Study Design Algorithm & Quiz
This interactive guide helps you rapidly identify the correct study design, sampling method, or randomization technique in exam questions. Start at Question 1 and follow the logic down the correct path.
Step 1: The Core Question
Ask yourself: Did the researchers actively assign an exposure or treatment to participants, or did they only observe what was already happening?
Option A: YES, they assigned an intervention (Experimental)
The Experimental Algorithm
You are here because the study involves a deliberate intervention (e.g., giving a drug, therapy, or procedure).
Question 2: How was the intervention assigned?
Was it assigned by a process of chance (randomly)?
YES ➜ The study is a Randomized Controlled Trial (RCT). Now, ask one more question to specify the type:
Did the SAME participants receive BOTH treatments?
- YES ➜ It is a Crossover Trial.
In this design, each participant receives all the interventions being tested, but in a randomized sequence (e.g., Treatment A then Treatment B). Crucially, there is usually a "washout period" between treatments to prevent carryover effects.
✅ Exam Hook:
"Each participant acts as their own control," "before and after in the same group," "washout period."
A researcher wants to compare treatment effects in the same group of individuals before and after an intervention. What study design is best?
Correct Answer (c): The phrase "in the same group of individuals before and after" points to a crossover design where patients serve as their own controls.
In a trial of a new hypnotic, each participant takes the active drug for 2 weeks, then after a washout period takes placebo for 2 weeks (order counterbalanced). What design is described?
Correct Answer (d): Same participant receives both treatments (often with washout) = Crossover trial.
A ward measures average incident reports for 3 months before introducing a new policy and for 3 months after. No separate control ward is used. Which option best fits?
Correct Answer (d): Same group/unit compared across two conditions/times (before vs after) = Crossover-style within-subject comparison.
A study compares two anxiolytics. Participants receive Drug A for 4 weeks, then after a washout receive Drug B for 4 weeks, with half starting on each drug. What design is this?
Correct Answer (c): Washout + same participants receive both treatments = Crossover trial.
In a crossover trial comparing two long-acting anxiolytics, participants receive Drug A for 4 weeks, then immediately switch to Drug B with no gap. The results appear inconsistent. What is the most likely methodological issue?
Correct Answer (c): Crossover without washout risks carryover effects contaminating the second treatment period.
- NO (different groups got different treatments) ➜ It is a standard Parallel Group RCT.
This is the gold standard for efficacy. Randomization minimizes selection bias, and blinding minimizes observer/participant bias.
✅ Exam Hook:
"Randomly assigned," "double-blind," "placebo-controlled."
A paper describes an antidepressant study: “Participants were allocated to escitalopram or placebo using a computer-generated sequence. Capsules were identical, and neither participants nor assessors knew group allocation.” What is the most appropriate classification?
Correct Answer (c): Random sequence + blinding of participants and assessors = Double-blind RCT.
In a drug trial, participants are unaware whether they are receiving active medication or placebo, but the clinician rating symptoms knows which treatment each participant receives. What is the best description?
Correct Answer (b): If only one party is blinded (usually participants), it’s Single-blind.
Participants are randomly assigned to receive either Medication X or Medication Y. Both clinicians and participants know which medication is prescribed. What is the best classification?
Correct Answer (c): Random allocation = RCT; blinding only decides whether it’s double-/single-/open-label.
Was randomization NOT used or NOT mentioned?
Yes ➜ It is a Non-Randomized Trial.
- If there is a control group, it's a "controlled clinical trial."
- If there is no control group at all (everyone gets the treatment), it is an Uncontrolled Trial (e.g., a case series with an intervention). This provides the weakest evidence for efficacy.
A new psychological intervention is delivered to 25 patients with chronic insomnia, and their sleep scores are measured before treatment and after 8 weeks. There is no control group and no randomisation. What is the most appropriate classification?
Correct Answer (a): Intervention given + before/after measurement + no control group = Uncontrolled trial.
Option B: NO, they only observed (Observational)
The Observational Algorithm
You are here because the researcher is only observing, not intervening. They are like a detective watching suspects.
Question 2: What is the study's starting point and direction of inquiry?
Did the study start by identifying people based on their OUTCOME (Disease)?
YES ➜ It is a Case-Control Study.
Example: Recruiting 100 patients with depression (Cases) and 100 people without depression (Controls) and asking them about past trauma.
- Direction: Looks backwards in time (Retrospective).
- Purpose: Investigating causes/risk factors for rare diseases.
✅ Exam Hook:
"Starts with cases and controls," "good for rare diseases," "recall bias is a major weakness."
Clinical Example: Autism & Stress
Researchers recruit 100 children with Autism and 100 neurotypical children. They interview mothers about prenatal stress. This looks backwards from outcome to exposure.
A researcher recruits 80 patients with new-onset psychosis and 80 matched individuals without psychosis, then interviews both groups about past cannabis exposure before illness onset. What design is this?
Correct Answer (c): Start with outcome (cases/controls) + look back for exposure = Case–control.
A researcher identifies people who attempted suicide in the last month and a matched group who did not. They then review records to determine prior exposure to SSRI treatment. What study design is this?
Correct Answer (b): Outcome-defined groups + retrospective exposure assessment = Case–control.
Investigators select patients who developed lithium toxicity and patients on lithium who did not then compare prior dehydration episodes in both groups. Design?
Correct Answer (b): Cases defined by outcome + look back for exposure = Case–control.
Researchers recruit 120 patients with a rare catatonia subtype and 120 matched controls without it. They interview families about past exposure to a suspected toxin. Which option best captures the key advantage and a key limitation?
Correct Answer (b): Case–control = best for rare diseases but vulnerable to recall bias.
A study recruits people with a rare dissociative disorder and controls without it, then asks both groups to recall childhood adversity in detail. Which bias is MOST likely to threaten validity?
Correct Answer (b): Case–control + retrospective exposure history = high risk of recall bias.
Did the study start by identifying people based on their EXPOSURE (Risk Factor)?
YES ➜ It is a Cohort Study.
Example: Recruiting 1,000 smokers (Exposed) and 1,000 non-smokers (Unexposed) who are all healthy at the start, and following them to see who gets cancer.
- Direction: Looks forwards in time (Prospective).
- Purpose: Determining the prognosis of a disease or the incidence of a disease linked to an exposure.
✅ Exam Hook:
"Two groups followed forward without randomization," "establishes temporality," "calculates incidence/absolute risk," "expensive and slow," "attrition bias."
A memory clinic compares agitation scores over 12 weeks in two groups of patients with Alzheimer’s disease: those whose consultant chose to start rivastigmine and those managed with usual care. The methods section does not mention random allocation or blinding. What is the most appropriate study design?
Correct Answer (b): Two groups defined by exposure + followed forward + no randomisation = Cohort study.
A paper states: “We compared sertraline versus placebo for PTSD symptoms over 8 weeks.” The methods do not describe randomisation or blinding, and allocation appears to follow clinician preference. Best classification?
Correct Answer (c): “Placebo” is not enough—without randomisation, exposure-defined groups followed forward = Cohort.
A sleep clinic identifies patients who choose CBT-I and those who decline it, then measures insomnia severity at baseline and again at 3 months. No randomisation is described. Best design?
Correct Answer (b): Exposure groups + baseline-to-follow-up outcome measurement + no randomisation = Prospective cohort.
A paper states: “We conducted a two-arm trial of lithium vs placebo for aggression.” Allocation was based on clinician decision and the methods do not mention blinding or randomisation. What design is most appropriate?
Correct Answer (c): “Two-arm trial” wording is a trap—without randomisation, it’s not an RCT; exposure-defined forward comparison = Cohort.
A researcher identifies people currently taking anticholinergics and those not taking them, then measures new onset constipation over the next 6 months. There is no randomisation. What design is this?
Correct Answer (c): Exposure groups + follow forward for new outcomes + no randomisation = Cohort.
A study recruits 2,000 adolescents with high social media exposure and 2,000 with low exposure, all free of depression at baseline, then follows them for 5 years to measure new-onset depression. Which statement best explains a major strength and a major weakness of this design?
Correct Answer (b): Cohort = temporality strength; biggest practical weakness = attrition bias from follow-up loss.
A researcher wants to measure how many new cases of panic disorder develop over 2 years among people exposed to a workplace trauma versus those unexposed. Which design best fits the aim?
Correct Answer (b): “New cases over time” (incidence) + exposure-defined groups = Prospective cohort.
A paper reports outcomes of patients with schizophrenia treated in routine clinical practice over one year. Treatments vary according to usual clinician choice, and there is no defined comparison of “exposed vs unexposed” groups at baseline. What is the best classification?
Correct Answer (a): Routine practice outcomes with no structured exposure groups = Naturalistic outcome study.
A study compares symptom scores in two separate groups: one group receives Therapy A, and a different group receives Therapy B. Outcomes are measured at baseline and 10 weeks. No participant receives both therapies. Which design best fits?
Correct Answer (b): Different people in different interventions (no one gets both) + follow-up = Cohort/parallel follow-up, not crossover.
Did the study measure both EXPOSURE and OUTCOME at the exact same moment?
YES ➜ It is a Cross-Sectional Study.
Example: A single survey asking junior doctors about their current work hours (exposure) and current burnout score (outcome).
- Direction: A single snapshot (no arrow of time).
- Purpose: Measuring the prevalence of a disease or risk factor.
✅ Exam Hook:
"Snapshot in time," "prevalence study," "cannot determine temporality" (the chicken-and-egg problem).
A service evaluation measures current antidepressant use and current PHQ-9 scores in all patients attending a clinic during one week. No follow-up occurs. What study design is this?
Correct Answer (a): One time-point measurement of exposure and outcome = Cross-sectional study.
A survey measures current benzodiazepine use and current falls in older adults attending a GP practice on a single day. Design?
Correct Answer (a): “On a single day” snapshot of exposure and outcome = Cross-sectional.
A clinic audits all attendees on one afternoon and records current opioid use and current sleep quality score. No groups are pre-selected by sleep problems. Design?
Correct Answer (b): Measured once in everyone (no follow-up, no case selection) = Cross-sectional study.
Did the study ONLY describe features of a few patients (No comparison)?
YES ➜ It is a Case Series (or Case Report if only one patient).
- Purpose: Generating hypotheses, identifying new phenomena (e.g., a new side effect).
✅ Exam Hook:
"Collection of anecdotes," "idea generator," "no comparison group."
A psychiatrist in an early intervention service notices that seven patients presenting with first-episode psychosis over the last month all report using the same newly marketed synthetic stimulant. The psychiatrist writes a paper describing the shared clinical features and the temporal link, but includes no comparison group. What is the most appropriate classification?
Correct Answer (c): A “collection of similar patients” with no comparison group = Case series.
Did the study use aggregated, population-level data?
YES ➜ It is an Ecological Study.
Example: Comparing antidepressant sales figures between different countries with their national suicide rates.
- Purpose: Generating hypotheses using group-level data.
✅ Exam Hook:
"Data from populations, not individuals," "risk of ecological fallacy."
A researcher compares national antidepressant sales figures across several countries with their national suicide rates and reports an association. What study design is this?
Correct Answer (a): Aggregated “country/ward/population-level” exposure and outcome data = Ecological study.
A researcher reports that countries with higher alcohol consumption have higher national rates of depression and concludes that individual heavy drinkers are therefore more likely to be depressed. Which statement best identifies the key methodological problem?
Correct Answer (c): Group-level association ≠ individual-level causation → Ecological fallacy.
Algorithm for Sampling & Randomization Methods
If the question is about how participants were selected or allocated, use this guide.
Part A: Sampling (Recruitment)
Question: How did the researchers get participants into the study?
- Is it a RARE disorder or a HARD-TO-REACH group (e.g., IV drug users), and were participants asked to REFER others they knew?
- YES ➜ Snowball Sampling.
In a study of a rare disorder, the researcher asks participants to suggest other individuals with the same condition to include in the sample. Which technique is being used?
Correct Answer (c): The chain-referral technique for rare disorders is the hallmark of snowball sampling.
A researcher studying a stigmatised, hard-to-reach group recruits a few initial participants and asks each to invite others they know with the same trait. What is this technique?
Correct Answer (d): Hard-to-reach group + participants recruit peers = Snowball sampling.
- Did they recruit whoever was easiest to access (e.g., first 60 people)?
- YES ➜ Convenience Sampling.
A community survey team recruits whoever is available in a shopping centre to complete a questionnaire about current alcohol use. Which sampling method is most likely?
Correct Answer (a): “Whoever is easiest to reach” in a survey = Convenience sampling.
A researcher stands outside an outpatient clinic and enrolls the first 60 people who agree to participate in a stress questionnaire. Which sampling method is described?
Correct Answer (c): First willing/available participants = Convenience sampling.
To recruit for a study on self-harm in university students, a researcher enrolls the first 100 students who respond to a poster in the campus café. Sampling method?
Correct Answer (a): First/easiest-to-recruit respondents = Convenience sampling.
- Did they select every nth person from a list?
- YES ➜ Systematic Sampling.
A clinic has an alphabetical list of 5,000 patients. To recruit 500 for a survey, the researcher selects every 10th name from the list after choosing a random starting point. What sampling method is this?
Correct Answer (b): “Every nth person from a list” = Systematic sampling.
A researcher has a list of 2,000 patients and recruits participants by selecting every 5th name after choosing a random starting point. What sampling method is this?
Correct Answer (b): “Every nth person” (with random start) = Systematic sampling.
- Did they divide the population into strata and randomly sample within each?
- YES ➜ Stratified Sampling.
A population registry lists all 20,000 adults in a region. To estimate the prevalence of depression, the researcher divides the population into men and women, then randomly selects equal numbers from each group to complete a questionnaire. Which method is being used?
Correct Answer (a): Divide population into strata + random sample within each = Stratified sampling (recruitment).
- Did they randomly select clusters (e.g., hospitals) and survey everyone inside?
- YES ➜ Cluster Sampling.
To estimate PTSD prevalence among hospital staff, a researcher randomly selects 6 hospitals from a region and surveys every staff member in those selected hospitals. What is this sampling method?
Correct Answer (a): Randomly pick clusters (hospitals) then include everyone inside = Cluster sampling.
To study burnout prevalence, a researcher randomly selects 10 hospitals, then surveys all doctors within those hospitals. No intervention is applied. What method is being used?
Correct Answer (b): Randomly choose clusters to recruit participants (no intervention) = Cluster sampling.
- Did they fill predetermined quotas non-randomly?
- YES ➜ Quota Sampling.
A researcher wants a community sample with 50% men and 50% women. They stand in a public square and recruit passers-by until the male quota and female quota are filled, without random selection. What sampling method is this?
Correct Answer (b): Set subgroup targets + recruit non-randomly until each target is met = Quota sampling.
A researcher wants exactly 60 men and 60 women for a community survey. They recruit passers-by until the targets are met, without random selection from any list. What sampling method is this?
Correct Answer (b): Predetermined subgroup numbers filled non-randomly = Quota sampling.
Part B: Randomization (Allocation in an RCT)
Question: In an RCT, how were participants assigned to their groups?
- Was the main goal to ensure a key characteristic (like gender, age, or disease severity) was PERFECTLY BALANCED between the groups?
- YES ➜ Stratified Randomization.
✅ Exam Hook:
"Ensures equal distribution of males/females," "balances an important prognostic factor."
A medication trial ensures the treatment and control arms have equal numbers of males and females by randomising separately within “male” and “female” groups. What randomisation method is used?
Correct Answer (c): Randomise within levels of a prognostic variable (sex/age/severity) = Stratified randomisation.
A clinical trial team is concerned that age may affect response. They split participants into two age bands (<40, ≥40) and randomise separately within each band. Method?
Correct Answer (b): Split into strata (e.g., age bands) + randomise within each = Stratified randomisation.
A trial team worries baseline severity will affect outcomes. They divide participants into “mild” and “moderate–severe” strata, then randomise within each stratum to active drug or placebo. What method is this?
Correct Answer (c): Balance an important baseline factor by randomising within strata = Stratified randomisation.
- Were participants randomized in "groups of a fixed size" or "blocks" to keep the number of people in each arm of the trial equal at all times?
- YES ➜ Block Randomization.
✅ Exam Hook:
"Randomly assigned to groups of four," "randomized in blocks."
During recruitment, participants are randomised in blocks of 4 to keep the numbers in each arm closely balanced throughout the trial. What method is this?
Correct Answer (b): Fixed-size blocks to maintain arm balance during recruitment = Block randomisation.
A medication study uses randomisation lists arranged in blocks of 8 to prevent large imbalances in allocation during recruitment. Best answer?
Correct Answer (b): Mention of fixed “blocks of X” to keep arm numbers balanced = Block randomisation.
A researcher states: “To ensure equal numbers of participants in each arm throughout recruitment, we used blocks of 6.” What is the method being described?
Correct Answer (b): “Blocks of X” to keep arms numerically balanced = Block randomisation.
- Did they randomize entire groups (e.g., wards, hospitals) instead of individuals?
- YES ➜ Cluster Randomization.
A trial evaluates a new de-escalation training programme by randomising entire wards (not individual patients) to receive the programme or continue usual practice. What randomisation approach is this?
Correct Answer (d): Randomise groups (wards/hospitals/schools), not individuals = Cluster randomisation.
A study evaluates a new handover checklist by randomising whole hospitals to either implement it or continue standard practice. What is this?
Correct Answer (c): Whole institutions/wards randomised as units = Cluster randomisation.
A trial tests a new ward-round checklist by allocating entire inpatient units to use it or not, because staff work across patients and contamination is likely. What is the randomisation type?
Correct Answer (d): If the unit of allocation is a ward/clinic to prevent contamination = Cluster randomisation.
- Did they use pure chance (e.g., coin toss, computer generator) without blocks or strata?
- YES ➜ Simple Randomization.
A researcher allocates participants to treatment or control by using a computer random number generator, with no grouping into fixed sets and no balancing for sex or age. Which randomisation method is described?
Correct Answer (a): “Pure chance allocation for each participant” with no blocks/strata = Simple randomisation.
Final Quick-Reference Table: Matching Clinical Question to Design
| If the clinical question is about... | The BEST study design is... |
|---|---|
| Treatment Efficacy (Does it work in ideal conditions?) | RCT (Explanatory) |
| Treatment Effectiveness (Does it work in the real world?) | Pragmatic RCT |
| Aetiology / Causation (Is this a risk factor for a disease?) | Cohort Study or Case-Control Study |
| Prognosis (What is the future for someone with this disease?) | Cohort Study |
| Prevalence (How common is this disease right now?) | Cross-Sectional Study |
| Diagnostic Accuracy (How good is this new test?) | Cross-Sectional (comparing to a gold standard) |
| Lived Experience (What is it like to have this disease?) | Qualitative Study |
Full Mock Exam (50 Questions)
A memory clinic compares agitation scores over 12 weeks in two groups of patients with Alzheimer’s disease: those whose consultant chose to start rivastigmine and those managed with usual care. The methods section does not mention random allocation or blinding. What is the most appropriate study design?
Correct Answer (b): This question is testing whether you can separate observational follow-up from experimental allocation. The key clues are: two groups defined by treatment decision, followed over time, and no randomisation/blinding described. A prospective cohort study identifies groups based on exposure status and follows them forward to observe outcomes. Because allocation is based on clinician choice (not random), it fits an observational cohort design.
A researcher recruits 80 patients with new-onset psychosis and 80 matched individuals without psychosis, then interviews both groups about past cannabis exposure before illness onset. What design is this?
Correct Answer (c): The examiner wants you to identify where the sampling begins: with the outcome or with the exposure. Case–control studies begin with the outcome (cases with psychosis; controls without) and then look backward to compare past exposure (cannabis).
A service evaluation measures current antidepressant use and current PHQ-9 scores in all patients attending a clinic during one week. No follow-up occurs. What study design is this?
Correct Answer (a): This is a classic “snapshot” stem—everything happens at one point in time. Cross-sectional studies assess exposure and outcome simultaneously (current antidepressant use and current PHQ-9) with no follow-up.
In a trial of a new hypnotic, each participant takes the active drug for 2 weeks, then after a washout period takes placebo for 2 weeks (order counterbalanced). What design is described?
Correct Answer (d): The examiner is looking for the phrase “each participant takes… then… washout… then…” which signals within-person comparison. In a crossover trial, each participant receives both interventions in sequence (drug and placebo), often separated by a washout; participants act as their own control.
A paper states: “We compared sertraline versus placebo for PTSD symptoms over 8 weeks.” The methods do not describe randomisation or blinding, and allocation appears to follow clinician preference. Best classification?
Correct Answer (c): This stem is designed to trap you with the word “placebo.” The real discriminator is whether allocation was random. Two groups defined by exposure (sertraline vs placebo as chosen/available) followed forward for PTSD symptoms without randomisation fits a cohort (observational comparison).
A researcher studying a rare neuropsychiatric syndrome asks initial participants to refer other people they know with the same condition to join the study. Which sampling technique is this?
Correct Answer (c): This is testing recognition of how participants are found, not how they are allocated in a trial. Snowball sampling starts with a few “seed” participants and expands through participant referrals, useful in rare or hard-to-reach populations.
A medication trial ensures the treatment and control arms have equal numbers of males and females by randomising separately within “male” and “female” groups. What randomisation method is used?
Correct Answer (c): The examiner wants you to see that the aim is balance on a specific characteristic, not just equal numbers overall. Stratified randomisation divides participants into strata (male/female) and randomises within each stratum, ensuring balance for that variable across arms.
During recruitment, participants are randomised in blocks of 4 to keep the numbers in each arm closely balanced throughout the trial. What method is this?
Correct Answer (b): This is a pure “block size” stem. Block randomisation uses fixed-size blocks (here, 4) so that within each block, allocation is balanced between arms, preventing drift over time.
A community survey team recruits whoever is available in a shopping centre to complete a questionnaire about current alcohol use. Which sampling method is most likely?
Correct Answer (a): The question is about how participants are selected, not how they are allocated. Convenience sampling selects participants based on availability and ease of access, such as recruiting whoever is present in a shopping centre.
A trial evaluates a new de-escalation training programme by randomising entire wards (not individual patients) to receive the programme or continue usual practice. What randomisation approach is this?
Correct Answer (d): The discriminator is the unit of randomisation. Cluster randomisation randomises pre-existing groups (wards, clinics, hospitals) rather than individuals, often used to avoid contamination between participants.
A researcher identifies people who attempted suicide in the last month and a matched group who did not. They then review records to determine prior exposure to SSRI treatment. What study design is this?
Correct Answer (b): This stem starts with outcome status (attempted suicide vs did not), then looks back for exposure (SSRI). Case–control selects cases (suicide attempt) and controls (no attempt) and then compares previous exposure (SSRI) between them.
A sleep clinic identifies patients who choose CBT-I and those who decline it, then measures insomnia severity at baseline and again at 3 months. No randomisation is described. Best design?
Correct Answer (b): The key features are exposure-defined groups and subsequent follow-up measurement. Participants are grouped by exposure (choose CBT-I vs decline) and followed forward to observe outcome differences—this is a cohort structure without random allocation.
A ward measures average incident reports for 3 months before introducing a new policy and for 3 months after. No separate control ward is used. Which option best fits?
Correct Answer (d): This is testing the “within-group comparison across time/conditions” idea: the same setting acts as its own control. The structure is a within-unit comparison (before policy vs after policy), where the same unit serves as its own control across conditions/times—this aligns with the crossover-style logic tested by Paper B stems about “same group before and after.”
A paper states: “We conducted a two-arm trial of lithium vs placebo for aggression.” Allocation was based on clinician decision and the methods do not mention blinding or randomisation. What design is most appropriate?
Correct Answer (c): The examiner is testing whether you demand randomisation before calling something an RCT. Two groups defined by exposure with forward outcome comparison, but without randomisation, fits an observational cohort comparison despite the word “trial.”
A researcher studying a stigmatised, hard-to-reach group recruits a few initial participants and asks each to invite others they know with the same trait. What is this technique?
Correct Answer (d): This is pure recognition of referral-based recruitment. Snowball sampling grows the sample through participant invitations/referrals, especially in stigmatised or hard-to-reach populations.
A clinical trial team is concerned that age may affect response. They split participants into two age bands (<40, ≥40) and randomise separately within each band. Method?
Correct Answer (b): The key is balancing a specific prognostic variable (age), not merely balancing numbers. Stratified randomisation involves creating strata (here, two age bands) and randomising separately within each, ensuring age distribution is balanced across arms.
A medication study uses randomisation lists arranged in blocks of 8 to prevent large imbalances in allocation during recruitment. Best answer?
Correct Answer (b): This is testing whether you recognise “blocks of 8” as a balancing method during enrolment. Block randomisation uses predetermined block sizes (here, 8) so that within each block, allocation is balanced between arms, limiting imbalance during recruitment.
A survey measures current benzodiazepine use and current falls in older adults attending a GP practice on a single day. Design?
Correct Answer (a): The stem explicitly signals a single-day measurement. Cross-sectional studies measure exposure and outcome at one point in time (current benzo use, current falls) without follow-up.
A study evaluates a new handover checklist by randomising whole hospitals to either implement it or continue standard practice. What is this?
Correct Answer (c): The unit of allocation is the giveaway. Cluster randomisation randomises entire groups (hospitals) to intervention or control, often to prevent contamination across individuals within the same institution.
A researcher stands outside an outpatient clinic and enrolls the first 60 people who agree to participate in a stress questionnaire. Which sampling method is described?
Correct Answer (c): This tests whether you recognise “first people who agree” as non-probability sampling. Convenience sampling recruits participants based on ease of access/availability—here, the first 60 willing participants.
A paper describes an antidepressant study: “Participants were allocated to escitalopram or placebo using a computer-generated sequence. Capsules were identical, and neither participants nor assessors knew group allocation.” What is the most appropriate classification?
Correct Answer (c): This stem explicitly includes the two defining features of a high-quality RCT: random allocation and blinding. Random allocation to escitalopram vs placebo plus identical capsules and concealment from both participants and assessors indicates a double-blind RCT.
In a drug trial, participants are unaware whether they are receiving active medication or placebo, but the clinician rating symptoms knows which treatment each participant receives. What is the best description?
Correct Answer (b): The examiner is testing whether you can identify who is blinded. When participants are blinded but the assessor/clinician is not, it is a single-blind trial.
A researcher states: “To ensure equal numbers of participants in each arm throughout recruitment, we used blocks of 6.” What is the method being described?
Correct Answer (b): This question checks whether you can distinguish allocation methods used in trials from sampling methods used in recruitment. Block randomisation uses predefined blocks (here, 6) to keep group numbers balanced over time.
A trial team worries baseline severity will affect outcomes. They divide participants into “mild” and “moderate–severe” strata, then randomise within each stratum to active drug or placebo. What method is this?
Correct Answer (c): The key is that allocation is designed to balance a prognostic variable across arms. Stratified randomisation means participants are grouped into strata (severity levels) and randomised within each stratum, ensuring balance on severity.
A study compares two anxiolytics. Participants receive Drug A for 4 weeks, then after a washout receive Drug B for 4 weeks, with half starting on each drug. What design is this?
Correct Answer (c): The stem contains the classic crossover identifiers: sequence and washout. In a crossover trial, each participant receives both treatments at different times, often with a washout, and acts as their own control.
Investigators select patients who developed lithium toxicity and patients on lithium who did not then compare prior dehydration episodes in both groups. Design?
Correct Answer (b): This is about where the study begins: outcome status first. Case–control selects cases (toxicity) and controls (no toxicity) and then looks back for prior exposure (dehydration).
A researcher identifies people currently taking anticholinergics and those not taking them, then measures new onset constipation over the next 6 months. There is no randomisation. What design is this?
Correct Answer (c): The key is follow-up for incident (new) outcomes. Exposure-defined groups (anticholinergics vs not) followed forward to measure outcomes without randomisation = prospective cohort.
To recruit for a study on self-harm in university students, a researcher enrolls the first 100 students who respond to a poster in the campus café. Sampling method?
Correct Answer (a): The examiner is testing recognition of availability-based recruitment versus referral-based recruitment. Convenience sampling recruits those easiest to access—in this case, the first respondents to a poster.
A trial tests a new ward-round checklist by allocating entire inpatient units to use it or not, because staff work across patients and contamination is likely. What is the randomisation type?
Correct Answer (d): The stem signals group-level allocation and gives the classic reason: avoid contamination. Cluster randomisation randomises pre-existing groups (units/wards/hospitals), especially when contamination would occur if individuals were randomised within the same setting.
A clinic audits all attendees on one afternoon and records current opioid use and current sleep quality score. No groups are pre-selected by sleep problems. Design?
Correct Answer (b): The examiner wants you to notice that everyone is measured once, without selecting cases/controls. Cross-sectional studies measure exposure and outcome at the same time in a defined sample (one afternoon audit).
A psychiatrist in an early intervention service notices that seven patients presenting with first-episode psychosis over the last month all report using the same newly marketed synthetic stimulant. The psychiatrist writes a paper describing the shared clinical features and the temporal link, but includes no comparison group. What is the most appropriate classification?
Correct Answer (c): This stem is testing whether you can recognise a descriptive design that is essentially a structured “signal” rather than a hypothesis test. A case series is a collection of similar cases described together. Its purpose is to highlight a possible new phenomenon (e.g., a new adverse exposure or association) and generate hypotheses, but it cannot test causation because there is no control/comparison group.
A researcher compares national antidepressant sales figures across several countries with their national suicide rates and reports an association. What study design is this?
Correct Answer (a): This question is about the unit of analysis: individuals versus populations. Ecological studies use aggregated, population-level data (e.g., country-level sales and country-level suicide rates). They are useful for hypothesis generation but risk the ecological fallacy (assuming group-level associations apply to individuals).
A new psychological intervention is delivered to 25 patients with chronic insomnia, and their sleep scores are measured before treatment and after 8 weeks. There is no control group and no randomisation. What is the most appropriate classification?
Correct Answer (a): The examiner is testing whether you can spot an “intervention applied” study that still lacks the defining safeguard of comparison. An uncontrolled trial is “experimental” in the sense that an intervention is applied, but it lacks a comparison/control group, so you cannot separate treatment effect from placebo effect, regression to the mean, or natural course.
A population registry lists all 20,000 adults in a region. To estimate the prevalence of depression, the researcher divides the population into men and women, then randomly selects equal numbers from each group to complete a questionnaire. Which method is being used?
Correct Answer (a): This question is designed to punish mixing up sampling (who enters the study) with randomisation (how they’re allocated within a trial). Stratified sampling is a probability sampling method where you divide the population into strata (e.g., sex) and perform random sampling within each stratum to ensure representation.
A clinic has an alphabetical list of 5,000 patients. To recruit 500 for a survey, the researcher selects every 10th name from the list after choosing a random starting point. What sampling method is this?
Correct Answer (b): The stem describes selection at regular intervals from a list. Systematic sampling selects individuals at regular intervals (e.g., every 10th person) from a list, usually after a random start.
To estimate PTSD prevalence among hospital staff, a researcher randomly selects 6 hospitals from a region and surveys every staff member in those selected hospitals. What is this sampling method?
Correct Answer (a): This is about how the sample is recruited, not how treatments are allocated. Cluster sampling is a probability sampling method where pre-existing clusters (e.g., hospitals) are randomly selected, and then all individuals within selected clusters are studied.
A researcher wants a community sample with 50% men and 50% women. They stand in a public square and recruit passers-by until the male quota and female quota are filled, without random selection. What sampling method is this?
Correct Answer (b): The key is “aiming for numbers of characteristics” but selecting non-randomly. Quota sampling aims to include a specified number of participants with certain characteristics (e.g., 50/50 by sex) but recruits them non-randomly, increasing bias risk.
A study recruits 2,000 adolescents with high social media exposure and 2,000 with low exposure, all free of depression at baseline, then follows them for 5 years to measure new-onset depression. Which statement best explains a major strength and a major weakness of this design?
Correct Answer (b): This stem describes an exposure-first, forward-followed design → cohort. Cohort studies measure exposure before the outcome, establishing temporality (cause precedes effect), and reduce recall bias. Their major weakness is being slow/expensive with loss to follow-up—attrition bias.
Researchers recruit 120 patients with a rare catatonia subtype and 120 matched controls without it. They interview families about past exposure to a suspected toxin. Which option best captures the key advantage and a key limitation?
Correct Answer (b): The examiner is testing whether you recognise the “rare disorder + cases/controls + look back” pattern. Case–control studies are efficient for rare diseases because you start by finding cases, then compare past exposures. A major limitation is recall bias because exposure history is reconstructed from memory/records.
In a crossover trial comparing two long-acting anxiolytics, participants receive Drug A for 4 weeks, then immediately switch to Drug B with no gap. The results appear inconsistent. What is the most likely methodological issue?
Correct Answer (c): This question checks whether you know the specific vulnerability of crossover designs when washout is missing. In crossover trials, effects of the first treatment can persist into the second period, distorting results. A washout period is used to prevent these carryover effects.
A researcher allocates participants to treatment or control by using a computer random number generator, with no grouping into fixed sets and no balancing for sex or age. Which randomisation method is described?
Correct Answer (a): This stem tests whether you can recognise pure chance allocation without extra structure. Simple randomisation assigns each participant independently by chance (e.g., computer-generated sequence/coin toss). It does not guarantee balance of numbers early in recruitment and does not ensure balance of characteristics like sex or age.
A researcher has a list of 2,000 patients and recruits participants by selecting every 5th name after choosing a random starting point. What sampling method is this?
Correct Answer (b): The stem is built around the phrase “every 5th”, which is the giveaway. Systematic sampling uses a fixed interval approach (every nth) after a random start, producing a structured probability sample.
A researcher wants exactly 60 men and 60 women for a community survey. They recruit passers-by until the targets are met, without random selection from any list. What sampling method is this?
Correct Answer (b): This stem tests whether you spot fixed subgroup targets achieved by non-random recruitment. Quota sampling fills predetermined numbers of key characteristics (e.g., sex) using non-random recruitment methods, increasing selection bias risk.
To study burnout prevalence, a researcher randomly selects 10 hospitals, then surveys all doctors within those hospitals. No intervention is applied. What method is being used?
Correct Answer (b): The examiner is checking whether you can separate sampling (recruitment) from randomisation (allocation in trials). Cluster sampling is a recruitment method where clusters (hospitals) are randomly selected, and individuals within selected clusters are included.
A researcher reports that countries with higher alcohol consumption have higher national rates of depression and concludes that individual heavy drinkers are therefore more likely to be depressed. Which statement best identifies the key methodological problem?
Correct Answer (c): This question tests whether you recognise the danger of moving from group-level to individual-level conclusions. Ecological fallacy occurs when you infer individual-level risk from group-level (country) associations. A country association does not prove the same relationship holds within individuals.
A study recruits people with a rare dissociative disorder and controls without it, then asks both groups to recall childhood adversity in detail. Which bias is MOST likely to threaten validity?
Correct Answer (b): The stem is a classic case–control exposure assessment using retrospective memory. Recall bias is common when participants must remember past exposures. Cases may search for explanations and recall exposures differently from controls, distorting exposure estimates.
A researcher wants to measure how many new cases of panic disorder develop over 2 years among people exposed to a workplace trauma versus those unexposed. Which design best fits the aim?
Correct Answer (b): The question is about new cases over time, which requires follow-up. A cohort design starts with exposure groups (trauma-exposed vs unexposed) and follows forward to measure incident (new) panic disorder cases.
A paper reports outcomes of patients with schizophrenia treated in routine clinical practice over one year. Treatments vary according to usual clinician choice, and there is no defined comparison of “exposed vs unexposed” groups at baseline. What is the best classification?
Correct Answer (a): The examiner is testing whether you notice the absence of a deliberately defined two-group exposure structure. Naturalistic outcome studies describe what happens in real-world practice where treatment decisions occur as usual, often with heterogeneous management and without a pre-specified exposed vs unexposed comparison framework.
Participants are randomly assigned to receive either Medication X or Medication Y. Both clinicians and participants know which medication is prescribed. What is the best classification?
Correct Answer (c): This stem is testing that randomisation defines an RCT, while blinding determines the subtype/quality. Random assignment to treatments makes it an RCT, but the absence of blinding means it is an open-label (non-blinded) randomised controlled trial.
A study compares symptom scores in two separate groups: one group receives Therapy A, and a different group receives Therapy B. Outcomes are measured at baseline and 10 weeks. No participant receives both therapies. Which design best fits?
Correct Answer (b): This item is built to trick you into choosing crossover just because baseline and follow-up are mentioned. The key discriminator is whether the same participants receive both treatments. Two separate exposure-defined groups (Therapy A vs Therapy B) followed forward from baseline to 10 weeks, with no randomisation mentioned, best matches an observational cohort-style comparison (or at minimum, a parallel-group follow-up design without crossover).