Breaking down the Atlantic

I know I've stated that I would not be doing math for this election, but I've noticed a trend that I thought I could help correct regarding the Atlantic.

Most polling firms will release federal polls divided into 6 areas. Ontario, Quebec, BC, Alberta, the (rest of the) Prairies, and the Atlantic. Some polling firms will split Saskatchewan from Manitoba, but few, if any, split the 4 Atlantic provinces.

This causes some problems when, as we've had during the past decade, certain parties do well or poorly in certain Atlantic provinces. Harper, in particular, was rather unpopular in Newfoundland. As a result, the 2015 election had the CPC on an appalling vote total in the province.

As such I've decided to grab this polling data, The winter, spring, and summer polls from the Atlantic provinces, and average them.

This is a one-firm average, but does average three polls. I've also included the 2015 result, so that I can get a ratio of the result. Take, for example, New Brunswick. The Liberals are, according to these polls, at 40%, whereas, last time, they were at 52%. The ratio is thus 76.9%, that is, I expect that the Liberals are going to take 769 votes for every 1000 votes they took last time in New Brunswick.

I've applied this to every riding, using the 2015 results. I've even looked at multi-election data; in particular for the riding of Avalon, but the 2011 pattern matched the 2015 pattern; IE the Independent candidate didn't have such an overwhelming impact as to make the 2015 numbers useless.

Regardless, when I applied the ratio to each riding, then averaged it out to ensure turnout is matched, I ended up with the following:

This is, of course, the pure math. I've warned about putting Egmont in the CPC column due to the lack of a former star candidate. The math, however, which is based on that star candidate, shows this as a win for team blue.

What you should notice is that there is a CPC win in Newfoundland. In fact, Newfoundland as a whole is quite competitive in places. New Brunswick is less blue than it would be if you'd not re-balanced the provinces, while Nova Scotia is more blue.

I hope that this can explain why my "gutomatic" projections in the Atlantic look so "weird". This is especially true when you consider that I am looking at current, not past candidates. The lack of a CPC star in Egmont reduces their vote there. A Green star candidate in Avalon increases the chance for a split vote. A Liberal star in Cumberland is balanced by the lack of a Liberal star in Kings-Hants.

As such, my current "gutomatic" projection for the Atlantic is as follows: