Flashcards in Exam 2 Deck (92):
A research design which you test a claim about a variable by exposing people to the variable of interest. You then note that these people feel,think or behave as expected.
Some people are just good at playing SAT, bowling etc. This deficiency in randomization makes it harder to rule out confounding variables and introduces new threats to internal validity
Problem with Pseudo Experiments
why is internal validity high in experiments?
Internal Validity is high in experiments because the only thing you are maniputaing is the IV and everything else is held constant.
A ___ has a large N, Control group, and random assignment of participants to different experimental conditions.
The extent to which the IV caused the change observed in your DV, extent to which your findings provide information about causality.
History & Maturation
Regression to the mean
Threats to internal Validity
the bias introduced by the selection of individuals, groups or data for analysis in such a way that proper randomization is not achieved, thereby ensuring that the sample obtained is not representative of the population intended to be analyzed. Sampling people from an unrepresentative samples
Type of selection bias
A type of selection bias which results from the fact that people who choose to answer surveys are systematically different from people who choose not to answer surveys. example (election of 1936 FDR vs Landon) Literary digest study
Changes that are occurring across the board in a very large group of people, changes could be due to historical event or IV. example study of PTSD during 9/11
Specific developmental changes that are occuring in a particular person or group or age cohort. Exmpl. Kids on pills start treatment at 2 at 3 child is maturin not sure if treatment or child is affecting results.
difference between maturation & history
Maturation is at an individual level, history is population level.
Tendency for people who receive high or low scores on a particular measure to score closer to the mean on later testing. example sophmore slump in sports. if you score at an extreme the first time the second time you are more likely to score closer to the mean.
Regression to the mean
two types of score
true score, observed score
a persons actual ability
persons measured ability (true score +error)
If you have lots of errors the 1st test, bound to be lest errors at the 2nd. at that point there is only one way for the score to go towards the mean. This is a better indicator of a persons
true score, actual ability
Tendency for participants to perform better on a test or personality measure the second time they take it
why people do better the second time they take a test?
Learning (IQ test)
figuring out what kinds of responses are sociably desirable (personality test)
Solution for repeated testing
pre and post assessment test
eliminate the pretest
Test for all participants with 1/2 randomly assigned to receive the treatment and 1/2 randomly assigned to not receive the treatment
pre and post assessment
problem with pre and post assessment?
you're not getting ride of repeated testing
randomly assign 1/2 people to get treatment and 1/2 not to get treatment. the random assignment will make it likely that the 2 groups were equivalent at pretest. (this relies on random assignment)
eliminate the pretest
dropping out, problem for longitudinal studies.
3 reasons why does attrition occur
people get bored, they die, they move
best kind of attrition, when people are dropping out of both groups. equal number of people dropping out of both groups (experimental & control)
when attrition differs between groups (10 in experimental, 2 in control) this one is bad because you dont know if your results are due to your IV or attrition.
How to avoid attrition?
1. communicate the importance of the study with authority and enthusiasm
2.warn people that the study may not be that exciting
3.Offer significant compensation $$$
is a type of ___ in which individuals modify an aspect of their behavior in response to their awareness of being observed. The act of studying people can dramatically change the way people behave. When people realize they're being studied they may behave in ways that they normally wouldn't.
Reaction Bias (Hawthorne Effect) name from hawthorne plant in chicago
three types of reaction bias
a) occurs when participants try to disconfirm the experimenter hyp.
b) this happens from the desire for autonomy (no one wants to be a puppet) the screw you attitude
a) Occurs when participants change their behavior because they want to be judged favorably by the experimenter. exp study of aggression display less aggressive behavior
Occurs when participants behave according to what they think the experimenter's hypotheses are. when a research subject or patient expects a given result and therefore unconsciously affects the outcome, or reports the expected result. Because this effect can significantly bias the results of experiments (especially on human subjects), double-blind methodology is used to eliminate the effect.
why does participant expectancies happens?
effort to please experimenter
to feel normal
Characteristics of the experiment itself that subtly suggest how people are expected to behave.
1. Guarantee anonymity or privacy of participants (tell them not to put their name on anything, ask them to seal their survey in an envelope)
2. Cover story- give participants the same "fake expectancy" one that has nothing to do with the real experiment
3. Unobtrusive observation- make observations without participants knowledge, use confederates
4.Make indirect measurements
Ways to reduce reaction bias
when the experimenter's expectations about the study biases his/her own experimental observations (2 types)
-Making biased observations in an experiment
-Treat participants differently based on how you expect participants to differ.
2 types of experimenter bias
see what you expect to see, and observe support for your hypothesis when an unbiased observer might not reach the same conclusion.
Making biased observations in an experiment
Can make the participants behave differently which serves to "confirm" the researchers hypothesis example. Teachers treating "bright" students differently when you actually treat participants differently. (teachers treated children differently because the child had been
said to be smart. participants were told the mice were really smart (had learned a maze before/ the other group dumb, “smarter mice completed the maze.
Treat participants differently based on how you expect participants to differ.
How to avoid experimenter bias?
deceive experimenter/ research assistants
everything is exactly the same for all participants.
stadardization (avoid experimenter bias)
You dont know what group a participant is in, have a buddy one that knows the condition and groups people are in another one who runs the subjects, double blind procedure.
Be Blind (avoid experimenter bias)
both the experimenter and the participants are unaware of the condition
tell the researcher assistant experiment is about one thing, which is not.
deceive experimenter (avoid experimenter bias)
any design problem in which some additional variable varies systematically along with the IV, due to design problems. You think the results are due to the IV but is in fact due to the ___. Usually occurs when the experiment didnt think about what other variables might be playing a role in participants behavior. (close cousin to third variable ) it varies systematically at the same time as IV
Ways to avoid confounds?
you cannot fully avoid them, but you can control for them
1. conduct true experiment (you're holding everything constant)
2. be careful ( give thought to what else might effect DV, ive thought to what else can have an effect.)
3. measure other variables- that might be confounded with the IV, you may be able to control for them
study contains only one IV
One way design
Study contains more than one IV.
Participants serve in one condition
Between subject Design
Participants serve in more than one condition
within subject design
simple experiment design on IV, two types (two group design , multiple group design.
One way design
Only one IV and this IV has two levels.
-experimental group vs. control group or
-two levels of the manipulation itself ( Group 1, receives 1 aspirin vs. Group 2 recieves 2)
two group design
only demonstrates a linear relationship. because you only have two groups you cannot see a nonlinear relation.
limitation of two group design
you only see one IV in the real world people are being effected by more than one IV.
Limitation of one way design
solves the major limitation on two group design (non linear relation limitation) you still have only one IV but the IV has multiple levels. example (IV= aspring, Levels: placebo 1, 2,3,4,5) alllows you to see non linear relations
Multiple Group Design
Contains 2 or more IV's that are completely crossed, meaning that every level of every IV appears in combination with every level of every other IV.
this is an example of
effects of light and water on plant growth
2IVs- light (alot, little) water (alot, little)
DV- plant growth
data from factorial designs is analyzed using a ___, number of IVs determines which one to use
ANOVA analysis of Variance
1. More efficient- allow us to look for more than one main effect at a time
2. More comprehensive- tell us more of the whole story behind a specific phenomenon (see different variables to influence a phenomenon)
3. More externally Valid- allow us to look at how things interact with each other since in the real world variables constantly interact.
advantage of factorial design
Participants are only in one cell, they serve in only one experimental condition
Between subject design
participants serve in more than on (perhaps even all conditions) all cells
Within subject design
Have one IV with 2 ore more levels each participants gets each level of the IV
Within-subject One way Design
Have 2 or more IV's with 2 ore more levels in each IV participant is exposed to each level of the IV.
Within-subject factorial design
How do we analyze within subject designs
Repeated Measures ANOVA
1. You dont have to recruit many people
2. you are controlling for individual differences
advantage of a within subject Design
Sometimes you want to have one of your IVs be a between subjects factor and another IV be a within-subject factor is called
Mixed Model Design
1. Order effect/sequence effect
2. Contamination effect
Disadvantage of a within subject design
people's psychological states changes as they work their way through the task (they get lazy, tired,excited to be almost done)
some aspects of peoples experiences in earlier conditions of the study influence their response in later conditions of the study. practice effect (people get better with practice) doing A influences how you do B
Solution of within subject disadvantage
A method of control in within-subject designs when you vary the order of presentation of the different conditions, you don't get rid of the order effects they may still exist but you have balanced htem across all codnitions.
Present every possible order of all conditions works when you have only 2 or 3 conditions.
complete counterblanace, incomplete counter balance are types of
If you have more than 3 conditions you don not generate all possible orders you use
2 types of ____ reverse counterbalance, partial counter balance
generate a single order, then reverse it.
The average position of any given condition in the study is exactly the same for all of the unique conditions
advantage of reverse counterbalancing
Does not control for how frequently all possible conditions occur first or last
Disadvantage of reverse counterbalancing
two types of _____ a) choose a limited number of orders at random from the pool of all possible orders (exp 10 or 12) limited number of orders b) latin square you arrange things so that each condition appears exactly once in each possible serial position
allows you to test whether two groups differ from each other. you use when you only have one IV with two levels in it.
when your t-value is larger than the critical value
reject the null hypothesis
when you t-value is smaller than the critical t value
fail to reject the null hypothesis
used to control for some variables you are afraid may be confounds, you "match" participants to each other on the variable you fear may give you some trouble. example match particiapns on smoke status and weight in an experiment effects of exercise on depression.
Matched participant procedure
Monkeys yoked with each other each time experintal monkey A chose to eat food experimentrs would feed control monkeys at the same amount of food at the same time
states that the groups are not different
any observed difference between the experimental and control groups are real and not simply due to chance.
statistics you get when you run an anova
types of ____ t-test & f-ratio, they tell you how your groups differ
false positive, Fall prey to that 5% allowance (flak alone) why replication is important, run it again unlikely you will have this error again
Type 1 error
When there is an actual difference and you do not findit, what happens is your N is probably smaller. run a power anlaysis to see if you can actually find an effect.
Type 2 error