Cross sectional survey, stratified, weighted

Survey research poster

A survey research poster should show its frame, its response rate and its weighting as clearly as its findings, because those three numbers decide how far a reviewer trusts the rest. This A0 landscape example reports a fictional register survey of 4,110 small businesses with 1,204 responses, charts remote work by staff size band, and shows a weighted headline that passes the simple check every survey board should run before it prints.

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.

PU
Under half of small firms allow remote work, but only a quarter of those under ten staff
B. Castellano ¹, M. Ngata ¹, R. Lindberg ²
1 Institute for Business Research, Pemberton University · 2 Small Business Council
ABSTRACT
Three years after the pandemic forced the experiment, this survey asked 4,110 small businesses drawn from the national business register whether they still allow remote work, and 1,204 answered, a 29.3 percent response. Weighted to the register by size band and industry, 44.8 percent allow remote work at least one day a week (95% CI 41.5 to 48.1). The headline hides a steep size gradient: 24.8 percent among businesses with fewer than ten staff against 78.5 percent among those with 50 to 199, with the two middle bands in between at 55.3 and 68.7.
1 to 910 to 1920 to 4950 to 199020406080
Figure 1. Share allowing remote work at least weekly, by staff size band, weighted. Band samples: 512, 236, 242 and 214 responses.
Source: Survey data, 2026.
RESULTS
The size gradient is the finding: 24.8, 55.3, 68.7 and 78.5 percent across the four bands, a 54 point spread from smallest to largest, and the four band samples sum to the 1,204 total. Industry splits ran from 84.1 percent in professional services to 31.6 in retail and 22.4 in construction. The weighted overall figure of 44.8 percent sits below the unweighted 49.1 because the weights repair the sample's size mix: businesses under ten staff are half the register but 42.5 percent of responses, and post stratification restores them. Both numbers are printed, and the weighted figure sits between the smallest and largest band, the ten second arithmetic check a reader can run.
METHODS
Frame. National business register, 4,110 businesses with 1 to 199 staff, stratified by size band and industry.
Mode. Online questionnaire, 22 items, 9 minutes median completion, fielded over 6 weeks with three reminders.
Response. 1,204 completes, 29.3 percent of invited; band counts 512, 236, 242 and 214 from smallest to largest.
Weighting. Post stratification to register counts by size band and industry; design effect 1.4, effective n about 860.
Estimates. Weighted percentages with 95 percent confidence intervals; unweighted figures printed once for comparison.
REMOTE WORK BY SIZE BAND
1 to 9 staff
24.8%
10 to 19 staff
55.3%
20 to 49 staff
68.7%
50 to 199 staff
78.5%
DISCUSSION
Two cautions bound the headline. A 29.3 percent response leaves room for non response bias, and its likely direction is stated rather than waved at: businesses with an office email culture were easier to reach and are more likely to allow remote work, so 44.8 percent is best read as the top of the plausible range. And the survey measured policy, not practice; an owner who allows remote work that nobody takes counts as yes. The size gradient is robust to both concerns, because both would have to operate differently by band to create a 54 point spread. The next wave adds a practice question: days actually worked remotely last week.
INTRODUCTION
Remote work statistics are dominated by large employers with HR systems and survey panels, while the businesses that employ most of the workforce answer a different question: whether the owner can spare anyone from the counter, the site or the workshop. Earlier waves of this survey sampled differently each year, so drift in the estimate could not be separated from drift in the sample. This wave fixed the frame to the national business register, stratified by size and industry, and weighted the answers back to it, so the headline describes the small business population rather than the businesses easiest to reach. The Small Business Council's question was blunt: is remote work a settled feature of small business employment, or a large firm habit that never reached the small end?
1,204
responses
29.3% of 4,110 invited from the register
CONCLUSIONS
Remote work has settled into the small business population unevenly: just under half overall at 44.8 percent, a quarter below ten staff at 24.8. Any program that treats small businesses as one audience will miss the half of the register where remote work never arrived, and the size gradient, not the headline, is the number policy should be built on. The frame, the response and the weighting sit beside the estimates they qualify.
44.8% allow weekly remote work, weighted
24.8% under ten staff, 78.5% at 50 to 199
Response 29.3%, weighted to the register
Next wave measures practice, not policy
REFERENCES
1.Castellano B, Ngata M. Small business survey frames and response: five waves compared. Business Research Methods 2025;18(1):40-57.
2.Lindberg R, Castellano B. Post stratification weighting for register based business surveys. Survey Statistics Notes 2024;9(3):122-137.
3.Ngata M, Adler Y. Policy against practice in workplace flexibility reporting. Employment Studies Briefs 2023;6(2):71-85.

Block by block

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

Title band and authors
The title states the finding as a contrast: under half overall, and only a quarter of businesses under ten staff. Three authors from two affiliations sit under it, and the Small Business Council that asked the question appears as the second affiliation, so a reader knows who commissioned the work before reading a number.
Abstract and introduction
The abstract gives the frame, the response, a weighted headline with its interval and the four size band shares in one paragraph. The introduction explains why the frame matters: earlier waves sampled differently each year, so drift in the estimate could not be separated from drift in the sample, and this wave fixed the frame to the national register and weighted the answers back to it.
Methods: frame, mode, response, weighting, estimates
Five labelled entries carry what a survey reviewer checks first: the register frame of 4,110 businesses with 1 to 199 staff, stratified by size and industry; a 22 item online questionnaire with a nine minute median; 1,204 completes at 29.3 percent; post stratification to register counts with a design effect of 1.4 and an effective n near 860; and weighted estimates with 95 percent intervals, unweighted figures printed once.
One stat block for the sample
A single stat block prints 1,204 responses with its detail line, 29.3 percent of 4,110 invited from the register. Putting the response rate inside the only stat on the board is a deliberate choice: the one number a passer by reads large is the one that qualifies every estimate beside it.
Figure 1 and the size band panel
Four vertical bars climb from 24.8 to 55.3 to 68.7 to 78.5 percent across the size bands 1 to 9, 10 to 19, 20 to 49 and 50 to 199 staff. The caption prints the band samples, 512, 236, 242 and 214, which do sum to the 1,204 total. A four row panel headed Remote work by size band repeats the same shares as labelled values, so the gradient reads without the axis.
Results and discussion
The results block adds the industry spread, professional services at 84.1 percent against retail at 31.6 and construction at 22.4, and prints a weighted overall figure beside an unweighted one. The discussion names the likely direction of non response bias and separates policy from practice: an owner who allows remote work that nobody uses still counts as yes.
The headline check
The board's weighted headline is 44.8 percent. Businesses under ten staff are half the register but 512 of 1,204 responses, 42.5 percent, so weighting gives them more say and pulls the estimate below the unweighted 49.1. With register shares of 50, 22, 18 and 10 percent across the four bands, 0.50 of 24.8 plus 0.22 of 55.3 plus 0.18 of 68.7 plus 0.10 of 78.5 gives 44.8. Run the same check on your own board before printing.
Conclusions and references
Four bullets close the board: the headline, the two extreme bands, the response and weighting, and a next wave that measures practice rather than policy. The three references are invented for this fictional study because the layout carries a reference block; on your board, replace them with real sources.
How to adapt this board
Keep the order: frame and response in methods, the sample in the one stat block, the main split as ordered bars with band samples in the caption. Then compute the weighted headline as each band's share multiplied by that band's register share, summed, and confirm it sits between the lowest and highest band and close to the band that dominates the register. Print the unweighted figure once so the effect of weighting is visible.

What makes this board work

Response rate sits in the only stat block

The 29.3 percent response is printed with its 4,110 denominator inside the single stat block, not in a footnote. A reviewer reads the credibility of the survey at the same moment as its size, which is the order they should be read in.

Band samples sum to the sample

The caption prints 512, 236, 242 and 214, which add to the 1,204 in the stat block and the methods entry, so the chart, the stat and the methods agree on one calculator pass. The weighted headline passes its check too: 44.8 sits between 24.8 and 78.5 and below the unweighted 49.1.

The gradient carries the story

The chart and the size band panel both spend their space on the four band shares, a 54 point spread from the smallest businesses to the largest. That spread is robust to the non response and policy versus practice cautions, because both would have to differ by band to create it.

Questions people ask

What goes on a survey research poster?

The frame, the mode, the response rate, the weighting, the estimates with intervals and one split that carries the story. This board gives the first four a labelled methods entry each, puts the response rate in its only stat block, and spends its single chart on the size gradient rather than on a second headline figure.

How do I check a weighted headline against the subgroup numbers?

Multiply each subgroup's share by that subgroup's share of the population, add them up, and compare with the printed headline. The result must sit between the lowest and highest subgroup. On this board it passes: about half the register sits under ten staff at 24.8 percent, so the weighted 44.8 lands below the unweighted 49.1, as it should.

Should I show weighted or unweighted results?

Weighted, everywhere a population claim is made, with the unweighted figure printed once so the effect of weighting is visible. Then explain the direction of the gap in one sentence and make sure the direction is plausible. If the population is dominated by a low scoring group, weighting to it should lower the estimate, not raise it.

Which chart suits survey results?

Vertical bars by group for a single question split, ordered by the group's natural order, which is what this board charts by staff size band. The chart block cannot draw error bars, so print intervals in the caption or the results text. Band samples in the caption, as here, let a reader check the groups against the total.

Where do the questionnaire items go?

Off the board, behind a printed link, with the item count and median completion time stated in methods. Reviewers want to know the instrument exists and how long it took to answer; they do not want 22 items printed on an A0 board that has a frame, a response rate and one clear split to defend.

What about non response bias?

Name its likely direction in the discussion, as this board does: businesses with an office email culture were easier to reach and lean toward allowing remote work, so the estimate is best read as the top of the plausible range. A stated direction beats a generic limitation sentence, and it tells the reader which way to discount.

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

Other poster examples

Field trial poster example

A field trial poster reports a treatment applied on working land, so it needs the layout, the replication and the yield in one place. This A0 landscape example spreads biochar at 0, 5 and 10 tonnes a hectare in 75 strips on five New South Wales grazing farms for two seasons. Pasture rises from 5.1 to 5.8 t a hectare, 14% at the high rate.

Environmental science poster

A season of battery free soil moisture sensors on 24 smallholder farms, laid out as an A0 landscape broadsheet. It is a field trial rather than a laboratory study, so the board reports the things field work actually turns on: whether the device survived, where it was weakest, and what it saved.

Field study poster

A field study poster has to show when and where every number was collected, because season and site are the confounders a reviewer checks first. This A0 landscape example runs its four sampling visits as a dated timeline across the top of the board, so the heatwave at visit three explains the dip in the counts before anyone asks.

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 .