Term
|
Definition
|
|
Term
|
Definition
|
|
Term
| An analysis of variance is conducted when |
|
Definition
| there are more than two experimental groups |
|
|
Term
|
Definition
¨Between-groups variability
Error variability
|
|
|
Term
| Between groups variability occurs (allegedly) because |
|
Definition
| of the independent variable |
|
|
Term
| Error variability occurs because |
|
Definition
of other differences within the groups (errors resulting from random assignment)
For this reason, error variability is also called within-groups variability |
|
|
Term
|
Definition
| we are only manipulating one independent variable |
|
|
Term
| If we had two or even three IVs |
|
Definition
| then we would be conducting a two-way ANOVA, or a multivariate ANOVA |
|
|
Term
|
Definition
Ex: ¨F(2, 21) = 4.71, p < .05
¨2 equals dfbg
¨21 equals dferror |
|
|
Term
| The effect size measure for ANOVA |
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
| are changes in the dependent variable based on the independent variable? |
|
|
Term
| Pre-testing lets us figure out |
|
Definition
| how much the IV actually affected the group |
|
|
Term
But, it presents some problems…. |
|
Definition
§Taking a pre-test is a form of practice, so isn’t it possible that practicing the test will affect the outcome too?
§Yes. |
|
|
Term
¡To get around the problem, we use a control group |
|
Definition
§We give one group a pre-test and then a post-test, but not the SAT class in the middle
§Then we compare how much the control group improved with how much the SAT class group improved |
|
|
Term
¡When working with single subjects, some version of AB designs are used |
|
Definition
¡A = baseline measure
B = measurement during or after treatment
¡So in it’s most simple form (AB design), we measure the baseline, introduce an IV, and then measure again |
|
|
Term
If B is different than A following treatment, how can we be sure it’s because of the IV (this is the question of internal validity)
|
|
Definition
¡We take the IV away, and then measure again
§This is now an ABA study
¡If the measurement has returned to baseline, we assume the IV had something to do with it |
|
|
Term
¡Quasi-experimental design
|
|
Definition
We can randomly select people in this case, but we cannot randomly assign
¡This violates one of the primary assumption of experimental psychology
§That the two groups are equal before the experiment starts |
|
|
Term
| Non-equivalent group design |
|
Definition
¡What I’d want to do is find two groups of people who were as similar as possible with the exception of health insurance
§Gender, SES, neighborhood, annual income, etc etc etc….
§Then I compare my treatment group (the one with insurance) to my control group (the one without) |
|
|
Term
| Non-equivalent group design(cont) |
|
Definition
¡Note that in this case, we did not assign variables randomly—we used pre-existing conditions
¡In fact, our first “observation” isn’t an observation at all, it’s just a logical way to match people
¡We must be very careful not to use the word cause in this situation
|
|
|
Term
|
Definition
4-group design includes every possible combination of pre-testing, post-testing, and IV introduction
R O O
R O X O
R O
R X O |
|
|
Term
| Interrupted Time-series design |
|
Definition
¡One way we can try to reduce the effects of a nuisance variables is with an interrupted time-series design
O O O X O O O
Basically these just involves a lot of observations before and after the manipulation of interest
Again—red-light cameras example
Problems—time consuming, expensive, etc. |
|
|
Term
|
Definition
|
|
Term
|
Definition
|
|
Term
| Population generalization |
|
Definition
| do the findings apply to people other than the ones we tested? |
|
|
Term
| Environmental generalization |
|
Definition
| do the findings apply in the “real world?” |
|
|
Term
|
Definition
| do the findings apply at different times of day/year/etc? |
|
|
Term
| Methods-Based Treats to Ex. Validity |
|
Definition
|
|
Term
•Interaction of testing and treatment |
|
Definition
▫Does the act of testing or measuring itself cause a change?
▫Pre-testing can bias performance
|
|
|
Term
•Interaction of selection and treatments |
|
Definition
| When a treatment effect is foundonly for a specific sample of participants |
|
|
Term
|
Definition
▫Is the experimental set-up so contrived that it causes people to behave unnaturally?
▫Hawthorne Experiments
▫Can people really act naturally when they know they’re being measured? |
|
|
Term
|
Definition
•Occur when participants behave based on their own agenda—they deliberately alter their behavior in some manner
▫This could be considered a threat to both a methods-based and a participant-based threat to validity—they cannot be controlled
▫Good subject, “screw you,” desirable responding, etc etc etc
|
|
|
Term
|
Definition
| The study of behavior in different species including humans |
|
|
Term
| How can we claim external validity |
|
Definition
•Replication with extension
|
|
|
