Slide 1Evaluating ResearchConstruct ValidityStatistical ValidityInternal ValidityExternal ValidityFrequency ClaimsAssociation ClaimsCausal ClaimsThe Dirty WindshieldThe Candle vs. the FlashlightThe Four ValiditiesWill WattlesFrancis Marion UniversityEvaluating ResearchConstruct ValidityStatistical ValidityExternal ValidityInternal ValidityAssociation ClaimsCausal ClaimsFrequency ClaimsThe CandleThe Dirty WindshieldConstruct Validity•The quality of measures and manipulations. •Important for all three claims•How well have you measured each of the variables in the study?–convergent validity, –predictive validity,–Types of reliabilityStatistical Validity•The degree to which statistical conclusions are appropriate.•Frequency claims:–margin of error of the estimate•Association and Causal Claims–is it significant?–what is the effect size?Internal Validity•The degree to which there are not alternative explanations.•Frequency claims–generally not claiming cause so internal validity not an issue•Association Claims•Causal claims–temporal precedence?–control for alternative explanations–experiment•random assignmentExternal Validity•To whom, what or where can we generalize.•Frequency claims–random sample? Representative of the population?•Association Claims–how representative in the sample?•Causal Claims–In theory testing modes not a major issueFrequency Claims•Hypertension affects almost 50 million people in the U.S.Association Claims•Religious people are happier than non-religious.Causal Claims•Clinical trial shows that reducing sodium lowers blood pressure.The Dirty Windshield•Unsystematic variability–aka noise•What does it mean to find a difference despite a lot of unsystematic variability in our data?The Candle vs. the Flashlight•Do well always need a large