Please enable JavaScript.
Coggle requires JavaScript to display documents.
Experimental Design (Research needs to be highly controlled to avoid the…
Experimental Design
Research needs to be highly controlled to avoid the effects of extraneous variables. Extraneous variables can be controlled in a study so that they're kept constant for all participants. Extraneous variables can also be eliminated altogether. There are many ways to eliminate extraneous variables.
Counterbalancing
Counterbalancing (mixing up the order of the tasks) can solve order effects in repeated measures designs.
Random Allocation
Random allocation means everyone has an equal chance of doing either condition. Random allocation should ensure the groups are not biased on key variables.
Standardised Instructions
Standardised instructions should ensure the experimenters act in a similar way with all participants. Everything should be as similar as possible for all the participants including each participant's experience in such studies.
Randomisation
Randomisation is when the material is presented to the participants in a random order. It avoids the possibility of order effects.
-
Validity
If an experiment shows that the results were caused by the manipulation of the variables, rather than the effect of something else, then it has internal validity.
If the findings can be generalised beyond the experimental setting, then the experiment has external validity.
Ecological validity is the measure of how true the results are to real life. So if something has high ecological validity, it means that the results are representative of what would happen in the real world. Lab studies have low ecological validity because they're artificial and so it's harder to generalise the results to the wider population.
Reliability
If a test is consistent within itself, it has internal reliability. The split-half technique assesses this. A questionnaire is randomly split into two. If all participants score similarly on both halves, the questions measure the same thing.
If the measure is stable over time or between people, then it has external reliability. This can be assessed by measuring test-retest reliability (if the same person always scores similarly on the test) or inter-rater reliability (do different assessors agree).
To help foresee any potential problems, a small-scale pilot study can be run first. This should establish whether the design works, whether participants understand the wording in instructions, or whether something important has been missed out. Problems can be tackled before running the main study, which could save wasting a lot of time and money.