Sampling Behaviour with TasksoAlternative to behaviours that can’t be observed directlyoCreate tasks that seem to sample behaviours of interestE.g. digit span for memory – give sequence of numbers and gradually increaseSelf-Reports
oChildren’s answers to questions about the topic of interest.oRetrospective reports are subject to response or memory bias and forgetfulnessPhysiological MeasuresoHeart rate portrays degree of attention; it slows down when paying attentionoCortisol levels in saliva, measure stressoBrain activity can be associated with reasoning, memory, emotions, etc.Evaluating MeasuresoValidity shows measurement measures what it is supposed to measureoConstruct validity – test measures the theoretical construct it is supposed to measureoConcurrent validity – when two forms of measurements correspond and validate each otherRepresentative SamplingoValid measurements also depend on thisoBroad groups of subjects are being sampled and have to be accurately representedCorrelational ResearchoRelation between variables as they exist naturallyEstablish a correlational coefficient btw -1 and 10 – no correlation1 – strong positive correlation-1 – strong negative correlation, i.e. inverse relationshipoStatistical research and survey studiesUsually one-on-one research instead of survey formsKids are more fond of truth-telling than adultsExpensiveoParent, teacher and peer reportsE.g. personality surveys (parents may lie about kids’ conscientiousness)E.g. popularity leveloRetrospective reportsProblem: hard to get info from subject about past due to memory loss and biasSolution: prospective studies (follow subject from young to older ages)E.g. studies show that feminine boys correlate with gay menoCan’t make causal statementsNeed independent variables that can be manipulatedExperimental ResearchoDependent and independent variables have to be definedIndependent variable is independent of subjectoRandom assignment – causationEnsures even study groupsYou can’t randomly assign age, parenting style, social environment on case studiesoInferential statistics allows conclusions to be drawnoNull hypothesis states that nothing the experimenter did had an effect on resultsThere has to be a 95% chance that results are gotten from experimental manipulation. i.e. 0.05 probability that chance was involvedoTypes of Experiments
True lab experiments – random assignmentField experimentsresearch manipulates I.V. in a natural setting for proper conclusionsdrawbacks: loss of control for researcher and unaccounted variablesQuasi experiments– random assignments not possible on at least 1 variable.