Going by the polls, Brad Wall of the Saskatchewan Party and Brian Pallister of the Manitoba Progressive Conservatives are on track to win the April elections in their respective provinces. But just how predictive are polls three months out?

In 32 provincial elections held since 2001 in which relevant polling data is available, the party leading in the polls with three months to go before the vote has won in 24 cases, or 75 per cent of the time.

The latest polls out of both provinces were conducted by Mainstreet Research for Postmedia. The poll in Saskatchewan, conducted on Jan. 4, put Wall's governing Saskatchewan Party ahead with 59 per cent to 28 per cent for the opposition New Democrats. The Liberals followed with seven per cent.

In Manitoba, Mainstreet's Jan. 7 poll pegged the PCs to have 44 per cent support, with the Liberals at 27 per cent and the governing NDP under Greg Selinger at 23 per cent.

Recent history suggests that Wall and Pallister are in a good position, but their margins are likely to narrow. Parties leading in the polls this far out from the next election have, on average, seen their lead decrease by five points.

The news is better for Pallister, who holds the narrower lead of the two, as in the past opposition parties have seen their edge in the polls narrow by an average of just two points, compared with an average drop of seven points for incumbent parties.

The size of a lead increases the odds of a party winning three months later. Parties holding a lead of five or more points have won more than four-fifths of elections, compared with a 50/50 record for parties with a lead of less than five points. Both Pallister and Wall are well above that mark.

The full list of cases can be found in the table following this article. Note that in 2006, the PCs in New Brunswick won the popular vote but lost the election.

Cautionary tales in B.C., Ontario, Alberta

Neither party leader should get complacent, however. One party that recently was in the same position as the Manitoba Tories was the B.C. New Democrats in 2013. Leading the Liberals by 17 points with three months to go, the NDP ended up losing the election by more than four points. The parallels do not end there as, just like the Manitoba PCs, the B.C. New Democrats had held a wide lead over the governing party for several years before coming up short.

The example of the 2011 Ontario Tories might also serve as a warning. The party was ahead by 13 points in the run-up to the election that year, before losing by just over two points.

In that case, the Tories were banking on the unpopularity of Dalton McGuinty's Liberal government to carry them over the top and campaigned primarily on change for change's sake. To avoid the same mistake, Pallister's Tories may need to promise more than just a change in the colour of government.

There are fewer examples for Wall to worry about. The closest would be the case of the Alberta Progressive Conservatives, who squandered a lead of 28 points last year. But Jim Prentice, leader of the PCs at the time, did not have the same personal approval ratings that Wall currently has. And Wall has already won two elections, whereas Prentice was taking his first kick at the can.

The 3-month forecast

Can we use the polls today to make a prediction of where Wall and Pallister will stand on election night?

While there is a correlation between election results and a party's standing in the polls three months earlier, it is nowhere near a perfect one. But a regression analysis of the data suggests that for every point that a party leads with in the polls, that party can expect to win by 0.7 points on election night.

That reduces the Saskatchewan Party's 31-point lead to a win by 21 or 22 points, more than enough to secure another majority government. The Manitoba Tories would see their lead drop to about 12 points, but that lead is a lot more precarious than it looks.

Because of the Manitoba PCs' strong levels of support outside Winnipeg and more moderate levels of support within that city, the party needs to build a sizable lead over the New Democrats in order to win. A lead of less than 10 points puts a PC majority in doubt. A lead of seven points or less puts victory in doubt. 

How the math works if the Liberals are in second place is harder to determine, as the party has not taken as much of the vote as the polls are currently awarding it in over 25 years.

But the sample size of cases is still quite small, and there is plenty of scope for the electorates in both Saskatchewan and Manitoba to buck the trends — in ways that may be positive or negative for either Wall or Pallister. Nevertheless, history is on their side.

The polls by Mainstreet Research were conducted for Postmedia, interviewing 1,508 Saskatchewanians on Jan. 4, 2016, and 1,552 Manitobans on Jan. 7, 2016, via interactive voice response. Both polls carry a margin of error of plus or minus 2.5 per cent, 19 times out of 20.

Poll leads three months out vs. election results

The table shows comparisons of the lead held by provincial parties in polls three months prior to an election to the vote share margin of victory or defeat for that party when the election was held (including all provincial elections with polling data available since 2000). (Éric Grenier)