You are viewing a free preview of this lesson.
Subscribe to unlock all 10 lessons in this course and every other course on LearningBro.
Before a psychologist collects a single piece of data, the study has already been won or lost on paper. A muddled aim, a hypothesis that cannot be tested, a variable that has not been pinned down precisely — any of these dooms a study before it begins. This lesson is about the planning skeleton of research: how a general curiosity becomes a testable prediction, and how the things we are interested in become measurable variables.
You will learn the difference between an aim and a research question, how to write null and alternative hypotheses, when a hypothesis should be one-tailed (directional) or two-tailed (non-directional), and how to define the independent and dependent variables through operationalisation. Finally we deal with the variables researchers do not want — extraneous and confounding variables — and how they are controlled. These skills are examined directly in Component 01 (often "write a suitable hypothesis for this study" or "operationalise the DV") and they underpin your own Practical Activities.
| This lesson covers | OCR H567 Component 01 sub-area | AO focus |
|---|---|---|
| Aims vs research questions | 1.2 Planning & conducting — aims; research question | AO1 |
| Null and alternative hypotheses | 1.2 — hypotheses | AO1; AO2 writing a hypothesis for a scenario |
| One-tailed (directional) vs two-tailed (non-directional) | 1.2 — directional/non-directional | AO1; AO2 |
| IV, DV and operationalisation | 1.2 — IV/DV; operationalisation | AO1; AO2 |
| Extraneous and confounding variables and their control | 1.2 — control of extraneous variables | AO1; AO3 evaluation |
Referenced descriptively; see the official OCR H567 specification document for exact wording. This lesson develops AO1 (defining aims, hypotheses and variables), AO2 (writing an operationalised, correctly-tailed hypothesis for a given study — a very common Component 01 task) and AO3 (evaluating how uncontrolled variables threaten a study's validity).
Research begins with a broad interest — "does caffeine affect memory?" The aim is a concise statement of the general purpose of the study: what the researcher intends to investigate. An aim is written as a statement, typically beginning "To investigate…" or "To examine…".
Aim: To investigate the effect of caffeine consumption on memory recall.
A research question poses the same intention as a question rather than a statement — "Does caffeine consumption affect memory recall?" Some researchers, especially in more exploratory or qualitative work, prefer a research question because it does not commit them to a specific predicted outcome. An aim and a research question do the same job (they orient the study) but neither is precise enough to be tested statistically. For that we need a hypothesis.
It helps to see the whole chain as a funnel of increasing precision, because every stage constrains the next and errors made early propagate to the end. At the widest point sits a broad area of interest ("memory and stimulants"). This narrows to an aim ("to investigate the effect of caffeine on recall"), which fixes the general purpose but leaves the details open. The aim then sharpens into an operationalised hypothesis ("participants given 200 mg of caffeine will recall significantly more words from a 20-item list than participants given none"), which specifies exactly what will be manipulated and measured. That hypothesis, in turn, dictates the design, the variables to control, and ultimately the statistical test used to evaluate it. Because each stage is derived from the one before, a woolly aim breeds a woolly hypothesis, which breeds an un-analysable study. Getting the aim clear and the hypothesis fully operationalised is therefore not pedantry — it is what makes everything downstream possible.
Note that the aim should describe the relationship under investigation, not the researcher's hopes or the practical motivation. "To help students revise more effectively" is a motivation, not an aim; "to investigate whether background music affects recall" is an aim, because it names the variables whose relationship the study will examine. Keeping the aim focused on the variables keeps it testable.
A hypothesis is a precise, testable statement predicting the relationship between variables. Unlike an aim, a hypothesis must be operationalised (variables defined in measurable terms) and must be falsifiable — capable, in principle, of being shown false. Every study has two hypotheses working as a pair.
The alternative hypothesis (often called the experimental hypothesis when the method is an experiment) predicts that there will be an effect, difference or relationship. It states what the researcher expects to find.
Alternative hypothesis: Participants who consume 200 mg of caffeine will recall significantly more words from a 20-word list than participants who consume no caffeine.
The null hypothesis predicts that there will be no effect, difference or relationship — that any difference observed is due to chance. It is the statement we actually test. Statistical testing works by assessing how likely our data would be if the null hypothesis were true; if that likelihood is very small (see the significance lessons), we reject the null hypothesis and retain the alternative.
Null hypothesis: There will be no significant difference in the number of words recalled between participants who consume 200 mg of caffeine and those who consume none; any difference is due to chance.
A common and heavily-penalised error is to write the null hypothesis as "caffeine will decrease recall". The null is not the opposite prediction — it is the statement of no effect. Note also the word significant: hypotheses in psychology predict a significant difference or relationship, because we test them against the possibility of chance.
Why bother with a null hypothesis at all, when it seems to state the boring outcome we usually hope to reject? The answer lies in the logic of scientific testing, which owes a great deal to Karl Popper's principle of falsification (developed fully in Lesson 10). We cannot prove a hypothesis true — no matter how many caffeinated participants recall more words, the next batch might not. What we can do is set up a precise, falsifiable statement of "no effect" and then ask how probable our data would be if that statement were true. If the data would be very improbable under the null (conventionally, less than a 5% chance), we are entitled to reject the null and provisionally accept the alternative. Science therefore advances not by confirming predictions but by failing to falsify them — and the null hypothesis is the falsifiable target that makes this possible. This is why every properly-designed study carries a null hypothesis even when the researcher fervently expects an effect: it is the statement the statistics actually put on trial.
Notice too that a hypothesis is not merely a guess but a bridge between an abstract theory and concrete, measurable data. Working-memory theory predicts that a stimulant should aid short-term recall; the hypothesis translates that abstract prediction into something a study can test — specified participants, a specified dose, a specified recall measure. This translation is where operationalisation (below) becomes indispensable: a theory can be vague, but a hypothesis that a study will test cannot be.
graph TD
A["Aim / research question<br/>(broad purpose)"] --> B["Alternative (experimental) hypothesis<br/>predicts an effect / difference / relationship"]
A --> C["Null hypothesis<br/>predicts NO effect — difference is due to chance"]
B --> D{"Is a direction predicted?"}
D -->|Yes| E["One-tailed / directional"]
D -->|No| F["Two-tailed / non-directional"]
C --> G["The null is the statement we statistically TEST<br/>→ reject it or retain it"]
style B fill:#27ae60,color:#fff
style C fill:#c0392b,color:#fff
style E fill:#2980b9,color:#fff
style F fill:#8e44ad,color:#fff
The alternative hypothesis can be written in two forms, and choosing correctly is an examined skill.
A one-tailed (directional) hypothesis predicts the direction of the effect — which condition will score higher, or whether a correlation will be positive or negative. You use a directional hypothesis when previous research or theory gives you good reason to expect a specific direction.
Participants who consume 200 mg of caffeine will recall more words than those who consume none. (predicts direction: more)
A two-tailed (non-directional) hypothesis predicts that there will be a difference or relationship but does not specify the direction. You use a non-directional hypothesis when there is no previous research, or when previous findings are contradictory, so you cannot justify predicting a direction.
There will be a difference in the number of words recalled between participants who consume 200 mg of caffeine and those who consume none. (predicts a difference, but not which way)
| One-tailed (directional) | Two-tailed (non-directional) | |
|---|---|---|
| Predicts | The direction of the effect | That there is an effect, not its direction |
| Signal words | "more than", "faster than", "positive correlation" | "a difference", "a relationship", "an effect" |
| Use when | Prior research/theory supports a direction | No prior research, or contradictory findings |
| Statistical note | Uses a one-tailed critical value | Uses a two-tailed critical value |
The choice matters again in Lesson 6/7, because whether a hypothesis is one- or two-tailed determines which column of the critical values table you read. Getting the tail wrong at the planning stage cascades into the wrong critical value at the analysis stage.
There is a genuine judgement to be exercised here, and examiners like to probe it. A directional hypothesis is bolder: it stakes a claim about which way the effect will go, and — because a one-tailed test concentrates all of the 5% significance region at one end of the distribution — it makes it slightly easier to reach significance if the effect is in the predicted direction, but leaves you unable to claim anything if a strong effect emerges in the opposite direction. A non-directional hypothesis is more cautious: it splits the 5% across both tails, making significance marginally harder to reach, but it will detect an effect whichever way it runs. The professional rule is therefore not "directional is better" but "predict a direction only when prior evidence justifies it". Where the literature is thin, mixed, or absent, the honest and defensible choice is non-directional. A candidate who can explain this reasoning — rather than mechanically matching signal words — is demonstrating the AO2/AO3 understanding that distinguishes the strongest answers.
A subtle related point concerns what happens when a directional prediction turns out to be wrong. If a researcher predicts "caffeine improves recall" but the data show caffeine impairing recall, a one-tailed test does not permit them to claim a significant impairment — they set out to test only one direction. This is why over-eager directional hypotheses can backfire, and why some methodologists advocate defaulting to non-directional tests unless the theoretical case for a direction is strong.
The independent variable (IV) is the variable the researcher manipulates (or, in a quasi-design, selects) — the presumed cause. The dependent variable (DV) is the variable the researcher measures — the presumed effect, which depends on the IV.
In our caffeine study: the IV is caffeine dose (200 mg vs 0 mg); the DV is memory recall.
A frequent stumbling block is telling the IV and DV apart in an unfamiliar scenario. A reliable technique is to fit the study into the sentence "the effect of [IV] on [DV]": the thing the researcher sets or changes is the IV, and the thing they measure to see what happened is the DV. In a study of "the effect of sleep deprivation on reaction time", sleep deprivation is set by the researcher (IV) and reaction time is measured (DV). Beware two traps: in a quasi-experiment the "IV" is a pre-existing characteristic the researcher merely selects (age, diagnosis) rather than manipulates, but it still functions as the IV in the analysis; and in a correlation there is strictly no IV/DV at all, only two measured co-variables, because nothing is manipulated. Mislabelling a correlation's variables as IV and DV is a common and avoidable error.
Operationalisation means defining a variable precisely in terms of how it is manipulated or measured, so that the study is replicable and unambiguous. "Memory" is not operationalised; "the number of words correctly recalled from a 20-word list after a two-minute delay" is. "Aggression" is not operationalised; "the number of physical acts directed at the Bobo doll in a ten-minute session" is (as Bandura did).
A vague variable is a fatal flaw: two researchers could not replicate the study, and the finding could not be checked. OCR frequently asks candidates to "operationalise the dependent variable in this study" — the mark is earned by turning an abstract concept into a specific, countable measure.
Operationalisation carries a hidden risk that repays A-Level-level scrutiny: the way you define a variable can quietly change what you are actually studying. If "aggression" is operationalised as "number of physical acts toward the Bobo doll", the study measures physical, imitative aggression — not verbal aggression, not real interpersonal aggression, and not the intent behind the act. The operational definition and the underlying concept are not the same thing, and the gap between them is precisely the issue of construct validity (Lesson 8): does the concrete measure genuinely capture the abstract construct it claims to? A study can be perfectly replicable (because its variables are tightly operationalised) yet still be measuring something narrower or subtly different from what its conclusions assert. Recognising that operationalisation buys replicability at some cost to breadth is a sophisticated evaluative move, and it explains why psychologists often use several different operational measures of the same construct and look for convergence across them.
The same care applies to the IV. In our caffeine study, "caffeine" must be operationalised as a specific dose (200 mg), delivered in a specified way, at a specified time before testing — otherwise "caffeine" could mean anything from a weak tea to several espressos, and the study could not be replicated or interpreted. Fully operationalising both variables is what turns a vague aim into a testable, repeatable study.
Subscribe to continue reading
Get full access to this lesson and all 10 lessons in this course.