135 representativeness of the G1000 was perceived by its participants. Did their perception differ greatly from the representativeness we assessed?
And do they think that the G1000 participants should be a good reflection of their local community? In the final sub-section of this chapter, we will first assess whether non-random sampling have contributed to or undermined the representativeness of the G1000 by comparing the socio-demographic characteristics and general attitudes and behaviour of the randomly selected G1000 participants with the non-randomly selected participants. Finally, we will assess whether self-selection undermined the representativeness of the G1000 initiatives by investigating how non-participation, no-shows and drop-outs influenced the composition of the G1000 initiatives? We also look at the question of who decided to end their participation in the G1000 initiatives and why?
7.1 Were the G1000 initiatives a good reflection
136
often took part in the deliberating event than women (34 %). These initial patterns did not change over time (see Table A6 in Appendix).1 Moreover, the participants in all three G1000 initiatives did not reflect the total population in terms of age.2 G1000 participants between 16 and 34 were under-represented at the first large-scale deliberation event, while participants over 65 were over-represented. In Enschede and Borne the pattern outlined above is maintained over time (Table A6 in Appendix).3 In Steenwijkerland, however, this pattern for the last G1000 event was somewhat different, as participants aged 35 to 54 were over-represented alongside the older generation.
Participants in all three G1000 initiatives also were on average far higher educated than the general population.4 In Steenwijkerland and Borne, the proportions of people with a higher level of education in 2017 were 12 and 16 per cent. By contrast, 72 and 65 per cent of the participants who participated in the deliberation event in Steenwijkerland and Borne had a higher level of education. In Enschede, the figures are less extreme, but the difference is still evident. This pattern did not change over time (Table A6 in Appendix).5
We finally examined whether ethnic minorities or citizens with foreign roots were also sufficiently present at the large-scale deliberation events of the three G1000 initiatives. According to Hoffmeyer-Zlotnik and
1 Chi-square test results: Checked-in: p Borne = .87, p Enschede = .15, p Steenwijkerland < .001; Follow-up: p Borne = .47, p Enschede = .26, p Steenwijkerland < .05; Final event: p Enschede = .41, p Steenwijkerland <
.001.
2 Chi-square tests were used to determine if there was a statistical difference between the G1000 populations and the local populations in terms of age categories. In all three initiatives, the differences were significant (p < .001).
3 In all three initiatives, the results of the Chi-square tests were significant in all phases of the G1000 process (p < .05).
4 Chi-square tests were used to determine if there was a statistical difference between the G1000 populations and the local populations in terms of educational background. In all three initiatives, the differences were significant (p < .001).
5 In all three initiatives, the results of the Chi-square tests were significant in all phases of the G1000 process (p < .05).
137 Warner (2013), the language most spoken at home is a “good indicator of the integration of ethnic minorities” (p. 234). As Table 7 shows, the proportions of participants in the three initiatives with an ethnic or bilingual1 background were comparable to the wider local populations.2 Looking at the proportion of G1000 participants with an ethnic or bilingual background over time, the proportion of Dutch speakers compared to non-Dutch speakers remains a representative sample of the total population (Table A6 in Appendix).3
Thus, the participants in the three G1000 initiatives did not reflect their local communities concerning some of the analysed socio-demographic characteristics (mainly age and education level). Moreover, the figures became more skewed over time. One could argue that these observed socio-demographic differences between G1000 participants and the wider population are not necessarily problematic if the G1000 participants are comparable to the wider population in their diversity of interests and attitudes (Fournier, et al., 2011). However, in terms of general attitudes and behaviour of participants, the G1000 participants did not reflect their local communities much better either.
The G1000 participants at the deliberative events were relatively well engaged in politics: all voted in the Second Chamber (national) elections in 2017. Turnout in these three municipalities differed somewhat but was substantially lower than that (see Table 8). Also, after the
1 Participants who stated that they speak another language besides Dutch at home were considered “bilingual”. Participants who did not state that they speak Dutch at home were conceptualized as “European” or “Non-European” ethnical minorities. For the population data we have used the following two variables from the European Social Survey data Round 8 (2016):
“Most spoken language at home: first mentioned” and “Most spoken language at home: second mentioned”.
2 A chi-square test was used to determine if there was a statistical difference between the G1000 populations and the local populations in terms of immigration background. In all three initiatives, the differences were insignificant (p > .05).
3 In all three initiatives, the results of the Chi-square tests were significant in all phases of the G1000 process (p > .05, α = .05).
138
deliberation event, this pattern remained (see Table A7 in Appendix).1 G1000 participants in all three initiatives were also more left-wing and progressively oriented than the wider population (see Figure A1 in Appendix): in all three G1000 processes voters of GroenLinks and PvdA were strongly overrepresented, while the voters of CDA, PVV and VVD were underrepresented.2
Regarding G1000 participants’ engagement in citizens’ initiatives, differences can be observed between the cases. In Enschede participants engagement in citizens’ initiatives did not differ from the local community, while in Borne and Steenwijkerland, G1000 participants were substantially more active in citizens’ initiatives.3 When looking at the level of engagement in citizens’ initiatives over time, it can be observed that the less active G1000 participants dropped out more, while the more active ones stayed on (see Table A7 in Appendix).
Also regarding G1000 participants’ level of engagement in non-electoral political activities, differences can be observed between the three studied cases. While the level of engagement of participants in non-electoral political activities was comparable to that of the wider populations in Borne and Enschede, G1000 participants in Steenwijkerland were far more active in non-electoral political activities than citizens in their local municipality (see Table 8).4 When looking at
1 A chi-square test was used to determine whether there was a statistical difference between the G1000 population and the wider population in terms of participation in the elections. For all three initiatives, the tests were significant at all stages of the G1000 process (p < .05, α = .05).
2 A chi-square test was used to determine if there was a difference between the G1000 population and the wider population in terms of participants’ voting behaviour. In all three initiatives, the tests were significant at all stages of the G1000 process (p < .05, α = .05).
3 A chi-square test was used to determine if there was a difference between the G1000 population and the wider population in terms of participants’ non-political activities. In Enschede, the test result was insignificant (p > .05, α = .05). In Borne and Steenwijkerland, the test results were significant (p < 0.01, α = .05).
4 A chi-square test was used to determine if there was a difference between the G1000 population and the wider population in terms of participants’ non-political activities. In Borne and Enschede, the test results were insignificant (p > .05, α = .05). In Steenwijkerland, the test result was significant (p < 0.01, α = .05).
139 the level of engagement in non-electoral political activities over time, it can be observed that the less active G1000 participants dropped out more and the more active ones stayed on (see Table A7 in Appendix).
140
1 Composition of the participants is measured at the deliberation events. 2 Difference between the G1000 population and the local population in %.
G1000 pop.1Pop. Diff.2G1000 pop.Pop. Diff.G1000 pop.Pop. Diff. Gender (in %) Male Female52 48+ 2 − 253 47+ 2 − 266 34+ 16 − 16 Age (in %) 16-24 25-34 35-54 55-64 65-84 85 + 5 3 29 21 41 1
− 6 − 10 − 4 + 4 + 18 − 2
6 11 38 19 25 1
− 11 − 6 + 6 + 5 + 7 − 1
5 8 30 24 33 0
− 7 − 4 − 2 + 6 + 10 − 3 Education (in %) Secondary education (havo, mavo, vmbo) Senior secondary vocational education (mbo)18 14 − 30 − 2225 16− 6 − 2115 16 − 32 − 25 Non tertiary education32 − 5240 − 2829 − 59 Higher professional education (hbo) Academic higher education (wo)43 26 + 30 + 2339 20 + 19 + 851 20+ 40 + 19 Tertiary education68 + 5260 + 2871+ 69 Language spoken at home (in %) Dutch Dutch, bilingual European Non-European
89 8 0 3
− 2 + 2 − 1 + 2
92 7 0 0 + 1 + 1 − 1 − 1 96 4 0 0 + 5 − 2 − 1 − 1
Table 7. Differences between the G1000 population and the local population in terms of socio-demographic characteristics of the participants in %.
141 A similar pattern can be found when looking at G1000 participants’
civic engagement level in their local communities. The participants at the deliberation event in Steenwijkerland were much more active in civic activities in their local communities, whereas the participants in Borne and Enschede did in this respect not differ from the wider population (see Table 8)1. When looking at the level of civic activity over time it can be observed that active participants stayed on, others dropped out (see Table A7 in Appendix).
Finally, G1000 participants in G1000 initiatives showed far more trust in the local government than their fellow citizens in the wider population (see Table 8).2 When looking at the level of trust over time it can be observed that the G1000 participants who showed less trust dropped out more, while the participants with more trust stayed on (see Table A7 in Appendix).
Summarizing, we can conclude that participants did not reflect their local communities very well regarding their attitudes and behaviour.
The participants were generally more left-wing and progressive than the general population and had more trust in the local government. They were also more involved in citizens’ initiatives and national elections than the general population, but in some cases, they reflected the general population in terms of their participation in non-electoral political activities and civic activities. However, about participants’
attitudes and behaviour, figures have become more skewed over time, as the less active participants dropped out over time, while the more active ones stayed on.
1 A chi-square test was used to determine if there was a statistical difference between the G1000 population and the wider population in terms of participants’ non-political activities. In Borne and Enschede, the tests were insignificant (p > .05, α = .05). In Steenwijkerland, the test was significant (p < 0.01, α = .05).
2 In all three initiatives, the results of the Chi-square tests were significant (p > .05, α = .05).
142
1Composition of the participants is measured at the deliberation events. 2 Difference between the G1000 population and the local population in %. 3 To get an indication of the engagement level of citizens in citizens initiatives 2017 in Borne, Enschede, and Steenwijkerland, we used population data (measured at local level) provided by Van der Meer and Van der Kolk (2016, p. 32) and Jansen and Denters (2019, p. 37). Based on the data provided for 2016 and 2019, we have calculated the average value for 2017. It should be noted that the average value used is only indicative, as no exact data were available for the three municipalities 2017. 4 To get an indication of the level of political engagement of citizens in 2017 in Borne, Enschede, and Steenwijkerland, we used population data (measured at local level) provided by Van der Meer and Van der Kolk (2016, p. 32) and Jansen and Denters (2019, p. 37). Based on the data provided for 2016 and 2019, we have calculated the average value for 2017. It should be noted that the average value used is only indicative, as no exact data were available for the three municipalities 2017. 5 To get an indication of the level of civic engagement of citizens in 2017 in Borne, Enschede, and Steenwijkerland, we used population data (measured at national level) provided by Van Houwelingen and Dekker (2017, p. 234) for the years 2016 and 2018. Van Houwelingen and Dekker (2017) used data collected by the two Dutch
Table 1. Differences between the G1000 population and the local population in terms of general attitudes and behaviour of the participants. BorneEnschedeSteenwijkerland G1000 pop.1Pop. Diff.2G1000 pop.Pop. Diff. G1000 pop.Pop. Diff. National election participation (in %) Yes No
100 0 + 14 − 1492 8 + 14 − 1497 3 + 11 − 11 Member of a citizen initiative (in %) Yes No32 68+ 263 − 267113 87+ 771 − 77124 76+ 18 71 − 1871 Political activity (in %) Active Non-active 7 93 − 94 + 97012 88− 470 + 47033 67+ 17 70 − 1770 Civic activity (in %) Active Non-active 57 43+ 125 − 127240 60− 572 + 57268 32+ 23 72 − 2372 Trust in local government (in %) Trust No-trust 94 6 − 40 + 40 78 22− 24 + 2487 13− 33 + 33
Table 8. Differences between the G1000 population and the local population in terms of general attitudes and behaviour of the participants.
143