How do you show results so people understand their limits?

Fix three misleading charts, write findings with their conditions, and brief a council and the public

Research studio R8 of Energy Systems, Step by Step: sort findings by whether they carry their conditions, diagnose and redesign three misleading charts drawn from EIA and SPP data in your browser, write one-sentence findings, and turn a study’s result into a briefing for a city council and a paragraph for the public, checked against the course rubric.
  • Researcher · L4
  • Research studio
  • 30 min guided · about 90 min practice
  • After R6

The task

  1. Produce: one key figure of your own study redesigned to pass the checklist, three one-sentence findings with their conditions, a briefing for a decision-maker and a public summary of at most 150 words.
  2. From: nine findings to sort, three misleading charts drawn in your browser from EIA and SPP data, and R6's result about Prairie Town's 2035 peak.
  3. Done when: the parts of the rubric's bar for Communication practised here are met: one key figure, a limits paragraph beside the result, and a 150-word public summary. The six-page paper comes in the capstone.

About 30 minutes guided here, then about 90 minutes on your own, on your own study:

  • redesign its key figure to pass the checklist;
  • write its findings, each with its conditions;
  • write a one-page briefing and a 150-word summary;
  • ask someone outside the field to read them back to you.

Three ways a result loses its limits

  • The axis: bars on a truncated axis make differences look bigger than they are.
  • The window: a few days or weeks cut from a year can show a trend the year does not have.
  • The average: an average hides the spread, and the busiest days are the ones that matter for planning.
Nine findings, sorted, with reasons
Finding Where it goes Why
The U.S. utility-scale wind fleet's capacity factor was 33.8% in 2025 (EIA). Holds as written A number, the fleet (plants of 1 MW or more), the year and the source: a reader can check it and knows where it stops. Capacity factor: output over full-power output.
In SPP's 2025 data, the busiest hour of demand began at 16:00 on 29 July. Holds as written Place, year, hour and data named; it says when, and claims nothing about why.
With central car uptake, the town's 2035 peak is 480–535 MW in about 8 of 10 model runs. Holds as written A range, the case it belongs to (central uptake: 15,000 electric cars in 2035) and what the range means. The model's other conditions belong in the next sentence.
Wind farms run at about a third of their capacity. Missing its conditions Which fleet, which year? True of the U.S. fleet in 2025 (33.8%), but it moved from 23.1% to 43.1% month to month.
Demand peaks in the afternoon. Missing its conditions In summer, yes: all 92 June to August days peaked between 12:00 and 18:00. In winter, 64 of 90 days peaked before noon (SPP, 2025).
Managed charging cuts the 2035 peak by 20 MW. Missing its conditions Only with fast car uptake, at the median, with demand added hour by hour. Adding the separate peaks ignores when charging happens, so it cannot show this effect.
Wind power is unreliable. Claims more than shown A verdict the data cannot carry. They show a capacity factor that moved between 23.1% and 43.1% by month, which planners can use.
Hot weather caused 2025's peak. Claims more than shown The data show when the peak came, not why. Heat is likely, but showing a cause needs weather data and a method that separates it.
Managed charging will solve the town's peak problem. Claims more than shown “Will solve” claims a certainty and a goal the study never set. It shows a cut of about 20 MW in one case.
Chart 1: “Wind collapses in late summer”

What makes the late-summer dip look so deep?

  • The months run in calendar order. Not it. Calendar order is right for a seasonal pattern; it misleads nothing.
  • The axis starts at 20%, so the bars' lengths no longer compare the values. Defensible. On this axis March's bar is 7.5 times September's; the values differ by a factor of 1.9. Bars show length, so they start at zero.
  • It shows the national fleet, not one farm. Partly. Worth saying on the chart, but it does not deepen the dip.

Fixes: Start the axis at zero (the key one); Name the fleet, the year and the source; Make the title say what the data show.

Month Capacity factor
January 37.4%
February 37.2%
March 43.1%
April 40.7%
May 32.1%
June 32.2%
July 27.5%
August 23.8%
September 23.1%
October 33.8%
November 35.4%
December 39.2%
Chart 2: “Wind output fell 80% in two weeks”

Why does wind seem to be falling?

  • Two weeks were cut from a year, starting on a windy day and ending on a calm one. Defensible. The window makes a fall look like a trend. Over the whole year, daily output swung between 13% and 91% of the windiest hour, averaging 53%.
  • Output is shown as a share of the year's windiest hour. Not it. A fair scale, if the axis says so. It does not create the fall.
  • Daily averages hide the hours. Partly. True, but the hours would show the same fall. The window creates it.

Fixes: Show the whole year (the key one); Mark the two weeks on it; Make the title say what is measured.

The whole year in numbers: daily means from 13% to 91% of 2025's windiest hour, averaging 53%; 12 September to 25 September fell from 71% to 14%.

Chart 3: “Prairie Town's demand peaks at 311 MW”

Why would planning for this line fail?

  • It is drawn in local time. Not it. Local time is right for a daily pattern; the hours match what people live by.
  • The line is too smooth to read. Partly. Smoothness is the symptom, not the cause: averaging 365 days removed the spread.
  • It averages 365 days; the busiest days go far higher. Defensible. On 1 day in 10 the 16:00 hour is above 399 MW, and the year's highest hour was 450 MW, against the average day's 311.

Fixes: Show the spread across days (the key one); Add the year's highest hour; Make the title name the line.

Hour Average (MW) P10 (MW) P90 (MW)
00:00 258 222 310
01:00 252 217 301
02:00 248 214 296
03:00 246 214 292
04:00 248 214 293
05:00 253 219 297
06:00 264 228 309
07:00 272 235 319
08:00 278 241 326
09:00 282 247 333
10:00 287 250 341
11:00 292 250 357
12:00 296 251 369
13:00 300 251 383
14:00 303 249 392
15:00 306 249 398
16:00 309 251 399
17:00 311 254 399
18:00 311 256 393
19:00 306 255 383
20:00 300 254 368
21:00 293 249 358
22:00 281 241 341
23:00 268 231 323

Highest hourly demand in 2025: 450 MW, at 16:00 on 29 July.

R6's result, for the briefing and the paragraph
The 2035 peak
480–535 MW with central car uptake (15,000 electric cars), 495–550 MW with fast uptake (30,000): P10–P90, about 8 in 10 of 1,000 runs. 450 MW in 2025.
Managed charging
A programme moving most charging after 23:00. With fast uptake, the median falls from 520 to 500 MW when demand is added hour by hour; adding the separate peaks ignores timing and misses this.
Inputs
Varied in the runs: growth (0.4–1.6% a year), the year's weather (±4%) and a car's daily energy (7–13 kWh). Fixed: 2025's hourly pattern; cars charging at home.
Not in the model
A new customer (a constant 60 MW would raise each 2035 figure by 60 MW), workplace charging, the programme's cost and sign-up. Outcomes outside the ranges remain possible.
Example findings, briefing and public paragraph
Chart 1, the wind fleet by month
In 2025, the U.S. utility-scale wind fleet's monthly capacity factor ranged from 23.1% in September to 43.1% in March (EIA).
Chart 2, SPP's daily wind
In SPP's 2025 data, daily wind output ranged from 13% to 91% of the year's windiest hour, averaging 53%; a two-week window cannot show a trend.
Chart 3, Prairie Town's average day
On Prairie Town's 2025 pattern, the average day peaks at 311 MW at 17:00, but on 1 day in 10 the 16:00 hour tops 399 MW, and the year's highest hour was 450 MW.
Briefing · The decision
Whether to fund a managed-charging programme before 2030.
Briefing · What we found
In about 8 of 10 illustrative model runs, the 2035 peak is 480–535 MW with central car uptake and 495–550 MW with fast uptake, against 450 MW in 2025.
Briefing · What the programme changes
With fast uptake the programme lowers the median by about 20 MW, when demand is added hour by hour.
Briefing · What it depends on
Growth, the year's weather and a car's daily energy, varied within illustrative ranges; 2025's daily pattern of demand; cars charging at home.
Briefing · What it leaves out
A large new customer (a constant 60 MW would raise each 2035 figure by 60 MW), workplace charging, and the programme's cost and sign-up; outcomes outside the ranges remain possible.
Briefing · What would change this advice
A connection request from a large customer, or car numbers on the fast path, would call for a new forecast.
For the public
Prairie Town uses the most electricity on summer afternoons. We looked at how that busiest hour could change by 2035 as more people drive electric cars. If they spread at a middle pace, about 8 in 10 of our model runs put it between 480 and 535 megawatts, compared with 450 in 2025. If they spread fast, the range moves up to 495 to 550, and higher values are possible. Charging cars late at night instead of in the early evening would keep it lower, by about 20 megawatts if car numbers grow fast. These numbers leave out a large new factory or business, which would add more, and we will update them as new car numbers come in.

With JavaScript on, you diagnose each chart, apply fixes that redraw it, and write your own findings, briefing and paragraph; your drafts stay in this browser.

Chart 1: EIA's 2025 monthly capacity factors of the U.S. utility-scale wind fleet (plants of 1 MW or more), drawn in your browser.
Charts 2 and 3: the course's cleaned and normalized series of SPP's 2025 hourly demand and wind (EIA-930). SPP, the Southwest Power Pool, runs the grid across much of the central U.S.
Cleaning: implausible demand hours and frozen wind readings replaced (choices like R4's), each series divided by its 2025 maximum, in thousandths; the day counts and extremes here describe this series.
Prairie Town: fictional; its demand is SPP's 2025 pattern scaled so the highest hour is 450 MW, as in lesson 0.
The briefing: R6's fictional study and model (1,000 runs, seed 6); its inputs are illustrative.

Two answers to the same task

Weak answer · Developing

Our analysis shows that the town's peak will grow to 515 MW by 2035 [1]. Managed charging is the solution and should be funded [2]. See the attached figures for details [3]. The model has been validated [4].

  1. Falls short: One number with “will”: no range, no case, no conditions.
  2. Falls short: Advice past the evidence: the study measured a peak, not the programme's cost or sign-up.
  3. Falls short: A council member cannot act on figures nobody walked them through.
  4. Falls short: Asserted, not shown; and checking a model does not remove the future's uncertainty.

Strong answer · Proficient

Decision: whether to fund a managed-charging programme before 2030 [1]. Finding: in about 8 of 10 illustrative model runs, the 2035 peak is 480–535 MW with central car uptake and 495–550 MW with fast uptake, against 450 MW in 2025 [2]. Programme: with fast uptake it lowers the median by about 20 MW, when demand is added hour by hour [3]. Depends on: growth, the year's weather and a car's daily energy, varied within illustrative ranges; 2025's daily pattern; cars charging at home [4]. Leaves out: a large new customer (a constant 60 MW would raise each figure by 60 MW), workplace charging, the programme's cost; outcomes outside the ranges remain possible [5]. Would change it: a connection request from a large customer, or car numbers on the fast path [6].

  1. Works: Opens with the council's decision, not the method.
  2. Works: Ranges with the case each belongs to, what a range means, and the 2025 value for scale.
  3. Works: The effect, with the case and the model condition it rests on.
  4. Works: The assumptions a member could challenge, with what the runs varied, in plain words.
  5. Works: The limits beside the result, not at the end.
  6. Works: Signposts: what would make the advice out of date.
Weak answer Strong answer
Opens with The model's result The council's decision
Numbers One, with “will” Ranges, each with its case and its meaning
Limits None Beside the result
Advice Beyond the evidence What the evidence shows, and what would change it

Three charts, what misled and the fix

Chart What misled The fix A finding with its conditions
Wind fleet by month An axis from 20% made a 1.9× difference look 7.5×. Bars from zero; fleet, year and source named. Monthly capacity factor ranged from 23.1% to 43.1% in 2025 (U.S. fleet, EIA).
SPP's daily wind Two weeks chosen from windy to calm looked like a fall of 80%. The whole year, with the two weeks marked. Daily output swung between 13% and 91% of the windiest hour through 2025; two weeks cannot show a trend.
Prairie Town's average day An average of 365 days peaked at 311 MW; no plan can use that alone. The spread across days, and the year's highest hour. The average day peaks at 311 MW; 1 day in 10 tops 399 MW; the highest hour was 450 MW.

Where the managed-charging finding holds

It holds when It leaves out
  • demand is added hour by hour on 2025's pattern, and every day's charging follows the same managed profile;
  • car uptake is fast (30,000 cars in 2035); with central uptake the cut is smaller;
  • growth, weather and a car's daily energy stay inside the ranges R6 sampled.
  • a large new customer;
  • workplace charging;
  • the programme's cost, and how many drivers sign up;
  • other years' patterns of demand.

Model: R6's, 1,000 runs, seed 6. Data: SPP's 2025 hourly demand (EIA-930).

Go deeperWords for chances, and pictures of uncertainty
  • Say what a range is: “8 of 10 model runs” or “in most of the futures we modelled”. A modelled range is not a promise about the world (R6).
  • Bands, fans and several futures side by side all show uncertainty; which works best depends on the readers' numeracy, and uncertainties that come from disputed knowledge remain hard to draw (Spiegelhalter et al., 2011, abstract).
Go deeperWhat this studio leaves out
  • Writing the paper itself: its structure, peer review and replies come with the capstone research project (CR).
  • Talking with the people a study is about: R9.
  • Testing a summary with real readers in a structured way: a topic for a science-communication course.

Check your own work against the rubric

Competence Developing Proficient (the bar) Strong
Communication
Look for: the parts R8 practises (one key figure that passes the checklist, findings with their conditions, limits beside the result, a briefing, and a public summary of at most 150 words).		 Results without limits. Six-page paper; one key figure; a limits paragraph; a 150-word public summary. A briefing a decision-maker could act on without misreading.

Before a result leaves your desk

  • One message: Each figure makes one point, and its title or caption says which.
  • Honest axes: Units on every axis; bars start at zero; any zoom is labelled.
  • The whole picture: The time window and any selection said; the context shown or one click away.
  • Spread, not only averages: A band or the range beside any average, and what the band means.
  • Findings with conditions: Each finding: a number with its unit, the place, the time and the main assumption.
  • Limits beside results: What the study leaves out, next to what it found, not at the end.
  • The decision first: A briefing opens with the reader's decision and ends with what would change the advice.
  • A public version: Plain words, at most 150, read back to you by someone outside the field.
  • Not colour alone: Colour-blind-safe colours, with labels or patterns that carry the meaning too.

Myth or fact? “A bar chart's axis should start at zero.”

  • Myth
  • Fact
Answer

Fact. Bars show length: on an axis from 20%, a difference of 1.9 times looked like 7.5 times.

Evidence: truncation makes readers see bigger differences even when the chart warns them (Correll et al., 2020, abstract).

Myth or fact? “Stating the uncertainty always makes people trust a result less.”

  • Myth
  • Fact
Answer

Myth. A review found some evidence that stating uncertainty does not necessarily affect audiences negatively, though effects vary between people and formats (van der Bles et al., 2019, abstract).

Myth or fact? “An average day tells you what a busy day needs.”

  • Myth
  • Fact
Answer

Myth. Prairie Town's average day of 2025 peaks at 311 MW; on 1 day in 10 the 16:00 hour tops 399 MW, and the highest hour was 450 MW.

Myth or fact? “A study's limits belong in a section at the end.”

  • Myth
  • Fact
Answer

Myth. A full list can go at the end, but the limit that changes a finding goes beside it: many readers stop after the result.

Go deeper · ResearcherFrom this studio to your capstone
  • The capstone research project (CR): the whole bar, a six-page paper with one key figure, a limits paragraph and a 150-word public summary.
  • Strong goes further: a briefing a decision-maker could act on without misreading it.
  • To test that: ask a decision-maker to tell you what your briefing says.

Sources and what’s next

Data and numbers on this page

Data or number What it is Source
23.1% to 43.1% Monthly capacity factor of the U.S. utility-scale wind fleet in 2025 (September and March); 33.8% for the year. Chart 1. EIA, Electric Power Monthly, Table 6.07.B · data for August 2026, released 23 October 2026; accessed 2 November 2026
2025 hours SPP's hourly demand and wind for 2025, in the course's cleaned and normalized series: charts 2 and 3, and the hour the year's demand peaked (16:00 on 29 July). EIA-930, six-month files, SWPP · accessed 2 November 2026
450 MW Prairie Town's highest hour of 2025: the course's scale for the fictional town (lesson 0). The course's own convention

Next

R9 · Research integrity, and working with the people your study is about

Sources and further reading

Spotted an error, or stuck? Write to hlli@ksu.edu. Corrections are made in place and dated here.

Text licensed CC BY 4.0, code MIT: teachers and students may reuse and adapt it, with credit to Honglin Li.