Exam 1 Flashcards
What t-test for a between subjects design
Independent t-test
What t-test for a within-subjects design
Dependent t-test
Between subjects Design
participants receive only one level of the IV
Within subjects design
participants receive every level of the IV
Non-experiment designs
What, Who, How much, How often
- interviews, observations, surveys
Manipulated variable
researcher controls, assigns people to conditions; experimental claims; IV
Measured variable
observed or recorded; used in frequency or association claims; DV
Association Claim
One level of a variable is likely to be associated with one level of another variable; 2 or more correlated variables
Frequency claims
particular rate or degree of a single variable; only one measured variable
- rates, percentages, proportions
Strength of Association
The closer to zero, the weaker the association. The closer to one(positive or negative) the stronger the association
Positive association
variables more in the same direction; both negative or both positive
-less sleep is associated with lower exam grades
Negative association
variables move in opposite directions
Zero association
variables move randomly, no link between them
- how many burgers you eat is not related to exam grades
Construct validity
how well did they measure the variables; two conceptual variables need to be operationalized appropriately
External validity
Does the claim generalize across populations, contexts, times, places
Statistical validity
extent to which the statistical conclusions are accurate and support the causal claim
Type 1
false positive (test says you’re pregnant when you aren’t)
Type 2
false negative(test says you aren’t pregnant when you are)
Three criteria for causation
covariance, temporal precedence, internal validity
Covariance
if there is no relationship, there cannot be an affect
Temporal precedence
Can establish that X causes Y because X comes before Y; manipulated variable comes before measured variable in time
True Experiments
researcher randomly manipulates the independent variables; randomly assigns participants
Quasi-experimental
random assignment not used; variables treated as IV even though the researcher didn’t manipulate
- we sometimes cannot control or ethically manipulate a variable(gender, relationship, status)
Control variable
experiment holds constant on purpose
-ensures that only thing influence the DV is the IV
Unsystematic variability
random/haphazard; affects both groups
Matched groups
better for small samples; match groups on some variable (IQ Example)
Independent Groups/Between Subjects Again
Posttest only/ pretestposttest
Posttest only
Participants are randomly assigned to IV groups and are tested on the DV once
Pretest/posttest
Randomly assigned to at least 3 different groups and are tested on the key dependent variable twice-once before and once after exposure to the IV
Within Groups Designs Types
Concurrent and repeated measures
Concurrent measures
participants are exposed to all levels of the IV at the same time, and a single preference is the DV
Repeated Measures
Participants are measured on the DV more than once after each exposure to the IV
Within Groups Advanttages
participants are their own control
- fewer participants
- researches have more power to notice differences
Within Groups Disadvantages
- potential order effects
- not practical or possible at times
- experimenting all levels of the IV changes how participants act
Demand Characteristics
participants pick up on cues that lead them to guess the hypothesis(within groups)
Manipulation Check
an extra Dv is used to convince that the manipulation worked
Pilot study
using a separate group of participants before or after the study
Internal validity is emphasized most
in order to achieve experimental control, researches sometimes conduct their studies in artificial laboratories that may not represent the real world
Effect size
- can help determine covariance
- the larger the effect size, the stronger the causal effect
- the coefficient quantifies how far apart two groups are on the DV
Three fundamental questions about internal validity
Were there design confounds?
If an independent groups was used, did the researches control for selection effects using random assignment or matching?
For within groups, did researchers control for order effects by counterbalancing?
Order effects
being exposed to one condition affects how participants respond to other conditions; avoided by counterbalancing
Counterbalancing
presenting levels of the IV in different orders
Full counterbalancing
all possible conditions are presented
- two conditions, 3 orders; 3 conditions, 6 orders etc
Partial counterbalancing
only some of the possible condition orders are used
Rules for hypotheses
- must be mutually exclusive(one is correct and one is wrong)
- encompass all outcomes