whom they meet more often, and so forth. As a result,the amount of
bonding social capital may be overstated. The ‘position generator’
partly corrects for this. In its pure form it does not create names
and it has no name interpreter. The respondents are asked whether
they know or do not know persons from a sample of occupations,
such as: ‘Do you know a primary school teacher’ or ‘Do you know
a judge’? However, a simple yes/no answer does not suffice for our
social capital measures. The name of the ‘teacher’ and so on was
therefore also recorded and a name interpreter was applied to each
of the names.
3.2.1. Name generator
A single name generator question may generate results biased
towards a single form of social capital. For instance the question
‘whom would you ask to borrow a large amount of money?’ will
probably reveal close relationships such as core family members
and ultimately result in a very large amount of bonding social capital
(Marin and Hampton, 2007). In the light of this, we applied
ten different name generators (see Box 1 in Appendix A for exact
wording). The name generator questions are all based on specific
resources, skills or knowledge that can potentially be exchanged
among rural people, such as borrowing of money or obtaining
information about formal credits. This leaves little room for the
respondents to interpret the questions differently. The specific
items, skills, or knowledge were determined during several group
discussions with farmers in northern Thailand/Vietnam. The name
generator questions ask only about areas important to rural inhabitants
and in which a more or less regular exchange is taking place.
We restricted the number of persons named per question to a maximum
of three to limit the interview burden on the respondent.
3.2.2. Position generator
The position generator was primarily applied to measure weak
ties. This data collection tool builds on a sample of occupations and
asks respondents to indicate contacts in each of the occupations.
The position generator utilises a person’s occupation as an indicator
of the resources available to that person. A person’s occupation is a
good indicator of his/her social roles and resources, and hence the
forms of assistance that s/he might be able to provide. The sample
of occupations should range widely in prestige and represent different
sectors of the economy in order to meet the theoretical goal
of measuring access to different parts of the social structure and
their differing resources. The occupations should have fairly large
populations since few people, if any, will know anyone in a very
rare occupation. The occupations should have clear titles that all
respondents will understand. If good census information is available,
one should always use occupational titles from the census
(Erickson, 2004). Erickson (2004) further points out that 15–30 different
occupations is a good number to obtain meaningful results.
For our sample we used the official translation of the National Statistical
Office in Bangkok. Occupational groups are classified in the
survey by the four-digit code ISCO-88 from ILO (NSO, 2007).10 In
Vietnamnodetailedlabour statistics were available. For this reason,
and to ensure comparability, we used the same list of occupations
in Vietnam as in Thailand. We asked a representative sample of
26 different occupations, selected from the national ‘labour force
survey’ in Thailand. The complete list of chosen occupations can be
seen in Box 2 in Appendix A