What makes a question worth researching?

Sort twenty questions, name their types, and rewrite three until a researcher could answer them

Research studio R1 of Energy Systems, Step by Step: sort twenty energy questions into researchable, too vague, already answered and not answerable with the data at hand; name each one’s type; rewrite three with an audience, a metric and a counterfactual; then check your own question card against the course rubric.
  • Researcher · L4
  • Research studio
  • 30 min guided · about 60 min practice
  • After lesson 13 and A22

The task

  1. Produce: three rewritten questions, each with an audience, a metric with units and a counterfactual, and one question card for the best of them.
  2. From: twenty illustrative questions about energy systems, which you sort and type first.
  3. Done when: your card reaches Proficient on Framing: a typed question with audience, metric with units, counterfactual, scale and boundary, and one falsification statement.

About 30 minutes guided here, then about an hour on your own: collect ten questions from your courses, your work or the news, sort and type them, and turn the best into a card.

The four bins

  • Researchable: specific, still open, and answerable with data you could get.
  • Too vague: no metric, no audience or no comparison.
  • Already answered: a lookup in a public dataset, or settled physics, is enough for this task. Re-checking or replicating a published number is research, with its own purpose and evidence plan.
  • Not answerable with the data at hand: the data the question names cannot show the answer.
The twenty questions, sorted, with reasons
Question Where it goes Why
In 2025, how many hours did Prairie Town's wind output stay below a tenth of its maximum, season by season? Researchable · descriptive Specific and answerable: the town's hourly 2025 wind data can be counted by season. Descriptive: it asks how often, not why.
Is solar power good? Too vague Good for whom, measured how, compared with what? No audience, metric or comparison.
How much electricity did the United States use in 2025? Already answered EIA publishes it: about 4.20 trillion kWh. A lookup, not a study; why use changed could be one.
Did a co-op's time-of-use tariff cut its members' evening peak demand, compared with similar co-ops that kept flat rates? Researchable · evaluative A named programme, a metric and a comparison group. Evaluative: it judges a past change against a comparison.
Using only each state's yearly electricity sales, at what hour of the day do electric-vehicle owners charge? Not answerable with the data at hand Yearly totals cannot show hours. Hourly charger or feeder data could.
What mix of rooftop solar and batteries should a school district buy to keep its 20-year cost lowest, compared with grid supply alone? Researchable · prescriptive A decision-maker, a metric and a comparison, with public prices and tariffs to work from. Prescriptive: it asks what a decision-maker should do.
How can we fix the grid? Too vague Fix what: outages, prices, emissions? Which grid, and who would act?
Why did heating demand fall more in some neighbourhoods than in others after the same retrofit programme? Researchable · explanatory An observed difference to explain, with meter and building data to test the reasons. Explanatory: it asks why an observed difference happened.
Does a battery give back less energy than it takes in? Already answered Yes: every charge and discharge loses some energy (round-trip efficiency, lesson 9). How much one battery loses can still be measured.
During a week-long cold snap, how would all-electric heating change the risk of elderly households losing heat, with and without backup? Researchable · predictive A stated event, a group, a risk to estimate and two cases to compare. Predictive: it asks what would happen under a stated event, with uncertainty.
Where in Prairie City are rooftop solar systems concentrated, compared with median household income by neighbourhood? Researchable · descriptive Specific and answerable from permit records and census income data. Descriptive: it maps where, without asking why.
What is the future of energy storage? Too vague No place, period, metric or audience: a topic, not a question.
Using one year of a utility's outage records, how will its customers' outage hours change by 2060? Not answerable with the data at hand One year cannot separate a trend from a bad year, and 2060 needs scenarios as well as records.
Did a rural co-op's community-solar programme lower renters' energy burden more than bill credits of the same cost would have? Researchable · evaluative A programme, a metric and a counterfactual on the same budget: the same money spent another way. Evaluative: it judges a past programme against an alternative.
What was the average capacity factor of U.S. utility-scale wind farms in 2025? Already answered EIA reports it: 33.8% (Electric Power Monthly). Why regions differ could be a study.
What will Prairie City's winter peak demand (MW) be in 2035 if half its homes switch to heat pumps, and with what range? Researchable · predictive A stated change, a metric with units (peak demand in MW) and a range to estimate. Predictive: it asks what a future value will be, with uncertainty.
Are heat pumps worth it? Too vague Worth what, to whom, over how long, compared with which heating?
Which pumping schedule should a Plains farm adopt to cut its power bill under the co-op's time-of-use tariff without lowering yield? Researchable · prescriptive A decision, a goal and a limit, with the tariff and pump data to work from. Prescriptive: it asks what the farm should do.
Using Prairie Town's hourly grid demand alone, why did each household's electricity use change? Not answerable with the data at hand Town-level demand mixes thousands of homes; household changes need meter data or a survey.
Why did some rural co-ops' customers lose power for longer than their neighbours' in the same ice storm? Researchable · explanatory An observed difference, with outage records and line data to test the reasons. Explanatory: it asks why the same storm had different effects.

Words in round 1: co-op · time-of-use tariff · round-trip efficiency

Words in round 2: energy burden · capacity factor · peak demand · heat pump · time-of-use tariff

Three vague questions, rewritten as examples
“Is solar power good?”, rewritten
Rewrite Did the co-op's community-solar array lower members' average yearly bill, compared with buying the same energy at wholesale prices?
Type Evaluative
Audience the co-op's board, which decides whether to build a second array
Metric average bill per member
Units $ a year
Compared with the same energy bought at wholesale prices
“Are heat pumps worth it?”, rewritten
Rewrite Over 15 years, would a cold-climate heat pump cost a Prairie Town homeowner less than a new gas furnace?
Type Predictive
Audience homeowners replacing a furnace, and the town's rebate office
Metric 15-year cost of heating: purchase, installation and bills, as a present value
Units $ in today's dollars
Compared with a new gas furnace installed at the same time
“How can we fix the grid?”, rewritten
Rewrite Which would cut Prairie Town's customer-hours of outage more per dollar over ten years: a second feeder or a battery at the substation?
Type Prescriptive
Audience the co-op's engineering committee
Metric customer-hours of outage avoided per $1 million spent
Units customer-hours per $1 million (one customer out for one hour is one customer-hour)
Compared with today's single feeder, with no upgrade

With JavaScript on, you sort and rewrite step by step, and your drafts stay in this browser.

The questions, rewrites, co-ops, Prairie Town and Prairie City are illustrative. Two answers quote public numbers (U.S. electricity use and the wind capacity factor, both for 2025), sourced at the end of the page.

Two answers to the same task

The task

Turn “Is community solar good for low-income families?” into a question card.

Weak answer · Developing

Question: Is community solar good for low-income families [1]? Type: not named [2] Audience: low-income families [3] Metric: savings [4] Compared with: (blank) [5] Scale and boundary: rural areas [6] I would be wrong if: (left blank) [7]

  1. Falls short: “Good” is not a metric: any result could be read as success.
  2. Falls short: Without a type, a survey, a model and a regression could all claim to answer it.
  3. Falls short: A broad group, with no use for the answer stated: who would do what with it?
  4. Falls short: No units and no period: dollars a month, or a share of income, over which years?
  5. Falls short: No counterfactual: savings compared with what?
  6. Falls short: A place, not a boundary: which programme, which members, which years, and what is left out?
  7. Falls short: No falsification statement: no result could show the expectation wrong.

Strong answer · Proficient

Question: Did a rural co-op's community-solar programme lower renters' energy burden more than bill credits of the same cost would have? Type: evaluative [1] Audience: the co-op's board, deciding whether to keep the programme [2] Metric: energy burden, yearly energy bills as % of household income [3] Compared with: bill credits of the same total cost, paid by the co-op over the same years [4] Scale and boundary: S1 community; the co-op's renter members, 2022–2025; landlords' costs left out [5] I would be wrong if: participants' burden fell no more than that of renters given credits, matched on income and housing type [6] Main uncertainty: renters who moved away during the four years

  1. Works: Names the type: a past programme judged against a comparison.
  2. Works: A specific audience and its use. Renters could be the audience of a question that serves their own choices.
  3. Works: A metric with units that bills and income data can compute.
  4. Works: Same budget, same payer, same years. Other differences between the groups that affect bills or income still need checking.
  5. Works: Says who and which years are counted, and what is left out.
  6. Works: Names the result that would show the expectation false, and how the two groups are made comparable.
Weak answer Strong answer
Type Not named Evaluative
Audience A broad group, no use stated A specific audience and its use
Metric and units “Savings”, no units Bills as % of income
Counterfactual None Credits: same cost, payer and years
Boundary “Rural areas” Members, years, what is left out
Falsification None A result that would show it wrong

Four tests before a question is worth researching

  1. Specific? If not, it is too vague: rewrite it.
  2. Is a lookup in a public dataset, or settled physics, enough? Use the answer; to re-check or replicate it, say why and with what evidence.
  3. Can the data at hand show the answer? If not, change the data or narrow the question.
  4. Then it is researchable: name its type.

Check your own work against the rubric

The question card

  1. Write the question as one sentence ending in a question mark.
  2. Add its type, its audience or decision-maker, a metric with units, and the counterfactual.
  3. Add the scale and boundary, “I would be wrong if…”, and the uncertainty that matters most.

With JavaScript on, a card form shows which parts are still empty.

Competence Developing Proficient (the bar) Strong
Framing
Look for: all eight parts of your card are filled in, each meets the rubric's words, and the boundary says what is left out.		 Question vague or untyped; no audience or counterfactual. Typed question with audience or decision-maker, metric with units, counterfactual, scale and boundary; one falsification statement. Also explains why the question matters to its audience and what answer would change a decision.

A checklist to reuse

  • One sentence: About one system and one period, ending in a question mark.
  • Typed: Descriptive, explanatory, evaluative, predictive or prescriptive.
  • Audience: Someone who would use the answer, and for what: to act on it, or to inform their own choices.
  • Metric with units: Something you could measure or compute, such as energy burden in % of income.
  • Counterfactual: What happens otherwise: the baseline you compare with.
  • More than a lookup: If a public dataset or settled physics already answers it, say what you add: a replication, a re-check or a new place.
  • Answerable: Data you can get show the effect, at the time step it needs.
  • Falsifiable: You can name a result that would show you wrong.

Myth or fact? “A question is worth researching only if nobody has written about it.”

  • Myth
  • Fact
Answer

Myth. Novelty can be a new question, new evidence, a new place or a new mix of methods, and it has to survive a literature search (R2).

Myth or fact? “A ‘what should … do?’ question needs an optimization model.”

  • Myth
  • Fact
Answer

Myth. Optimization is one way; decision analysis and multi-criteria methods also answer prescriptive questions (R3 compares the families).

Myth or fact? “If the data at hand cannot answer a question, the question is useless.”

  • Myth
  • Fact
Answer

Myth. Change the data, with meter readings or a survey, or narrow the question; a good question can outlive one dataset.

Myth or fact? “Naming the counterfactual is part of framing, not an extra.”

  • Myth
  • Fact
Answer

Fact. Every effect is measured against something; without a stated comparison, almost any result can be read as success.

Go deeper · ResearcherFrom a card to a proposal

In the research capstone (CR), your proposal opens with this card.

  • R10, the question studio, helps you generate candidate questions.
  • R2 tests what is already known.
  • R3 picks the system, the boundary and the method.
Go deeperWhat this studio leaves out
  • Sorting is a judgement: a few questions sit near a border, and a mentor may place them differently. The reasons matter more than the bins.
  • The studio checks the form of a question, not whether it matters. Whether it matters is a conversation with the people who would use the answer.
  • The twenty questions are illustrative and tidy; real ones arrive half-formed, and the same rewriting applies.

Sources and what’s next

Numbers quoted on this page

Number What it means Source
4.20 trillion kWh U.S. electricity use in 2025: a lookup, so “already answered”. EIA, Energy Explained · data for 2025; checked 2 November 2026
33.8% Capacity factor of the U.S. utility-scale wind fleet in 2025: also a lookup. EIA, Electric Power Monthly, Table 6.07.B · final 2025 values; checked 2 November 2026

Next

R10 · What could you study? The question studio

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.