What Is a Research Hypothesis? Definition, Types, Examples and How to Write One

A research hypothesis is a specific, testable statement that predicts the relationship between two or more variables before data is collected. It names an independent variable and a dependent variable, is grounded in theory or prior evidence, and must be falsifiable. For example: customers who receive same-day support replies will give higher satisfaction scores than those who wait longer.

A hypothesis turns a loose curiosity into something you can check. This guide explains what a research hypothesis is, the types of hypothesis, how to write one and how it gets tested, with examples from psychology, marketing, health and consumer research.

What Is a Research Hypothesis?

To define hypothesis in research terms: it is a precise, testable prediction about the relationship between variables, made before data collection. That research hypothesis definition applies to qualitative and quantitative work alike, and to everything from lab experiments to consumer insights studies.

A hypothesis starts with a question the researcher wants to answer. It states the expected answer, which tells you what to measure and how to design the study. This fits designs that test relationships. A descriptive research design, by contrast, measures what is happening without predicting why.

Hypotheses do not appear from nowhere. They usually come from one of four places:

• Theory: an established explanation that predicts a specific outcome

• Prior research: earlier findings you want to extend or challenge

• Observation: a pattern you noticed in real behavior

• Exploratory findings: early interviews or surveys that suggest a relationship worth testing

Independent and Dependent Variables in a Hypothesis

Every hypothesis in research rests on variables, which are traits or conditions that can change.

• Independent variable (IV): the factor that is manipulated or controlled to see its effect.

• Dependent variable (DV): the factor that is measured to assess the impact of the independent variable.

• Control variables: factors you hold constant so they do not distort the result.

• Confounding variables: outside factors that influence both the IV and the DV and can create a false link.

Here is a short example. Suppose you test whether a new onboarding tutorial raises feature adoption. The tutorial is the IV, and feature adoption is the DV. User experience level is a control variable, so you compare only new users. Seasonality is a possible confounder, because a holiday sales spike could raise adoption regardless of the tutorial.

Characteristics of a Good Research Hypothesis

The characteristics of a good hypothesis are easy to check. A strong, testable hypothesis has these traits:

• Testable: you can collect data that bears on it.

• Falsifiable: a result could prove it wrong.

• Specific: it names the variables and the expected direction or relationship.

• Grounded: it builds on theory or evidence, not a guess.

• Clearly defined variables: each can be observed and measured.

• Feasible: you have the time, sample and tools to study it.

Here is a quick pass-and-fail check.

Why Hypotheses Matter in Research

The importance of hypothesis in research is practical. A hypothesis forces clarity before you spend time and money on data. It also supports the shift Gartner's 2025 data and analytics trends describe, from a data-driven to a decision-centric approach, because a hypothesis ties every measurement to a decision it should inform.

A hypothesis does seven jobs:

• Guides the research process. It sets direction and shapes your questions and methods.

• Defines the variables. It specifies what you manipulate and what you measure.

• Enables testability. It gives you a statement that data can support or contradict.

• Reduces bias. A specific prediction pushes you to rely on evidence, not opinion.

• Promotes critical thinking. Writing one requires a literature review, which is just as true when you follow qualitative research best practices.

• Structures analysis. It tells you which tests to run and how to read the results.

• Drives progress. Supported or refuted, a tested hypothesis creates the next question.

Types of Research Hypotheses

The classification of hypothesis in research varies by textbook, and the labels overlap. The easiest way to remember the types of research hypothesis is as pairs, and each pair describes a different feature of the same statement. These are the hypothesis types in research methodology you will meet most often.

You may also see the "7 types of hypothesis" list: simple, complex, directional, non-directional, null, alternative and causal. The pairs below cover all of these, plus associative, empirical, statistical and logical hypotheses.

Simple vs Complex Hypothesis

A simple hypothesis links two variables. A complex hypothesis links more than two.

• Simple: "Increased exercise leads to improved cardiovascular health."

• Complex: "The interaction between genetic predisposition, diet and exercise influences longevity."

Directional vs Non-Directional Hypothesis

A directional hypothesis predicts which way the effect goes. A non-directional hypothesis predicts only that a relationship exists. The choice affects the statistical test, because directional hypotheses use a one-tailed test and non-directional hypotheses use a two-tailed test.

• Directional hypothesis example: "Higher levels of education lead to higher income levels."

• Non-directional hypothesis example: "There is a relationship between social media use and levels of anxiety."

Associative vs Causal Hypothesis

An associative hypothesis says two variables are related without claiming one causes the other. A causal hypothesis claims that a change in one variable produces a change in another.

• Associative: "There is a correlation between income level and access to healthcare services."

• Causal: "Increased consumption of sugary drinks causes an increase in body weight."

Causal hypotheses usually require an experimental design with random assignment. Otherwise, a confounding variable can mimic the effect, which is why it helps to understand correlation vs causation before you word the claim.

Empirical, Statistical and Logical Hypotheses

These types describe how the hypothesis is used.

• Empirical (working) hypothesis: a prediction tested through observation or experiment. The term connects to empirical research, which relies on direct evidence. Example: "Employees who work remotely two days a week will report higher job satisfaction."

• Statistical hypothesis: a claim about a population parameter, written so a statistical test can evaluate it. Example: "The mean satisfaction score of group A is greater than that of group B."

• Logical hypothesis: a proposed explanation supported by reasoning but not yet testable with data. Example: "If memory capacity is limited, long onboarding forms will lower completion."

Null Hypothesis vs Alternative Hypothesis

The null hypothesis (H0) states that there is no significant relationship or difference between variables. It proposes that any observed effect is due to chance or sampling error. The alternative hypothesis (H1) states that a relationship or effect exists. In null hypothesis vs research hypothesis terms, the research hypothesis is usually the alternative hypothesis.

Here is a null hypothesis example and an alternative hypothesis example for each field.

You never "accept" the null hypothesis. When the data are unlikely under the null, you reject it. When they are not, you fail to reject it. In plain language, "reject" means the evidence points against no effect. "Fail to reject" means the evidence was not strong enough to rule out no effect, which is not the same as proving there is none.

Research Hypothesis Examples

Here is an example of hypothesis in research for each field below, with the IV and DV labeled.

Psychology

• Hypothesis: Individuals who practice mindfulness meditation daily will report lower stress levels than those who do not.

• IV: Mindfulness meditation practice. DV: Stress levels.

Marketing

• Hypothesis: Consumers exposed to advertisements with emotional appeals will have higher purchase intent than those exposed to rational appeals.

• IV: Type of advertisement appeal (emotional vs rational). DV: Purchase intent.

Education

• Hypothesis: Students who receive personalized math tutoring will perform better on standardized tests than those who do not.

• IV: Personalized tutoring. DV: Test performance.

Technology

• Hypothesis: Users who receive personalized recommendations on a streaming platform will spend more time watching content than users who do not.

• IV: Personalized recommendations. DV: Time spent watching.

Healthcare

• Hypothesis: Patients who receive appointment reminders by text will miss fewer appointments than patients who receive none.

• IV: Text reminders. DV: Missed appointment rate.

Environmental science

• Hypothesis: Streams with riverside tree cover will have lower summer water temperatures than streams without it.

• IV: Riverside tree cover. DV: Summer water temperature.

Customer experience

• Hypothesis: Customers who receive same-day support replies will give higher satisfaction scores than those who wait longer.

• IV: Reply time. DV: Satisfaction score.

Weak vs Strong Hypothesis Rewrites

Rewriting a hypothesis is often the fastest way to improve it.

Hypothesis Examples in Consumer and UX Research

A hypothesis in market research looks the same as any other, and consumer and UX teams write them constantly, often without calling them that. Here are four examples with a way to test each.

• Ad testing: "Ads that open with a human face will hold attention longer in the first three seconds than ads that open with a product shot." Test it with a concept test or an ad test, as covered in this guide to behavioral research for ad testing.

• Packaging: "Shoppers will choose the redesigned pack over the current pack more often when both are shown side by side." Test it with a monadic or paired concept test.

• Pricing: "Raising the price by 8% will not lower purchase likelihood among loyal customers." Test it with a survey experiment or dedicated price testing.

• App onboarding: "Showing a three-step checklist will raise first-week activation compared with no checklist." Test it with an A/B test.

How to Write a Research Hypothesis

Knowing how to write a hypothesis comes down to a repeatable sequence. These steps also answer how to write a hypothesis in a research paper.

Step 1: Start with a research question. Define the problem you want to investigate. A good starting point is an effective research question that is specific enough to answer with data.

Step 2: Review the literature. Learn what is already known so your hypothesis builds on evidence and fills a real gap.

Step 3: Identify your variables. Decide on the independent and dependent variables and how you will measure each.

Step 4: Write a testable statement. Turn the question into a clear prediction that names the variables and the expected relationship.

Step 5: Make sure it is falsifiable. Ask what result would prove you wrong. If nothing could, rewrite it.

Step 6: State the null and alternative versions and plan the test. Write H0 and H1, then decide which design, sample and statistical test you will use before you collect data.

Here is the customer satisfaction (CSAT) example run through the steps:

• Question: Does reply speed affect customer satisfaction?

• Variables: IV is reply time, DV is CSAT score.

• Hypothesis: Customers who receive same-day support replies will give higher CSAT scores than those who wait longer.

• H0: Reply time has no effect on CSAT scores.

• Plan: Compare CSAT between customers who get same-day replies and those who wait more than 24 hours.

Hypothesis Format and Templates

The research hypothesis format you choose depends on your field and test. These three templates cover most cases. Here is one research question converted into each.

Research question: Does offering free shipping increase checkout completion?

• If-then: If a store offers free shipping, then checkout completion will be higher than when it does not.

• Significant difference: There will be a significant difference in checkout completion between shoppers offered free shipping and shoppers not offered it.

• H0 and H1 notation: H0: Free shipping has no effect on checkout completion. H1: Free shipping increases checkout completion.

To learn how to write a hypothesis statement, fill in the blanks of a template and then check it against the characteristics above.

Common Mistakes When Writing a Hypothesis

• Vague variables. "Better experience" cannot be measured. Define what better means.

• Untestable claims. If no data could contradict it, it is not a hypothesis.

• Multiple predictions in one statement. Split them so each can be tested on its own.

• Writing a question instead of a statement. "Does price affect sales?" is a research question. "Higher prices will reduce unit sales" is a hypothesis.

Hypothesis vs Research Question vs Prediction vs Theory

These terms are related but not interchangeable. Here they are applied to one topic: sleep and exam performance.

The hypothesis vs prediction question comes down to scope, and hypothesis vs theory comes down to breadth. In short, a theory explains, a hypothesis proposes a testable link, a prediction states the expected result of one test, and a research question asks what you want to find out.

How a Research Hypothesis Is Tested

What is hypothesis testing? It is the process of using data to decide whether the evidence supports the hypothesis. Teams that run this at scale often field studies through consumer research platforms, but the logic is the same on any tool:

1. Choose a design. Pick an experiment, quasi-experiment or survey suited to the claim.

2. Set a significance level. This is the threshold for how unlikely the data must be under the null before you reject it. A common choice is 0.05.

3. Collect data. Gather it from a sample that fits your question.

4. Run the test. Choose a test that matches your data and your hypothesis type, using a one-tailed or two-tailed test as appropriate.

5. Interpret the p-value. The p-value shows how likely a result at least this extreme would be if the null were true. A small p-value is evidence against the null, not proof of your hypothesis.

For causal claims, experimental research with random assignment is the strongest design. If you cannot randomize, a quasi-experiment can still give useful evidence, but you should state its limits plainly.

For early-stage ideas, concept testing lets you test how an audience reacts before you build anything.

Digital teams apply the same logic daily. The control is the current experience and the variant is the change, so the comparison tests your hypothesis directly. In practice, A/B testing works best when each test carries one clear hypothesis.

Hypotheses are supported, not proven. The Open Science Collaboration replicated 100 psychology studies and found that 97% of the original studies reported significant results, but only 36% of the replications did. A single significant result is therefore a starting point, not a final answer.

The pattern holds beyond one field. In a Nature survey of 1,576 researchers, more than 70% said they had tried and failed to reproduce another scientist's experiments. The practical lesson is to treat a supported hypothesis as provisional, replicate important findings and report how you tested them.

Do Qualitative Studies Need a Hypothesis?

Often they do not. Qualitative and exploratory studies usually generate hypotheses rather than test them. Instead of predicting a relationship, they aim to understand experiences, language and motivations. That is why exploratory research typically comes first, and the hypotheses of the study in research come out of what it finds.

Here is how that handoff works. Interviews and open-ended responses are coded, and thematic analysis groups the codes into themes. A theme such as "billing surprises erode trust" becomes a testable hypothesis for a quantitative follow-up: customers who receive a renewal reminder will cancel less often than those who do not. That is a good example of a qualitative research hypothesis being built from the data.

The qualitative phase does not have to be large. A systematic review of 23 studies by Hennink and Kaiser found that saturation was reached within 9 to 17 interviews or 4 to 8 focus groups for relatively homogeneous samples. That is a manageable amount of fieldwork to produce hypotheses worth testing.

The same pattern appears in earlier work. In a study of 60 interviews, Guest, Bunce and Johnson found that saturation occurred within the first twelve interviews. Small, focused qualitative studies can therefore feed a hypothesis pipeline quickly.

Testing Consumer Hypotheses With Behavioral Data

Many consumer hypotheses depend on what people feel, not only on what they say. Take the hypothesis that emotionally framed ads increase purchase intent. A survey can ask people how likely they are to buy, but stated intent is a weak signal. A stronger test measures emotional and attention responses while people watch the ad, then compares them across versions.

That is the approach behind a modern consumer insights platform. It lets teams measure responses directly and run the same test across audiences.

Decode supports this kind of testing in three ways:

• Emotional response: facial coding at 90%+ accuracy, built on 62 facial expressions, shows how viewers react moment by moment.

• Attention: eye tracking at 96% accuracy shows where viewers look and for how long.

• Multi-market testing: with 70+ languages supported, the same hypothesis can be tested across markets with one study design.

The platform is backed by 17 patents and trusted by 150+ global brands. Decode by Entropik is built so the evidence behind a hypothesis is measured, not assumed.

Frequently Asked Questions

1. What is a research hypothesis?

A research hypothesis is a specific, testable statement that predicts the relationship between two or more variables before data is collected. It names an independent and a dependent variable and must be falsifiable.

2. What is the null hypothesis?

The null hypothesis (H0) states that there is no significant relationship or difference between the variables. Researchers test against it and either reject it or fail to reject it.

3. What makes a good research hypothesis?

A good hypothesis is testable, falsifiable, specific, grounded in theory or evidence, and built on clearly defined variables. It should also be feasible to study with the time and resources you have.

4. Can a hypothesis be proven true?

No. A hypothesis can be supported by evidence or rejected, but it is not proven. New data could always contradict it, which is why replication matters.

5. What are the 7 types of hypothesis?

The commonly listed seven are simple, complex, directional, non-directional, null, alternative and causal. Associative, empirical, statistical and logical hypotheses are also used.

6. What is an example of a research hypothesis?

"Customers who receive same-day support replies will give higher satisfaction scores than those who wait longer." The independent variable is reply time, and the dependent variable is satisfaction score.

7. What is the difference between a hypothesis and a research question?

A research question asks what you want to find out. A hypothesis is a testable statement predicting the answer. For instance, "Does reply speed affect satisfaction?" becomes "Faster replies will raise satisfaction."

8. Does qualitative research need a hypothesis?

Usually not. Qualitative and exploratory studies tend to generate hypotheses rather than test them. Findings from interviews and thematic analysis can become hypotheses for later quantitative testing.

Related Topic:

• Responsible and Ethical Research Practices: A Guide

• Saturation in Qualitative Research: What It Is and How to Reach It

• Selection Bias in Research: What It Is and How to Reduce It

• Subjective vs Objective Research: A Clear Comparison

• The False Consensus Effect in Research and How to Manage It