Haney et al. (1973): Stanford prison

Deindividuation is the credited concept; "they all got into it" earns nothing. Define it — losing your personal identity in a group or uniform, so you feel less responsible and act as the role demands — then tie it to the study's uniforms, sunglasses, smocks and ID numbers. A definition with no link to those features is knowledge without application.

Two figures are examined and routinely guessed: 75 replied to the advert (not the number who took part) and 24 were selected after screening (not the 75). Duration is the other trap — planned two weeks but stopped after six days, often given the wrong way. Three more marks: pay was 15 dollars a day, arrests were by real police, and five were released early.

A result is what was observed — guards gave more punishments as the days passed; five prisoners were released early. A conclusion is what it means — people take on the behaviour their role demands; the situation, not the men's personalities, caused the harm. Rewording the observation is not a conclusion. When conclusion is asked, give the inference, not the event.

Identification is only half the mark. "A weakness is that it was unethical" names a point and stops. Use point, why, consequence: the men were arrested without warning so could not give informed consent, and five were so distressed they were released early, so the harm outweighed what the study showed. Do the same for strengths — say why random allocation matters.

A 2-mark describe needs two unique points — the same point reworded is credited once, so do not restate the first. Watch the neighbouring command word: an improvement asks what you would change if the study ran again (an independent researcher running the prison), not another weakness. A weakness given where an improvement was asked scores nothing.

On a scenario, three things earn no AO2: the character's name ("Priya conformed"), repeating the stem, and simply describing the study. Application means taking the scenario's own detail — a shared uniform, a badge number, a group no one can be singled out of — and matching it to a named feature of the study. Without AO2, the AO3 that depends on it is lost too.

Two errors. The guards were not told to be cruel — they were told to keep order but use no violence, so the punishments were their own, which makes the finding striking. This is conformity to a role (no order given), not obedience (a direct order from an authority figure, as in Milgram). An answer with an experimenter instructing anyone is the wrong study.

It shows the situation matters more than assumed but proves nothing about everyone — the men were young male students from a prison-life advert (weak generalisability) and not every guard was harsh. Ethics is not all: screening and random allocation rule out disposition, and the recorded procedure supports reliability. It shows power, not that anyone turns tyrant.