Experimental Designs: Between and Within Subjects Design: Chapter 8 & 9 Flashcards
between groups
- 2+ groups are formed at random from a
pool of subjects - each group receives a different experimental treatment (value of the IV)
- scores for the groups are
compared
between groups: obtaining results
- 1 score per individual
- these are independent measures
do scores vary within groups
yes
systematic variance
difference in the DV
looking for an effect of IV
to look for an effect of IV, compare mean scores (DV) for each group
determining statistical significance
compare between groups variance to within groups variance
F = between-group variance / within group variance
When there is a large within-group variance, it is
difficult to see an effect, we want to ____
minimize it
large versus small variance in groups
- large variance in between groups (BG) = good
- large variance in within groups (WG) = bad
keeping within group variance low
limit individual differences
- Standardizing procedures
- Holding a participant variable constant
- Increase sample size
individual differences in between-groups are usually ____
always a potential confounding variable for this design
making between groups as equal as possible
- created equally
- treated equally, except for IV
- composed of equivalent individuals
randomization
- participants randomly assigned to groups to
ensure groups are as equal as possible before
treatment or intervention - most powerful technique to control for the effect
of pre-existing differences
randomization vs random sampling
randomization:
- random assignment of Ss to experimental
or control groups in a particular study
random sampling:
- random selection of Ss from a larger
population to participate in a study
free random assignment
- groups are based on chance
- if 2 groups +, table of random numbers is used to guard against repetition
- should lead to equality, but no guarantee
randomization: matched group
- participants matched on critical variables that may act as important confounds
- inherent in within-groups designs.
- ex:
1. intelligence
2. gender
3. age
4. severity of illness
randomization: randomized blocks
groups of individuals are matched in blocks
matched groups
critical variables that can act as confounds
- intelligence
- gender
- age
- severity of illness
matching procedure
- Rank subjects on the variable for which
control is desired. May require pretest. - Segregate subjects into matched pairs
on that variable. - Randomly assign pair-members to the
conditions.
advantages of between-groups
- very simple design
- no carryover effects
disadvantages of between-groups
- requires many participants
- individual differences & environmental differences
- groups must be equivalent before the manipulation
within-groups design
- only one treatment group and each subject is
given all levels (or conditions) of the IV - comparison = between scores obtained at
different levels of the IV for same participants - each participant serves as their own control
categories of within-groups
- concurrent measures
- repeated measures
concurrent measures
- all levels of IV are present at the same time – choice paradigm
- subjects choose the value of the IV they prefer
Harlow: monkeys cloth mother
(comfort) versus wire mother
(food).
Measured time spent at
each mother.
This is an example of ______
concurrent measures
repeated measures
- most common approach for within-subjects
- every subject receives all values of the IV
- participant’s performance is the basis of comparison
what does the repeated measures accomplish
- equating groups (by using same participants)
- reducing within-group variance (by controlling
for individual differences)
advantages of repeated measures
- increases sensitivity
- increases ability to detect a treatment
effect - error variance is reduced considerably because the participants become their own control (individual differences are eliminated)
main issue with repeated measures: carryover effects
- effects that one treatment may have on another treatment
- includes: practice effects, fatigue, boredom,
interference - exposure to one manipulation may produce consequences influencing response to manipulations
solutions to carryover effects
- randomization
- counterbalancing
counterbalancing
- all possible treatment orders are used equally
- equal numbers of participants in each treatment condition
counterbalancing: examples
2 treatments
- 2 x 1
3 treatments
- 3 x 2 x 1
4 treatments
- 4 x 3 x 2 x 1
product is the total amount of combinations
latin square design
- each treatment (A, B, C & D) occurs equally often in each position in the experiment
- refer to chapter 8&9, slide 32
reversibility in within-groups
- within-subjects designs = not adequate IF experimental conditions produce lasting effect on
the participant (cannot be reversed). - IVs permanently alter the development or state
of participants - have irreversible carry-over effects
examples of irreversible carryover effects
- physiological damage (brain lesions)
- interventions that improve or worsen a skill, such as learning or memory
- habituation
reversal design: ABA design
- allows verifying the presence/absence
of carry-over effects
Condition A = measure behavior at baseline
Condition B = measure during intervention
Condition A = measure after intervention stopped
Does behaviour return to the baseline?
YES: no carryover effect
NO: carryover effect
carryover effect desirability
- carryover effect can be desirable
advantages of within groups
- fewer participants required
- greater sensitivity to treatment effect (elimination individual differences)
- good when participants are hard to find
- each participant acts as his own control
- powerful design under suitable conditions
disadvantages of within groups
- not suitable when carryover effects are permanent
- participant attrition may be a problem
