Week 3 - Experimental and quasi-experimental designs

Question 1

What is an experiment?

Answer 1

• A true experiment is randomised, has a control group and consists of manipulation.
• We will focus on randomised controlled trials as one example of an experimental design.

Question 2

What is randomisation

Answer 2

Aim to make allocation to the two groups (control and experimental) random so there is an equal probability for an individual to be assigned to any given group.

Question 3

Why do we do Randomisation

Answer 3

Allocation concealment: to prevent selection bias groups should be equal at the outset (consider specific variables for homogeneity)

Question 4

So who should randomise?

Answer 4

Blinding: Double blinding: neither participant nor researchers know who is in what group desirable but not always possible.

Question 5

how to control Randomized controlled trails?

Answer 5

Where there is an experimental group – we also need a control group (sometimes called comparison)

Question 6

Manipulation

Answer 6

Manipulation of independent variable

Question 7

what is a randomized controlled trial (RCT)?

Answer 7

• Includes three aspects: randomisation, control and manipulation
• Data collection (outcome measures) at certain point over time
• RCTs in health care can be;
  -> “simple”: one component eg medication
  -> “complex”: several interacting components eg behavioural interventions

Question 8

When Randomized control trials (RCT) are the “right” design

Answer 8

• To test cause and effect of a treatment and outcome
• Highly important for medical interventions, particularly medication research where the focus is on efficacy of a treatment

Question 9

Are Randomized controlled trails always the golden standard?

Answer 9

• Over reliance on RCTs and discounting other evidence/ knowledge
• May not answer the questions of how and for whom a certain treatment is best
• Some interventions cannot and should not be verified by RCTs (ethical and pragmatic reasons)
• Transferability to other populations
• Even field tests may not represent true behaviour (Hawthorne effect)
• Complex interventions
• Consider mixed methods/ embedded process evaluations

Question 10

Confounding

Answer 10

A confounder is a variable that influences both the dependent variable and independent variable, causing a spurious association;
- hence correlation does not imply causation.
- Often a problem with uncontrolled experiments.

Question 11

Clinical equipoise

Answer 11

There is not one “better” intervention present during the design of an Randomized Control Trial.
- exists when one as no good basis for a choice between two or more care options/ interventions.
- It provides an ethical basis for RCTs.

Question 12

Intention to treat

Answer 12

Refers to analysis of the results of an experiment
- based on the initial treatment assignment
- not on the treatment eventually received.

Question 13

Hawthorne effect

Answer 13

Participants modify their behaviours because they are aware of being watched/ observed. It can also contaminate an intervention study if one of the control groups changes its behaviour because it is being observed more frequently than the other.

Eg: A study of hand-washing among medical staff found that when the staff knew they were being watched, compliance with hand-washing was 55% greater than when they were not being watched (Eckmanns 2006).

Question 14

Loss to follow up

Answer 14

Participants who at one point in time were actively participating in a clinical research trial, but have become lost (either by error in a computer tracking system or by being unreachable) at the point of follow-up in the trial.