structure of an experiment
- manipulating the IV
- randomly assign to conditions
Treatment A Treatment B
2.5 control/eliminate extraneous variables
3. measure DV from given scores
4. compare
extraneous variables
any variable that has some relationship to the DV but is not an IV or DV
example:
- couples
- IV: manipulating stress levels in one partner
- DV: measure how the other partner comfort them socially
- extraneous variable: how long they’ve been together affects how well they can comfort their partner
gender also affects how well they can comfort their partner
confound
alternative explanation for relationships between IV and DV
design confound
mistake in the design of the IV, such that the second variable
- systematically with the IV
- example:
study:
selection effects
participants in one level of the IV are systematically different from those in the other
- example:
assigning people to eat a big bowl of pasta for breakfast
note: not many people eat pasta for brekkie/there might have been experimenters who offered more pasta
systematic differences
- men might choose bigger than women
- people might have come in hungry or not
what is the difference between extraneous variables and confounds?
extraneous variables are any variable in the context of the study that has some relationship to DV but is not IV or DV
varies randomly
confounds: varying systematically with levels of IV
varies systematically
ways to avoid selection effects
- random assignment
- matched groups
random sampling
everyone has a chance of being selected
matched groups
participants are assigned by variables that match each other
structure:
group 1 group 2
similar characteristic 1
similar characteristic 2
similar characteristic 3
between subjects design
different groups of participants only see ONE condition or the other
- cons: there are less people in each group because its split into two
within subjects
one group of participants completes all conditions
- compares scores from the same person
posttest-only design (between subjects)
participants are randomly assigned to a condition and tested on the dependent variable once
- preferable when pretesting would affect results
example:
experimenting on how a new study program affects children’s intelligence
posttest-only design: giving them a test afterwards to see the results
pretest-posttest design (between subjects)
participants were randomly assigned to a condition and tested on DV twice: before and after exposure to IV
- preferable when ensuring groups were equivalent at the start
- preference when showing improvement over time
advantages of within subjects designs
- participants in the two groups will be equivalent
-requires fewer participants than between subjects
disadvantages of within subjects
- order effects
- may not be possible
- participants may figure out what the experiment is about
-> leading to demand characteristics
order effects
exposure to one level of the IV influences responses to the next level
- practice effects: participants get better or worse with fatigue
- carryover effects: come contamination carries over from one condition to the next
avoid by counterbalancing by changing the order for participants
example: have some people run without pepsi or with pepsi first
demand characteristics
cues that lead participants to guess experiment’s hypothesis
sources of confounds
- environment
- individual differences
- time
environment as a source of confound
setting or context differs across treatment conditions -> design confound
- control by holding everything except IV constant across conditions
example:
study about effects of stress in crowded rooms
- the crowded condition accidentally studied how room temp affects stress
individual differences as a source of confound
- assignment to conditions result in groups with different characteristics -> selection effects
- people are different (knowledge, mindset, experiences
- control with random assignment
time as a source of confound
- treatment conditions occur at different times and experience over time causes change in the DV
- often avoided with between subjects design
- if within subjects -> counterbalance
maturation
time related confound
- subjects naturally change physiologically or psychologically between treatment conditions
use comparison/control groups to control for changes in maturation
history
time related confound
- an outside event occurs between treatment conditions and affects the DV in conditions subsequent to event
use comparison/control groups
regression to the mean
time related confound
- extreme scores in first treatment condition statistically likely to become less extreme in subsequent conditions
