What Is a Dissertation? Meaning, US vs. UK Usage and Chapters
What is a dissertation? Get the definition, how US and UK usage differ from a thesis, real word limits and time to degree, and the standard chapters.
Read MoreLearning how to write a hypothesis comes down to one move: turn your research question into a specific, testable prediction about how two variables are related. A strong hypothesis names what you will change or compare (the independent variable), what you will measure (the dependent variable), and the result you expect, in words clear enough that data could prove it wrong.
This guide covers the definition, the if-then and null/alternative formats, research versus statistical hypotheses, and how to tell a testable statement from an untestable one. It ends with 25 hypothesis examples by field, a table of null and alternative pairs, and the mistakes instructors mark down most often.
A hypothesis is a proposed explanation for a narrow set of observations, based on what you already know from background reading, earlier studies, or your own preliminary observations. The University of California Museum of Paleontology's Understanding Science project defines it in almost exactly those terms. In practice, it is the answer you expect to your research question, written so that an experiment or study can check it.
Three terms get mixed up constantly:
One more point that changes how you write: a hypothesis is never proven. Data can support it or fail to support it. Even strong results leave room for a better explanation later, so write "the results support the hypothesis," not "the results prove the hypothesis."
Before you write a single sentence, identify these pieces:
Can't name the independent variable and the dependent variable yet? Then you're not ready to write the hypothesis. Go back to the question.
The steps below show how to write a hypothesis from scratch. One running example carries through all six, so you can watch the wording tighten as you go.
Students who ask how to formulate a hypothesis for a class project often skip step 2. The background reading is what turns a guess into an educated, defensible prediction, and it is what your instructor will look for in your introduction.
Your instructor or field usually expects one of these formats. Pick the one that matches the assignment, and use the examples under each as wording models.
The if-then format is standard in school science and lab courses. The "if" clause states the change or relationship you are testing, and the "then" clause states the measurable result you expect. Many instructors also want a "because" clause that gives the reason.
Not every if-then sentence is a hypothesis, though. A biology course handout from California State University, Bakersfield uses "If I play the lottery, then I will get rich" as a non-example: it is a wish dressed up as a prediction, with no tested relationship between variables.
Statistics courses, and most quantitative research papers, use a pair of statements instead of one sentence:
OpenStax's Introductory Statistics sets out exactly this rule, and notes that many researchers write the null with a plain = even when the alternative is one-sided. The statistical test then decides whether to reject the null hypothesis or fail to reject it. You don't "accept" the null hypothesis, because failing to find an effect isn't proof that no effect exists.
Examples for the caffeine study:
A directional hypothesis predicts which way the effect goes ("scores will be higher"). It matches a one-tailed test, where the alternative hypothesis uses < or >. A non-directional hypothesis predicts only that a difference exists ("scores will differ"). It matches a two-tailed test, where the alternative hypothesis uses ≠. Penn State's STAT 200 notes pair them the same way. Use a direction only when theory or earlier evidence justifies it, and decide before you collect data.
A research hypothesis is the plain-language prediction in your introduction. A statistical hypothesis is the same idea translated into a null and alternative pair about population parameters, ready for a specific test. Most papers need both.
| Research hypothesis | Statistical hypotheses | Likely test |
|---|---|---|
| Students who use flashcards with self-quizzing will score higher on a vocabulary test than students who reread the word list. | H0: μ quiz = μ reread; Ha: μ quiz > μ reread | Independent-samples t-test (one-tailed) |
| Daily screen time is related to hours of sleep among high school students. | H0: ρ = 0; Ha: ρ ≠ 0 | Pearson correlation |
| Preferred study location differs by class year. | H0: location and class year are independent; Ha: they are not independent | Chi-square test of independence |
| Mean fill weight of the cereal boxes differs from the labeled 500 g. | H0: μ = 500; Ha: μ ≠ 500 | One-sample t-test (two-tailed) |
The test listed is a typical choice, not the only one. Your sample size, data type, and whether assumptions such as normality hold will decide the final test.
A hypothesis is testable when you can measure the variables and a real result could contradict it. Philosophers of science call that second property falsifiability. The examples below show the difference. Untestable statements usually fail for one of four reasons: they involve values or opinions, they are too vague to measure, they can never be checked, or they explain every possible outcome.
| Untestable | Problem | Testable rewrite |
|---|---|---|
| Dogs are better pets than cats. | "Better" is an opinion. | Adults who adopt a dog will report more minutes of daily walking after three months than adults who adopt a cat. |
| Music affects plants. | Vague variables, no direction. | Bean seedlings exposed to 3 hours of classical music per day will not differ in height after 21 days from seedlings kept in silence. |
| Some people have a sixth sense that cannot be measured. | Cannot be observed by definition. | Participants will guess the color of a hidden card at a rate above chance (25% with four colors). |
| Students do worse when they are stressed or relaxed. | Every result fits. | Students who rate their pre-exam anxiety above 7 out of 10 will score lower on the exam than students who rate it 3 or below. |
| The new app will improve learning. | "Improve" and "learning" undefined. | Ninth graders who use the app for 15 minutes a day for four weeks will gain more points on a standardized algebra quiz than students who do not use it. |
These hypothesis examples follow the same rules: one independent variable, one measurable dependent variable, a clear population, and a direction when the background supports one. Use them as models for wording, not as finished answers for your own study.
Look at the wording in the social media example. That study is correlational, so the hypothesis says "associated with" rather than "causes." Use causal language only when you will manipulate the independent variable in an experiment.
Use this table when your assignment asks for both statements. Each alternative hypothesis is the claim you actually expect, and each null hypothesis is the "no effect" statement the test evaluates.
| Research question | Null hypothesis (H0) | Alternative hypothesis (Ha) |
|---|---|---|
| Does a new fertilizer change mean tomato yield? | Mean yield with the fertilizer equals mean yield without it. | Mean yield with the fertilizer differs from mean yield without it. |
| Does a tutoring program raise exam scores? | Tutored students' mean score is less than or equal to untutored students' mean score. | Tutored students' mean score is greater than untutored students' mean score. |
| Is study time related to GPA? | There is no correlation between weekly study hours and GPA (ρ = 0). | There is a correlation between weekly study hours and GPA (ρ ≠ 0). |
| Does a new drug lower systolic blood pressure? | The mean change in systolic pressure is zero or higher. | The mean change in systolic pressure is below zero. |
| Do men and women prefer different phone brands? | Brand preference is independent of gender. | Brand preference is not independent of gender. |
| Does a website redesign change the checkout rate? | The checkout rate is the same before and after the redesign. | The checkout rate differs before and after the redesign. |
If you know how to write a hypothesis but still get it marked down, the cause is usually one of these wording problems.
Knowing how to write a hypothesis includes knowing where it goes. In a lab report, the hypothesis usually closes the introduction, right after the background that justifies it, and the discussion section returns to it to say whether the results supported it. Our lab report format guide shows where each part sits. In a research paper, the hypothesis typically follows the literature review that motivates it. If you're still collecting that background, our literature review example shows how studies are grouped into themes that point toward a prediction, and our guide on how to write a research paper covers the full structure around it.
Don't confuse a hypothesis with a thesis statement. A thesis statement argues a position you defend with evidence and reasoning, as in a history or English essay. A hypothesis predicts a result you test with data. Our guide on how to write a thesis statement covers the argument-based version.
A simple example of a hypothesis is: "If radish seeds are watered with nitrogen fertilizer, then the seedlings will have greater dry mass after 14 days than seedlings given plain water." It names the independent variable (fertilizer), the dependent variable (dry mass), and an expected direction.
No. Evidence can support a hypothesis or contradict it, but it cannot prove it beyond any future challenge. In statistics, you either reject the null hypothesis or fail to reject it.
That description is common in middle school, but it undersells the idea. A hypothesis is a testable explanation grounded in background knowledge, and it must be specific enough that data could show it is wrong.
Usually one sentence, two at most. If yours runs longer, you probably have more than one prediction, and each should be stated separately.
No. Experiments and most quantitative studies use one. Exploratory, descriptive, and qualitative studies often use research questions instead, because they aim to describe or understand rather than test a prediction.
The hypothesis states the relationship or explanation ("light intensity affects photosynthesis rate"). The prediction states the specific result you expect in your setup ("more bubbles at 10 cm than at 40 cm").
Once you know how to write a hypothesis, the hard part is usually the rest of the project: designing the method, running the analysis, and writing up what the data show. If you want a second pair of eyes on a draft, a tutor from our research paper writing service can review your hypothesis, check that your variables are measurable, and suggest clearer wording before you submit.
Get affordable and top-notch help for your essays and homework services from our expert tutors. Ace your homework, boost your grades, and shine in online classes—all with just a click away!
Fast, secure, and handled by vetted experts.
What is a dissertation? Get the definition, how US and UK usage differ from a thesis, real word limits and time to degree, and the standard chapters.
Read More
Discover the implications of failing an Advanced Placement (AP) exam with this comprehensive guide. Learn about GPA impact, college credit loss, and how to address failure on college applications.
Read More
Explore ideas for an importance of education essay: what education means, the difference between formal and informal learning and how ancient cultures taught.
Read More