What happened in Wentworth?

The 20 October 2018 by-election in the NSW division of Wentworth, following the resignation of former Prime Minister Malcolm Turnbull, was notable for a number of reasons. These included the large number of candidates, the high personal vote of former member Malcolm Turnbull, and the victory of an independent candidate in the Liberals’ fifth safest seat. It was also notable for how the seemingly convincing margin of victory early on election night narrowed markedly over the course of the count as pre-poll and postal votes were included.

‘Calling’ an election on the basis of partial results is essentially a matter of maths. As counting is completed in each polling place and the vote tally added to the Australian Electoral Commission’s (AEC) website, the difference between the results at the election and previous elections (the ‘swing’), and the share of total votes that the returned polling places account for, is used to project the election result.

The chart below has been calculated by the Library from the AEC’s progressive results. The share of first preference and two candidate preferred (TCP) votes of independent candidate Kerryn Phelps and Liberal Party candidate Dave Sharma had stabilised by around 8 pm on election night. At this point many media commentators called the election for Phelps, given that her TCP vote was around 55 per cent and Sharma’s primary vote was well under 40 per cent. The vote stayed at this level fairly constantly until 10 pm, with 37 of the 43 polling places returned.

At around 11.30 pm on election night, the margins rapidly narrowed. Approximately 6,500 votes were added to the count from the new Rose Bay pre-poll voting centre (PPVC). These votes heavily favoured Sharma in TCP terms, making the race much closer and dropping Phelps’ TCP vote to 52 per cent. The gap was further narrowed on Sunday morning when the AEC started adding postal votes, which also favoured Sharma, to the count. As at publication the final margin for the by-election was 2.42 per cent, or 1,849 votes.

Open graph in new window.

It is not unusual for support for candidates to be highly localised within an electoral division, and Wentworth is particularly striking in respect of the TCP vote across what is a geographically small electorate. Sharma’s vote was heavily concentrated in Rose Bay, Watson’s Bay and Vaucluse, whereas Phelps’ vote was stronger in Bondi, Bronte and Paddington.

Open map in new window.

Compounding the variation in results between 20 October and 21 October was the difference in vote distributions between different types of votes. Votes cast on election day (election day ordinary votes) convincingly favoured Phelps, which was reflected in the results forecasts on election night.

Open graph in new window.

Pre-poll votes cast at one of the PPVCs—which, due to the size of the voting centres, were not reported until after 11 pm—favoured Sharma. This led to the narrowing of the margin once they were entered into the count.

Even among the pre-poll ordinary votes, however, there was considerable variation in candidate support. The Haymarket and Paddington PPVCs favoured Phelps, Waverley was almost tied, and Rose Bay heavily favoured Sharma.

Open graph in new window.

Postal votes, some of which were entered into the count on Sunday 21 October and others in the days following the by-election, also heavily favoured Sharma with 60 per cent to Phelps’ 40 per cent. There was even speculation at one point that Sharma might take the lead on postal vote preferences.

However, while early postal votes gave him a 64 to 36 per cent TCP lead, the postal votes counted on Monday 29 October (which may have been posted later in the campaign or spent longer in transit) had Phelps with a TCP lead of 56 to 44, according to Library analysis. 

Open graph in new window.

Although Sharma did well in terms of first preferences and TCP votes from postal votes, only around 10 per cent of the votes cast at the by-election were postal votes. 

Open graph in new window.

Of the 16 candidates who contested the by-election, only four (Phelps, Sharma, Murray from the ALP and Kanak from the Greens), received at least four per cent of the vote, which is the threshold required to claim back the $1,000 nomination deposit. Three of those four candidates suffered a swing in primary vote terms. As Phelps has not contested Wentworth before, it is not possible to calculate a TCP swing—there is no previous result to compare.

At a little over 19 per cent, the first preference swing against the Liberal Party at the Wentworth by-election is well above the average 6.3 per cent swing against governments in government-held seats at by-elections since 1949. It is also well above the average 7.1 per cent swing against governments in government-held seats at by-elections caused by resignations, according to the Parliamentary Library’s analysis.


Flagpost is a blog on current issues of interest to members of the Australian Parliament

Parliamentary Library Logo showing Information Analysis & Advice

Filter by



Tag cloud