Chapter 8: Experimental Designs: Between Subjects Design Flashcards
a _____ design uses different participants compared over each treatment condition (levels of the IV). in this type of design, only 1 score per participant is obtained because they are only exposed to one condition
in a between subjects design..
in a between subjects design, each participant is exposed to only _____ of the IV
each participant is exposed to ONLY ONE LEVEL of the IV (only one condition)
How are time related factors or confounds not an issue in between subjects design tests?
because each participant is being exposed to only one condition, experiments can be run all at the same time and the participants will not be fatigued/maturation/history will not impact them because the data is being collected once.
Advantages and Disadvantages of a between subjects design?
Pros; - can be used for a variety of research questions, -
time related factors not an issue,
-each individual score is independent and the resulting measurement is NOT CONTAMINATED from other treatment factors or carryover effects.
-Each participant enters the study fresh and naive with respect to the procedures being tested.
Cons: - requires a large group of participants- problem if working with a small population.
- May have PRE-EXISTING INDIVIDUAL DIFFERENCES. Individual or extraneous variables
- may produce CONFOUNDS or produce HIGHLY VARIABLE SCORES.
What are individual differences
personal characteristics that can differ from one participants to another (IQ, sex, hair color, experience). May pose as an extraneous variable tht can systematically differentiate the groups and thus confound the process.
How does assignment bias make individual differences into a confounding variable?
when assignment of individuals to treatment conditions produces groups with different characteristics. ex/ one treatment group is a lot older than the other treatment gorup. age thus becomes a confounding variable because it changes systematically as the IV changes (the treatment conditions change)
How can assignment bias be controlled in order to avoid individual differences becoming confounding variables?
by using EQUIVALENT GROUPS. groups making up the treatment conditions need to be:
1) created equally
2) treated equally
3) Composed of equivalent individuals.
Name methods that would help limit confounds by individual differences
1) random assignment or restricted random assignment or block randomization
2) matching groups
3) holding a variable constant.
What is restricted random assignment?
random assignment that ensures equal group size (so they’re not randomly placed in the same group)
How does random assignment or restricted random assignment help with limiting individual differences as a confounding variable?
random assignment allows for individual characteristics to be randomly distributed across groups. This minimizes potential for confounding because it is unlikely that any group is inherently systematically older, smarter or more feminine than another.
What is block randomization
a procedure for ensuring that each condition has a participant randomly assigned to it before any condition is repeated a second time
name the advantages and disadvantages as a method for limiting individual difference effects
pros: fair and unbiased, easy to perform, doesn’t require measure or direct control of extraneous variables
cons: disadvantages of random assignment: does not guarantee perfectly balanced outcomes.
- uses chance
- needs large sample
- may place some limitations on extent of CONTROL
How does group matching help with limiting individual differences as a confounding variable? What is the process of group matching?
group matching ensures that treatment groups are all matched on particular variables
Process:
1) identify variables that are likely to influence the DV
2) measurement of the matching variables
3) assignment to groups by means of the RESTRICTED random assignment to balance groups
pros and cons of group matching as a method for limiting individual difference effects
prod: easy and effective, smaller groups samples can be sued, reduces random error
cons: - must know what variables to match
- must measure variables first
- practical to match only a few variables
How does holding a variable constant help with limiting individual differences as a confounding variable?
it identifies an extraneous variable that influences the dependent variable and keeps it level for all groups. ex/ same age, same occupation.
Pros and cons of holding a variable constant as a method of limiting individual difference effects in an experiment
advantages: easy and effective, small samples can also be used, reduces random error
cons: threatens EXTERNAL validity, hard to generalize because you are control attributes of the population. also requires prior measurement of the extraneous variable.
T/F Individual differences do not contribute to the variability in scores
false: they produce high variability in scores, and can produce or obscure the treatment effects, thus undermining the purpose of the study,
the _____ of the difference must be evaluated in relation to the variance of the scores in order to see the magnitude of true effect
the absolute size of difference (ex/ percent variance, size effects)
What is the difference between within-group variance and between-group variance? In an ideal experiment, what should the relationship between these two ideas be?
between group variance: experiemental variance due to the IV, or extraneous variance due to confounding variables
within group variance: due to chance factors and individual differences.
Ideally, a research outcome should have LITTLE VARIANCE within treatments, and BIG differences BETWEEN treatments.
name 3 ways that you could limit within-group variance
1) standardize procedures to avoid chance
2) reduce individual differences via holding a variable constant, restricting a range of individuals etc.
- creates equivalent groups
- reduces variances within groups
con: limits external validity
3) sample size: larger samples may have an effect on within group variance.
T/F using random assignment and group matching will help with within-group variances
False: random assignment and matching does not affect within group variance because there is still difference within the groups. If you were to use matching so that all groups had 5 girls and 5 boys, there are still individual differences in the group itself. The participants are not uniform.
Name other threats to internal validity besides individual differences to a between subjects design
1) Differential attrition: more people may drop out of the study in one group than the other.
2) diffusion or imitation of treatment: spread of treatment effects from the experimental group to the control group– maybe the groups interact with each other and find out the differences in their treatments.
3) compensatory equilization: occurs when untreated groups learn about the treatment being received in the other group and demand the same or equal treatment
4) Compensatory RIVALRY: occurs when untreated groups learn about the treatment being received in the other group and then work EXTRA HARD TO SHOW they can perform just as well “we can be just as good”
- level of motivation is thus a confounding variable
5) Resentful DEMORALIZATION:occurs when untreated groups learn about the treatment being received in the other group and then become LESS PRODUCTIVE and LESS MOTIVATED “we give up”
- may make treatment appear more effective than it actually is.
In order to stop compensatory equalization, rivalry, demoralization or diffusion of treatment information researchers need to keep participants :
keep participants in different treatment groups away and UNAWARE from each other
What is a 2 group mean difference design?
a single-factor design that manipulates only 1 IV at 2 different levels, usually control and experimental group. Means are then computed for each group and analyzed for significance via T TEST
ex/ difference in running speed with cocaine and no cocaine
Advantages and Disadvantages of 2-group mean difference analysis
pros: simple, easy to set up, easy to interpret results, opportunity to MAXIMIZE DIFFERENCES BETWEEN TREATMENTS, increased likelyhood of obtaining significant difference.
con: little information, only 2 data points for comparison, limits options when wishing to compare treatment group to control group
- sometimes you may need multiple control groups (control no treatment and placebo fake treatment)
What statistical test is applied when calculating variances when your single-factor experiment has more than 2 treatment conditions?
ANOVA
After conducting an ANOVA test on your single-factor-multiple-treatment group experiment, it shows up as statistically significant. What test do you then perform?
a post hoc test: determines exactly which groups are significantly different from one another
Advantages and disadvantages to using a single-factor-multiple-treatment group experiment instead of a 2 group mean difference single factor experiment
pros: can discover the full and functional relationship between IV and DV, provides stronger evidence for a real causal relationship between variables to be formed
cons: possible to have too many groups
- levels should be sufficientlly different to allow for substantial differences for DV.
example of a single-factor-multiple group design experiment
a researcher may want to compare 5 different dosages of a drug to evaluate the relation between dosage and activity level for laboratory rats.
How would you compare proportions of nominal or ordinal scale data? you cannot take the mean or descriptive data.
you can classify each participant into a descriptive category, and count the participants in each category. Then use a CHI-SQUARE TEST FOR INDEPENDENCE to compare the proportions in one treatment with proportions in the other treatment conditions.