School trial, 18 schools, two years

Education research poster

Q: What goes on an education research poster?

The problem in classroom terms, the design and who was in it, what each arm actually received, the outcome with its interval, and the resource cost. Effect size alone does not tell a school whether it can afford the change.

Q: How do I compare two approaches on a poster?

Use a table with one row per dimension you are comparing and one column per approach. Keep the rows to things that actually differ, and put the resource use and the cost in the same table as the result. This board runs from session length and group size down to adult hours, mean gain, effect size and cost per child, so each result sits beside what it took.

Q: Should fidelity be reported?

Yes, whenever the intervention is delivered by ordinary staff rather than by the research team. This board states that teaching assistants had four days of training and that fidelity was checked by observation twice a term. That line is what separates a trial of an approach as schools would run it from a trial of an ideal version nobody can staff.

Q: Can I edit the table after generating?

Yes. The table editor in the inspector lets you retype any cell and add or delete rows and columns, and column widths can be dragged in edit mode. Every cell is left aligned, with no per column alignment setting. The block keeps its place and size on the board, so check that a longer table still fits before you export.

Q: How many rows can the comparison table hold?

The design is drawn for six rows and three columns with a header row, which is what generation aims for. This board carries eight rows because the table editor can add rows beyond that. More rows mean less space per row inside a block of fixed size, so after adding any, check the table at full size and keep the rows that carry your primary outcomes.

Q: Should I report effect sizes or raw scores?

Report both, because they answer different questions. Raw scores show the scale, and the standardised effect lets a reader compare with other studies. This board puts both in the table, the mean gains of +7.4 and +2.1 points beside effect sizes of 0.31 and 0.09, and gives the stat block to the +5.3 point difference with its 95% interval of 3.1 to 7.5.

Q: What if my two approaches were not randomised?

Say so in the first entry of the methods block, name what the groups were matched on, and write difference rather than effect in the results and the conclusion. The comparison table still works exactly as it does here; only the claim on the board has to be smaller, and the key number should be labelled as an observed gap.

A two year school trial comparing twenty minutes of small group phonics against whole class revision, laid out as an A0 landscape board built around a comparison table. It reports the reading gain, the adult hours and the cost per child together, because those are the three numbers a head teacher decides with.

Create a poster with OneCraftA0 Landscape, printed at 1189 by 841 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.

Twenty minutes of small group phonics against whole class teaching: a two year trial

R. Whitfield ¹, A. Boateng ¹ ², S. Kaur ²

1 School of Education, Northgate University · 2 Meadowbank Teaching School Alliance

BACKGROUND

Schools in this alliance were spending an extra forty minutes a week on whole class phonics revision for children who had fallen behind by the end of the first year. Teachers reported that the children who needed it most were the least able to use it, because a whole class session moves at the pace of the class. The alternative on offer was twenty minutes of small group teaching, three times a week, delivered by a teaching assistant with four days of training. It costs less staff time in total but it costs a trained adult, which is the resource schools have least of. The alliance agreed to test the two directly rather than adopt the small group model on the strength of the literature, because the literature is mostly from trials with researcher delivered teaching, and those effect sizes have not survived contact with ordinary school staffing. We also wanted the comparison stated in the terms a head teacher actually decides in. An effect size on its own does not answer whether to fund a teaching assistant for a year, so we recorded adult hours per child and cost per child alongside the reading score, and we report all three together rather than leading with the one that flatters the intervention.

TRIAL METHODS

Design. Two arm trial across 18 schools, randomised at school level, two academic years.

Participants. 1,412 children in the lowest attaining third at the end of year one, aged 5 to 6.

Arms. Small group: 20 minutes, three times weekly, groups of four. Whole class: 40 minutes weekly revision.

Delivery. Teaching assistants with four days of training, fidelity checked by observation twice a term.

Outcome. Standardised reading score at the end of year two, marked externally and blind to arm.

The two approaches

What differs	Small group	Whole class
Session length	20 minutes	40 minutes
Frequency	Three times weekly	Once weekly
Group size	Four children	Whole class, 28 to 31
Delivered by	Trained teaching assistant	Class teacher
Adult hours per child, per term	4.0	0.4
Mean gain, standardised score	+7.4 points	+2.1 points
Effect size against no revision	0.31	0.09
Cost per child, per year	$118	$14
The two arms as delivered, with the cost and the measured gain side by side.

+5.3

points ahead, small group

95% CI 3.1 to 7.5, p<0.001

Figure 1. Gap in mean standardised reading score, small group minus whole class, by term.

Source: Trial dataset, 2026.

CONCLUSIONS

Small group teaching gained 5.3 points more, at eight times the cost per child.

Small group ahead by 5.3 points, 95% CI 3.1 to 7.5

Effect size 0.31 against 0.09 for whole class

Costs $118 per child per year against $14

Gain held through the summer term, no fade

Block by block

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

Title band and authors: The trial title across the band with the university and teaching alliance affiliations.
Background: What the schools were already doing, why the children who needed it most could not use it, and why the alliance tested rather than adopted on the strength of the literature.
Trial methods: Design, participants, the two arms as delivered, who delivered them and how fidelity was checked, and the externally marked outcome.
The comparison table: Eight rows covering session length, frequency, group size, who delivers, adult hours per child, mean gain, effect size and cost per child.
Key number: A stat block carrying the 5.3 point difference with its confidence interval and p value.
The gap by term: A figure plotting small group minus whole class across autumn, spring and summer, showing the gap opening rather than fading.
Conclusion: One line stating both halves of the result: the gain, and the eight times higher cost that comes with it.
Takeaway points: Four checked points carrying the difference, the effect sizes, the costs and the fact that the gain held through the summer term. This layout has no reference block, so if your sources must be on the board, cite them inside the introduction text.
How to adapt this board: Put one row per outcome in the comparison table and keep the columns as what differs, arm one, arm two. Put the gains and the effect sizes in the table, and give the key number block to the headline difference with its interval. Keep the cost in the last table row so it sits directly beside the gain it bought. That pairing is the whole argument of a school trial: a head teacher is deciding whether the effect is worth the staff time, not whether it is statistically significant. Keep the term by term chart if your outcome was measured more than once, because it answers the fade question before anyone asks it.

What makes this board work

The table is the argument

Eight rows put session length, group size, adult hours, gain, effect size and cost against each other. A comparison written as two paragraphs makes the reader hold both in their head; a table does not.

Cost sits beside the effect

The gain is 5.3 points and the cost is eight times higher per child. Education posters routinely report the first and omit the second, which makes them useless to the person deciding.

The figure shows the gap, not the scores

Plotting both arms on a standardised scale made the bars look identical. Plotting the gap by term shows what the trial actually measured, and shows it opening rather than fading.

Questions people ask

What goes on an education research poster?

The problem in classroom terms, the design and who was in it, what each arm actually received, the outcome with its interval, and the resource cost. Effect size alone does not tell a school whether it can afford the change.

How do I compare two approaches on a poster?

Use a table with one row per dimension you are comparing and one column per approach. Keep the rows to things that actually differ, and put the resource use and the cost in the same table as the result. This board runs from session length and group size down to adult hours, mean gain, effect size and cost per child, so each result sits beside what it took.

Should fidelity be reported?

Yes, whenever the intervention is delivered by ordinary staff rather than by the research team. This board states that teaching assistants had four days of training and that fidelity was checked by observation twice a term. That line is what separates a trial of an approach as schools would run it from a trial of an ideal version nobody can staff.

Can I edit the table after generating?

Yes. The table editor in the inspector lets you retype any cell and add or delete rows and columns, and column widths can be dragged in edit mode. Every cell is left aligned, with no per column alignment setting. The block keeps its place and size on the board, so check that a longer table still fits before you export.

How many rows can the comparison table hold?

The design is drawn for six rows and three columns with a header row, which is what generation aims for. This board carries eight rows because the table editor can add rows beyond that. More rows mean less space per row inside a block of fixed size, so after adding any, check the table at full size and keep the rows that carry your primary outcomes.

Should I report effect sizes or raw scores?

Report both, because they answer different questions. Raw scores show the scale, and the standardised effect lets a reader compare with other studies. This board puts both in the table, the mean gains of +7.4 and +2.1 points beside effect sizes of 0.31 and 0.09, and gives the stat block to the +5.3 point difference with its 95% interval of 3.1 to 7.5.

What if my two approaches were not randomised?

Say so in the first entry of the methods block, name what the groups were matched on, and write difference rather than effect in the results and the conclusion. The comparison table still works exactly as it does here; only the claim on the board has to be smaller, and the key number should be labelled as an observed gap.

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 one like this

Sources

Written and checked by the OneCraft team. Last checked 29 September 2026.