08 May 2019

Election 2019 - First Poll #nlpoli

While the MQO poll shows the Liberals in the lead, a new poll from Abacus  - released with an hour of this appearing - will likely show that since the debate, the political landscape of the province has changed dramatically.
The latest poll from MQO showed that after a week and bit of campaigning, voters were almost precisely where they were at the end of March when MQO conducted a poll for NTV.

When asked which party they would vote for in an election, 47% made no choice compared to 56% in March.  More chose Liberals (25/21) than Conservatives (20/18) or New Democrats (6/4). One percent chose the NL Alliance and one percent chose the Green Party.  Neither registered support in March.

The gap between Liberals and Conservatives is within the margin of error.

Detailed poll data courtesy MQO Research
Click image to enlarge
The variation between March and April results is also within the margin of error for the poll (plus or minus four percentage points 19 times out of 20). The margin of error for the “decideds”-only sub-sample, widely reported by news media, would be higher since it based on only half the total sample.

NTV included a question in the MQO quarterly omnibus.  They asked which party people thought would win the election.  46% of respondents chose the Liberals, 21% the Conservatives and 30% made no choice. This is a useful question since research suggests the response to this question frequently closely matches the popular vote result in an election.

MQO completed two thirds of the survey before the debate, according to NTV’s Mike Connors.
Turn-out/No choice:  The very large number of people who made no choice can be regarded as “none of the above”.  Since the number remains largely unchanged after two to three weeks of campaigning, it is likely that these people represent individuals who will not vote. 

MQO’s last two polls would suggest a turnout between 44% based on March and 53% based on this survey.  That would be consistent with the downward trend noted in elections since 1993.  

Sex:  Support for the two major parties was as expected, with Liberals having stronger support among women (L27/C14) with Conservatives having stronger support among men (C26/L23).  Most respondents made no choice (M46/F49).

Age:  Liberals were favoured by those under 34 three to one (L30/C10).  Conservatives had the support of 24% of respondents aged 35 to 54. Liberals had 22%.  Over 55 years of age, Liberals had slightly more support (25%) than Conservatives (21%) Across all three age cohorts, most people made no choice (45/46/50).

This may be significant since older people tend to vote while younger people do not. While it is a crude calculation, if we exclude those under 34 years of age, Liberals have 24% of the older age cohorts, compared to 22% for the Conservatives.

Geographical Distribution:  The 2015 seat apportionment distributed half the seats in the legislature to the Avalon peninsula with the other half west of Goobies.  This is consistent with the population distribution.

As in 2015, support for the Liberals is stronger west of Goobies while they are tied, according to MQO, on the Avalon.  In metro St John’s, the Liberals enjoy a slight lead (L22/C19/U46).  As MQO’s Stephen Moore noted, the strength of Liberal support in the West/Labrador (L30/C18/U47) and in Central (L29/C17/U50) may produce Liberal inefficiency in its vote.  That is, the party may rack up sizeable majorities in these areas without winning more seats while east of Goobies, Liberals will have to ensure every vote gets to the polls.

Ground-Truth:  Observations or conclusions made by analysis of data should be tested against direct observation.  With political polling,  any attempt to translate the results to seat counts, for example, should be adjusted using knowledge of the local conditions.  

Projections of seat results in this election should be adjusted using knowledge of specific districts or of conditions not captured by polling that would affect outcomes.  This would include, for example, the difficulty all parties are apparently having in finding volunteers needed to conduct voter identification and get-out-the-vote operations.

Such difficulties would affect the ability of all parties to get their vote to the polls.  This is less of an issue for Liberals west of Goobies, for example, but would pose a greater problem on the Avalon where it is crucial for every vote to get to the polls.  

This imbalance between resources and need can be addressed by an assessment of campaign logistics, or how campaigns deploy resources.  In this case, it would be the the ability of campaigns to redeploy resources away from districts where opposition is weak to areas that need reinforcement. 

If we assumed that the Liberals and Conservatives have identical resources (even though they are not), their ability to redeploy is not symmetrical given their seat counts in the House and relative strengths.  This is similar to the concept of Interior Lines/Exterior Lines.  

The scope of their respective campaign effort is the difference between the number of safe seats they hold versus the number required for a majority government (21).  If the parties have equal resources,  the one having to go further - to fight more battles simultaneously - is at a disadvantage compared to the other.  

In this case, the polls suggest  the Liberals have an inherent advantage since they are politically strong in an area where they hold most of the seats.  If they held 15 seats and were strong there,  the Liberals could deploy some resources from their strong seats and redeploy them to the seats they need to win (minimum six).  The Conservatives on the other hand must have resources to fight to win in a minimum of 13 seats, assuming they held eight already.