Research Hypothesis Examples: Null, Alternative and Directional, 20 Examples by Field (2026)
A hypothesis is the sentence in Chapter 1 that a committee will hold you to in Chapter 4. Every quantitative thesis needs at least one, most need several, and each must be written so that a specific statistical test can confirm or reject it. This guide gives you twenty research hypothesis examples, one per field, each shown as the full set a committee expects: the research question, the alternative hypothesis, and the null.
The examples follow the conventions in Creswell and Creswell’s Research Design (6th ed., SAGE, 2023), the text most US and international methods courses assign: quantitative studies state hypotheses or research questions, hypotheses are written in null and alternative form, and the alternative may be directional when theory or prior evidence predicts a direction. As Regent University’s dissertation guidelines put it, “Hypotheses are tested, while research questions are answered.”
The null hypothesis (H0) states that there is no effect, difference or relationship. The alternative hypothesis (H1) states that there is one. A directional alternative predicts the sign of the effect (“higher,” “lower,” “positively related”) and is tested one-tailed; a non-directional alternative predicts only that a difference exists and is tested two-tailed. Write the alternative first, then negate it precisely to get the null.
The Four Forms and When to Use Each
| Form | What it claims | Use it when | Signal words |
|---|---|---|---|
| Null (H0) | No difference, no relationship, no effect | Always; it is what the test evaluates | no significant difference, is not related to |
| Non-directional alternative | A difference or relationship exists, sign unspecified | Prior evidence is mixed or absent | differs from, is associated with |
| Directional alternative | A difference or relationship exists in a stated direction | Theory or consistent prior findings predict the direction | higher than, lower than, positively, negatively |
| Research question instead | Asks rather than predicts | Descriptive, exploratory or qualitative designs | to what extent, how do |
Two rules save most rewrites. First, the alternative hypothesis must name both variables and the population. Second, the null must be the exact negation of the alternative, not a vaguer version of it. If H1 says “higher,” H0 says “not higher” (equal or lower), which is why directional hypotheses are tested one-tailed.
Twenty Hypothesis Examples by Field
Each set gives the research question (RQ), the alternative (H1) and the null (H0). Directional sets are labelled; the rest are non-directional. Populations and settings are illustrative; replace them with your own.
1. Psychology (directional)
RQ: Does a brief online self-compassion intervention reduce dissertation anxiety in final-year undergraduates?
H1: Final-year students who complete a two-week self-compassion intervention will report lower dissertation anxiety at post-test than waitlist controls.
H0: Post-test dissertation anxiety will not be lower in the intervention group than in the waitlist group.
2. Nursing (directional)
RQ: Is longer emergency-department boarding associated with lower pressure-injury bundle adherence?
H1: Bundle adherence will be negatively associated with boarding duration among admitted patients boarded in the emergency department.
H0: Bundle adherence will not be negatively associated with boarding duration.
3. Education (non-directional)
RQ: Do Grade 8 mathematics scores differ between students taught with and without retrieval-practice quizzes?
H1: Mean end-of-unit mathematics scores will differ between classes using weekly retrieval-practice quizzes and classes using standard homework.
H0: Mean end-of-unit scores will not differ between the two conditions.
4. Business and Management (directional, moderation)
RQ: Does the manager’s own working pattern moderate the link between hybrid days and engagement?
H1: The positive relationship between employees’ weekly hybrid days and work engagement will be stronger when the line manager also works a hybrid pattern.
H0: The relationship between hybrid days and engagement will not differ by the line manager’s working pattern.
5. Marketing (directional)
RQ: Do sustainability labels increase purchase intention for mid-priced apparel?
H1: Purchase intention will be higher for apparel shown with a third-party sustainability label than for identical apparel shown without one.
H0: Purchase intention will not be higher for labelled apparel than for unlabelled apparel.
6. Economics (directional)
RQ: Are first-home-buyer grants capitalised into regional house prices?
H1: Median sale prices in regional local government areas will rise by more than in metropolitan areas following an increase in the grant threshold.
H0: The post-threshold change in median prices will not be greater in regional areas than in metropolitan areas.
7. Public Health (non-directional)
RQ: Does attendance after booking a health check differ by deprivation quintile?
H1: The proportion of booked health checks that are attended will differ between the most and least deprived quintiles.
H0: The attended proportion will not differ between the most and least deprived quintiles.
8. Sociology (non-directional)
RQ: Is neighbourhood tenure mix associated with residents’ reported social trust?
H1: Residents’ social-trust scores will be associated with the proportion of owner-occupied dwellings in their neighbourhood.
H0: Social-trust scores will not be associated with the proportion of owner-occupied dwellings.
9. Computer Science and HCI (directional)
RQ: Do LLM-generated unit tests detect fewer injected faults than developer-written tests?
H1: The mutation score of LLM-generated test suites will be lower than that of developer-written suites for the same repositories.
H0: The mutation score of LLM-generated suites will not be lower than that of developer-written suites.
10. Engineering (directional)
RQ: Does measured seasonal heat-pump performance fall below rated values in retrofitted houses?
H1: The measured seasonal coefficient of performance in retrofitted detached houses will be lower than the manufacturer-rated value used in the rebate model.
H0: The measured seasonal coefficient of performance will not be lower than the rated value.
11. Environmental Science (directional)
RQ: Does riparian buffer width predict stream macroinvertebrate diversity?
H1: Macroinvertebrate diversity will be positively related to riparian buffer width across sampled agricultural catchments.
H0: Macroinvertebrate diversity will not be positively related to buffer width.
12. Sports and Exercise Science (non-directional)
RQ: Do two warm-up protocols produce different sprint times in youth footballers?
H1: Mean 20-metre sprint time will differ between a dynamic warm-up and a combined dynamic-plus-potentiation warm-up.
H0: Mean sprint time will not differ between the two warm-up protocols.
13. Linguistics (directional)
RQ: Does captioned video improve listening comprehension for intermediate learners?
H1: Intermediate learners who watch captioned video will score higher on a listening comprehension test than learners who watch the same video uncaptioned.
H0: Comprehension scores will not be higher in the captioned condition.
14. Political Science (non-directional)
RQ: Is turnout in local elections associated with the timing of the election relative to national contests?
H1: Local-election turnout will be associated with whether the election is held concurrently with a national election.
H0: Local-election turnout will not be associated with concurrent scheduling.
15. Criminology (non-directional)
RQ: Does the use of out-of-court disposals for comparable offences differ between two police forces?
H1: The proportion of eligible cases receiving an out-of-court disposal will differ between the two forces.
H0: The proportion will not differ between the two forces.
16. Finance (directional)
RQ: Do firms with higher board independence show lower earnings management?
H1: Discretionary accruals will be negatively related to the proportion of independent directors among listed firms in the sample.
H0: Discretionary accruals will not be negatively related to board independence.
17. Biology (directional)
RQ: Does elevated water temperature reduce hatching success in a freshwater fish species?
H1: Hatching success will be lower in eggs incubated at the elevated temperature than in eggs incubated at the control temperature.
H0: Hatching success will not be lower at the elevated temperature.
18. Communication and Media (non-directional)
RQ: Does message framing change willingness to share a public-health post?
H1: Willingness to share will differ between gain-framed and loss-framed versions of the same post.
H0: Willingness to share will not differ between the two frames.
19. Social Work (directional)
RQ: Is peer-support participation associated with placement stability for kinship caregivers?
H1: Kinship placements in which the caregiver attends a peer-support group will show fewer disruptions over twelve months than placements without peer-support attendance.
H0: Placements with peer-support attendance will not show fewer disruptions.
20. Information Systems (directional, mediation)
RQ: Does perceived usefulness mediate the effect of training on intention to use a new records system?
H1: The positive effect of training hours on intention to use the system will be mediated by perceived usefulness.
H0: Perceived usefulness will not mediate the effect of training hours on intention to use.
Psychology students who need worked sets tied to named instruments (PHQ-9, GAD-7, BFI-2) should read our companion piece on research questions, hypotheses and variables for a psychology dissertation, which extends examples 1 and 4 into full proposal pages.
Write Your Hypotheses Where They Stay Connected to Your Methods
Tesify keeps your research questions, hypotheses and analysis plan in one structured workspace so Chapter 1 and Chapter 4 never drift apart. More than 15,000 chapters have been written with it, every word by the student.
Which Statistical Test Each Hypothesis Implies
A well-written hypothesis names its own test. The verb and the variables tell the reader whether you are comparing groups, relating variables, or testing a conditional effect.
| Hypothesis pattern | Examples above | Typical test |
|---|---|---|
| Two groups differ on a continuous outcome | 1, 3, 5, 12, 13, 17 | Independent-samples t-test (one- or two-tailed to match) |
| Two proportions differ | 7, 15 | Chi-square test of independence or two-proportion z-test |
| Two continuous variables are related | 2, 8, 11, 16 | Pearson or Spearman correlation; regression with covariates |
| A measured value is below a benchmark | 9, 10 | One-sample t-test or paired comparison, one-tailed |
| Change over time differs between groups | 6 | Difference-in-differences regression |
| Effect depends on a third variable | 4 | Moderated regression (interaction term) |
| Effect passes through a third variable | 20 | Mediation analysis with bootstrapped indirect effect |
| Event counts differ | 19 | Poisson or negative binomial regression |
If you are unsure which family your hypothesis belongs to, our quantitative research methods guide walks from design to test selection, and the sample size and power analysis guide shows how a directional hypothesis changes the power calculation.
Fill-In Template
RQ: Does [independent variable] [affect, predict, differ by] [dependent variable] among [population] in [setting]?
H1 (directional): [Population] who [level of IV] will show [higher, lower] [DV] than [population] who [other level of IV].
H1 (non-directional): [DV] will differ between [level 1 of IV] and [level 2 of IV] among [population].
H0: [Exact negation of the H1 you chose: “will not be higher,” or “will not differ.”]
Number hypotheses H1, H2, H3 in the order your research questions appear, and keep the numbering identical in the results chapter. The hypotheses grow out of the problem statement; if yours does not yet name a gap that a hypothesis could close, start with our problem statement examples by discipline.
Five Mistakes Committees Flag
- A null that is not the negation of the alternative. H1 “scores will be higher” paired with H0 “scores will differ” is logically incoherent. Negate precisely.
- A directional hypothesis with no justification. Choosing one-tailed testing halves the p-value threshold on one side; committees expect a theoretical or empirical reason stated in the literature review. Without one, use the non-directional form.
- Two claims in one hypothesis. “Training will increase usefulness and intention” is two hypotheses. Split them, or state a mediation hypothesis explicitly as in example 20.
- Variables that are not measurable as written. “Students will be more motivated” needs an instrument. Write the hypothesis in terms of the measured variable (a named scale score) or add an operational definition in Chapter 1.
- Treating a non-significant result as a failed thesis. Failing to reject the null is a finding. Report the effect size and confidence interval and interpret it, as explained in our guide on what to do when results are not significant and the companion on effect sizes and confidence intervals.
Frequently Asked Questions
Do I need a null hypothesis for every alternative hypothesis?
Conventions differ. Many US dissertation templates require both forms to be written out for each hypothesis. UK and Australian theses often state only the alternative and treat the null as implied. Follow your programme’s handbook, and if it is silent, write both: it costs one sentence and removes ambiguity.
When should I use a directional hypothesis?
When a theory or a consistent body of prior findings predicts the direction of the effect. If the literature is mixed, or the study is the first in its setting, a non-directional hypothesis with a two-tailed test is the safer and more defensible choice.
Can a qualitative thesis have hypotheses?
Qualitative designs use research questions rather than hypotheses, because they explore rather than test. Mixed-methods studies typically state hypotheses for the quantitative strand and questions for the qualitative strand.
How many hypotheses should a master’s thesis have?
Usually between two and five, one per research question. More than that suggests the study is trying to do too much, and each extra test raises the risk of a chance finding unless you correct for multiple comparisons.
Is a hypothesis the same as a prediction?
A research hypothesis is a testable statement about a relationship between variables; a prediction is the specific observable outcome you expect if the hypothesis is true. In a thesis the two are often written as one sentence, but the hypothesis must always name the variables.
Where do hypotheses go in the thesis?
In Chapter 1, after the problem statement and purpose and beside the research questions, and again at the start of the results chapter where each is tested in turn. Some programmes also restate them at the opening of the methodology chapter.
Next Step
Choose the example closest to your design, rewrite the alternative with your own variables and population, then negate it exactly to produce the null. If you want the hypotheses, the analysis plan and the results write-up to live in one place while you draft, start your thesis in Tesify.
Write your thesis with AI
Structure, draft, cite, and format your thesis faster with Tesify’s AI writing tools, automatic bibliography, and plagiarism checker. Free to start, no credit card required.
Leave a Reply