Seven independent variables were assessed:
1) & 2) An independent variable of primary interest was current place of residence. The 10 initial categories (Ha’avakatolo, Sopu, Fo’ui. Nuku’alofa, Ha’apai, Vava’u, Tongatapu, New Zealand, United States and Australia) were reduced to three. These categories are: Nuku’alofa (Sopu, Nuku’alofa), the rest of Tonga (Ha’apai, Vava’u, Ha’avakatolo, Fo’ui) and overseas (New Zealand, US, Australia). Thirty one per cent of respondents (n=214) resided in Nuku’alofa, 26.5 per cent resided overseas (n=187) and 42.5 per cent (n=290) resided in Tonga, but outside Nuku’alofa. In order to assess these data via the technique of a multiple regression equation, two dummy variables were computed.[8] One new variable, labelled Current Residence 1, compares those living in Nuku’alofa to all others (Nuku’alofa=1, others=0). The second new variable, Current Residence 2, compares those living overseas with all others (overseas=1, others=0).
3) Church membership was dichotomised based on whether a denomination was Tonganised or not.[9] The initial variable was comprised of seven categories, specifically, Free Weslyan, Free Church of Tonga, Latter Day Saints, Church of Tonga, Roman Catholic, Tokaikolo and other. The majority (83.7 per cent) of respondents identified themselves as belonging to Tonganised churches (all categories shown above except Latter Day Saints) and 16.3 per cent indicated that they belonged to a non-Tonganised church (Latter Day Saints).
4) Time spent away from Tonga was measured on a four-point ordinal scale. About one quarter of respondents (26.5 per cent) had not left Tonga, slightly more had been overseas for less than 12 months, about one quarter had been overseas for between one and 10 years, and one fifth (19.8 per cent) had spent at least 10 years away from Tonga.
5) Gender split was almost 50–50, with 50.8 per cent being males.
6) Mean age was 34.4 years (sd=14.3 years).
7) Education was measured on a four-point ordinal scale, asking respondents for the highest level of schooling obtained. Slightly less than half of respondents had not completed high school (47.5 per cent), one-in-four had finished high school (26.3 per cent), 17.8 per cent finished a tertiary degree or diploma, and 6.5 per cent had finished a university degree
[8] The new variables, technically referred to as ‘dummy’ variables, were computed to satisfy the statistical requirements of multiple regressions. Specifically, all the variables in a multiple regression are assumed to be measured at the interval level. The technique is robust enough to handle this conversion.
[9] This division of churches is somewhat problematic, as it effectively analyses the LDS as distinct from all others in the sample. Further, it could be argued that significant elements of LDS practice are in fact Tonganised. We recognise these issues and make only a weak claim here that the Tonganised/ non-Tonganised distinction captures a meaningful axis of variation.