1/47
Looks like no tags are added yet.
Independent variable
manipulated(x)and random assignment
Levels
The specific variation of that factor that is actually tested. (treatment vs control)
Dependent variable
measured (Y) aka data
Between-subjects
each experience only one level of iv
within-subjects
everyone experiences all levels
random sampling
How you choose participants
random assignment
each participant has equal chance of being assigned to each group
pros - shorter, prevents carryover effects,
cons- not always possible, more people, might not be equivalent groups
Between subjects
Between
Posttest-only & Pretest-posttest
Within
Repeated-measures & Concurrent-measures
Cohen’s d values
weak(.2), moderate(.5), and strong (.8)
Counterbalancing
levels of iv are presented in different orders for different participants
Why do good experiments support causal claims?
They show the three rules of causation (temporal precedence (x before y), Covariance, internal validity)
Factorial design
2x3 (number of levels each IV has)
Main effects
averaging across levels of other ivs ( mean- add then divide by number)
Interaction
when effect of iv on dv depends on level of other iv (moderation)
Confound
A factor that systematically varies with the independent variable, so its effect cannot be separated from the manipulation (for example, coffee in the morning and decaf in the afternoon).
Selection effects
Participants at one level systematically differ from participants at another level (for example, the first 10 get caffeine and the next 10 get decaf).
Order effects
One level of the independent variable influences behavior at later levels (for example, caffeine is still in the system).
Maturation
A change in behavior that emerges spontaneously over time, such as from getting older or better (for example, brain training effects as people grow).
History
A change in behavior due to a historical or external factor that affects all participants (for example, a change in weather or air conditioning use).
Regression to the mean
An extreme finding at pretest is likely to be closer to the mean at posttest.
Attrition
When a certain kind of participant drops out of the experiment (for example, the most depressed participants).
Testing threat
A change in behavior that results from taking the test (DV) more than once, including practice effects and fatigue effects.
Instrumentation
Measurement equipment changes over time, such as a scale becoming less accurate with usage.
Observer bias
Observers see what they expect to see.
Demand characteristics
Cues in the experiment that lead participants to guess its hypothesis.
Placebo effects
A change in behavior caused only by participants believing they are receiving the treatment.
Quasi-experiment
Similar to a true experiment, except the researcher does not have full experimental control and actively selects the groups for the independent variable.
Pros of quasi-experiments
Enhanced external validity, because researchers can study real-world events and phenomena.
Cons of quasi-experiments
Increased threats to internal validity.
Small-N design
A design that gathers a lot of data from a few cases.
Pros of small-N designs
The ability to make causal claims and to control who, what, and how the study is conducted.
Cons of small-N designs
It is not always ethical or practical to manipulate some variables.
Practical significance
Whether the size of the effect is large enough to matter in the real world; consider the effect size.
r values (strength of correlation)
Weak (±.1), moderate (±.3), and strong (±.5).
Direct replication
Repeating the original study as closely as possible. It does not help external validity.
Replication-plus-extension
Repeating the original study but with an added variable.
Conceptual replication
Testing the same research question with different procedures or operationalizations.
Meta-analysis
Mathematically averaging the results of many different studies that tested the same basic question.
File-drawer problem
Researchers often do not publish studies that fail to find the results they want or find no correlation. Publishing only the positive results overestimates the effect.
Generalization
Results that apply to other people or contexts, which is important for most research.
Prioritizing goals for generalization
external validity > internal validity.
Prioritizing goals for theory testing
internal validity > external validity.
Ceiling effect
Everyone scores at the top of the test, often because the test was too easy.
Floor effect
Everyone scores at the bottom of the test, often because the test was too hard.
Descriptive statistics
Summarize outcomes using central tendency (such as the mean) and variability (such as the standard deviation).
Inferential statistics
Use data to make inferences about a population, such as null hypothesis testing and p values.