Hypothesis examples: 50 testable hypotheses by type and field

A hypothesis is a testable statement that predicts how two or more variables are related, written before you collect data. A simple example, number 21 below: “If tomato seedlings get 16 hours of light a day instead of 8, then they will be taller after three weeks.” A good hypothesis names what changes, what will be measured and who or what is studied, and some possible result would count against it.

The 50 examples below are original and grouped by field. Each is labeled with its type and its variables, so you can see how the parts fit before you write your own. They are predictions written to be tested, not findings: some echo questions researchers have studied with mixed results, so don't cite them as evidence. One study is worked through in full first; after it, start with the types if your assignment asks for a particular kind, or go straight to your field:

One study, from question to prediction

Here is one study written out before any data are collected, with each part labeled. The middle rows, the variable you change, the comparison and the measurement, are what every hypothesis in the list names.

Two distinctions help with the list. A hypothesis is the relationship you propose, with its reason: reminders reduce missed visits because patients forget. A prediction is what one particular test should show if the hypothesis is right: a lower no-show rate in the reminder group over these three months. Course assignments usually ask for the hypothesis in this testable, predictive form, and the examples below are written that way.

The second distinction is the design. This hypothesis is experimental: the clinic assigns the reminders, so a difference can be traced to them. A correlational hypothesis predicts that measured variables vary together, as in “patients who say they read their appointment texts miss fewer appointments.” Nobody assigns anything there, and something else, such as how organized a patient is, could explain both. The list labels the two causal and associative, the terms Barroga and Matanguihan use (see the types below); in methods courses you may meet the same split as experimental and correlational, after the designs that test them.

In a paper written in the IMRaD format, the hypothesis closes the introduction, and the methods section explains how the variables were measured (see this methodology section example). If you are still choosing a subject, start from the research paper topics and narrow from there.

Types of hypotheses

Hypothesis types come in pairs, and one hypothesis usually carries several labels at once: number 14 is simple, directional and causal. The definitions below follow the list in Edward Barroga and Glafera Janet Matanguihan's guide to writing research questions and hypotheses in the Journal of Korean Medical Science (2022).

The same article lists a few more labels you may meet. A working hypothesis is one you accept for now to guide further research. A statistical hypothesis restates the research hypothesis as a claim about population values, such as a mean or a proportion, so that a test can be run on it. In practice, the null and the alternative are the statistical pair: the alternative is usually your research hypothesis, and the null is what the test tries to reject.

Whether you can test a causal claim depends on the design as much as the wording. If you assign people to get eight hours of sleep or five, you can test it. If you only ask people how long they slept, the same sentence can be tested only as an association, because something else, such as stress, could explain both the sleep and the result.

The if–then format

The if–then form is the classic shape for an experimental hypothesis. The if clause holds the change you make, the independent variable. The then clause holds the result you expect to measure, the dependent variable. Adding a because clause states your rationale: the reason, from theory or earlier studies, that you expect this result rather than another.

If [the independent variable changes in this way],
then [the dependent variable will change in this direction],
because [the reason you expect it].

If–then suits experiments, where you control the if. In an observational study, where you can't, write the relationship instead: “Neighborhoods with more native plant cover have more bird species at feeders” or “Hours of paid work are negatively associated with hours of sleep.” Forcing an if–then onto data you only observed suggests a cause you can't show.

Psychology hypothesis examples

Behavior, memory, emotion and attention, measured with tasks, timers and standard scales.

1. If participants are told a puzzle measures intelligence, then they will give up on it sooner than participants told it is a game. Causal, directional. IV: how the puzzle is described; DV: minutes before giving up.
2. People who sleep less than six hours the night before a memory test will recall fewer words from a 30-word list than people who sleep eight hours or more. Simple, directional. Causal if sleep is assigned; associative if it is only reported.
3. Stress at the end of the workday will differ between commuters who drive and commuters who take the train. Non-directional, associative. Variables: how people commute, stress score.
4. The more often employees check work email in the evening, the lower they will score on a scale of psychological detachment from work. Associative, directional. Variables: evening email checks, detachment score.
5. Drivers in a simulator who hold a hands-free phone conversation will miss more road signs than drivers who listen to the radio. Causal, directional. IV: phone call or radio; DV: signs missed.
6. Regular exercise and weekly social contact will each reduce depressive symptoms in adults over 65, and together they will reduce them more than either alone. Complex, causal, directional. IVs: exercise, social contact; DV: depression score.
7. There is no difference in reaction time between participants who drink caffeinated coffee and participants who drink decaf 30 minutes before the test. Null. Variables: type of coffee, reaction time in milliseconds.

Education hypothesis examples

Teaching methods, schedules and support, measured in scores, attendance and enrollment.

8. If third graders read aloud to a trained therapy dog for 15 minutes twice a week, then their reading fluency will improve more over a semester than that of classmates who read aloud to an adult. Causal, directional. IV: who listens; DV: words read correctly per minute.
9. Students who get the grading rubric before they write will score higher on organization, but not on grammar, than students who first see it with their grade. Complex, causal. IV: when the rubric is shared; DVs: organization and grammar scores.
10. Class size is negatively associated with the number of questions each student asks per lesson. Associative, directional. Variables: class size, questions per student.
11. Students who handwrite lecture notes and students who type them will differ in their scores on conceptual questions a week later. Non-directional, causal if note-taking method is assigned. DV: conceptual score.
12. Schools that serve breakfast in the classroom will record fewer late arrivals than schools that serve it in the cafeteria before the bell. Associative, directional. Variables: where breakfast is served, late arrivals.
13. If high school seniors complete the federal student aid application in a session with a counselor, then a higher share will enroll in college the next fall than seniors who get an informational email. Causal, directional. IV: type of support; DV: fall enrollment.

Health and medicine hypotheses

Treatments, habits and exposures, measured in clinical readings, visits and outcomes. Numbers 18 and 19 are a null and an alternative for the same study.

14. If adults with high blood pressure walk briskly for 30 minutes five days a week for 12 weeks, then their systolic blood pressure will fall more than that of adults who keep their usual routine. Simple, causal, directional. IV: walking program; DV: systolic pressure.
15. Adults who live within 300 meters of a public park will report more minutes of physical activity a week than adults who live more than a kilometer from one. Associative, directional. Variables: distance to a park, weekly activity minutes.
16. A text-message reminder sent the day before an appointment will reduce missed appointments at a community clinic compared with no reminder. Causal, directional. IV: reminder or none; DV: missed appointments.
17. Age, smoking status and daily step count together predict lung function in adults over 50, and step count adds to the prediction after age and smoking are accounted for. Complex, associative. Variables: three predictors, lung function.
18. There is no difference in days absent with respiratory illness between classrooms with hand-sanitizer dispensers at the door and classrooms without them. Null. Variables: dispenser or none, days absent.
19. Classrooms with hand-sanitizer dispensers at the door will have fewer days absent with respiratory illness than classrooms without them. Alternative, directional, causal if dispensers are assigned by classroom.
20. Reported pain two days after knee surgery will differ between patients who receive a nerve block and patients who receive oral pain medication only. Non-directional, causal. IV: type of pain control; DV: pain score.

Biology and ecology hypotheses

Plants, animals and habitats, the home ground of the science fair. Most need nothing more than a ruler, a thermometer and patience.

21. If tomato seedlings get 16 hours of light a day instead of 8, then they will be taller after three weeks. Simple, causal, directional. IV: hours of light; DV: height in centimeters.
22. Bean plants watered with a 1% salt solution will have fewer leaves after four weeks than bean plants watered with tap water. Simple, causal, directional. IV: type of water; DV: number of leaves.
23. The number of earthworms per square meter will differ between soil under oak trees and soil under pine trees in the same park. Non-directional, associative. Variables: tree type, earthworm count.
24. The number of bird species seen at a feeder is positively associated with the amount of native plant cover within 50 meters of it. Associative, directional. Variables: plant cover, species count.
25. Water temperature and dissolved oxygen together predict the number of mayfly larvae per square meter of streambed, with more larvae in colder, better-oxygenated water. Complex, associative, directional. Variables: two predictors, larvae count.
26. Brine shrimp eggs will hatch faster in salt water kept at 28 °C than in salt water kept at 20 °C. Simple, causal, directional. IV: temperature; DV: hours to hatching.
27. If ladybugs are released on rose bushes infested with aphids, then the number of aphids per stem will fall faster than on bushes without ladybugs. Causal, directional. IV: ladybugs or none; DV: aphids per stem.

Physical science hypotheses

Chemistry, physics and earth science, where the variables are easy to control and the hypotheses can be very precise.

28. If a pendulum's string is made longer, then its swing will take longer, but changing the mass of the weight will not change the swing time. Complex, causal. IVs: string length, mass; DV: seconds per swing.
29. Ice cubes sprinkled with a teaspoon of salt will melt faster than ice cubes sprinkled with a teaspoon of sand. Simple, causal, directional. IV: salt or sand; DV: minutes to melt.
30. If the water is heated from 10 °C to 40 °C, then an antacid tablet dropped into it will dissolve faster. Simple, causal, directional. IV: water temperature; DV: seconds to dissolve.
31. Cardboard model houses with black roofs will reach higher inside temperatures under a heat lamp than identical models with white roofs. Simple, causal, directional. IV: roof color; DV: inside temperature.
32. The bounce height of tennis balls stored at 0 °C will differ from that of balls stored at 30 °C. Non-directional, causal. IV: storage temperature; DV: bounce height.
33. Streams whose watersheds have more paved surface will reach higher peak flows after storms of the same size. Associative, directional. Variables: share of paved surface, peak flow.

Business and economics hypotheses

Prices, offers and policies, often tested with A/B experiments or before-and-after data.

34. If an online store shows free shipping on the product page instead of only at checkout, then more visitors will complete a purchase. Causal, directional. IV: where shipping is shown; DV: purchase rate.
35. Customers who join a loyalty program will make more store visits in the next six months than similar customers who don't join. Associative, directional: customers choose whether to join, so the design can't show cause.
36. Price and the number of customer reviews together predict a product's weekly sales, and reviews matter more for products under $20. Complex, associative. Variables: price, review count, sales.
37. Employees moved to a four-day week at full pay will report different job satisfaction from employees who stay on a five-day week. Non-directional, causal if the schedule is assigned. DV: satisfaction score.
38. Restaurants that print calorie counts on their menus will sell fewer high-calorie desserts than similar restaurants that don't. Associative, directional. Variables: calorie labels, dessert sales.
39. There is no relationship between the share of a company's staff who work remotely and its revenue per employee. Null. Variables: share remote, revenue per employee.
40. When a city raises its downtown parking meter rates, the share of spaces occupied at noon will fall. Causal, directional. IV: meter rate; DV: occupancy at noon.

Social science hypotheses

Voting, trust and public life, usually studied with surveys, records and field experiments.

41. Registered voters who get a text on election morning with their polling place address will vote at a higher rate than voters who get no text. Causal, directional. IV: text or none; DV: turnout.
42. The number of events a neighborhood association holds each year is positively associated with residents' reported trust in their neighbors. Associative, directional. Variables: events per year, trust score.
43. Support for a four-day school week will differ between parents who work outside the home and parents who don't. Non-directional, associative. Variables: work status, support.
44. Age and daily hours on social media together predict how often people share news links they haven't opened, and the effect of hours is stronger among younger users. Complex, associative. Variables: age, hours, unread shares.
45. Cities that post council agendas a week before each meeting, rather than two days before, will receive more public comments per meeting. Associative, directional. Variables: notice period, comments per meeting.

Technology and computing hypotheses

Interfaces, performance and software teams, where experiments are cheap and samples large.

46. If a sign-up form asks for three fields instead of eight, then more of the visitors who start it will finish it. Simple, causal, directional. IV: number of fields; DV: completion rate.
47. Pages that load in under two seconds will have a lower bounce rate than pages that take more than five. Directional. Causal if you add the delay yourself; associative if you compare existing pages.
48. Code changes reviewed by two people will have fewer bugs reported in the month after release than changes reviewed by one. Associative, directional. Variables: number of reviewers, bugs reported.
49. After two weeks of practice, typing speed will differ between people using a split keyboard and people using a standard one. Non-directional, causal. IV: keyboard; DV: words per minute.
50. There is no difference in the share of messages misclassified by a spam filter retrained every week and one retrained every month. Null. Variables: retraining schedule, misclassification rate.

What makes a hypothesis testable

The philosopher Karl Popper made falsifiability the line between science and everything else in Logik der Forschung (1934), published in English as The Logic of Scientific Discovery (1959): a claim is scientific if some possible observation could contradict it. A hypothesis that every result would confirm tells you nothing. Popper also held that a test that goes your way corroborates a hypothesis but never proves it, which is why careful writers say the results “support” a hypothesis rather than prove it.

The last item is a matter of honesty as much as method. Barroga and Matanguihan advise developing the questions and hypotheses before the study begins. Writing the hypothesis after you have seen the results and presenting it as if it came first has a name: HARKing, hypothesizing after the results are known, from Norbert Kerr's 1998 article in Personality and Social Psychology Review.

If you draft in Strokes, version history keeps earlier versions of the document on your device, so you can check how each hypothesis read before the data came in. It is a private record, not a public preregistration.

Hypothesis, prediction and theory

The worked study above separated the hypothesis from the prediction; theory is a third, much larger idea. UC Berkeley's Understanding Science glossary defines a hypothesis as “a proposed explanation for a fairly narrow set of phenomena.” A prediction is what you would expect to observe in a particular test if the idea were right. A theory explains far more: many related phenomena at once, usually by tying together many hypotheses that have held up under testing.

A thesis statement sits at the other end of a project: it is the claim you argue once the evidence is in. The guide on how to write a thesis statement covers that step.

Common hypothesis mistakes

• Writing a question. “Does music help plants grow?” is a research question. The hypothesis is your answer to it, stated in advance.
• Words you can't measure. Better, affects, improves, healthy and successful all need replacing with the thing you will count, time or score.
• Causal words for an observational design. “Causes,” “leads to” and “improves” promise an experiment. If you only observe, write “is associated with” or “predicts.”
• Two predictions with no plan for either. A complex hypothesis is fine when the design tests each part. Two unrelated predictions in one sentence are two hypotheses; number them H1 and H2.
• Promising proof. A study can support a hypothesis or contradict it. Write “the results support the hypothesis,” never “prove it.”
• Hedging until nothing can fail. “Light may affect growth in some plants” is true whatever happens, so it can't be tested. Commit to a direction or to a difference.

A hypothesis that isn't supported is still a result. The guide on writing a research paper conclusion shows how to report one plainly.