Thursday, 30 July 2015

Party Vote Demographics: Part II

In the last post, I explained how we’re using electorate census data and electorate voting data to find statistical relationships between electorate demographics and party vote. There are plenty of limitations associated with using this data in this way, so all statistics should be interpreted with caution. Just in case you’d forgotten, I’ll say it again: correlation does not imply causation. This post will look at ethnicity, families, and immigration.

“Māori just vote for the Māori Party and Mana”
In electorates where there are more Māori…
… of any age, a lot more people voted for the Māori Party (r≈+0.76) and InternetMANA (r≈+0.68)
… aged 15-44 years old, less people voted for the Conservatives (r≈-0.50)

Discussion: It should be noted that any statistics involving Māori ethnicity are heavily skewed by the inclusion of the Māori electorates in the analysis. The Māori Party received 10-20 times more votes in Māori electorates than General electorates (InternetMANA’s variation was a little less).

“Polynesian families vote Labour and Asians vote right-wing”
In electorates where there are more…
… Pacific Peoples, a lot more people voted for Labour (r=+0.708), and fewer people voted for ALCP (r=-0.251)
… Asians, a lot more people voted for ACT (r=+0.773), slightly fewer people voted for InternetMANA (r=-0.167) and Māori (r=-0.253), and fewer people voted for NZ First (r=-0.546) and ALCP (r=-0.656)

Discussion: Labour attracted much higher party votes in the Pacific Ms – Mana, Mangere, Manukau, Manurewa, and Maungakiekie. ACT attracted more votes in electorates with more Asians - New Lynn, Mt Roskill, Botany, and Pakuranga (and of course Epsom). These areas (which also happen to all be suburban areas in Central and East Auckland) gave NZ First fewer votes, plausibly on the back of the perception within the Asian community that NZ First is a xenophobic party, and also gave ALCP fewer votes, plausibly due to the stronger anti-drug stances held by most south-east Asians. I should note that individuals can choose more than one ethnicity in the census, so they may be counted more than once in the ethnicity statistics but are only responsible for one (or no) vote in the election.

“The bigger the family, the more likely they’ll vote Labour (Working for Families, etc.)”
In electorates where there are more females with…
… no children, more people voted for the Greens (r=+0.539) and the Civilian Party (r=+0.301), slightly more people voted ACT (r=+0.185), and fewer people voted NZ First (r=-0.568)
… one child, slightly more people voted ALCP (r=+0.268), ACT (r=+0.141)
… two children, more people voted for the Conservatives (r=+0.528) and National (r=+0.437), slightly more people voted for United Future (r=+0.153), and fewer people voted Labour (r=-0.538)
… three children, a lot more people voted for NZ First (r=+0.711), and more people voted for the Conservatives (r=+0.386), DSC (r=+0.353), National (r=+0.197), and the Māori Party (r=+0.178)
… four children, a lot more people voted for NZ First (r=+0.616), and fewer people voted National (r=-0.168)
… five children, more people voted for NZ First (r=+0.403), and fewer people voted for the Greens (r=-0.520)
… six or more children, slightly more people voted Labour (r=+0.277)
… an objection to answering about how many children they have, more people voted for NZ First (r=+0.303)

Discussion: It should be noted that these statistics relate to the “number of children born alive” by females, which isn’t necessarily a direct match for family size. For example, “no children” includes single people and young people, who are less likely to have been pregnant. Therefore it seems more reasonable for there to be a correlation between electorates with more people with no children and party vote for the Greens and Civilian, not because the supporters of those parties are opposed to having children, but simply because those parties may be more popular with young people.

I think in general what these statistics show is that the number of children is probably a very poor indicator of party vote. It could possibly be argued that electorates with women with more children vote more left-wing than right-wing, but I think that would be a pretty tenuous argument based on these statistics. I’m not sure why people would object to Statistics NZ knowing anonymously how many children you’ve had (55,199 individuals), but electorates with more of those people also voted more for NZ First, which perhaps suggests that Winston supporters tend not to trust the government with information about them. Maybe.

“Immigrants vote for [insert various statements here]”
In the general electorates where there are more people…
… born in Asia, greatly more people voted for ACT (r=+0.854), and fewer people voted for NZ First (r=-0.620) and ALCP (r=-0.692)
… born in Middle East and Africa, more people voted for ACT (r=+0.679), slightly more people voted for National (r=+0.160), and fewer people voted for ALCP (r=-0.466)
… born in Australia, more people voted for the Greens (r=+0.496) and National (r=+0.443), slightly more people voted for United Future (r=+0.225), the Conservatives (r=+0.142), and the Civilian Party (r=+0.168), and fewer people voted for Labour (r=-0.539)
… born in the Pacific Islands, greatly more people voted for Labour (r=+0.753), and fewer people voted for ALCP (r=-0.436)
… born in North America, greatly more people voted for the Greens (r=+0.840), more people voted for the Civilian Party (r=+0.368) and United Future (r=+0.330), and slightly more people voted for National (r=+0.143)
… arrived to NZ within the last 2 years, more people voted for ACT (r=+0.482), and fewer people voted for ALCP (r=-0.575) and NZ First (r=-0.638)
… arrived to NZ within the last 3-9 years, more people voted for ACT (r≈+0.63), and fewer people voted for the Maori Party (r≈-0.22), NZ First (r≈-0.61), and ALCP (r≈-0.72)
… arrived to NZ within the last 10-19 years, a lot more people voted for ACT (r=+0.834), and fewer people voted for the Maori Party (r=-0.269), NZ First (r=-0.589), and ALCP (r=-0.721)
… born overseas, more people voted for ACT (r≈+0.7), slightly fewer people voted for the Maori Party (r≈-0.2), and a lot fewer people voted for NZ First (r≈-0.6) and ALCP (r≈-0.7)

Discussion: The census reports data for people who were born overseas only for the General Electorates, which we can use as a proxy for immigration. Firstly, electorates with more Asians, Middle East and Africans voted more for ACT, and really didn’t like NZ First and ALCP. (Electorates with more) Australians voted more for the Greens and National and a lot less for Labour (perhaps a contagion effect from the Labor party across the ditch). As covered previously, Pacific Islanders voted more for Labour (and less for ALCP), and (electorates with more) North Americans really liked the Greens! Electorates with recent immigrants (and also less recent immigrants) liked ACT, plausibly due to the more recent influx of Asian, Middle Eastern, and African immigrants, who also heavily disliked NZ First and ALCP. That trend holds for all immigrants in general as well.

Coming up – even MORE demographics (the final dataset I used had over 1,800 variables)!

No comments:

Post a comment