experimental design Flashcards
what is an experimental design
it describes the way participants are allocated to experimental groups.
what is the point of an experimental design
they are a set of procedures used to control extraneous variables
what are the three types of experimental design
independant groups, repeated measures, matched pairs
what is an independant groups design
different participants complete in each of the two or more conditions of the experiment. participants are randomly allocated to each condition (to avoid researcher bias).
what kind of data does independant groups design produce
unrelated data
this means individual data points in one condition cannot be paired with any data points in the other condition
what are the strengths of an independant groups design
- reduced demand characteristics participants are less likely to work out the aims of the experiment as they only take part in one condition
- no order effects as participants take part in one condition. no one gets better through practice (learning effect). no one gets worse due to boredom (fatigue effect)
what are the weaknesses of an independant groups design
- need 2x more participants (more expensive)
- no control over participant variables since participants take part in only one condition
what is the solution to overcome the problems of an independant groups design
- randomly allocate participants to each condition
- be prepared to spend money
what is a repeated measures design
the same participants complete in each of the two (or more) experimental conditions
what kind of data does a repeated measures design produce
related data
this means each participants score/data points in one condition can be paired with a data point (in their own score) in the other condition
what are the strengths of a repeated measures design
- need 1/2 of the participants for the same amount of data making it quicker and cheaper
- participants variables between the conditions is not a problem as participants take part in both conditions
what are the weaknesses of a repeated measures design
- one condition may be harder than another and would affect the accuracy of the results
- participants are more likely to work out the aim as they take part in both conditions increasing demand characteristics.
- order effects can affect performance (learning and fatigue effect)
what is the solution to overcome the problems with repeated measures design
- make tests equivalent to make both conditions equal
- use single blind trials/tests
- use counterbalancing
what is counterbalancing
mixing up the order of the tasks
how does the AB/BA way of counterbalancing work
divide participants into two groups
group 1 - condition A then condition B
group 2 - condition B then condition A
how does the ABBA way of counterbalancing work
all participants take part in all conditions
trial 1 - A (morning)
trial 2 - B (afternoon)
trial 3 - C (afternoon)
trial 4 - A (morning
you would compare scores on trials 1&4 with trials 3&2
which design fixes the order effects problem in RMD and the participants variables in IGD
matched pairs design
what is a matched pairs design
different participants complete in each of the two (or more) conditions of the experiment.
what is the procedure of a matched pairs design
participants are first assessed and ranked on a characteristic (e.g. aggression) and then the top two participants (then each following two) are randomly allocated in to separate conditions
what are the two separate conditions in a MPD
one is an experimental group the other is a control group
what type of data does a matched pairs design produce
related data
this means each participants score in one condition can be paired with a data point (the participant matched to them) in the other condition
what are the strengths of a matched pairs design
- reduced participant variables as participants are matched on a relevant characteristic
- no order effects as participants take part in only one condition
- less participants needed so quicker and cheaper
what are the weaknesses of a matched pairs design
- very time consuming to match participants and probably have to start with large groups which can be expensive
- need twice as many participants as a repeated measures design
- participants are similar not identical so there may still be some participant variables between conditions that influence the dependant variable. they can only match variables known to be relevant
what is the solution to overcoming the problems in a matched pairs design
- restrict matching variables to make it easy
- conduct pilot study to consider key variables
