Estimated ageing effect on the LNP 2PP% vote: 2026-2028

By Mark McGrath, 25 May 2026.

All other factors being neutral, this study shows that the LNP 2PP% vote will degrade by an average of 0.26% per annum over the next 3 years.

Approach

I could have based the calculations on Essential's age-based polling, focusing on their latest poll conducted on 24-27 Apr 2025. They polled respondents in three age groups (18-34, 35-55 and 55+). But being only 3 age groups covering 81+ years of voter age, I considered that to be too much of a blunt instrument to do any meaningful estimations on how 2PP voting may change over the next 3 years due to ageing effects (with all other factors being neutral). Extracting a linear (straight line) regression from 3 data points wasn’t very sophisticated (or exciting) and unlikely to produce reliable estimates. So I decided to look further afield.

In my research looking for data on how voting behaviour changes with age, I came across ANU’s THE 2022 AUSTRALIAN FEDERAL ELECTION: Results from the Australian Election Study. On page 22 of this report, there is a graph (Figure 4.4) showing the relationship between birth year and (primary) vote choice in elections from 1987 to 2022. Here it is below.

In this graph, you can clearly see the relationships are not linear; they are curved, where as age increases, the rate of voting change increases. 

I also discovered this very good resource, the ABS’s Population Clock & Pyramid, that allows you to extract estimates of the Australian population for each year, broken down by age.

Fitting an age-based voting curve to the national 2PP result

This led me to the idea of fitting a polynomial curve to the 2025 voting population by age that produces the current national ALP 2PP% result. I decided on using a 2nd order polynomial for two reasons:

1. The curve of a 2nd order polynomial could closely match what was observed in the ANU study (no inflection points, but a fairly gentle curve where the drop-off of the ALP 2PP% increases with older age).
    
2. Occam’s razor: when faced with multiple options to solve a problem, the simplest option is most likely to be correct.

Power and logarithmic functions were tested, but the curves these functions produced did not align with the Essential age polling. In fact, they produced opposite trends; a rapid decrease in the ALP 2PP vote in young voters, tapering off to a slow decrease in the ALP 2PP vote in older voters.

Here are the results of the regression:

How accurate is this forecast? Without any definitive data to test this against it's impossible to know (unless someone can point me to some reliable age-based exit polling).

But we do have the Essential age-based polling mentioned above. So I’ve also included adjusted 2PP results (adding the 2.72% they under-estimated the actual 2PP result) from the Essential poll, with error bars calculated from their 1,500 sample. As you can see, the adjusted Essential poll results line up quite well with my fitted curve.

Now, with estimated ALP 2PP% by each age (18-99+) produced by this regression, we can then apply these to future population estimates for each age for the next 3 years (sourced from the ABS Population Clock & Pyramid) to forecast how the 2PP% values would change over the next 3 years.

Assumptions on future voting intentions

I assumed that voters would retain their original voting intention in 2025 when migrating to older age groups over the next 3 years. But before we get to the results, a bit of discussion on an alternative assumption

An alternative assumption would be that voting intention by age changes to a degree over time, which you could reliably model and predict.

The ANU study quoted above provides some fertile ground to explore this further. If you look at Figure 4.5 on page 23 of the report (as per the screenshot below), you will see a collection of graphs that show the voting trend of the major generation types across elections from 1987-2022.

What strikes me about this collection is that from the Silent Generation onwards, the latest gap between the ALP & LNP vote across the generations is widening, with the internal trend of each generation also showing a widening between the ALP & LNP, in the ALP’s favour. The only exception here is Gen Z, whose LNP trend is in the opposite direction, but as the report notes, it has a relatively small sample size and starts with the LNP much further behind the ALP compared to earlier generations at the same age.

What this suggests is that a likely trend would be younger voters skewing further to the ALP than they already are now. But to model and test the predictability of this would require the data behind these graphs and quite a bit of backtesting of possible models (a future project perhaps).

Results

What's happening here is that the older LNP voters, who are passing on, are not being replaced at the same rate by younger LNP voters.

Discussion

Determining the accuracy of this analysis is difficult. The Australian Electoral Commission does not collect voting data by party choice or age because our elections are secret ballots. The Australian Election Study may provide some survey data that I can test these results against. I'm currently exploring this avenue and will report any progress here on this front.

But it is an interesting exercise estimating how much the LNP will fall away over the coming years simply through natural attrition of an ageing supporter base, and what looks like a failure in recent years to remain relevant and attract younger voters.

Methodology

1. Using ABS 2025 population by age data and AEC voting data, calculate valid votes as a % of the estimated voting age population (18+). The calculated percentage here was 71.57%.

2. Use the Valid votes % calculated in 1 to estimate the valid votes in 2025 for each age group from 18-99+.

3. Fit a 2nd order polynomial curve to the estimated valid votes in each age group to produce an overall 2PP result that equals the current national ALP 2PP result (54.82% at the time of calculation). For the record, the formula derived from this regression was:

ALP 2PP% = (-0.00004x² + 0.001000313x + 0.614000007) * 100

Where x = Age.

4. For each subsequent year (2026-28), using forecasted population by age data from ABS, calculate expected votes for each age group by:

a) Applying the Valid votes % calculated in 1 to the estimated population for each age group for each year.

b) Applying the ALP & LNP 2PP%’s from their original 2025 age group to the forecasted population of the proceeding age groups. For example, for 19-year-olds in 2026, use the ALP 2PP% in 2025 for 18-year-olds and apply this to the estimated valid votes for 19-year-olds in 2026.*

c) Total all votes for ALP & LNP and from this data, calculate a 2PP% for each party.

* Note; for 18-year-olds in 2026-28, I used the formula to calculate 2PP% values for 15,16 & 17-year-olds in 2025; ie 2026 18 year old = 2025 17-year-old, 2027 18-year-old = 2025 16-year-old and so on.