Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
98 Cards in this Set
- Front
- Back
verification
|
involves operationalizing variables
|
|
naturalistic observation
|
meets descriptive goals
|
|
descriptive goals
|
ex. narratives, DSMIV criteria, naturalistic observation
|
|
goal of understanding
|
ex. CAUSES of attitude change
|
|
systemic observer error
|
caused by telling your assistant what you are proving
|
|
descriptive stats vs. inferential stats
|
descriptive = summary
inferential = hypothesis testing |
|
quasi experiment
|
done when you went to make a causal inference of the effect of the IV on the DV but can't control many of the variables
|
|
correlation
|
-is the covariation between naturally occuring variables
-measured by surveys |
|
if you can make a causal inference, what kind of validiity do you have?
|
Internal validity
|
|
internal consistency reliability
|
consistency of items in a multi item scale as shown as the average of correlation.
|
|
What does representativeness allow you to do?
|
generalize your findings
|
|
natural groups design looks like one thing but actually is another. What are these things?
|
natural grps design looks experimental but is correlational
|
|
features of repeated groups design
|
-more sensitive
-can involve practice effects |
|
longitudinal studies good for..
|
measuring change in an individual
|
|
differential transfer
|
manipulation in one condition carries over
|
|
ceiling effect
|
dependent variable reaches a maximum
|
|
what is the major threat to sucessive indep samples?
|
nonrepresentative successive sampling
|
|
differential transfer
|
manipulation in one condition carries over
|
|
parallel lines on a graph means
|
no interaction
|
|
a circle on a graph means
|
correlation value of zero
|
|
If a prof is teaching a class 2x per day, and uses an experimental technique in only one of the classes, what is the problem with his experiment?
|
intact groups
|
|
definition of interaction
|
effects of one IV on the DV
differs depending upon the level of another IV |
|
*ceiling effect of floor effect*
|
make you conclude that an INTERACTION exists when it doesn't
|
|
placebo controls are used to minimize
|
demand characteristics
|
|
double blind controls are used to minimize
|
experimenter effects
|
|
Natural groups IV
|
ex.participants peronality (introvert/extrovert)
|
|
Random groups IV
|
ex. level of arousal (ie high, medium, or low caffine)
|
|
advantages of case study method (5)
|
-develop hypotheses
-tenative evidence -evidence against a theory -useful for studying rare event -useful for trying new intervention |
|
disadvantages of case study method (5)
|
-oberver bias
-experimenter effects -very limited internal validity -limited external validity -incomplete methods |
|
ways to strengthen case studies (3)
|
-use several ways of data collection
-conuduct multiple case studies, either by replication of by finding cases that vary on IV (to see DV effects) or cases varying on DV (to see IV effects). -expand a case study with intervention to a single case experiment or interrupted time series design |
|
explanations for lack of return to baseline in ABAB (reversal) design (3)
|
-effect of treatment remains even after it is withdrawn
-confounding variable caused the effect, not the treatment -new variables might emerge to effect the behavior (ie, behavior becomes self-reinforcing) |
|
definition of multiple baseline design
|
-establish multiple baselines of a behavior,usually in diff settings
-introducte treatment in one setting at a time |
|
What evidence supports the effects of treatment in a multiple baseline design? (2)
|
- baseline remains stable in a setting before treatment applied in that setting
-behavior changes as predicted after treatment applied to that setting |
|
Does a quasi experiment involve random assignment to conditions?
|
No
|
|
Does a quasi experiment have a control group?
|
No, but sometimes a pseudo control, called a "comparison group" is used.
|
|
Does a quasi experiment involve a change in an IV?
|
yes
|
|
Quasi Experiments
|
-include an intervention (change in IV)
-no control, but comparison grp -no random assigmnemt -scarce control of potentially confounding variables |
|
How do quasi experiments handle the dependent variable?
|
Quasi exper include measures of the dependent variables taken after (and possibly before) the intervention).
|
|
History
|
A threat to internal validity in which:
-an event is confounded with treatment -ex. Hawthorne effect -History =Could something else have happened at the same time to cause the effect? |
|
Maturation
|
-A threat to internal validity in which natural changes occur in the participants over time.
-If maturation occurs, you'll see a gradual change in both exper & psedo controls |
|
Testing
|
-threat to internal validity in which taking a pretest effects the posttest.
-If you don't see a change in the comparison group from pre-to post-, it's probably not an issue. |
|
Instrumentation
|
-threat to internal validity in which measurement instruments change over time.
|
|
Two categories of Quasi experimental design
|
1. nonequivalent control group design (with or without pretest)
2. interrupted time series design |
|
Regression
|
-A threat to internal validity in which extremem scores on the pre-test are less extreme on the post-test.
|
|
If you pick the sample on the basis of extreme scores, what problem are you likely to have?
|
Regression to the mean
|
|
Selection
|
A threat to internal validity in which there are differences between the treatment and comparison group due to lack of random assignment
- If there's no comparison group, selection will not be an issue. -Selection is a big problem in nonequivalent grps design, both with and without pretest. |
|
Are the threats a problem for internal or external validity?
|
Internal
|
|
Threats cause you to believe that:
the intervention has an effect when it doesn't or the intervention does not have an effect when it does |
Threats cause you to believe that: the intervention has an effect when it doesn't
|
|
Most plausible threats to a One-group pretest posttest design (O1 * O2)
|
history, testing, mortality
|
|
Interaction of selection with other threats....
|
only happens when you have 2 grps (I'm assuming experimental and comparison).
|
|
Least plausible threats for one grp pretest/post test design.
|
maturation, instrumentation, selection, interaction of selection with other threats
|
|
Most plausible threats to a nonequivalent control grp w/out pretest
|
history (not as a main effect but as selection by history), *selection* (be/c 2 grps here), mortality.
|
|
selection is a prob be/c of random assignment but it can only occur when...
|
2 groups exist (like in nonequivalent control grps designs).
|
|
Least probable threats to int validity in a nonequiv control grp w/out pretest
|
maturation, testing (no be/c there's no pretest), regression, mortality won't be a prob if people don't leave differentially
|
|
The "testing" threat to int validity is only a major threat when...
|
A pretest and post test exist.
|
|
If people in the treatment grp leave in greater numbers than the comparison grp, the threat that is a prob is...
|
Mortality
|
|
If you choose groups based on their feelings about a subject, the threat to int validity that exists could be..
|
Regression to the mean
|
|
The most plausible threat(s) in a non-equivalent control group with pretest are...
|
history
|
|
The least plausible threats to a non-equivalent control grp with pretest?
|
selection (as long as the grps are similar on the pretest measure), maturation (be/c you'd see an effect in control and comparison if this occurs), mortality
|
|
Interupted time series
|
-quasi experiment that looks like a simple baseline design
-many measures of dv, treatment, and then more dv measures |
|
What are the major threats to interupted time series designs?
|
history and instrumentation
|
|
What are the least plausible threats to an interrupted time series design?
|
maturation, testing, regression are not threats because they would not explain the abrupt discontinuty often present
SELECTION does not apply to an interupted time series design because there are not two grps |
|
Which threat is the interrupted time series with nonequiv control grp design unique in reducing? Why
|
history. History threat is reduced with this design because an event that effects both groups cannot be responsible for the treatment effects
|
|
What are the four questions in program evaluation?
|
N. Poe
-Needs -process -outcome -efficiency |
|
Label this question of program eval:
Is the program meeting the needs of the people that it serves? |
needs
|
|
Label this question of program eval:
Is the program being implemented as planned? |
process
|
|
Label this question of program eval:
Has the program met its goals? |
outcome
|
|
Label this question of program eval:
Is the program cost-efficient relative to alternative programs? |
efficiency
|
|
What type of testing would be used to eval needs?
|
survey
|
|
What type of testing would be used to eval process?
|
observation
|
|
What type of testing would be used to eval outcome?
|
experiment or quasi experiment
|
|
What type of testing is used to evaluate efficiency?
|
experiment or quasi experiment
|
|
nomothetic approach
|
an approach to research that seeks to establish broad generalizations or laws that apply to large groups (populations) of individuals; the average, or typical, performance of the group is established
|
|
idiographic approach
|
general laws plus individual uniqueness
|
|
Kuhn
|
-scientific revolutions are due to anomolies that accumulate during normal science
-theory, concept, method |
|
using recording equipment in observational research can do one of two things (or both). What are they?
|
1.It may help deal with observer bias
2. may only postpone the prob |
|
data vs. results
|
data= responses of individual participants, issues of confidentiality are important here
results=summary of data from which conclusions are drawn |
|
Examples of direct observational methods
1.w/out intervention 2.w/intervention |
1.naturalistic observation
2.participant observation, structured observation, field experiment |
|
Examples of indirect observational methods
|
1. physical traces
-use traces=unintended -products 2.archival records -running=continuous -episodic |
|
Biases in archival records-
selective deposit vs. selective survival |
deposit = prob in production
survival = upkeep |
|
Reactivity
|
knowledge of being observed affects behavior.
People use demand cues to guide their behavior |
|
solns to reactivity (4)
|
1. allow participants to adapt to observer
2.use unobtrusive meas where the researcher isn't present 3. used disguised participant observation 4. conceal the observer |
|
observer bias
|
Systematic error due to observer expectations
ie, you're more likely to notice behaviors that you expect |
|
solns to oserver bias
|
1. automatic recording equipment
2. awareness of possible bias 3. blind observers who are not aware of the hypothesis. |
|
If the goal of your research is to describe, what kind of design will you use?
|
observational or survey
|
|
If the goal of your research is to predict, what kind of design will you use?
|
correlational design
|
|
If the goal of your research is to understand, what kind of design will you use?
|
experimental or quasi-experimental design
|
|
Reliability vs. Validity
|
reliability= consistency of measures across time (test-retest reliab) and items (internal reliability)
validity = measure assesses what is intended convergently (correlates w/ other measures of construct) and divergently (correlates poorly with unrelated measures) |
|
Survey resesarch can be used to describe or predict. What are the three types of survey research designs?
|
1. cross sectional - can't asses change be/c done at one time
2. sucessive independent samples - cross sections of diff samples; describes changes in a population 3. longitudinal design- same sample over time; describes individuals. It is the best way to asses change but people may drop out, responses may be artificially consistent, takes time and can be expensive. |
|
True experiments include...(5 things)
|
1. 2 conditions
2. random assignment 3. control of extraneous variables 4. measure effects on DV 5. must have 1 IV w/ 2 levels & 1 DV. |
|
There are three requirements for internal validity. What are they?
|
1. elimination of alternative causes
2. IV before DV (time order relationship) 3. Covariation (DV varies w/ IV) |
|
3 types of indep groups design
|
1. random grps design
2. matched grps design 3. nat'l groups - based on individual diff variable, not a true experiment |
|
Repeated measures are ____ ___ designs and indep grps designs are ____ ____ designs.
|
repeated = within
indep = between |
|
advantages of repeated measures designs
|
1. requires fewer participants, so more efficient
2. more sensitive |
|
disadvantages of repeated measures designs
|
1.can't be used if differential transfer occurs
2. practice effects (ANY change over trials) |
|
2 X 2 represents
|
2 factor design
2 IVs? 4 conditions two main effects - effect of each IV on DV 1 interaction - effect on one IV depends upon the level of the other IV |
|
How to id a main effect
|
main effects
-represent the effects of each IV considered seperately -effect of a single IV on a DV |
|
How to identify an interaction effect
|
Interaction effects:
-occur when the effect of one IV on the DV differs depending on the level of another IV -"both IVs have an effect" is NOT an interaction |