Difference in differences, 312 regional cafes, 24 monthly waves

Economics research poster example

A difference in differences study of a state minimum wage rise, laid out as an A0 portrait board built around its results table. At a metre you read the estimate, the table under it and the interval, and only then the method that produced them.

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.

WU
Did the 2025 minimum wage rise cut casual hours? Evidence from 312 regional cafes
R. Halloran ¹, P. Naidu ¹, S. Whitlam ¹ ²
1 School of Economics, Wattle University · 2 Regional Labour Research Unit, Wattle University
Abstract
A state minimum wage rise took effect on 1 January 2025 in one state and not in its neighbour, which gave us a clean border to compare across. We tracked rostered casual hours in 312 regional cafes for 24 monthly waves, 156 in the state that raised the floor and 156 matched on turnover, town size and staff count across the border. Mean weekly casual hours in treated cafes fell from 28.4 in 2024 to 26.9 in 2025. Control cafes moved from 28.1 to 28.0 over the same period. The raw difference in differences is 1.4 hours a week; with cafe and month fixed effects the adjusted estimate is 1.5 hours (95% CI 0.6 to 2.4). Head counts did not move.
METHODS
Design. Difference in differences across a state border, 24 monthly waves from January 2024 to December 2025.
Sample. 312 cafes: 156 in the state that raised the floor, 156 matched controls within 90 km of the border.
Matching. Nearest neighbour on 2023 turnover band, town population, casual head count and opening hours.
Measure. Rostered casual hours per week from the cafes own scheduling exports, not from payroll totals.
Analysis. Cafe and month fixed effects, standard errors clustered at the town level, 41 towns.
Why this question
Employers told the wage review that a rise of this size would be absorbed by cutting shifts rather than jobs, and the review recorded the claim without evidence either way. Regional cafes are a good place to test it because they roster almost entirely casual staff, they sit close enough to a state border that two otherwise similar towns can face different wage floors, and their rostering software keeps hours at the shift level rather than the pay period. The question this board answers is narrow on purpose. Not whether the rise was warranted, but whether rostered casual hours fell, by how much, and in which cafes.
Discussion
A cut of 1.5 hours a week is a little under 5% of the roster, and it landed on hours rather than on jobs. That is the shape employers described, but the size is smaller than the submissions implied, and it is not evenly spread. Cafes with fewer than six casual staff account for most of the effect, at 2.3 hours, while cafes with ten or more show 0.6 hours and a confidence interval that crosses zero. The plain reading is that a small roster has nowhere to absorb a wage rise except the roster itself, whereas a larger one can move hours between people, trim a supplier cost or lift a price. Two limits are worth stating on the board rather than in a footnote. Rostered hours are not paid hours, so unrostered work would not appear here at all. And the border makes the two states comparable on wages without making them comparable on tourism, which was stronger in the treated state in the summer of 2025 and would, if anything, push measured hours up rather than down.
Hours before and after
Measure
Treated (156)
Control (156)
Difference
Mean weekly casual hours, 2024
28.4
28.1
0.3
Mean weekly casual hours, 2025
26.9
28.0
1.1
Change across the two years
1.5 fewer
0.1 fewer
1.4 fewer
Casual staff per cafe, 2025
4.6
4.5
0.1
Raw means over 24 monthly waves. The adjusted estimate appears at right.
2024
2025
Treated
Control
08162432
Figure 1. Mean weekly casual hours before and after the rise, by group.
Source: cafe rostering exports, 24 monthly waves, 2024 to 2025.
Results
Treated and control cafes tracked each other to within 0.5 hours in every quarter of 2024, which is the parallel trend the design needs. From the first quarter of 2025 they separate and stay separated. The gap opens fastest in the first two quarters after the rise and then flattens, so the cut looks like a one off reset of the roster rather than a slow drift. Casual head counts moved by 0.1 staff per cafe, well inside the noise.
1.5 hours
fewer casual hours a week per cafe
Adjusted difference in differences, 95% CI 0.6 to 2.4, 312 cafes, 24 monthly waves, errors clustered over 41 towns
Conclusion
Rostered casual hours fell by about 1.5 a week per cafe after the floor rose, head counts did not move, and the effect is concentrated in the smallest rosters. The adjustment came out of shifts, and mostly out of the shifts of cafes with the least room to move.
Casual hours fell 1.5 a week, 95% CI 0.6 to 2.4
Control cafes moved 0.1 hours over the same 24 waves
Cafes with under six casuals lost 2.3 hours a week
Casual head count per cafe did not change
REFERENCES
1.Amarasinghe, K. and Fowler, D. (2024) Rostering responses to wage floors in food service. Journal of Regional Labour Studies, 18(2), 114 to 139.
2.Byrne, T. (2025) Border designs in minimum wage research: what the matching can and cannot fix. Applied Economic Methods, 9(1), 44 to 61.
3.Ochoa, M., Reddy, S. and Lang, H. (2023) Hours, head counts and the small roster problem. Wattle Economics Working Paper 2023-07.

Block by block

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

Title band, three authors and a QR code
The title states the question and the sample in one line: did the 2025 minimum wage rise cut casual hours, across 312 regional cafes. Three authors carry superscripts generated from the two affiliations, and the QR code sits at the right of the band.
Abstract
A state raised its wage floor on 1 January 2025 and its neighbour did not. Treated cafes moved from 28.4 to 26.9 mean weekly casual hours, controls from 28.1 to 28.0. The raw difference in differences is 1.4 hours and the adjusted estimate is 1.5.
Why this question
Employers told the wage review that a rise would be absorbed by cutting shifts rather than jobs, and the review recorded the claim with no evidence either way. Regional cafes roster almost entirely casual staff and sit close enough to a border for two similar towns to face different floors.
Methods as a labelled protocol
Five labelled entries: the design, the 312 cafe sample, the nearest neighbour matching on turnover band and town population, rostered hours taken from scheduling exports rather than payroll, and cafe and month fixed effects with errors clustered over 41 towns.
Results table and figure
The table runs the full width: mean weekly casual hours in 2024 and 2025 for both groups, the change within each, and casual staff per cafe. Under it a grouped bar chart shows the same two groups in both years, with the source line naming the rostering exports.
The headline number
One key number block carries 1.5 hours a week, with the label naming what it is and the detail line giving the adjusted difference in differences, the 95% CI of 0.6 to 2.4, the 312 cafes, the 24 waves and the 41 town clusters.
Discussion, conclusion and references
The discussion splits the effect by roster size, at 2.3 hours for cafes with under six casuals against 0.6 for those with ten or more, and states the two limits plainly. Four checked points close the board above three numbered references.

What makes this board work

The table is the centre of the board

Four rows carry every number the argument needs: both group means in both years, the within group change and the raw difference of 1.4 hours. A reader who only reads the table still leaves with the result.

Raw and adjusted are kept apart

The table shows the raw difference in differences of 1.4 hours; the key number block shows the adjusted 1.5 with its 95% CI of 0.6 to 2.4. Collapsing the two into one figure is the fault this board avoids.

The chart carries two bars, not twenty four

Twenty four monthly waves are summarised as two grouped bars, treated and control, in 2024 and 2025. A poster figure read at a metre can hold four bars; it cannot hold forty eight.

Questions people ask

What goes on an economics research poster?

The question, the identification strategy, the sample and how it was matched, a table of the estimates, one figure, and the headline effect with its confidence interval. Economics posters live or die on the table, so give it the widest block on the board and keep the prose around it short.

How large should the type be at a metre?

Body text on this A0 portrait board sits near 24 pixels on a 1656 by 2340 working canvas, which prints at roughly 12 millimetres tall. That is readable at a metre. The title runs much larger, and anything under about 8 millimetres printed will be skipped by people walking past.

How many words should a poster like this carry?

Enough to fill every block without padding. This board runs a little over 2,000 words across abstract, introduction, methods, results, discussion and conclusion. The failure to avoid is a half empty section box, which reads as an unfinished study rather than a concise one.

Portrait or landscape for an economics conference?

Check the call for papers before you design anything, because the board size is set by the venue and not by you. Portrait suits a board that runs a table down the middle, which is what this one does. Landscape suits a wide figure with columns either side.

Can I print this at 300 DPI?

The export runs at twice the working canvas, so an A0 portrait board comes out at 3312 by 4680 pixels, which is about 100 DPI at A0 size. That is normal for large format printing, where viewing distance is a metre rather than 30 centimetres. There is no bleed, crop mark or CMYK option.

Do I have to keep the table?

No. Every block is a separate component. You can retype the table, add or remove rows, swap the chart between ten chart types, move a section or delete it. The board recalculates nothing for you, so if you change a number in the table, change it in the abstract too.

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 economics research poster example

Other poster examples

Criminology research poster example

A matched cohort study of youth diversion and twelve month reoffending, laid out as a classic A0 portrait research board. Three key numbers sit across the top of the results, and every one of them can be traced back to a count in the flow below.

Media studies research poster example

Four thousand news headlines coded ten years apart, laid out as an A0 landscape board built around a two column comparison table. The table is the argument: every measure is shown for both years with the change spelled out beside it.

Legal research poster example

Sentence length for one offence category across five courts, laid out as an A0 landscape board with the table given the widest block on the design. Empirical legal work is read through its table, so the table is placed where the eye lands first.

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 .