Nice. Completely agree re: interpretation of the South Auckland cluster too.
It seems simpler to change the voting age from 18 to 15 to accomplish exactly the same result; but what I like about your tweak is that the principle also could be (re)applied to the group of prisoners whose right to vote was removed in 2010 (i.e. those serving a prison sentence with less than three years remaining at the time of the election).
The same underlying principles apply in both cases: these are people we want to get involved in the process, and who will be directly affected by the actions of the incoming government, at a time when they will be eligible to vote.
If you look directly at flag referendum turnout it is the Māori electorates & a few others in one cluster
as noted above (with scatter plot attached)
Certainly National party vote is the single strongest direct predictor out of the data available (it gives the single highest correlation coefficient, with a nicely linear graph, and as you show, other party votes don’t account for the residual variation), and there may even be a direct causal connection; but there is probably also an indirect (causal) connection buried within that, in that
(i) National party vote is correlated with 2014 voter turnout (on the grounds that many people didn’t support National but couldn’t see anything positive to vote for among the other parties);
(ii) 2014 voter turnout is correlated with referendum voter turnout (on the grounds that people who are more likely to vote are more likely to vote); and
(iii) flag referendum voter turnout is correlated with desire to change the flag.
Teasing that out of the data would be challenging, though!
There’s also a question of whether to include the Māori electorates within this analysis: they seem to constitute an internally-consistent set of outliers on all of these scales, with possibly different causal motivations (e.g. more systematic opposition to the Lockwood design in particular because it fails to reflect any element of Māori culture or design).
You might suspect that voter turnout was correlated with support for Lockwood. This is broadly true across all electorates; but the graph suggests a major divide between two clusters of electorates: a smaller cluster with turnout below 52% and support below 35% (which includes all of the Māori electorates); and a larger cluster with turnout above 55% and support in the 36-52% range. Within each cluster, the trend is less clear.
Not too surprisingly, the Lockwood flag was strikingly unpopular in the Māori electorates – accounting for 7 of the 8 electorates where it had lowest support among those who voted. The lowest support rates for the Lockwood by electorate were: 21.1% in Te Tai Tokerau; 22.1% in Tāmaki Makaurau; 22.6% in Ikaroa-Rāwhiti; 23.8% in Waiariki; 25.4% in Hauraki-Waikato; 25.9% in Te Tai Hauāuru; 28.9% in Māngere; 31.9% in Te Tai Tonga.
Considering the implications of the partnership between Crown and Māori, I suggest that any future attempt to change the flag should require the replacement to receive a majority of votes both in general and (also, specifically) in Māori electorates. A new flag has to be something that Māori can accept as representing and including themselves; otherwise, how can it possibly function as a symbol of national unity?
Conversely, the Lockwood flag actually achieved a majority of votes cast only in the following six electorates:
Clutha-Southland (50.4%); Ilam (50.8%); East Coast Bays (51.1%); Bay of Plenty (51.4%); Selwyn (51.7%); Tāmaki (51.9%).
Regardless of the high voter turnout in solidly National electorates, I would suggest that the overwhelmingly consistent direction of results should not be that encouraging for the National party.
his expression makes reality, oh yeah:
like a larva spinning yarn out of its rear.
Now I’m confused.
In the teapot case, was Key sued for defamation as a Member of Parliament?
As long as the Left remembers to roll out policy plans in an order
sorted for ease and whizz.