Four anatomical model brains arranged in a cross on an orange background
Photo: Aakash Dhage / Unsplash

Build the sentence from the measurement backwards

Start by finding the dependent variable, then make it measurable. Words such as attention, memory, stress or aggression name concepts, but they do not tell a researcher what to record. A measured outcome might be the number of target symbols correctly crossed out from a sheet of 24, the score on a named scale, or the time taken in seconds to complete a task. The how to write a hypothesis guide calls this operationalising the DV.

Next identify the independent variable and write both conditions. Suppose earlier evidence suggests that a short rhythmic movement improves visual attention. The conditions could be completing four minutes of paced finger tapping and sitting quietly for four minutes. Writing only the tapping condition leaves the comparison unfinished.

Now join the pieces into a directional prediction: 'Participants who complete four minutes of paced finger tapping will correctly cross out more target symbols from a sheet of 24 than participants who sit quietly for four minutes.' The sentence predicts a direction, states both conditions and gives an exact measure. It can be tested without the marker having to invent any missing detail.

Choose directional, non-directional or null wording

A directional hypothesis says which condition will produce the higher, lower, faster or slower result. Use it when the question asks for one or when the information supplied gives a reason to predict the direction. Words such as more, fewer, higher, lower and faster make the direction visible, but they still need to be attached to operationalised variables.

A non-directional hypothesis predicts a difference without saying which condition will produce which result. For a study with no earlier evidence, a suitable sentence could be: 'There will be a difference in the number of puzzle pieces correctly placed during a six-minute task between participants given a colour-coded diagram and participants given a black-and-white diagram.' Both conditions and the DV are still precise, even though the direction is open.

A null hypothesis predicts no difference between the conditions, with any observed difference attributed to chance. Using the tapping study, it would state that there will be no difference in the number of target symbols correctly crossed out from a sheet of 24 between participants who tap for four minutes and participants who sit quietly for four minutes. Keep all the operational detail when changing the prediction to no difference.

The wider research methods guide places hypotheses alongside variables, experimental design and test selection. That context matters because directional wording leads to a one-tailed test and non-directional wording to a two-tailed test, while a null hypothesis gives the statistical test a precise no-difference claim to assess.

Correlational hypotheses predict a relationship

A correlation has no manipulated IV and no experimental conditions. It has two measured co-variables, so the hypothesis must predict a relationship between them. Writing that one group will differ from another changes the study into a design it does not have.

For example: 'There will be a positive correlation between the number of times a student checks their phone during a 30-minute homework period and the number of arithmetic errors they make on a 25-item exercise.' Both co-variables are operationalised and the direction is positive. A non-directional version would predict a relationship without using positive or negative.

A correlational null hypothesis predicts no relationship between the two operationalised co-variables, with any relationship found attributed to chance. Avoid causal verbs such as causes, improves or reduces, because measuring two things together does not show that one produced the other.

Try repairing a hypothesis for this plainly fictional correlational study. Each participant reports the number of days in the previous 14 days on which they practised a tile-matching puzzle for at least ten minutes. Each then completes the same 30-pair tile-matching task. The researcher records the time in seconds from the first tile being turned over until all 30 pairs have been matched. No practice is assigned. The task asks for a directional correlational hypothesis, with the expected direction supplied as more qualifying practice days being associated with shorter completion times. This is an imagined prediction; no study results are supplied.

The first attempt is: 'Participants given daily tile-puzzle practice will make fewer matching errors than participants given no practice.' The word 'given' turns measured practice into an assigned condition and invents two groups. The researcher instead records two co-variables in the same people. The attempt also substitutes an error count for the recorded completion time. Neither adding a comparison group nor replacing seconds with errors repairs this study's hypothesis.

Keep both measures as supplied: the reported number of days in the previous 14 days with at least ten minutes of tile-puzzle practice, and the time in seconds from the first tile turn until all 30 pairs are matched. A shorter completion time is a lower number of seconds. The expected pattern therefore pairs a higher day count with a lower time, so the direction is negative. Calling it positive because more practice is expected to go with better performance would confuse performance with the actual time measure.

The complete directional prediction is: 'There will be a negative correlation between the reported number of days in the previous 14 days on which a participant practised a tile-matching puzzle for at least ten minutes and the time in seconds from the first tile being turned over until all 30 pairs have been matched on the same tile-matching task.' This predicts a relationship between the two operationalised co-variables. It does not claim that practice causes a shorter completion time.

Use the 3, 2, 1, 0 marking ladder

This ladder belongs to Revision Library's hypothesis practice exercise. The numerical examples below illustrate that local scheme, not a universal exam-board marking rule. For an exam question, use its own instructions and mark scheme.

For the site's three-mark hypothesis practice, 3 marks means the response is the right type, is written as a prediction, names both IV conditions, and operationalises the DV. A 2-mark response keeps the prediction and both conditions but leaves one variable vague, such as saying participants will 'perform better' without stating the score, count or time recorded.

A 1-mark response identifies the variables but operationalises neither. 'The type of activity before the task will affect visual attention' points to an IV and a DV, yet gives no conditions and no measurement. It cannot be repeated precisely from that sentence.

A response falls to 0 when it is an aim, a question, the wrong type of hypothesis for the instruction, or a prediction that names only one IV condition. This is why a detailed sentence beginning 'To investigate whether...' still does not score as a hypothesis: it states a purpose instead of a predicted result.

Use the ladder as a final audit. Underline the prediction type, box the first condition, box the comparison condition, and circle the DV measure. If one of those parts cannot be marked on the page, revise the sentence before moving on.

Repair an incomplete hypothesis step by step

Consider this hypothetical experiment. Participants are allocated to read the same instructions for assembling a 12-piece model either as a numbered list of steps or as one continuous paragraph. Each participant then has five minutes to assemble the model. The researcher records the number of pieces placed in the wrong position, out of 12, at the end of the five minutes. The task asks for a directional hypothesis predicting fewer errors after reading the numbered steps; this is an imagined prediction, not a finding from real research.

The first attempt reads: 'Participants who read numbered steps will make fewer errors.' It already predicts a direction, but fewer than whom? The paragraph condition is missing. 'Errors' also leaves the DV incomplete: it could mean misplaced pieces, skipped steps or the number of attempts needed. The scenario supplies a particular count and time point, but the sentence does not state them. Under this site's practice ladder, the attempt receives 0 because it names only one IV condition, even though the intended direction is clear.

First restore the comparison: 'Participants who read the assembly instructions as numbered steps will make fewer errors than participants who read the same instructions as one continuous paragraph.' That fixes the omitted condition. It would receive 2 in this site's exercise because it now predicts a difference between both named conditions but still leaves the error measure vague. Adding the comparison alone has not completed the repair.

Now replace 'make fewer errors' with the recorded outcome. The complete prediction is: 'Participants who read the assembly instructions as numbered steps will have fewer of the 12 model pieces placed in the wrong position at the end of five minutes than participants who read the same instructions as one continuous paragraph.' It receives 3 under this site's exercise scheme: the requested directional prediction is present, both instruction conditions are named, and the DV states exactly what is counted and when.

Keep the measure tied to this study. Replacing the count with 'finish faster' would introduce completion time, which the scenario does not say is recorded. Here a lower count means fewer misplaced pieces; the five minutes specifies when the count is taken. Repair the missing details using the supplied procedure rather than inventing a different outcome.

Our own resource

Revision Library

Best for free printable practice with mark schemes

Everything above can be practised with the step-by-step hypothesis guide and its printable worksheet. The guide separates aims from predictions, shows how to select directional, non-directional and null wording, and makes the marking ladder explicit.

The new pack sits in the A Level Psychology library with a full mark scheme. Use the Psychology study tools to move between research methods and the rest of the course, or take the Psychology diagnostic first if the weak point is unclear.

After short hypothesis practice, the A Level Psychology mock papers put research methods back into longer mixed work. That progression helps because writing one clean prediction is a small skill, but it must still be recognised and applied when a paper changes the study context.

Practise writing hypotheses

In the numbers

packs
102
question pages
388

A Level Psychology worksheet packs with full mark schemes

How to choose

Use a short cycle for this skill: identify the study type, write one sentence, then mark it against the site's practice ladder. The printable Psychology practice guide explains where this kind of written correction fits beside essays and application questions.

If several parts of the subject need attention, the A Level Psychology revision method separates research methods from study recall and evaluation. The Psychology revision resources guide is for choosing a broader set of materials, not for learning this single sentence structure.

Move to full papers after you can produce the right hypothesis type without prompts. The A Level Psychology past-paper guide explains how to match papers to the student's board before using a marked paper to choose follow-up practice.

Finish each attempt by reading only what you actually wrote. Do not give yourself credit for a condition or measure that was present in the scenario but never made it into your hypothesis.

Frequently asked questions

Can a hypothesis begin with 'To investigate'?
No. That opening normally produces an aim because it states the purpose of the study. A hypothesis states the result the researcher predicts, using a difference, a relationship or no difference.
Do I always need a directional hypothesis?
No. Use the type requested by the question. A directional hypothesis predicts which way the result will go, while a non-directional one predicts a difference or relationship without choosing a direction.
Why must I name both IV conditions?
A difference needs a complete comparison. Naming only one condition leaves the reader unable to tell what that condition is being compared with. In this site's hypothesis practice exercise, that omission receives no marks under the local ladder; an exam question has its own mark scheme. The research methods material shows how the two conditions fit into the wider plan of a study.
What should a null hypothesis include?
It should predict no difference between the two named conditions, or no relationship between two measured co-variables. Retain the operational detail and state that any result found is due to chance.

Revision Library resources in this article

How we wrote this

A student request prompted the new hypothesis worksheet and guide published on 23 September 2026. The method below was checked against that guide, the site's broader research-methods guide, and all of the worksheet's marking guidance. The examples here are new and are not copied from either source.