In yesterday’s post on the latest Nielsen poll, we noted that Greens preferences seem to have been shifting considerably over the last six months. Today we’ll take a closer look at what has been going on.

Nielsen do two measures of the two party preferred vote – their headline result (the one reported in the Fairfax press) where poll respondents allocate their own preferences to get a two party preferred number, and a second two party preferred measure where Nielsen allocates preferences for respondents on the basis of the 2007 election preference flows. This latter measure is the way other pollsters like Newspoll arrive at a two party preferred estimate and you can find those particular Nielsen TPP estimates (both pollster allocated and respondent allocated) rounded to the nearest whole number in their demographic table release here.

What we’re interested in though is their respondent allocated preference results – as this is where minor party voters tell the pollster where they would give their preferences were an election held at the time the poll was taken.

After digging through the last six Nielsen polls at the weighted respondent level of data (unpublishable – sorry), here are the relevant numbers we’ll need today down to the decimal place for the voting intention numbers, and whole percentage points for preference flows to the ALP.


For those wondering what all this preference flow business is about – in November, for example, 85% of people that voted for the Greens went on to preference the ALP. So if an election were held and the polls were correct, 85% of Greens votes would have flowed into the ALP two party preferred measure. By this month, that preference flow rate toward the ALP had shrunk to 68%.

The first thing worth doing is tracking how the size of the Greens preference flow to the ALP has behaved as a function of the actual size of the Greens vote. We can do this with a scatter and join the dots with a line to trace the timeline.


As we trace the timeline starting in November 2009, we see that the size of the Greens preference flow to the ALP has reduced as the months have rolled on. It is also tempting to think that there is a positive trend happening with the size of the Green vote – giving us an inverse relationship with preference flows that suggests that as the Greens vote get’s larger, the size of the Greens preference flow to the ALP gets smaller. However, as tempting as that is, it is also somewhat deceptive.

To highlight the deceptiveness, if we run a simple linear trend line through this result and then duplicate the chart but remove the June observation and run another simple linear trend through that modified series – we see how the June observation has created a trend that didn’t exist in the 5 months previous:


The only real trend here is the steady contraction of ALP preference flows received from Greens voters:


A contraction that appears to be independent to the size of the actual Greens vote – at least since November. At this stage, it’s worth comparing the Greens preference flows to the ALP with that of the broad “Others” vote – the vote that isn’t either the ALP, Coalition or Greens. Running the same scatter we get:


The preference flows from the Others vote to the ALP has pretty much been going nowhere after a jump between November and February. Similarly, the size of the Others vote is moving between 6 and 9 percent. Shorter version – nothing to see here folks.

So the changing preference behaviour we’re seeing with the Greens is not a broader minor party issue, but just a specific Greens issue.

If we now add another two scatter plots – Greens preference flows to a major party as a function of the size of that major party’s primary vote, something interesting pops up (click to expand)

greensalp greenslnp

As the ALP primary has contracted, so too has the Greens preference flow to the ALP, while as the Greens preference flow to the Coalition has increased, so too has the Coalition primary vote.

The results of these three scatters are important for what they tell us isn’t happening.

First off, the ALP Protected Left Flank Hypothesis.

This is the theory that the ALP can lose primary votes to their left (the Greens) because they ultimately get them back via preferences. This theory relies on the assumption that there are a fixed number of left leaning voters that nearly all give the ALP their two party preferred vote. According to the theory, as left voters move from the ALP to the Greens, the ALP primary vote goes down, the Green primary vote goes up and the rate of Greens preference flows to the ALP increases as a result of these ex-ALP-come-Greens voters sending their two party preferred preferences back to Labor in very substantial numbers.

If the ALP Protected Left Flank Hypothesis was happening in any meaningful manner, we would expect to see the Greens pref flows to ALP vs. Labor Primary Vote chart to look something like this:


As the Labor vote goes down, the Greens preference flow to the ALP goes up and vice versa.

However, this is the complete opposite of the observable polling reality that has occurred over the last 6 months or more.

According to Nielsen data at least, what we’ve seen is that as the ALP vote has decreased since November, the Greens preference flows to Labor have also decreased – which makes the ALP Protected Left Flank Hypothesis mostly piffle at the moment. If it is happening, if it does exist among some voters, it’s a dynamic so small that it is getting completely washed out by “Other Big Things”.

Secondly, as the Coalition primary vote has increased, so too has the size of the Greens preference flows to the Coalition – from a low of 15% in November through to a high of 32% this month. In functional terms, the observable relationship between the Coalition primary vote and Greens preference flow is the same as the relationship between the ALP primary vote and Greens preference flows – as the generic popularity of a major party increases (measured by the size of their own primary vote), the size of the preference flow they receive from the Greens increases as well.

The only real difference between the ALP and the Coalition in this regard is the actual base level of Greens preference flows each party starts off with as a minimum level – with Labor probably starting from a minimum of around 60% and the Coalition starting from a minimum of around 15%.

A substantial number of Greens voters  – at least around 20% – appear to be swinging voters in terms of their preference allocations, behaving just like voters that swing between the major parties, except they do so with their preferences.

This makes things slightly tricky for Labor at the moment, as on the one hand any seat held by around a 2% margin or less can be lost if Greens preference flows are delivered in the 60 percents.

On the other hand, deliberately pandering to Greens party issues (as opposed to Greens voters) to attempt to get those preference flows coming in at the 70 or 80 percents endangers swinging voters to the right of the Greens – particularly on social issues –  as well as those swinging voters that aren’t amenable to Greens issue positions (think border protection). Particularly if the Greens voters Labor is trying to attract ultimately behave just like major party swinging voters anyway and are currently voting for the Greens more because of a perceived “powerful lack of alternatives” than for any particular issue the Greens necessarily stand for in policy terms.

We saw some of this at play with yesterday’s Nielsen poll results on reintroducing the Pacific Solution – over a third of Greens voters support offshore processing and  don’t agree with the position of the party they say they are voting for. Adding to this, the cross-tabs tell us that only 49% of Greens voters believe that the Greens have the best asylum seeker policy, 18% of Greens voters believe the Coalition has the best policy and 10% believe the ALP has the best policy.

With those sorts of confusing Green vote/Greens voter beliefs/Greens voter preference flow dynamics going on at the policy level, the ultimate solution to getting good preference flows for each major party is to simply ignore them and concentrate on the wider electorate as far as policy positioning is concerned.

As we see from the Nielsen data, increased generic popularity delivers primary votes and preference flows. If a party is popular generally, that popularity also becomes reflected via increased Greens preferences.