Turn a research question into a testable hypothesis, written as an if-then prediction or a null-and-alternative pair for statistics and science research.
You are a research methods tutor who helps students and researchers turn a research question into a hypothesis precise enough to test, not a restated guess dressed up as one. My research question is [RESEARCH_QUESTION], in the [FIELD:select:General or Any Field,Psychology or Behavioral Sciences,Education,Business or Management,Health or Nursing Sciences,Social Sciences or Sociology,Biology or Life Sciences,Engineering or Computer Science,Humanities] field at the [ACADEMIC_LEVEL:select:middle or high school,undergraduate,graduate,doctoral or professional research] level, since a science fair project and a doctoral proposal both need a hypothesis but hold it to different standards of precision. If I already know my independent variable, what I'm changing or comparing, and my dependent variable, what I'm measuring, here they are: [VARIABLES?]. If I left that blank, work out the most sensible independent and dependent variables from my research question yourself and state them plainly before writing the hypothesis, so I can check your reasoning. Write the hypothesis in [FORMAT:select:if-then format,null and alternative hypothesis,both formats,not sure which I need] form. The if-then format states "if [independent variable changes], then [dependent variable will change], because [reasoning]," and is what most middle and high school science classes expect. The null and alternative pair states a null hypothesis, that no relationship or difference exists between the variables, and an alternative hypothesis, the specific relationship or difference I actually expect, in the formal notation statistics and social science courses expect. If I chose "not sure which I need," pick the format that matches my [ACADEMIC_LEVEL] and [FIELD], usually if-then below the college level or for a general science topic, null and alternative for statistics, psychology, or any study heading toward a significance test, explain in one sentence why you picked it, then give me that version. If I chose "both formats," give me both so I can see the same prediction in each register. Before finalizing anything, check the hypothesis against what actually makes one testable: the variables have to be specific enough to measure or observe, not just named in passing, and the prediction has to be falsifiable, meaning a real experiment or dataset could prove it wrong. A hypothesis like "social media affects mental health" fails both tests, since "affects" isn't measurable and nothing about the claim could turn out false. If my research question or variables are too vague to clear that bar, don't quietly write a vague hypothesis anyway. Tell me exactly what's missing, for example which mental health measure I mean, and rewrite the variables into something measurable before you produce the final hypothesis. Don't invent a specific numeric prediction, effect size, or cited study to make the hypothesis sound more rigorous than what I gave you supports. If a directional prediction, which way the effect will go, isn't clear from my input, default to the non-directional version and say so, rather than guessing a direction I never gave you.
Use this prompt anywhere
10,000+ expert prompts for ChatGPT, Claude, Gemini, and wherever you use AI.
Get Early AccessA hypothesis has to do two things most drafts miss: name variables specific enough to measure, and predict something an experiment could actually prove wrong. Restating your topic as a guess, like 'social media affects mental health,' isn't a hypothesis. It's an opinion with no way to fail.
This tool turns your [RESEARCH_QUESTION] into a hypothesis that clears that bar. Give it your independent and dependent variables if you already know them, or just describe what you're studying and it works the variables out for you before writing the prediction. Pick the [FORMAT] you need: the if-then structure most science classes expect, the null-and-alternative pair statistics and psychology courses use, or both if you want to see the same prediction in each register.
Before it hands back a hypothesis, it checks the prediction against what actually makes one testable: measurable variables, a specific claim, and a real chance of being proven wrong. If your research question is too vague to clear that bar, it tells you exactly what to sharpen instead of writing a vague hypothesis anyway.
Set your [FIELD] and [ACADEMIC_LEVEL] so the vocabulary matches a middle school science fair or a doctoral proposal. Run it in the Dock Editor to draft and revise your hypothesis, or paste the same input into ChatGPT, Claude, or Gemini. Once it holds up, the research design and methodology explainer helps you match it to a study that can actually test it.
Start in the Dock Editor or your assistant of choice (ChatGPT, Claude, Gemini), then fill in [RESEARCH_QUESTION] with what you're investigating. If you already know your independent variable (what changes) and dependent variable (what you measure), add them to [VARIABLES]. Leave it blank and the tool works them out for you.
Choose [FIELD] to match your discipline and [ACADEMIC_LEVEL] so the vocabulary fits a middle school lab report or a doctoral proposal. These two settings decide how technical the output sounds.
Set [FORMAT] to if-then for a school science hypothesis, null and alternative hypothesis for a statistics or social science study, both formats to compare them, or not sure which I need for a recommendation with a one-line reason.
Read how the output justifies each variable as measurable and each prediction as provable wrong. If it flagged something as too vague, sharpen that detail and run it again before you drop the final hypothesis into your paper or proposal.
Get an if-then hypothesis for a science fair project or class lab report, with the reasoning that ties the prediction to what you'll actually measure.
Set [FORMAT] to null and alternative hypothesis to see the pair written in the formal notation a first research methods course expects.
Turn a working [RESEARCH_QUESTION] into a hypothesis precise enough to anchor a proposal or thesis chapter, checked against testability before you commit to it.
Generate model hypotheses at different [ACADEMIC_LEVEL] settings to show students the line between a testable prediction and a vague guess.
Discover more prompts that could help with your workflow.
Analyze historical documents, letters, artifacts, and original texts using established historical methodology including OPVL framework, contextual analysis, and bias evaluation
Write compelling research grant proposals with proper structure for NIH, NSF, and foundation funding including specific aims, significance, innovation, approach, and budget justification
Create a properly formatted academic curriculum vitae for researchers, professors, and PhD students with comprehensive sections for publications, grants, teaching, and service
10,000+ expert-curated prompts for ChatGPT, Claude, Gemini, and wherever you use AI. Our extension helps any prompt deliver better results.