Did the Parliament shooting boost the Conservatives in the polls?

Photo Reuters

Coup d'oeil sur cet article

One of the most important (and tragic) events of 2014 was of course the shooting that occurred at the House of Commons on October 22nd. And with such a big media coverage, one could wonder if it affected the political landscape at the federal level.

A couple of days ago for instance, I caught a Twitter conversation about that between Ekos’ Frank Graves (who thinks it helped the CPC among other things) and Ipsos’ Darrel Bricker (who thought the upward trend for the Tories was there before). So let’s take a look.

First of all, let’s look at the federal polls of 2014. Here is the graph with some added trends. The Conservatives are naturally in blue, the Liberals in red and NDP in orange.

Statistiques - Bryan Breguet

As you can see, the situation has been relatively stable during the last year with no crazy jump or fall (as opposed to 2013 with the arrival of Justin Trudeau for instance). However, if the Liberals had a comfortable lead at the beginning of the year, the race was a lot closer towards the end. But is it really due to the shooting and (potentially) an increase over security concerns (something that should help the Conservatives over the Liberals)?

First of all, let’s notice that the trend did start before. The CPC was higher in June than in January for instance. We of course need to be careful because we only get a limited number of polls every month (around 3-4 in average). So one “bad” or “good” poll can significantly change the average. Also, the clearest trend is actually for the NDP who kept falling during last year.

In June, the Liberal’s lead over the Tories was of 3.5 points. Then it jumped to almost 11 in July before going back to 7-8 in August-September. This same lead decreased to 3.6 and 2.7 in November and December respectively. So the shooting does seem to have had an impact. But a limited one. The Tories had been within 3-4 points of the Grits before (April to June for instance).

If we take an average of the polls 2 months before and 2 months after the shooting, we find the following:






 2 months before




 2 months after





There again, we observe small changes and a lot tighter race since the end of October. Is it significant? Well it depends how we look at it. These two averages are based on 12 polls each, for a total sample size of around 32,000 for the polls done after and 17,000 for the ones before (the difference is mostly due to some giant polls from Angus-Reid in December, including one with over 6000 observations). If we completely ignore the other sources of uncertainty in polls (methodology of each firm, weighting, etc), the difference of 1.4 points for the CPC is indeed statistically significant (it needed to be at least 0.9 point). But again, this is assuming that all the firms are using a similar and comparable methodology. We know it’s not the case. Some firms poll by phone, some use internet. And this is only one of the many differences. If we instead look at the volatility of the polls, we find that the margins of error for the CPC after the shooting is actually around 3.7 points. In other words, there was quite a lot of volatility. Therefore, it is quite hard to conclude that the change was significant. And even if we could, it wouldn’t mean it was due or caused by the shooting. Other things happened during the last few months, including some policy announcements from Stephen Harper.

Of course, if we add the slight increase of the CPC and the drop of the Liberals, the difference becomes a lot more important. But as for the Conservatives, the trend for Justin Trudeau and his party started before the shooting. And there are polls that actually showed the Liberals increasing after the shooting. Also, the sum of the drops for the Liberals and NDP is much higher than the increase of the Conservatives. So at least one other party increased since then (the Green in particular).

Let’s take a final look at this. Before the shooting, the CPC had been ahead in only 3 polls, all from Angus-Reid (and one tie from Ipsos). Since the shooting, the Tories have been first in only one poll (from Abacus) and one draw (Angus-Reid). The race is therefore closer but the situation is still similar. Justin Trudeau was first and favorite during the first half of 2014, he still was during the last half and even the last two months. His chances of winning are naturally less (and the probabilities of a majority are becoming quite slim), but if effect there was from the shooting, it was small.

At the same time, some polling firms did observe a sharp shift. This is the case for Ekos. It’s no wonder Frank Graves is so convinced the shooting had an effect (it’d be weird to see him not trust his own numbers). So here are the averages 2 months before and 2 months after for the CPC and LPC by firm.




























*only one poll since shooting




























As you can see, all the main firms mostly have the same qualitative effects (the Liberal’s lead has decreased). The change is especially important with Ekos and the least visible with Forum. Ekos is also the only one with such a big jump for the Tories.

In conclusion, the polls suggest a tighter race in the 2 months since the incident at the Parliament. Quantitatively, the lead for the Liberals has shrunk from around 7-8 points to 3-4. This is the result of two modest but opposing effects: an increase for the Conservatives and a drop for the Liberals. Individually, these two effects are small (and potentially not significant, especially for the CPC). But put together, I think it is fair to conclude there was indeed a change. Is it a result of the shooting? Or of the policies (and tax cuts) announced by the government? The Conservatives are also running a lot more ads (on the radio for instance). We’d need to look more into the data. Finding causality is never easy. As is usually the case, the answer is probably a mix of the above mentioned factors.

Also, it is important to notice that the Conservatives were already increasing slowly during the last year. And Trudeau started to decrease before the shooting as well. So again, the incident at the Parliament probably influenced voting intentions but not as much as what Ekos observed for instance. If you want an answer to the question I asked in the title, it’d be: yes, a little bit. Maybe.

Finally, voting intentions being a zero-sum game, the constant decrease of the NDP (both before and after the shooting) should probably be studied more. The overall data for the last year show that the Conservatives are the one benefiting (if we look at the net effects) from the decrease of the NDP while the Liberals have been mostly constant (look at the trends). I can’t explain this phenomenon right now but I’ll look into it soon.