1. Research Design Flashcards
When is an experimental study used?
To investigate interventions e.g., compare patient response after a treatment vs control
When is an observational study used?
When no intervention takes place. This is used to enable inferences about diseases through natural observation of groups defined by their exposure or disease status
What is a dependent variable?
What we measure. The “outcome” or “criterion” variable.
What is an independent variable?
What we manipulate or vary systematically. The “predictor” variable.
What are the pros and cons of between-subjects designs?
Advantages: comparison of different groups (e.g., gender, age, ethnicity), eliminates risk of practice or order effects.
Disadvantages: unidentified differences in groups may affect results, not always clear if difference is due to IV or a confounding variable.
What are the pros and cons of within-subjects designs?
Advantages: useful to monitor the effects on same individuals, controls for individual differences.
Disadvantages: Order/practice effects can confound results (reduced with counterbalancing).
What is a control condition?
It is identical to other conditions, except the proposed causal factor (IV) is not present. Acts as a baseline for comparison with the causal condition.
What are single and double blind trials?
Single blind: either the investigator, patient, or assessor is blind to the allocation.
Double blind: participant and researcher do not know which treatment the ppt is taking.
Aim is to reduce selection and observer bias.
What is an explanatory RCT?
Tests efficacy with carefully selected participants in highly standardised conditions
What is a pragmatic RCT?
Tests effectiveness in real life practice under more flexible conditions
What are possible sources of bias with RCTs?
Blinding: clinicians/ researchers are often unavoidably aware of treatment allocation, which could cause bias
Drop out: non-random drop out could bias results
Treatment adherence: participants may stop, miss sessions or receive additional treatment
What are the pros and cons of RCTs?
Advantages: avoids confounding variables as comparison groups should only differ in intervention/exposure status, causal hypothesis can be tested, effect of intervention can be measured.
Disadvantages: costly and difficulty, ethical issues, efficacy vs effectiveness issues
What are the pros and cons of experimental studies?
Advantages: repeatability due to controlled conditions, controllability.
Disadvantages: artificial, representativeness of participants, limited number of variables can be manipulated, control of relevant influencing variables.
What are the three types of observational studies?
Cross-sectional, case-control and cohort.
What is a cross-sectional study used for?
Used to estimate prevalence (point, period, lifetime). And used to measure association.
What are the pros and cons of cross-sectional studies?
Advantages: good at estimating prevalence, maximise completeness of data, useful for describing distribution of continuous variables in a population.
Disadvantages: cannot establish causality, cannot establish incidence (new cases), generalisability is limited by size and representativeness of sample.
What is a case-control study used for?
To help determine if an exposure is associated with an outcome. Compares prior expose and frequency. Cases and controls can be matched on an individual or group basis on characteristics that need to be controlled for (e.g., age, sex).
What are the pros and cons of case-control studies?
Advantages: useful for rare outcomes and outcomes with long latency, can examine multiple exposures, current hypotheses can be tested.
Disadvantages: prone to selection bias and information bias, temporal relationship between exposure and outcome may be difficult to determine, not suitable for studying rare exposures, cannot establish incidence.
What is a cohort study?
This is to observe a cohort over time. They are usually a better option than case-control studies as the exposure is measured before the outcome, avoiding recall bias and “reverse causality”.
What are the types of cohort studies?
Prospective: cohort is formed and followed over time, temporal relationship between exposure and outcome can be recognised.
Retrospective: data gathered for a cohort formed in the past, outcome is determined by exposure status.
Population: large cohort is recruited and followed e.g., birth cohort, allows a large number of common exposures to be considered in relation to multiple outcomes (test multiple hypotheses).
What are the pros and cons of cohort studies?
Advantages: can establish incidence, suitable for rare exposures, multiple outcomes can be examined, participants in a cohort can be matched.
Disadvantages: can be expensive and take many years, attrition bias, not ideal for outcomes with long latency (unless using retrospective design), unsuitable for rare outcomes, new knowledge can develop which is not accounted for in the design.
Why is measuring mental health challenging?
Study and outcome are only as strong as the measures used. Measurement is difficult in MH as is influenced by subjective experience, there is considerable variability in clinician’s diagnoses, little/no biological markers, reliance on questionnaire-based assessments.
What is reliability?
The consistency of a measure when applied repeatedly under similar circumstances.
What is validity?
An expression of the degree to which a measurement measures what it purports to measure.
How can reliability be assessed?
Test-retest, calculating inter-rater or inter-observer reliability (or inter-interviewer); operationaling behaviours, Cronbach’s alpha, split-half method (for questionnaires).
Explain stratified sampling
- Identify the different types of people that make up the target population
- Calculate the proportions required to ensure the sample is representative
- Create a variable list where any factors that could impact outcome are listed
- The relative percentage of each group in the sample is calculated as per the target proportion
Sample is highly representative and therefore generalisable to target population
Explain opportunity sampling
Selects ppts from target population as they are available and willing to take part (convenience sampling).
Advantages: straight forward and quick
Disadvantages: less likely to be representative
Explain systematic sampling
- Calculate total number in target population
- Decide sample size
- Select every ‘N’th number
Provides a representative sample but requires a large target sample
What are the four types of randomisation?
Simple, block, stratified, covariate adaptive
Explain simple randomisation
Roll of a dice, toss of a coin etc.
Valid for trials where there are more than 200 participants, but problematic for less than 100 due to risk of sampling error
Explain block randomisation
divide participants into subgroups (blocks), ensuring that the variability of characteristics is similar, so that the groups do not differ systematically.
Explain stratified randomisation
Similar to stratified sampling. Ppts are allocated to blocks, and blocks are randomly allocated to treatment/control group.
Explain covariate adaptive randomisation
Allows continuous allocation to groups, by keeping track of prior participants and their characteristics. Probability of being allocated to a group changes depending on the current participants. Participants are assigned by calculating the least imbalance across groups in terms of covariates by using an algorithm.