Variation, Validity,& Variables
Lesson 3
Research Methods & Statistics
nIntegral relationship
lMust consider bothduring planning
nResearch Methods
lHow data are collected
lWhat kind of data
nStatistics
lAnalysis & interpretationdepends on data & how itis collected ~
Scientific Validity
nScientific conclusions
lAbout relationships b/n variables
nValidity
lSoundness, legitimacy, truth
nInternal validity
lAbout cause & effect
nExternal (ecological) validity
lAbout broad applicability ~
How are data collected?
n2 scientific approaches
lSame or similar statistical analysis
lNOT same  confidence in conclusions
nObservational methods
lObserve co-occurrence of variables
lNaturalistic observation, case studies,archival research, surveys, etc.
nExperimental method
lManipulate a variable  observe effecton another variable ~
The Experimental Method
nAt least 2 variables:
lIndependent  (IV) & Dependent (DV)
nAt least 2 groups (levels of IV)
lcontrol group - no treatment
lexperimental - receives treatment
lrandom assignment to groups
nControl extraneous variables
lWhich might also affect DV
lWeakens internal validity  ~
Experimental Variables
nIndependent  (IV)
lPredictor (or cause)
lManipulated
nDependent (DV)
lOutcome (or effect)
lMeasured
nExtraneous variables
lOr confounding
lMight also affect outcome (DV) ~
Variation within an Experiment
nSystematic
lVariation due to manipulation of IV
lDifference between groups
nUnsystematic
lIndividual differences
lVariation due to random oruncontrolled variables
lPotentially confounding variables ~
Variation within an Experiment
Internal Validity
nLegitimacy of conclusions
labout cause & effect
nHigh internal validity
lConfident that only changes in IVcause change in DV
nLow internal validity
lConfounding variables influenceoutcome ~
Randomization
nImportant for validity
lHelps avoid bias
nRandom sampling (or selection)
lSelection of participants for study
lRepresentative sample from population
l external validity
nRandom assignment to condition (groups)
lMinimize biasing of groups
l internal validity ~
Observational vs. Experimental
nInternal vs External validity
lInverse relationship based on control
nObservational?
l internal vs  external
lCannot determine causality
nExperimental
l internal vs  external
lEstablishes cause & effect relationships
nFor useful conclusions need both  ~
Observational vs. Experimental:Statistical Methods
nMisperception
lObservational  only correlational
lExperiment  hypothesis tests
lMethod not sole determinant of analysis
n Strength of cause & effect conclusions
lObservational  weaker
lExperiment  stronger ~
Planning Research
nObservational or experimental research
nResearch design
lBetween-groups or within-subjects
nOperational definition of variables
lData categorical or quantitative
nStatistical analysis
lDepends on all of the above ~
What are data?
nInformation from measurement
ldatum = single observation
nVariables
lDimensions that can take ondifferent values
uIQ, height, shoe size, hair color
lIs not the same for all individualsbeing measured ~
Measuring Variables
nOperational definitions
lVariables often abstract
lIntelligence, anxiety, fitness, etc.
lNeed to objectively measure
nHypothesis: Exercise increases fitness
nIndependent: Exercise
lOperational definition?
nDependent: fitness
lOperational definition? ~
Levels of Measurement
nLimits type of statistical analysis possible
nQualitative
lCategorical
lFrequency data
lDiscrete: only whole numbers
nQuantitative
lContinuous or discrete
lrepresents magnitude
linfinite # intermediate values ~
Levels of Measurement: Categorical
nNominal scale
lcategorical
lorder NOT meaningful
lcan assign arbitrary values
nOrdinal scale
lCategorical + meaningful order
lNo info about magnitude ofdifferences
lIf assign numerical value, mustreflect order ~
Levels of Measurement: Quantitative
nInterval scale (numbers)
lContinuous or discrete
lEqual intervals  equal differences
nRatio scale
lsame characteristics as interval
lRatios of values must be meaningful formagnitude
lscale must have true zero point
nMost statistics: interval/ratio treated thesame ~
Levels of Measurement: SPSS
nVariable view tab
lFormatting of variable
lMeasure
nNominal scale
nOrdinal scale
nScale
lInterval & ratio
nReminder: IV must be nominal for moststatistical tests ~
Measurement Error
nDiscrepancy
lbetween actual value of observationand the reported value
nSources of measurement error
lSensitivity of measuring instrument
lConscientiousness of observer
lSurveys: inaccurate or untruthful
lLow reliability of instrument
n unsystematic variation ~
Reliability & Validity
nAccurate measurement requires both
nReliability
lConsistency of measurement
nCriterion validity
lExtent instrument actually measureswhat it claims to measure
lScore on IQ test measures intelligence?
l  pulse rate a measure of fear?
nImportant for internal & external validity ~