Join the panel!


A week is a long time in politics...

6 Aug 20

Credit: The Spinoff
A week is a long time in politics...
Judith Collins regained some ground, but National still faces the task of winning back those of its 2017 voters who intend to vote for Labour, ACT

By Grant McInman, Manager, Horizon Research

Horizon Research includes questions on voting from time to time in its surveys – for both forthcoming referenda and general elections. 

For general elections, we have a high interest in switching patterns (party vote in the previous election versus current intentions) rather than the overall result, and the results from the June 2020 and July 2020 survey were ones we thought we should comment on.

We had noticed in a survey taken in mid-June 2020, three weeks after Todd Muller had been selected as National Party leader, that around 11% of National voters in the 2017 general election were now expressing an intention to vote for ACT.   While not all of them were 100% likely to vote, the implication was that a significant number of the party’s 2017 support base were considering switching their allegiance.

Our July 2020 online survey was taken from 14–19 July 2020.  Like other political polls, the survey data was weighted to align with Stats NZ population counts.  It was also weighted to align with Elections NZ’s party vote statistics for the 2017 General Election.

Our July 2020 survey started on the same day that Todd Muller resigned as Leader of the National Party.  Judith Collins was selected as Leader that night.  We were able to compare the results from 14 July with those from the remaining days.

The comparison indicated that the potential loss to ACT had reached 14% by the time that Mr Muller resigned as leader.  The selection of Ms Collins as leader changed the apparent decline in the National Party’s potential vote, lifting it by nearly 5% overall and retaining 7% more of National’s 2017 voters.  This was primarily by stemming the potential loss to ACT and some other minor parties and reducing the uncertainty among those who had voted for National in 2017 about which party they would vote for in 2020.

As this chart shows, between July 14-19 National had potentially lost 11% of its 2017 party voters after the election of Judith Collins (down from 14% under Todd Muller). It had lost 19% directly to the Labour Party (the same under Muller). Overall, 56% of National’s 2017 voters were loyal (49% under Muller).

Our overall party vote for National was 25%, the same as Reid Research’s for 3 News/Newshub.  The Reid poll started on 16 July, the day after Ms Collins selection as National’s leader (and the third day of the Horizon survey), and ran until 24 July.  As noted above, the Horizon Research survey covered the last day of Todd Muller’s leadership and the period when National Party voters from 2017 were adjusting to the change of leadership.

“A week is a long time in politics”

Attributed to 1960s British Prime Minister Harold Wilson, this comment could definitely apply to the 2020 New Zealand election.  Six days after our July survey closed, Colmar Brunton started its latest poll for 1 News.  Results were released on 30 July.

Their results show that National’s party vote was now 32%, 7% more than Reid Research’s and our own results.  Half of that growth appears to have come from those who had inteded to cast their party vote for ACT.  Taken together, ACT and National have 37% of the vote.  The parties that form the current Government have 60%.

On Horizon’s results, Labour could possibly govern alone, but not if there is a sizeable “overhang” (an “overhang” occurs when a party wins more electorate seats than the total share of seats it would otherwise be allocated based on its share of party votes. The overhang seats (the number above the party vote entitlement) are added to the usual 120 seats until the following general election).  If a sizeable overhang does occur, then Labour would need to either form a coalition or a minority government with support for confidence and supply.

But on Colmar Brunton’s results, taken a week later, Labour would have 67 seats in a 120-seat parliament (i.e. no “overhang”), and could govern alone.

Will there be an “overhang”?

That depends on the number of electorate seats won by National in comparison with its entitlement when all party votes are counted – and the party votes of ACT, the Green Party and potentially New Zealand FIrst. 

In 2017, National won 41 electorate seats out of 71, but was entitled to 56 seats.  Labour won 29 electorates but was entitled to 46.  ACT won one electorate seat, and its party vote level restricted it to the one seat in parliament.  New Zealand First and the Green Party did not win any electorate seats but were entitled to 9 and 8 seats respectively.

If National's party vote tally in the 2020 election is at 32% or higher, there is unlikely to be an “overhang”.  If National wins all of the electorate seats it currently holds and wins less than 32% of the party vote, there is likely to be an overhang.  The amount of overhang will depend on how much less than 32% of the party vote National wins.  

Election outcome?


The selection of Ms Collins has seemingly staunched some of the bleeding from the 2017 National party vote.  On the day that Mr Muller resigned, 76% of the potential 2020 National Party voters had voted for National in 2017.  Following Ms Collins selection as leader, that figure grew to 84%.   But will it be enough?

Our data shows a greater loss for National of its 2017 voters than a gain of 2017 voters from other parties or from new voters. Strategically, National may not be too worried if it shares its 2017 voters with ACT.  But it should be concerned about the potential loss of 2017 votes to Labour.   If it is to form the Government after the election, it needs to retain its 2017 voters and attract others from those who voted for the Labour, Green and New Zealand First parties in 2017.  Our data indicates that, currently, just 4.7% of its intending voters come from those who voted for those three parties.

Will that be enough?