Quantitative Research: Group Design Flashcards
Group Design
Quantitative data, obtained through a group of subjects, determine cause-effect relationship or descriptive association between variables use 1) within subject, 2) between subject, and 3) mixed design.
Group Design- broad classification
1 or more groups of subjects, exposed to 1 or more levels of IV, driven by research question, choose in regards to a certain criteria
Single-subject design- broad classification
Focus on behavior of individual subjects, sometimes more than 1 subject examined, summarize individual performance on dependent variable.
Between-subjects design
performances of seperate groups measured, groups exposed to diff treatments/levels of IV, performance of DV compared with groups.
control group
does not receive tx, gives ability for researchers to compare results, helps most with COMPARISON
experimental group
IV is implemented (experimental tx implemented)
bivalent
2 groups (1 exp and 1 control)
multivalent
more than 1 exp group (diff. IV or diff. levels of IV) class example- freq of treatment and length of session.
Group equivalence
ensuring difference between EG and CG is effect of IV and not due to difference btw subjects in both groups
Randomization and Matching
Randomization: each particpant equal chance/probability of being assigned to one group or the other (best option because no research bias, balanced, but need a lot of people, 2 groups may not always be equivalent but research bias is reduced)
Matching: purposefully match members of 2 groups on all extraneous variables, 2 groups equivalent with respect to extraneous variables
Within-subject design
only 1 group, performance of same subjects on different conditions/levels of IV, repeated measures, subject serves as his/her control- evaluating effect of condition, requires less participants, all subjects are exposed to all levels of the IV (ex. longitudinal developmental studies. EQUIVALENCE is not a concern but there are 3…
3 concerns for within-subject design
sequencing effect, order effect, and carryover effect,
Sequencing Effect
several levels of the IV, performance in an earlier condition may influence performance on a subsequent level
Order Effect
performance +/- occur btw the beginning and end of the experiment, familiarity with tasks may influence performance
Carryover Effect
result of 1 condition may carry over to the next condition (because completed 1 task, could do better on others)
Counterbalancing
Experimenter arranged all possible sequences of condition and randomly assigns participants. (EPD, EDP, PED, PDE, DEP, DPE,)
Longitudinal Study
following same participants for period of time
Correlational Study
Diff measure obtained from same group of participants, determine relationship btw variables
Mixed Design
Both types of comparisons in same study- both WSD and BSD. design driven by research question
Single-group Design
presence of a single group, in two or more conditions, sometimes weak- no control group usually, ex- WSD
2-group design
observations of 2 groups of participants, usually assigned different levels of the IV (Tx A and Tx B)
Parallel design- 2 groups, participants are assigned to 1 of 2 groups, can be control or experimental or both tx.
cross-over design- 2 groups, participants alternate btw tx/control. each participant is their own control (more economical, less # of participants, risk of attrition)
Experimental Design
control group and randomization
Quasi Experimental Design
naturally occuring, no randomization but match them instead,
Internal Validity
Accounting for and controlling factors- (does this research answer posed questions? or provide credible evidence)
degree tp which relationship btw IV and DV is observed without the influence of extraneous variables,
control of randibility
confounding/extraneous variable (other factor which can influence extra factor)
External Validity
how much can we generalize this results? extend to other stroke patients with dysphagia?
the ability of study to extend conclusions from specific environment to other individuals with similar conditions.
Threats to internal validity
history, maturation, reactive pretest, instrumentation, differential subject selection, attrition, statistical regression, Researcher bias
History
events occuring btw 1st and 2nd measurement in addition to experimental variable
Maturation
changes in subjects- can’t be controlled by experimentor, account for neurological development, cognitive changes that could have accounted for language performance
Reactive Pretest
practice, familiarity, reduction of test anxiety, counterbalance tests, time to account for change,
Instrumentation
faulty, inadequate, calibration of equipment, assessed- method section, tools- rating scales, standardized language tests, adequacy of the survey instrument, detailed description of instrumentation- calibration techniques employed,
Differential Subject Selection
differences in subjects between groups, subject randomization, matching,
Attrition
experimental mortality (start with 5 to 10 more particpants than needed)
differenctial loss of subjects,
very common in survey research
Statistical Regression
participants selected on basis of atypical low or high scores, incorrect conclusion - tx produced change but really just regression, scores seems somewhat better or poorer than original, error of measurement, one should have good mix towards means,
Researcher Bias
subjectivity and bias, member checks, summary of points, triangulation (not just using one type of method)
External Validity
ability to generalize results to bigger population with similar condition results from sample can be extended to the population.
Threats to external validity
subject selection- characteristics of participants, representative of population to which researchers would like to generalize
independent variable- describe IV- describe levels as well, ensure replication, experimenter effects,
measurement of dependent variable- operational definition, clear definition (what do I mean by dysphagia? what are the characteristics?)
Pilot Research
small # of participants, trial run to ensure feasibility of methods, discover flaws, ensure no major surprises, details