This article is from the source 'guardian' and was first published or seen on . It last changed over 40 days ago and won't be checked again for changes.
You can find the current article at its original source at http://www.theguardian.com/australia-news/2016/apr/01/political-polls-matter-but-the-overall-trend-is-what-to-look-at
The article has changed 2 times. There is an RSS feed of changes available.
Previous version
1
Next version
Version 0 | Version 1 |
---|---|
Political polls matter, but the overall trend is what to look at | Political polls matter, but the overall trend is what to look at |
(5 months later) | |
There will be a lot of polls over the next three months, from Galaxy, Essential, Morgan, Newspoll, Ipsos, Reachtel, Lonergan and others. | There will be a lot of polls over the next three months, from Galaxy, Essential, Morgan, Newspoll, Ipsos, Reachtel, Lonergan and others. |
There is real value in looking at these polls, but it’s dangerous to focus too much on any one poll. | There is real value in looking at these polls, but it’s dangerous to focus too much on any one poll. |
Each poll has a margin of error, and beyond the margin of error you would expect about one in 20 polls to be more inaccurate. For this reason, readers shouldn’t put too much trust in one poll that shows a dramatic shift compared to other recent polls. | Each poll has a margin of error, and beyond the margin of error you would expect about one in 20 polls to be more inaccurate. For this reason, readers shouldn’t put too much trust in one poll that shows a dramatic shift compared to other recent polls. |
For this reason, it’s better to look at how the polls shift gradually, and average out all of the different polls. This “polling aggregation” method is at the heart of FiveThirtyEight’s US election predictions. | For this reason, it’s better to look at how the polls shift gradually, and average out all of the different polls. This “polling aggregation” method is at the heart of FiveThirtyEight’s US election predictions. |
In Australia, William Bowe at Crikey’s Poll Bludger blog has been running his Bludger Track polling average for a few years. Metapoll has just launched its own polling average. | In Australia, William Bowe at Crikey’s Poll Bludger blog has been running his Bludger Track polling average for a few years. Metapoll has just launched its own polling average. |
At the moment, Bludger Track is giving the Coalition 51.3% of the two-party-preferred vote, which is down from over 54% in January. Metapoll’s first polling average gave the Coalition 52.4% of the two-party-preferred vote. | At the moment, Bludger Track is giving the Coalition 51.3% of the two-party-preferred vote, which is down from over 54% in January. Metapoll’s first polling average gave the Coalition 52.4% of the two-party-preferred vote. |
These polls should give us a good sense of the national political climate, but polls are less able to give us a sense of how the parties are faring in individual electorates. | These polls should give us a good sense of the national political climate, but polls are less able to give us a sense of how the parties are faring in individual electorates. |
Polling individual seats is a relatively new phenomenon in Australia, and it’s still a work in progress. Seat specific polls tend to have smaller sample sizes (and thus larger margins of error) and haven’t always accurately predicted the local result. | Polling individual seats is a relatively new phenomenon in Australia, and it’s still a work in progress. Seat specific polls tend to have smaller sample sizes (and thus larger margins of error) and haven’t always accurately predicted the local result. |
Pollsters and the media rarely have the capacity to repeatedly poll the same seat, so we often see a single seat poll without any trend, or any comparison to help us understand whether it is an accurate reflection of the electorate. | Pollsters and the media rarely have the capacity to repeatedly poll the same seat, so we often see a single seat poll without any trend, or any comparison to help us understand whether it is an accurate reflection of the electorate. |
We may see more of these seat polls produced, as robo-polling reduces costs, and electorate-level polling should become more accurate, but we’re not there yet. | We may see more of these seat polls produced, as robo-polling reduces costs, and electorate-level polling should become more accurate, but we’re not there yet. |
As it stands, seat polls are one useful bit of evidence, but seat polling should never be considered independently of other evidence in assessing a local race. | As it stands, seat polls are one useful bit of evidence, but seat polling should never be considered independently of other evidence in assessing a local race. |
It can be difficult for pollsters to predict support for smaller parties – naming a party in a poll can prompt the voter in a way that doesn’t reflect how the actual voting experience would work. There is some history of polls overstating the Greens vote, and for smaller parties their vote can be lost in the margin of error. | It can be difficult for pollsters to predict support for smaller parties – naming a party in a poll can prompt the voter in a way that doesn’t reflect how the actual voting experience would work. There is some history of polls overstating the Greens vote, and for smaller parties their vote can be lost in the margin of error. |
This is a particular problem when it comes to polling the senate. Senate polls are generally more difficult because voters aren’t primarily thinking about their senate vote, and because relatively small shifts in votes can produce very different senate results. | This is a particular problem when it comes to polling the senate. Senate polls are generally more difficult because voters aren’t primarily thinking about their senate vote, and because relatively small shifts in votes can produce very different senate results. |
If the next election is a double dissolution, parties could win seats on as low as 3% of the statewide vote – and it’s hard for pollsters to accurately capture voters’ intentions at that level of precision. | If the next election is a double dissolution, parties could win seats on as low as 3% of the statewide vote – and it’s hard for pollsters to accurately capture voters’ intentions at that level of precision. |
We’re yet to see any senate polling, and we’ve only seen a few seat-level polls, so for now we’re just watching the general trend in national polls as the volume of polls increases. | We’re yet to see any senate polling, and we’ve only seen a few seat-level polls, so for now we’re just watching the general trend in national polls as the volume of polls increases. |
The Coalition shot into the lead on the two-party-preferred vote after Malcolm Turnbull replaced Tony Abbott as prime minister, but since the beginning of this year we’ve seen Coalition support decline along with Turnbull’s personal popularity. | The Coalition shot into the lead on the two-party-preferred vote after Malcolm Turnbull replaced Tony Abbott as prime minister, but since the beginning of this year we’ve seen Coalition support decline along with Turnbull’s personal popularity. |
In recent weeks we’ve seen a number of polls putting Labor and the Coalition on 50-50, but we still see other polls giving the Coalition a lead. | In recent weeks we’ve seen a number of polls putting Labor and the Coalition on 50-50, but we still see other polls giving the Coalition a lead. |
It’s worth watching to see if the Coalition can stabilise its position, or if we start to see Labor break into the lead. As election day gets closer, we’ll should have some more state-level polling and local polling to give us a more detailed picture of how the campaign is playing out. | It’s worth watching to see if the Coalition can stabilise its position, or if we start to see Labor break into the lead. As election day gets closer, we’ll should have some more state-level polling and local polling to give us a more detailed picture of how the campaign is playing out. |
Previous version
1
Next version