A weekend press release from Roy Morgan Research breathlessly informs us that:

If a Federal Election were held today the LNP would win the key New South Wales seats of Macarthur & Robertson and the seat of Hasluck in WA could go either way. Combined with the results of Thursday’s special Queensland ALP marginal seat poll which showed the LNP picking up the Queensland seats of Flynn, Longman and Dawson the Opposition is close to winning the nine seats from the ALP it needs to form the next government …

Now let’s think about this. How many seats does the Coalition need to gain to win government?

There are two ways of answering. One is to look at the pendulum – here’s Antony Green’s, and here’s Malcolm Mackerras’s (their figures differ slightly, but not in a way that matters for this topic) – which shows the margins for every seat, adjusted for the recent redistributions. On that score, the opposition (including the three independents) hold a notional 62 seats. A majority is 76, so they need to win another 14.

The other way to look at it is to observe that there are five very marginal notional Labor seats (with margins under 1%) that have Liberal sitting members: redistribution has wiped out their Liberal majority, but only just. Since incumbency conveys an advantage, and adjusting for boundary change is an inexact science in any case, it could be argued that it makes sense to count those five as still being Liberal-held. (A sixth, Greenway, also has a Liberal member, but the redistribution there has given Labor a quite healthy margin of 5.7%, so it would be perverse to count that as anything other than a Labor seat.)

If we do that, the target comes down to a gain of nine seats, which is what Morgan said. If the Coalition picks up six, that would seem to put them well on the way.

Trouble is, one of Morgan’s six – Macarthur, in NSW – is one of the five Liberal-held but notionally Labor seats that we just mentioned. If it’s already counted as a Liberal seat, then it can’t also be a Liberal gain. But if it counts as a Labor seat, then the target is fourteen, not nine.

Either way of calculating the target is quite legitimate (although I prefer the first). What you can’t do is use both at once: use one method to minimise the size of the target, and then in the same paragraph use the contrary method to maximise the likely gain to the opposition.

Morgan want to eat their cake and have it too.