Plumbline Report exists to tell the difference between opinion changing and the sample changing. A publication with that job has to be exact about its own method. Several earlier posts on this blog were not, and this post sets the record straight.
What we got wrong
"Weighted and unweighted figures are published together." They are not. Our figures are unweighted. We do not yet collect the information a weighting scheme needs, such as education or how people voted last time. Whatever we publish describes how this panel moves, not the Victorian electorate as a whole. If we add weighting, we will publish the scheme first and show weighted and unweighted figures side by side.
"We collect education and past-vote recall." We don't. We ask one question a day, about your first preference, and nothing else about you.
"Location is recorded by region." It isn't. When you vote, your browser may ask to share your location. If you allow it, we check in memory whether the reading is in Australia, keep only that, and throw the coordinates away. We don't keep your state or region, and we don't use location to filter results.
"The most accurate" and similar claims. We have not yet published a single reading, so there is nothing to be accurate or inaccurate about. We shouldn't have said it.
Posts about tracking the Legislative Council. We don't. The panel asks one question about your first-preference vote. It doesn't measure the upper house separately.
Posts suggesting parties or campaigns could use our data. They can't. We don't share panellists' information with parties, campaigns or candidates, in any form that could identify anyone.
These posts weren't checked closely enough against what the site actually does. Where anything on this blog disagrees with the method page, the method page is right. It now carries a dated corrections notice, and we will add to it whenever we get something wrong.
Where the panel really is
The panel is open for the Victorian state election on 28 November 2026, and it is small. Fewer than 30 people vote on any given day, which is below the point where we will publish a breakdown at all. Below 30, a party breakdown describes identifiable people rather than an electorate, so we withhold it, party names included. The results page says that plainly rather than showing numbers that mean nothing.
That is the honest state of things. A daily panel only becomes useful once enough people stick with it, and we are not there yet.
What else changed this week
We used the same review to tighten how the site treats panellists:
- Joining no longer reveals who is on the panel. Whether or not an email address or mobile number is already registered, the join page gives the same answer. Only the email that follows is different, and only the owner of that inbox can read it.
- Deleting your account goes further. It now removes the network details stored with each answer, and withdraws any answer you gave that day, which hasn't been published yet.
- Analytics only see public pages. Google Analytics runs only on public pages for visitors who are not signed in. It never runs on your account pages or the links in our emails.
- Small groups stay hidden. Parties with too few responses are combined rather than named, so nothing about a small group can be worked out by subtraction.
- Unsubscribing works in one click from Gmail and other mail apps.
The details are in the privacy policy and the new terms of participation.
Who runs this
Plumbline Report is published by Synap AI Pty Ltd (ABN 86 695 003 653). Questions, corrections and criticism are welcome through the contact form. If you think we have got something wrong, we would rather hear it from you than find it ourselves later.
How you can help
It takes about ten seconds a day. Join the panel, answer one question each day until the election, and if it seems worthwhile, send the link to someone who might do the same. Every person who stays makes the readings mean more.