Stats Flashcards
Define: Control Group
A group that is similar to the group receiving the intervention.
Types of Experiment?
Completely Randomised Block Design
Randomised Block Design
Define: Completely Randomised Block Design
Treatments are allocated entirely by chance.
Define: Randomised Block Design
Group paprticipants according to some characteristic and then randomise within each block to balance out unknown factors.
Define: Single Blinding
Either the participant knows or the researcher knows who has received which treatment.
Define: Double Blinding
Neither the researcher nor the participant knows who has received which treatment. In this situation participants are randomly allocated to treatment groups for the researchers by a thrid party.
Only after measurements are treatment groups revealed.
Double Blinding > Single Blinding
Define: Placebo
This is essentially no treatment, but a participant will not know that they have not received any treatment.
Define: Placebo effect
This is an observed actual response to receiving a placebo.
Define: Response Variables
Variables of interest in your experiment. The values of these variables will be measured during the experiment.
I.e. Response to a pill or diet might be measured.
Define: Explanatory Variables
The variable that attempts to explain or cause the changes observed in the response vairable.
I.e. This might be a new drug or diet. All other variables must be unchanged.
Define: Treatment
The intervention that the experiment is attempting to observe.
I.e. Drug dosage, food supplement, fertiliser.
Define: Randomisation
Ensures that all participants are equally likely to receive treatments.
Define: Confounding or Lurking Variable
A variable that is not under scrutiny but affects the results of the experiment.
I.e. “People who drink more alcohol hace a shorter life expectancy”, confounding variable could be smoking.
Define: Replications
The number of times an experiment is repeated.
What makes a good Experiment?
- Random Allocation to treatment groups
- Use of a Control Group
- Use of a Placebo
- Use of Blinding
Define: Random Allocation
Used in experiments to ensure that all experimental unit have the same probability of receiving each treatment option.
Define: Hawthorne Effect
The Hawthorne effect is a type of reactivity in which individuals modify an aspect of their behavior in response to their awareness of being observed.
Types of Observational Studies
Cross-sectional
Longitudinal
Define: Cross-sectional
This type of study provides a snapshot of a group and is largely descriptive in nature.
Define: Longitudinal
A study which observes a group of people or other units over a long period of time. usually at regular intervals. A longitudinal study can be considered as a sequence of cross-sectional studies.
Potentioal Problems with Obervational Studies
- Confounding Variables
- Extending results inappropriately
- Using the past as a source of data.
Define: Confounding Variables
Confounding variables and implications for causation- in an observational study it is impossible to serparate out all potential confounding factors as random allocation of treaments has not been used.
Thus a Causal link cannot usually be claimed from an observational study.
Define: Extending results inappropriately
Many observational studies use convenience samples which cannot be regarded as representative of a wider population. results of a study can only be extended to a larger population if the sample is representative of that larger population.
Define: Using the past as a source of data.
Any study that uses the past as a source of data is known as a retrospective study. Such studies rely on people’s ability to recall information about past events accurately - which many people find hard to do! One option to overcome this is to use medical record, if appropriate, as a data source. An additional problem is that overtime the influence of confounding variable may change.