Criminological Psychology: Key Question & Practical

The Edexcel specification asks students of each applied topic to do two further things beyond learning the core content: to analyse a key question of societal relevance using the topic's concepts and theories, and to design (and understand) a practical investigation relevant to the topic — a small-scale, ethical study modelled on real research methods. This lesson does both for criminological psychology. First, it takes the key question "How reliable is eyewitness testimony in securing safe convictions?" — a genuinely consequential issue, because mistaken identification has contributed to documented miscarriages of justice — and analyses it using the topic's theories of memory, misleading information, anxiety and jury decision-making. Second, it models an ethical practical investigation: a laboratory experiment testing whether a leading question distorts eyewitnesses' recall of a filmed event, worked through in full — hypothesis, variables, design, sampling, procedure, ethics, and analysis, including a named descriptive statistic and a named inferential test (the chi-square test, $\chi^2$χ2). Together these show criminological psychology not as abstract theory but as a discipline with real social stakes and its own rigorous methods.

Key Definition: A key question is a socially or personally significant issue that can be analysed using a topic's psychological concepts; a practical investigation is a small, ethical study students design and understand to demonstrate research-methods skills relevant to the topic.

---

Spec Mapping

This lesson addresses the key question and practical investigation requirements of Edexcel 9PS0 — Paper 2, Topic 6: Criminological Psychology, using the reliability of eyewitness testimony as the key question and a leading-question experiment as the modelled practical.

• AO1 (knowledge and understanding): the nature of a key question and a practical investigation; research-methods concepts — hypotheses, independent and dependent variables, experimental design, sampling, ethics, and descriptive and inferential statistics (including the chi-square test).
• AO2 (application): applying topic theory (reconstructive memory, leading questions, anxiety, jury bias) to the real-world question of safe convictions, and applying research-methods concepts to design and analyse an appropriate study.
• AO3 (analysis and evaluation): evaluating the reliability of EWT as courtroom evidence, and evaluating the strengths, weaknesses, validity and ethics of the modelled practical investigation.

Connects to…

• Eyewitness testimony (Loftus & Palmer): the key question and the practical both build directly on the reconstructive-memory and leading-question research, applying it to the real-world problem of wrongful conviction.
• Jury decision-making: the "safe convictions" question depends not only on how accurate EWT is but on how much weight juries give it, connecting memory research to the courtroom-persuasion factors studied in the jury lesson.
• Contemporary study (Valentine & Mesout): the finding that high anxiety impairs identification feeds directly into the reliability question, since many real witnesses are frightened victims.

---

Part A: The Key Question — How Reliable Is Eyewitness Testimony in Securing Safe Convictions?

Why This Is a Question of Societal Relevance

Eyewitness testimony carries enormous weight in the criminal justice system. A confident witness identifying a defendant in court is among the most persuasive evidence a jury can hear, and for many crimes it is the primary evidence available. Yet the stakes of getting it wrong are as high as they come: a mistaken identification can convict an innocent person and leave the true offender free. This is not hypothetical. Analyses of wrongful convictions later overturned by DNA evidence — most prominently by the Innocence Project in the United States — have repeatedly found that mistaken eyewitness identification is the single most common contributing factor, present in a large majority of the cases. The question of how far EWT can be trusted to secure safe convictions is therefore of direct and serious societal importance: it bears on individual liberty, on public confidence in justice, and on how courts and juries should be directed.

Key Definition: A safe conviction is one that can be relied upon as correct because it rests on sound, properly weighed evidence. An unsafe conviction is one where doubt about the evidence — such as the reliability of an identification — means the verdict cannot be trusted.

Analysing the Question with Topic Concepts

Criminological psychology supplies a rich set of concepts for analysing EWT reliability, and a strong analysis draws them together rather than listing them.

Memory is reconstructive, not reproductive. The foundational insight is that recall rebuilds an event from fragments using schemas and currently available information, rather than replaying a stored recording. This is why an honest, confident witness can be wrong: their memory has been reconstructed, sometimes with contaminating material woven in. For safe convictions, this means confidence cannot be taken as a proxy for accuracy — a caution juries find counter-intuitive.

Leading questions and post-event information distort recall. Loftus and Palmer (1974) showed that a single word in a question can bias a witness's report and even alter the stored memory (the delayed "broken glass" finding). In real cases, suggestive police questioning, media coverage, and discussion between witnesses (memory conformity) can all contaminate testimony before it reaches court. A conviction resting on testimony gathered through leading questions is correspondingly less safe.

Anxiety can impair identification. Valentine and Mesout (2009) found that high state anxiety sharply reduced correct identification (~17% vs ~75%). Because many real witnesses are victims of frightening, often violent crimes, the very witnesses whose testimony courts most rely on may be operating at the impairing end of the arousal curve — a serious concern for the safety of convictions that hinge on their identifications.

Estimator and system variables. A useful framework distinguishes estimator variables — factors outside the justice system's control that were fixed at the time of the crime (lighting, distance, duration, the witness's anxiety, whether a weapon was present, cross-race identification) — from system variables — factors the justice system does control (how line-ups are constructed and administered, whether questions are leading, whether witnesses are kept apart). This distinction is powerful for the key question because it identifies exactly where reform can help: system variables can be improved (double-blind, sequential line-ups; non-leading interviewing; separating witnesses) even though estimator variables cannot.

Juries may over-weight confident testimony. The reliability question is not only about how accurate EWT is, but about how much weight juries give it. Because jurors reasonably but mistakenly treat confidence as a guide to accuracy, a confident-but-mistaken witness is especially dangerous — which is why the reliability problem is inseparable from the jury-decision-making topic.

graph TD
    A[Reliability of EWT for safe convictions] --> B[Estimator variables<br/>fixed at the crime]
    A --> C[System variables<br/>controllable by justice system]
    B --> B1[Lighting / distance / duration]
    B --> B2[Anxiety: Valentine & Mesout]
    B --> B3[Weapon focus / cross-race]
    C --> C1[Leading questions: Loftus & Palmer]
    C --> C2[Line-up design & administration]
    C --> C3[Keeping witnesses apart]
    A --> D[Jury weighting of confidence]
    style A fill:#2980b9,color:#fff

A Balanced Conclusion to the Key Question

The evidence does not support the conclusion that EWT is worthless — Yuille and Cutshall (1986) showed that real witnesses can be highly accurate, and much testimony is broadly correct. But it strongly supports the conclusion that EWT is fallible and improvable, and that convictions resting solely on a single, confident identification — especially one obtained under poor estimator conditions or through flawed system procedures — cannot be regarded as reliably safe. The psychologically informed position is therefore neither to dismiss EWT nor to trust it uncritically, but to (i) improve the system variables through evidence-based identification and interviewing procedures, (ii) direct juries realistically on the weak link between confidence and accuracy, and (iii) treat uncorroborated identification with appropriate caution. Analysed this way, the key question becomes a driver of concrete reform rather than a counsel of despair.

---

Part B: The Practical Investigation — Do Leading Questions Distort Eyewitness Recall?

Having analysed the reliability question, we model a practical investigation that tests one of its central mechanisms: whether the wording of a question can implant a false detail in a witness's memory. This is a small-scale, ethical adaptation of the Loftus-and-Palmer paradigm suitable for a student practical.

Aim and Hypotheses

Aim: to investigate whether a leading question containing a false presupposition increases the likelihood that eyewitnesses "remember" a non-existent detail in a filmed event.

• Alternative (experimental) hypothesis (H₁): Participants asked a leading question that presupposes a broken headlight will be more likely to report seeing a broken headlight than participants asked a neutral question — the difference being greater than would be expected by chance. (This is a directional / one-tailed hypothesis, justified because prior research, Loftus and Palmer 1974, establishes the direction of the effect.)
• Null hypothesis (H₀): There will be no difference in the proportion reporting a broken headlight between the leading-question and neutral-question groups; any difference is due to chance.

Variables

Variable	Definition / operationalisation
Independent variable (IV)	The wording of the critical question, with two levels: a leading version ("Did you see the broken headlight?", presupposing one existed) versus a neutral version ("Did you see a broken headlight?", making no such presupposition). There was no broken headlight in the film.
Dependent variable (DV)	Whether the participant reports seeing a broken headlight — a categorical (yes/no) response, recorded as the number of participants in each group answering "yes".
Key controls	The same film clip, viewing conditions, delay before questioning, and question order for all participants; the critical question embedded among neutral filler questions so its purpose is not obvious.

Design, Sampling and Procedure

Experimental design: an independent-groups (between-subjects) design — each participant experiences only one level of the IV (leading or neutral). This design is chosen to avoid the order and demand effects that a repeated-measures design would introduce (a participant asked both versions would immediately detect the manipulation), at the cost of individual differences between groups, which random allocation addresses.

Sampling: an opportunity sample of 40 adult volunteers (for example, sixth-form students aged 16–18 and staff), randomly allocated to the two conditions (20 per group) by drawing lots. Opportunity sampling is practical and cheap for a student project but limits representativeness — a limitation carried into the evaluation. Random allocation ensures that participant variables (e.g. memory ability) are, on average, evenly distributed across the two groups so they do not confound the IV.

Procedure (standardised for all participants):

1. Participants are briefed, give informed consent, and are told they will watch a short film and answer questions about it (the true focus — the leading question — is withheld to avoid demand characteristics; this partial concealment is addressed at debrief).
2. All participants watch the same ~30-second film clip of a minor traffic incident, containing no broken headlight.
3. After a short filled delay, participants complete a 10-item questionnaire. Item 7 is the critical question, phrased according to the participant's randomly allocated condition (leading vs neutral); the other nine items are neutral fillers.
4. Responses to the critical question (yes/no) are recorded.
5. Participants are fully debriefed: the true aim is explained, the absence of a broken headlight is revealed, the reason for the mild concealment is justified, and the right to withdraw their data is reaffirmed.

graph LR
    A[40 volunteers<br/>informed consent] --> B[Random allocation]
    B --> C[Leading group n=20<br/>'THE broken headlight?']
    B --> D[Neutral group n=20<br/>'A broken headlight?']
    C --> E[Same film clip<br/>no broken headlight]
    D --> E
    E --> F[Record yes/no to<br/>critical question]
    F --> G[Full debrief +<br/>right to withdraw data]

Ethics

The study is designed to meet the standard ethical requirements, and the mild concealment it involves must be justified and mitigated:

• Informed consent and the right to withdraw: obtained before participation, and the right to withdraw (including withdrawing data after the debrief) is made explicit.
• Deception (minimised and justified): participants are not told the specific focus on the leading question, because doing so would create demand characteristics and invalidate the study. This is a mild, low-risk concealment — participants are not misled about anything harmful — and it is disclosed at debrief. The scientific necessity is weighed against the small cost to the participant, as ethical guidelines require.
• Protection from harm: the film shows a minor incident with no gory or distressing content, so the risk of psychological harm is negligible — a deliberate design choice that distinguishes this ethical student practical from studies that induce genuine anxiety.
• Debriefing and confidentiality: a full debrief restores complete knowledge, and all responses are anonymised so no participant can be identified.

Analysis: Descriptive and Inferential Statistics

Descriptive statistics. Because the DV is categorical (a yes/no count), the appropriate descriptive summary is a frequency table and the percentage of each group reporting the false detail — the percentage being the natural descriptive statistic for nominal data (a mean would be meaningless here). Suppose the results were:

Condition	Reported broken headlight (yes)	Did not (no)	% reporting "yes"
Leading ("the")	11	9	55%
Neutral ("a")	4	16	20%

The descriptive pattern — 55% of the leading group versus 20% of the neutral group reporting a non-existent headlight — is consistent with the hypothesis that the leading wording implants a false detail. A simple bar chart of the two percentages would display this difference clearly (percentages, not raw counts, being the fair comparison when both groups are equal-sized).

Inferential statistics. To decide whether this difference is statistically significant (unlikely to be due to chance) rather than merely apparent, we need an inferential test. The appropriate test is the chi-square ($\chi^2$χ2) test of association (independence), because it tests for an association between two categorical variables — here, the condition (leading vs neutral) and the response (yes vs no) — using frequency (count) data from an independent-groups design. These are exactly the conditions the chi-square test requires, which is why a test of means (such as a t-test) would be inappropriate for this DV.

The test compares the observed frequencies with the expected frequencies that would occur if there were no association between condition and response, using the statistic:

$\chi^2 = \sum \frac{(O - E)^2}{E}$χ2=∑E(O−E)2​

where $O$O is each observed frequency and $E$E is the corresponding expected frequency. The calculated value is compared with a critical value from chi-square tables at $p \le 0.05$p≤0.05 with degrees of freedom $df = (\text{rows} - 1)(\text{columns} - 1) = (2-1)(2-1) = 1$df=(rows−1)(columns−1)=(2−1)(2−1)=1. If the calculated $\chi^2$χ2 equals or exceeds the critical value, the null hypothesis is rejected and the difference is deemed significant; if not, we retain the null hypothesis. (For the illustrative data above, the difference is sizeable, so a significant result would be plausible — though the exact decision depends on the computed statistic against the critical value.)

Note: Match the test to the data. Chi-square = categorical IV and DV with frequency data and independent groups. A Mann–Whitney U test would instead be the choice if the DV were ordinal (e.g. a rating of confidence 1–10) with independent groups; a related design measuring the same participants twice would call for a Wilcoxon test. Naming the correct test and justifying it from the design is the discriminator examiners reward.

---

Evaluation (AO3)