L5 14/02 Flashcards
For a randomised controlled trial or any experiment gathering new data, what form of data is relied on?
Quantitiative
What are experimental designs?
- Dominant method in health research
- Used to test as hypothesis and explain variable relationships
3 key features/assumptions of experimental designs
Objectivity, determinism, reductionism
Objectivity definition
Reality exists beyond social construction and can be measured
Determinism definiton
That the (health) world is governed by natural laws, which are predictable and can be generalised, reasons why everything happens
Reductionism definition
The phenomena of interest can be reduced to standard units of measurement
What are the 3 research validities?
Internal validity - the study design
External validity - sampling techniques, generalisability
Measurement validity - data collection methods
Why are randomised control trials (RTCs) essential?
If the research is explanatory and needs to demonstrate a causal relationship between variables
What do RCTs use to eliminate the effects of confounding variables and biases?
Probability sampling, random allocation, and blinding (people aren’t entirely sure which group they are in) to eliminate the effects of confounding variables and biases
What do RCTs aim to do?
- The study design and procedures aim to control for all possible alternative explanations of observed changes in dependent variables
What kind of studies are included in RTCs?
- Field experiments
- Cohort studies
- Case-control studies
What are some threats to internal validity in quasi-experiments?
- History: events occurring between pre and post-test measures – only difference between pre and post should be the intervention
- Testing: the effect of repeated measures on subject responses
- Regression to the mean: equalisation of measures over time
- Instrumentation: changes in the measuring tools (or measurers)
- Selection and interaction biases
- Mortality: differential ‘drop-out’ rates – will affect ability of data to demonstrate causality
What is a sample?
A subset of the population which it is feasible to study. A sample can also be the unit of analysis: a person, an organisation, a body part, anything. The sample is chosen to represent a population.
Give 2 examples of sampling threats
- Sample bias (a population characteristic is over- or under-represented in the sample e.g., volunteer bias)
- Sample size (too small a sample gives false (unrepresentative) results, unable to make generalisations about whole population)
Definition of random sampling
Drawing a specific number from a list of the whole population
Definition of random allocation
- Assigning people in the sample into treatment and control groups
- Can use a random or non-random sample from the population
- Random number generator
Definition of blinding
Where the subjects and / or the researchers do not know who is in the experimental and control groups e.g., placebo or drug
Definition of double blinding
When BOTH subjects and researchers don’t know who is in which group
3 types of experimental design
- Randomised control trials
- Cohort studies
- Case control studies
RCTs aim to…
Eliminate bias and provide valid conclusions
What are some key features of RCTs?
- Random allocation to groups
- Patients and researchers unaware of which treatment was given until the study completed (blinding)
- Both groups are treated identically except for the experimental treatment
- Analysis estimates the size of the difference in outcomes between groups AND the statistical power of the effect
In quasi-experiments….
Participants are NOT randomized to treatment or control groups
In cohort studies…
- ABA process
- Participants are studied over time
- The study population shares common characteristics
In case control studies…
- Studies that address questions about harm or causation
- Compares different cohorts
- Investigates why some people develop a disease or behave the way they do vs others who do not
What are 7 key factors to consider in experiments?
- Recruitment methods
- Sample size
- Randomisation method
- Controls and confounding factors
- Blinding
- Data collection methods
- Validity of outcome measures
What assumptions are relevant to all forms of experimental designs?
- Objectivity
- Determinism
- Reductionism
Does a randomised controlled trial NOT require a placebo intervention?
False, it is a key attribute of RCT - it does require it
Is internal validity in a RCT increased by use of randomisation and blinding?
Yes
When participants in both intervention and control groups are unaware of which group they are in the study it can be described as double blinding?
No it cannot
Does a field experiment GUARANTEE causality whereas a RCT can only INFER it?
No, a RCT can guarantee causality but field experiment can only infer
Would a valid measurement tool will produce the same results on different occasions?
No, that’s reliability
Does probability sampling decreases the chance of a sample being representative of the population?
No
What are potential threats to the internal validity of an experimental design?
- Selection bias
- External events affecting pre and post testing
- Participant response to being tested