Randomised trial, 18,400 voters, marked roll outcome

Political science research poster example

A randomised trial of one administrative text message and early voting turnout, laid out as an A0 portrait board with a strip of four numbers across the top. The strip carries the whole result, and the columns under it explain how it was obtained.

Create a poster with OneCraftA0 Portrait, printed at 841 by 1189 millimetres

The whole board

The poster at full size, exactly as it prints. Every number, citation and caption on it was written for this example, so the layout is being judged on real content.

FU
One text message, three days out: a randomised trial of reminders and early voting turnout
C. Meiring ¹, V. Domingo ¹
1 Department of Political Science, Fairhaven University
18,400
voters randomised
One to one, 9,200 per arm, from a single electorate roll
61.8%
turnout, reminder arm
5,686 of 9,200 voted at an early centre or on the day
58.4%
turnout, control arm
5,373 of 9,200, no message sent
3.4 points
turnout difference
95% CI 2.0 to 4.8, 313 extra voters
Abstract
Reminder messages are the cheapest instrument an electoral commission has, and the published estimates of their effect range from nothing to eight percentage points. We ran a pre registered trial on one state electorate at a scheduled by election. From the roll of 18,400 voters with a mobile number on file, 9,200 were randomly assigned to receive a single SMS three days before polling day naming their nearest early voting centre and its hours, and 9,200 received nothing. Turnout was taken from the marked roll, not from self report. In the reminder arm 5,686 voted, which is 61.8%, against 5,373 in the control arm, 58.4%. The difference is 3.4 percentage points, 95% CI 2.0 to 4.8, or 313 additional voters for 9,200 messages. The effect falls from 5.8 points among voters under 35 to 1.0 points among voters over 70.
Why one message
Most published reminder trials test a campaign rather than a message: several contacts, sometimes across channels, often carrying a persuasive frame as well as the practical detail. That design answers a question a commission cannot act on, because a commission may only send administrative information and may only send it once without inviting a complaint. This trial was built to answer the question the commission actually faces. One message, no persuasion, no repeat, sent at the time the roll data allows, with the effect measured against the marked roll rather than against a survey. The message was drafted with the commission and contained the voter address of the nearest early centre, its opening hours and nothing else. If a single administrative SMS moves turnout at all, that is a result an electoral office can spend money on, and if it does not, the commission can stop paying for one.
METHODS
Design. Pre registered before the roll data was received, individually randomised one to one, and single blind at the level of the voter.
Population. All 18,400 enrolled voters at one state by election who had a mobile number recorded on the roll, with no further exclusion.
Intervention. A single SMS three days before polling day naming the voter nearest early voting centre and its opening hours, and nothing else.
Outcome. Voted or did not vote, taken from the marked roll after the declaration of the poll rather than from a post election survey.
Analysis. Risk difference with a normal approximation interval, plus four age band subgroups specified in the pre registration rather than chosen afterwards.
REFERENCES
1.Barnaby, R. and Owusu, F. (2025) Administrative reminders and turnout: a review of single contact trials. Electoral Studies Review, 22(1), 61 to 89.
2.Domingo, V. (2024) Measuring turnout from the marked roll rather than the survey. Journal of Political Methodology, 17(4), 402 to 421.
3.Meiring, C. (2026) Pre registration and the by election turnout trial. Fairhaven Political Science Working Paper 2026-03.
pts
Under 55
55 or over
02468
Figure 1. Turnout difference in percentage points, reminder arm against control, by age.
Source: marked roll, one state by election, 18,400 voters.
Results
The overall difference of 3.4 points is concentrated in the youngest band, where turnout rose 5.8 points, and it falls steadily with age to 1.0 points among voters over 70, whose turnout was already 84% in the control arm. Grouped, the difference is 4.9 points among the 5,060 voters per arm aged under 55 and 1.7 points among the 4,140 aged 55 or over. Weighted by the 2,300, 2,760, 2,300 and 1,840 voters per arm in the four bands, the band estimates reproduce the headline figure of 3.4. No band showed a negative point estimate. Twenty seven voters asked to be removed from the message list, and eleven complaints reached the commission, all about the source of the number rather than the content. Early centre use rose more than polling day attendance, which is what a message naming an early centre should do: 2,204 of the reminder arm voted early against 1,933 of the control arm.
What this does and does not show
A 3.4 point rise for one message is at the upper end of the published range, and the most likely reason is the by election itself: turnout was low, information about early centres was thin, and the marginal voter was therefore easy to move. The same message at a general election, where every letterbox carries the same information, would be expected to do less. The result also says nothing about who those 313 extra voters are beyond their age band, and a reminder that moves young renters more than older home owners is not a neutral instrument even when its content is purely administrative. The cost side is unambiguous: at the commission rate for bulk SMS the intervention cost about 2.4 cents a message, or roughly 70 cents per additional vote cast.
Conclusion
A single administrative text message three days before polling day raised turnout by 3.4 percentage points, 95% CI 2.0 to 4.8. The effect fell from 5.8 points among voters under 35 to 1.0 points among voters over 70, and cost about 70 cents per additional vote. The gain came mostly through early voting rather than polling day attendance, so the message appears to have moved when people voted as well as whether they did.
One SMS raised turnout 3.4 points, 95% CI 2.0 to 4.8
5,686 of 9,200 voted with a reminder, 5,373 of 9,200 without
Largest effect under 35, at 5.8 points
About 70 cents per additional vote cast

Block by block

What each block on the board is for, in the order a reader walks it.

Title band, two authors and a QR code
The title names the intervention and the timing: one text message, three days out, and a randomised trial of reminders and early voting turnout. Two authors share one department affiliation, with the QR code at the right of the band.
The four number strip
Eighteen thousand four hundred voters randomised one to one; 61.8% turnout in the reminder arm, 5,686 of 9,200; 58.4% in the control arm, 5,373 of 9,200; and a difference of 3.4 percentage points with a 95% CI of 2.0 to 4.8, or 313 extra voters.
Abstract
The abstract states the design, the message content, the outcome source and both arms, and closes with the cost framing: 313 additional voters for 9,200 messages, with the effect falling from 5.8 points among voters under 35 to 1.0 among voters over 70.
Why one message
Most published reminder trials test a campaign rather than a message, which answers a question a commission cannot act on. A commission may send administrative information once. This trial was built to answer the question the commission actually faces.
Methods as a labelled protocol
Five entries: pre registered and individually randomised, 18,400 enrolled voters with a mobile number on the roll, one SMS three days out naming the nearest early centre, turnout taken from the marked roll, and a normal approximation interval with pre specified age subgroups.
Results and figure
The figure shows the turnout difference in percentage points for voters under 55 and 55 or over, at 4.9 and 1.7. The results block gives the four band estimates behind it, the 27 opt outs and the eleven complaints, all about the source of the number.
Discussion, conclusion and references
The discussion places 3.4 points at the upper end of the published range and explains why a by election makes the marginal voter easy to move. Four checked points and three numbered references close the board.

What makes this board work

Four numbers carry the whole trial

The strip reads 18,400 randomised, 61.8%, 58.4% and 3.4 points, and each has its denominator underneath. A reader who stops for five seconds leaves knowing the sample, both arms and the effect.

The outcome source is named on the board

Turnout comes from the marked roll rather than from self report, and the methods block says so. Self reported turnout overstates by a wide margin, so naming the source is the difference between a checkable result and a claim.

The cost is on the board, not in the paper

About 2.4 cents a message and roughly 70 cents per additional vote cast. An electoral office reading this poster is deciding whether to fund the thing, and the cost line is the number that decision turns on.

Questions people ask

What goes on a political science research poster?

The intervention, how it was randomised, the sample and its arms, where the outcome came from, the effect with an interval, and any pre specified subgroups. If the study was pre registered, say so on the board, because it changes how a reviewer reads every other number.

How do I lay out a trial result at A0?

Put the numbers first. A stat strip across the top of the board gives each arm and the difference its own block with room for the denominator, and leaves the columns underneath free for abstract, method, results and discussion. It reads at five metres and rewards a closer look.

Should turnout come from a survey or the roll?

From the roll wherever you can get it. Self reported turnout is biased upward, and a poster that does not say which source it used leaves a reviewer unable to judge the size of the effect. This board names the marked roll in the methods and again in the abstract.

How large should the numbers in a stat strip be?

Large enough to read across the aisle, which on an A0 board means the figure itself several times the size of the body text. The label and the denominator underneath can be much smaller, because they are read by somebody who has already stopped.

How many words does a poster like this need?

This one runs a little over 2,000 words across abstract, introduction, methods, results, discussion and conclusion, and every block is full. Aim to fill the blocks your layout gives you rather than to hit a word count, and cut rather than shrink when a block overflows.

Can I change the four numbers in the strip?

Yes. Each is its own block with a value, a label and a detail line, and you can retype all three parts or delete a block entirely. The rest of the board reflows around the space, and nothing recalculates, so keep the strip and the abstract in step by hand.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make my political science research poster example

Other poster examples

Human geography research poster example

A two wave household panel measuring what a rail extension did to the commute, laid out as an A0 landscape board with one wide hero figure. The figure plots commute time against distance from the nearest station, so the saving and the point where it disappears are both visible at once.

Archaeology research poster example

Nine shell middens dated along four kilometres of coastal dune, laid out on an A0 portrait board with a numbered rail running down the left. The rail turns the board into a reading order, so a visitor who starts at node one finishes at the conclusion without being told where to go.

Linguistics research poster example

A vowel merger measured across 120 speakers in one coastal city, laid out as an A0 portrait board where a single hero figure takes the top half. The figure plots formant difference by birth decade, so the change reads as a trend rather than as two bars.

Want the steps in the builder? Read Add charts and diagrams, then choose the template, theme and size. For everything this generator can do, see the poster maker.

Sources

Written and checked by the OneCraft team. Last checked .