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;
64 Cards in this Set
- Front
- Back
3 Scientific Method Steps
|
1. Hypothesis (propose relat bw 2 variables - A effect B)
2. Obj define (specify what should be observed) 3. Collect and analyze data |
|
Independent variable
|
Experimental - IV - Input - event or tx manip by researchers
Correl - no manip - Predictor |
|
Dependent variable
|
Experimental - DV- Output - What is hypo to change as result of manip to IV
Correl - Criterion |
|
Research question
|
What effect does (IV) have on (DV)?
|
|
Levels of the IV
|
One tx (IV) and 3 levels (drug, psychtx, dru plus psychtx)
|
|
Multiple IVs - Factorial Design
|
Combine every level of one IV with another IV and its levels - 2 levels of scz with 3 levels of tx (2 x 3 factorial design)
|
|
Internal Validity
|
Its possible to determine whether a causal relation exist bw the IV and DV (rather than extraneous) - or if a study does NOT have a causal relat
|
|
One group, pre/post test design
|
Very poor internal validity bc effects are largely due to extraneous variables (AKA confound or study is confounded)
|
|
Major principle of ensuring validity
|
Equivalence
|
|
Threats to internal validity: History
|
Any external event that effects scores or status on DV
|
|
Threats to internal validity: Maturation
|
Internal (bio or psy) change that occurs in subj while the exp is in progress and exerts a systematic effect on DV.
Ex. Fatigue, boredom, hunger, physical or intellectual devel |
|
Threats to internal validity: Testing
|
Pre/post test in one group - Experience on pre effects post not IV
|
|
Threats to internal validity: Instrumentation
|
Change in DV bc of nature of measuring instrument has changed (Asmt ability improved) Use highly reliable meas instru
|
|
Threats to internal validity: Statistical Regression
|
Tendency of extreme (high/low) scores fall closer to the mean upon re-testing - dep appear less dep after re-testing due to stat regres
|
|
Threats to internal validity: Selection
|
Pre-existing subj factors (motiv, self-esteem, intell)
|
|
Threats to internal validity: Differential Mortality
|
2+ groups - drop out differs in systematic ways from people who remain in study
|
|
Threats to internal validity: Experimental Bias
|
1. Communicate expectations
2. Pygmalion effect (teachers preconceived notions of students ability results in increased grades and IQ) 3. Makes errors in direction to confirm hypoth - double blind tech - control this problem |
|
Equivalence Techniques: Random Assignment
|
Most powerful method
The prob of being assigned to a group is the same for all subj in experiment - all extraneous variable are distributed equally |
|
Equivalence Techniques: Matching
|
Id subj who are similar in terms of status on extra variables (pretest) then grouping similar and randomly assign members of matched group to tx groups.
Used when sample size is small and random cannot ensure equiv |
|
Equivalence Techniques: Blocking
|
Making an extraneous variable an IV - how they work in combo - interaction and how they work separately
|
|
Equivalence Techniques: Holding the extraneous variable constant
|
Using only people who are homogenous in the terms of extra variable in control and experimental groups. Therefore diff not attrib to extra
Disadv: cannot be generalized to populations not sampled |
|
Equivalence Techniques: ANCOVA (analysis of variance)
|
Statistically adjusting data - post hoc matching - scores are adjusted so that subj are equalized - only controls ID extra variables
|
|
External Validity
|
Generalizability of results of study - to settings, time, people
|
|
Threats to external validity: Interaction bw Selection and Tx
|
"Interaction" variable has a effect on one but diff on other - not generalizeable.
Tx would not gen to other pop |
|
Threats to external validity: Interaction bw History and Treatment
|
Do not generalize beyond setting, time period which experiment was done
|
|
Threats to external validity: Testing and treatment
|
Not gen in where pretests not used - Pretest sensitization (sensitize to purpose of research and increase response to tx)
|
|
Threats to external validity: Demand characteristics
|
Cues in research setting to guess hypoth - behave diff than in real world to prove or disprove hypoth
|
|
Threats to external validity: Hawthorne Effect
|
Know they are participating and bx diff
|
|
Threats to external validity: Order Effects
|
Repeated measures design - expose to more than one tx
|
|
Ways to increase External Validity: Random Selection
|
Ensure can be gen to pop
Stratified Random Sampling: Take random sample from subgroup of target pop Cluster Sampling: Selecting from natural group of indiv. |
|
Ways to increase External Validity: Naturalistic Research
|
Obs and recorded in nat setting or similar to natural - lack internal validity
Analogue research - make analogies about real world phenom from lab -lack external valid |
|
Ways to increase External Validity: Single and double blind research
|
Single: Sbj not informed
Double: experimenter and and sbj not informed - demand and hawthorne effect |
|
Ways to increase External Validity: Counterbalancing
|
Controlling order effects - diff sbj or groups receive tx in diff order.
Latin square design each order appears once. |
|
True experimental research
|
Random assign and diff levels of IV - Greatest internal validity (experimenter has control over conditions)
|
|
Quasi-experimental research
|
No random assign - use existing groups - 2nd best internal validity
|
|
Correlational Research
|
Variables are measured not controlled
Does NOT have internal validity - cannot infer causal relationship - researchers has no control Association is bw variables is point of interest for the purpose of prediction -status of one variable (predictor) can predict the status of another (criterion |
|
Ways to increase External Validity: Single and double blind research
|
Single: Sbj not informed
Double: experimenter and and sbj not informed - demand and hawthorne effect |
|
Ways to increase External Validity: Counterbalancing
|
Controlling order effects - diff sbj or groups receive tx in diff order.
Latin square design each order appears once. |
|
True experimental research
|
Random assign and diff levels of IV - Greatest internal validity (experimenter has control over conditions)
|
|
Quasi-experimental research
|
No random assign - use existing groups - 2nd best internal validity
|
|
Correlational Research
|
Variables are measured not controlled
Does NOT have internal validity - cannot infer causal relationship - researchers has no control Association is bw variables is point of interest for the purpose of prediction -status of one variable (predictor) can predict the status of another (criterion) |
|
Correlational Research
|
Collect data on variables and apply correlational statistics.
|
|
Developmental Research: 3 types
|
Longitudinal, cross-sectional, and cross sequential
|
|
Longitudinal Design
|
Same people studied over long time.
Problems: High cost, drop out, practice effects, low level of age related decline bc those that stay in have higher ability, true age-related declines in perf may be masked. |
|
Cross-Sectional Design
|
Diff groups of sbj divided by age are studied at same time.
Problem cohort effects (diff may have to do w/ experience) - overestimate true age-related declines in perf |
|
Cross-Sequential Design
|
Combines Longit and cross sect. Diff groups assessed on 2+ occassions.
Control for cohort effect, less time and drop out |
|
Time Series Design
|
Taking mult measure over time (diff pre-post test)
Rule out threats to internal valid: maturation, regression and testing Threat: is history - control using 2 group (control group). |
|
Single subject design
|
Bx modif: variability in target bx threat
AB, Reversal, and Multiple baseline designs |
|
AB design
|
Single baseline and single tx phase - compounded
|
|
Reversal (W/D) Design
|
controls extra factor. Tx is w/d to see if return to baseline to be sure if tx works - ABA, ABAB (adv: confirm tx effic, if end in baseline leaving sbj in original condition.
|
|
Multiple Baseline design
|
Used when reversal is not possible - cant w/d tx
Apply tx sequentially across diff baselines - diff bx, diff settings, diff sbj |
|
Qualitative (descriptive) Research
|
Theory devel from data - pilot study
Methods: participant obs, nonparticipant, obs, interview, surveys, case studies |
|
Surveys
|
Attitude meas, consumer pref, worker satis.
3 Tech: Personal interview, telephone, mail. Internal valid: not prob bc not causal relat is hypoth External valid: random sampling, mail is biased |
|
Case Studies
|
Can be viewed as an example of a more general class.
No relation bw variables - lack in/external valid Useful as pilot |
|
Protocol Analysis
|
Collect and analy verbatim reports - based on interp
|
|
Greatest Threat to Interval Validity when a Mail Survey is conducted?
|
Selection: Self select to reply
|
|
Regarding the diff types of devel research this is most true
|
Combining the method of cross sect and longit - cross sequent reduce many prob assoc w/ both
|
|
Defining feature of a true exp design is
|
Random selection of sbj from pop
|
|
Francis Galton's concept of regression to the mean is best expressed by this statement
|
Short fathers have taller sons - Regress - extreme obs to be less extreme upon re-test or re-obs. Indiv
|
|
Conclusion of Hawthorne's Study is best described by
|
Any change in the working env that results in increased attention for workers may result in increased productivity
|
|
Is not true about multiple baseline designs
|
Involve admin and w/d of tx
coorect sequent applic of tx across diff baselines (bx, setting, sbj) |
|
Major Threat to Internal Validity in a one group time series design is
|
History - involves admin mult pre/post test - any external event that occurs while tx is admin
|
|
Major adv of case studies
|
Used to ID variables for future research
|
|
Major disadv of case studies is
|
Do not permit conclusion to be made about the causal relationship bw variables
|