2008). If you have no time to go through all this, my recommendation is to just use 'number of accessed occupations' and 'average accessed prestige' as measures.


So far, the Position Generator has been applied in many different social capital studies. A comprehensive selection of such studies is listed in this overview. The 2008 Lin and Erickson volume 'Social capital: An International Research Program' features many of them. If you know of other important Position Generator studies not listed here (perhaps your own) let me know.

related paper