Quant Research Exam 1

Term Definition
internal validity results can be interpreted accurately with no plausible explanations
external validity results are generalizable to broader populations and/or conditions
positivist paradigm scientific method, cause-effect, "value free"
post positivist paradigm not "value free", eliminate alternate explanations, probability, relationships among variables influenced by external factors
constructionist paradigm emerges from data, reality is socially constructed, focus on context, own values influence (more qualitative)
transformative paradigm political nature, power, justice, change-oriented, emphasis on diversity
law of parsimony simplest form to adequately explain phenomenon
null hypothesis a hypothesis of no difference
good educational research has these qualities empirical, systematic, valid, reliable
internal reliability extent to which data collection, analysis, and interpretations are consistent given the same conditions
external reliability whether or not independent researchers can replicate results in similar settings
qualitative research inductiveunderstanding social phenomenaatheoreticalholistic inquirycontext-specificobserver-participantnarrative description
quantitative research deductiveidentifying relationships, effects, causestheory-basedfocused on specified individual variablescontext-freedetached researcher rolestatistical analysis
experimental (quant research) at least one variable manipulated to determine cause/effect, random assignment; always an independent variable
quasi-experimental (quant research) at least one variable manipulated to determine cause/effect, naturally occurring groups
non-experimental (quant research) incidence, relationships, distributions of variables studied in natural setting, no manipulation
theory generalization or series of generalizations by which we attempt to explain some phenomena in a systematic manner
constant characteristic or condition that is the same for all participants in a study
variable characteristic that takes on different values or conditions for different participants in the study
independent variable cause, influence, or affect outcomes; treatment, manipulated, predictor, classifying
dependent variable depend on independent variable; outcome or result of the influence of the independent variable; criterion, outcome, effect, response
control IV other than primary IV that may affect DV and therefore built into design
confound IV not built into design that may be affecting the DV (sometimes a limitation)
moderator IV that may affect the direction/strength of relationship between IV and DV
hypothesis conjecture or proposition about the solution to a problem, the relationship of two or more variables, or the nature of some phenomenon; needs to be justifiable, states a relationship or effect between variables, concise and testable
substantive hypotheses tentative statements about expected outcomes for variables
statistical hypotheses translations of the research hypothesis using statistical terms
directional hypothesis derived from theory or previous evidence
R randomly assigned
G group
X treatment (IV)
O outcome (DV)
levels groups or treatment
criteria for well-designed experiments adequate experimental controlpractical applicabilitybasis for comparisonadequate, uncontaminated datafreedom from confounding of relevant variablesrepresentativenessparsimony
participant use this term instead of subject because it's more humanized
construct validity clear, accurate operational definitions of constructs (IVs and DVs)
statistical conclusion validity accuracy of statistical decisions about whether the experimental and control groups differ
instrumentation different ways of measuring
maturation changing that's different
history different experience in the past
mono-operation bias only one form of IV (common threat to construct validity)
mono-method bias only one form of DV (common threat to construct validity)
confounding variables things that cause change in dependent variables
low "power" too small sample size for group comparison
data fishing plugging in various dependent variables in statistical conclusion validity threats
nominal data naming things (ex: Caucasian, Latino, etc.)
continuous data numbers of measurement (ex: temperature)
construct validity threats inadequate operational definitions, mono-operation bias, mono-method bias, participants guessing hypotheses and behaving differently, confounding variables
statistical validity threats low "power", violating assumptions of statistical tests, data fishing, technical reliability of statistical measures
Solomon four-group design combo of the posttest-only and pretest-posttest control group designs
interaction effect on the DV such that the effect of one IV changes based on the levels of the other IV
quasi-experimental research participants aren't randomly assigned, intact groups used, often more practical
reasons to use quasi-experimental design more affordable, convenient, and realistic; gives initiative for larger study/proposal
nonexperimental research ex post facto, survey research, most common in educational research; variables aren't deliberately manipulated, and random assignment is difficult
ex post facto research variables studied after the fact to examine potential relationships and effects
causal-comparative research exploring effects of and between variables in nonexperimental setting
correlational research realtionships between variables occurring in natural setting
correlation strength of relationship between variables
longitudinal survey research collection of data over time at specified points in time
cross-sectional survey research collection of data at one point in time from random sample representing the target population at that time
cohort study specific population studied over time (gender, culture, political party) and random samples drawn at various times
trend study general population over time, attitudes and opinions over time
panel study same sample measured at least twice over time, allows for study of individual changes over time, random sample selected at outset
parallel-samples design collection of data at one point in time from random sample representing a population at that time–or from two or more samples representing two or more populations

Leave a Reply

Your email address will not be published. Required fields are marked *