The latest ReachTEL poll has Labor in a winning position with 52 per cent of the national two-party preferred vote, compared with the Coalition on 48 per cent: a four point lead. This is the third poll in a row where ReachTEL has allocated preferences using respondent preferences, rather than preference flows from the previous election. The respondent preferences appear to favour Labor by around a percentage point on average (but contained largely to the first and third polls). In past elections, respondent preferences have proved unreliable.
Because of this methodological difference, I have decided to treat these latest ReachTEL polls as a new series (labeled RT2 on the charts). This raises continuity issues: as it changes the property of the sum to zero assumption, moving it slightly to Labor. As a consequence, I will use the TPP model that is anchored to the result at the last election from here on to the election.
The sum-to-zero TPP model has the Coalition on 49.3 per cent.
The TPP model anchored to the outcome of the previous election has the Coalition on 50.2 per cent.
Of note: both charts suggest that the Coalition held steady early in the campaign, bur has declined in the last week or so.