Clinical Psychology: Key Question & Practical

The Edexcel specification asks students, for each topic, to consider a key question of societal relevance — a real-world issue that the topic's concepts can illuminate — and to design a practical investigation using the topic's methods. This final lesson does both for clinical psychology. Part (a) takes the key question "How reliable and valid is the diagnosis of mental illness, and what are the effects of labelling and stigma on people with mental health conditions?" and analyses it with the concepts built up across the topic — reliability, validity, comorbidity, culture and gender bias, Rosenhan's labelling findings, and the network model of causes. Part (b) models an ethical practical investigation relevant to the topic: a study of public attitudes to mental illness, using a questionnaire and, as an alternative design, a content analysis, worked through as a full method — aim and hypotheses, design, sampling, procedure, the heightened ethical sensitivity clinical topics demand, and analysis using a named descriptive statistic and a named inferential test (chi-square and, for an alternative design, Mann-Whitney). Because this is clinical content, the whole lesson keeps mental illness firmly in view as a matter affecting real people, and treats the design of research about it as an ethical responsibility, not merely a technical exercise.

Key Definition: A key question in the Edexcel scheme is a question of contemporary societal importance that can be examined using the concepts of the topic. A practical investigation is a small-scale study, designed and (where appropriate) carried out by the student, that applies the topic's research methods ethically to a suitable question.

---

Spec Mapping

This lesson addresses the Edexcel 9PS0 — Paper 2, Topic 5: Clinical Psychology requirements to (i) describe and analyse a key question of societal relevance using concepts from clinical psychology, and (ii) design a practical investigation using an appropriate research method from the topic, including its aim and hypotheses, design and sampling, procedure, ethics (with the heightened sensitivity required for mental-health topics), and analysis — a descriptive statistic and a named inferential test with its rationale. It draws together the whole topic: the reliability-and-validity concepts, Rosenhan's labelling findings, culture and gender bias, and the biological model of schizophrenia. In assessment-objective terms, you should be able to describe the key question and a suitable study design (AO1), apply clinical concepts and research-methods knowledge to the question and to the design (AO2), and evaluate the reliability, validity, ethics and generalisability of both the argument and the proposed study (AO3).

Connects to…

• Reliability and validity of diagnosis (this topic) — the key question is, at its core, the reliability-and-validity debate applied to society; comorbidity, symptom overlap and cultural bias are the analytic tools.
• The classic study, Rosenhan (1973) (this topic) — the "sticky label" and depersonalisation findings are central evidence for the stigma half of the key question, and the practical investigation extends Rosenhan's concern with labelling into a measurable study of attitudes.
• Research methods (whole course) — questionnaires, content analysis, sampling, operationalisation, descriptive statistics and the chi-square and Mann-Whitney tests are general Paper-2 methods applied here to a clinical question.

---

Part (a): The Key Question

Stating the question and why it matters

"How reliable and valid is the diagnosis of mental illness, and what are the effects of labelling and stigma on people with mental health conditions?"

This is a question of real societal importance for several reasons. A diagnosis is not a neutral description: it grants or withholds access to treatment and benefits, shapes how the person is seen by employers, insurers, courts and their own family, and — as Rosenhan showed — can reorganise the interpretation of everything the person subsequently does. If diagnosis is unreliable (clinicians disagree) or invalid (the category does not correspond to a real, distinct condition), then people may be mislabelled, denied appropriate help, or given the wrong treatment. And even when a diagnosis is accurate and helpful, the stigma attached to mental-illness labels can cause harm in its own right — through discrimination, social exclusion, reluctance to seek help, and internalised shame. Roughly one in four people experiences a mental-health problem in any given year, so this is a question that touches a very large proportion of the population, directly or through family and friends.

Analysing the question with topic concepts

The concepts developed across this topic give a structured way to analyse the question rather than merely opine on it.

Concept from the topic	How it bears on the key question
Reliability (inter-rater / test-retest)	If clinicians do not agree, or diagnoses change over time for a stable condition, the same person may be labelled differently by different services — a direct societal harm (inconsistent access to care). Operationalised criteria and structured interviews have improved this.
Validity (descriptive / predictive / aetiological)	If a category does not predict outcome or treatment response, the diagnosis offers little practical benefit and may mislabel. Validity, not reliability, is the harder and more contested property.
Comorbidity & symptom overlap	Where disorders routinely co-occur or share symptoms, the neatness of a single label is questionable — the person's difficulties may be forced into a category that fits imperfectly.
Culture & gender bias	Diagnostic disparities (e.g. higher rates of some diagnoses in particular ethnic groups without correspondingly higher incidence) show that who is assessed can affect the label — an equity and justice issue at the level of society.
Labelling & stigma (Rosenhan)	The "sticky label" and depersonalisation findings show that a diagnosis, once applied, can dominate identity and treatment; stigma then adds discrimination and reluctance to seek help.
Biological reality (Carlsson et al.)	The network model reminds us that schizophrenia has a genuine neurobiological basis — so the answer is not that mental illness is a mere label, but that a real condition must be diagnosed carefully and destigmatised.

Putting these together yields a balanced position, which is what a strong answer to a key question requires. On the one hand, diagnosis is far more reliable than in Rosenhan's day, thanks to operationalised criteria, and disorders such as schizophrenia have a real biological basis (Carlsson et al.), so wholesale scepticism about diagnosis is unwarranted and can itself be harmful (denying that conditions are real is a form of stigma). On the other hand, validity remains genuinely contested, cultural and gender biases persist, and the social consequences of a label — stigma, discrimination, self-stigma, reduced help-seeking — are serious and partly independent of whether the diagnosis is medically accurate. The defensible conclusion is that improving diagnosis (reliability and validity) and reducing stigma are both necessary, and that they are connected: more valid, culturally sensitive diagnosis, communicated well, is part of how stigma is reduced.

The effects of labelling and stigma

It is worth drawing out the stigma strand explicitly, since it is the most socially consequential.

• Public (social) stigma — negative stereotypes held by others (e.g. that people with schizophrenia are dangerous or incompetent), leading to discrimination in employment, housing and relationships.
• Self-stigma — the internalisation of those stereotypes by the person themselves, producing shame, lowered self-esteem and the "why try?" effect that undermines recovery.
• Structural stigma — disadvantage built into institutions and policy (for example, historically poorer physical-healthcare access for people with severe mental illness).
• Help-seeking effects — anticipated stigma deters people from seeking a diagnosis or treatment at all, worsening outcomes and, paradoxically, keeping the conditions less visible and less understood.

Rosenhan's study is the classic empirical anchor for this strand: it demonstrated the "sticky" label and the depersonalising effect of the psychiatric setting. Anti-stigma initiatives (public-education campaigns, contact-based interventions in which the public meet people with lived experience, and careful, non-sensational media reporting) follow directly from this analysis — and their evaluation is itself a suitable subject for a practical investigation, which is where Part (b) begins.

---

Part (b): A Practical Investigation

Rationale and choice of method

A practical investigation for this topic must be ethical, feasible for a student to design, and genuinely informative about a clinical question. Studying patients or manipulating anything to do with real diagnosis or treatment would be both unethical and impractical for a student. A safe and revealing alternative is to study attitudes towards mental illness in the general (adult, non-vulnerable) population — the stigma strand of the key question — using a self-report questionnaire. This is low-risk, uses a core Paper-2 method, and produces data suitable for a straightforward statistical test. Below, the questionnaire study is developed in full; an alternative content-analysis design is then outlined, because the specification values understanding more than one method.

Aim and hypotheses

Aim: to investigate whether prior personal contact with someone who has a mental-health condition is associated with more positive attitudes towards mental illness, testing the "contact reduces stigma" idea that follows from the labelling analysis.

The variables must be operationalised (defined in measurable terms):

• Independent variable (IV): whether the respondent reports prior close personal contact with someone with a diagnosed mental-health condition — coded as "contact" or "no contact".
• Dependent variable (DV): attitude towards mental illness, measured by a validated-style attitude questionnaire (e.g. a set of Likert items about willingness to work with, live near, or befriend someone with a mental-health condition), producing a total score; for the chi-square analysis this score is then categorised as a "more positive" or "less positive" attitude using a pre-registered cut-off (e.g. a median split).

Alternative (directional) hypothesis (H₁): Adults who report prior close personal contact with someone with a mental-health condition will show more positive attitudes towards mental illness than adults who report no such contact. A directional (one-tailed) hypothesis is justified because the labelling/contact literature consistently predicts this direction.

Null hypothesis (H₀): There will be no association between prior personal contact with someone with a mental-health condition and the positivity of attitudes towards mental illness; any difference is due to chance.

Design and sampling

Design feature	Choice and justification
Method	Self-report questionnaire — efficient, standardised, generates quantitative attitude data and categorical contact data suitable for a test of association
Experimental design	Independent groups / quasi-experiment — the IV (contact vs no contact) is a pre-existing characteristic, not manipulated, so this is a quasi-experiment using naturally occurring groups
Sampling	Opportunity or (better) stratified sampling of adults. Opportunity sampling is realistic for a student but risks bias; stratifying by age and sex improves representativeness. Exclude under-18s and anyone for whom the topic might be distressing
Materials	A standardised questionnaire: demographic items, a single clear contact question, and a set of Likert-scale attitude items (e.g. 1 = strongly disagree to 5 = strongly agree), some reverse-scored to reduce response bias
Controls	Identical wording and order for all respondents; anonymous completion to reduce social-desirability bias; clear instructions; a mix of positively and negatively worded items

Operationalisation and standardisation are the design priorities: the contact question must be unambiguous, and the attitude items must be piloted so that they measure attitude consistently (supporting the study's reliability and validity).

flowchart TD
    A["Key concept: contact reduces stigma<br/>(from labelling analysis)"] --> B["Operationalise IV & DV<br/>IV: contact / no contact · DV: attitude score → positive / less positive"]
    B --> C["Design: quasi-experiment,<br/>independent groups, questionnaire"]
    C --> D["Sample adults (exclude under-18s /<br/>anyone likely to be distressed)"]
    D --> E["Ethics: consent, right to withdraw,<br/>signposting, confidentiality"]
    E --> F["Collect data → descriptive stats<br/>(percentages, contingency table)"]
    F --> G["Inferential test: chi-square<br/>(association, categorical data)"]
    style E fill:#e74c3c,color:#fff
    style G fill:#2563eb,color:#fff

Ethics — with heightened sensitivity for a clinical topic

Research touching on mental illness demands more than the routine ethical checklist, because the topic can be personally painful for respondents (many will have lived experience or bereavement) and because careless design can itself reinforce stigma. The BPS-style principles are applied here with that heightened sensitivity:

Ethical issue	Standard requirement	Heightened, topic-specific safeguard
Informed consent	Explain the study's general purpose and gain agreement	State clearly that the questionnaire concerns attitudes to mental health before consent, so no one is ambushed by a sensitive topic
Right to withdraw	Participants may stop at any time	Emphasise it explicitly and allow withdrawal of data afterwards, recognising some may feel differently once they have seen the items
Protection from harm	Avoid distress	Word items respectfully and non-sensationally; avoid graphic or stigmatising language; provide signposting to support services (e.g. a helpline) at the end for every respondent
Confidentiality & anonymity	Keep data confidential	Collect anonymously — no names — since attitudes and personal contact with mental illness are sensitive; store data securely
Non-stigmatising design	(Beyond the standard list)	Ensure the questionnaire itself does not imply that mental illness is shameful or that people with conditions are "other"; pilot with this in mind
Vulnerable groups	Exercise special care	Exclude under-18s, and make clear no one is obliged to disclose their own diagnosis

The guiding principle is that a study about stigma must be scrupulously careful not to cause distress or reinforce the very attitudes it examines — an ethical standard higher than for a neutral topic.

Analysis: descriptive statistics

Descriptive statistics summarise the data before any test. For this categorical design:

• Percentages / frequencies are the appropriate descriptive statistic: for example, the percentage of the "contact" group whose attitudes fall in the "more positive" category, compared with the percentage of the "no contact" group.
• The data are best displayed in a contingency table (a 2×2 table of counts) and a suitable chart (a bar chart of the percentage "more positive" in each group). Because the DV, once categorised, is nominal, the mode or the percentage in each category is the meaningful measure of central tendency — a mean of the raw scores could be reported for the pilot, but the analysis proper is categorical.

A worked illustration of the kind of table the study would produce (figures illustrative only):

	More positive attitude	Less positive attitude	Row total
Contact	34	16	50
No contact	20	30	50
Column total	54	46	100

Here 68% of the contact group but only 40% of the no-contact group show more positive attitudes — a descriptive difference that the inferential test then evaluates for significance.

Analysis: the inferential test — chi-square

The appropriate inferential test is the chi-square (χ²) test of association. The choice is justified by three standard criteria:

1. The study tests a difference/association, not a correlation, between two conditions.
2. The design is independent groups (each respondent is in only one group — contact or no contact).
3. The data are nominal (categorical) — each respondent is classified into one cell of the 2×2 table (contact/no-contact × more/less positive).

Chi-square compares the observed frequencies with the expected frequencies (those predicted if there were no association), using the statistic:

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

where $O$O is each observed cell frequency and $E$E the corresponding expected frequency. The expected frequency for a cell is $E = \dfrac{(\text{row total}) \times (\text{column total})}{\text{grand total}}$E=grand total(row total)×(column total)​. For a 2×2 table the degrees of freedom are $df = (\text{rows}-1)(\text{columns}-1) = 1$df=(rows−1)(columns−1)=1.