SelectionTypes of AssessmentBiographical DataSlide 4InterviewsWork SamplesAssessment CentersDrug TestingLetters of RecommendationPsychological Test CharacteristicsPsychological Test TypesIntegrity TestsPersonality TestsThe Big Five InventoryIntelligence TestsSlide 16Determining “Test” UtilityReliability and ValidityFactors Influencing Selection QualitySelection DecisionsEffect of Selection Ratio on Predictor UtilityEffect of Predictor Validity on Predictor UtilitySelection StrategiesSelectionTypes of AssessmentBiographical InformationInterviewsWork SamplesLetters of RecommendationPsychological TestsBiographical DataGood questions are about events that are: historicalexternaldiscretecontrollable (by the individual)verifiableequal accessjob relevantnon-invasive(Mael, 1991)Rationale vs. empirical methodBiographical DataStrong criterion validitydrug use, criminal history predicts dysfunctional police behavior (Sarchione et al., 1998)not redundant with personality (McManus & Kelly, 1999)Measurement issuesGeneralizabilityFakingFairnessPrivacy concernsInterviewsStructured vs. UnstructuredInfo. gathering vs. interpersonal behavior sampleSituational interview“How would you handle a circumstance in which you needed the help of a person you did not like?”Measurement issuesstructured has more criterion related validityvalue of unstructured? Illusion of validityGuidelines for structured interviewsWork Samplesperform a task under standardized conditionshistorically were for blue collar jobse.g. use of tools, demonstrate driving skillswhite collar examplesspeech interview for foreign worker, test of basic chemistry knowledge, Measurement issueshigh criterion validity if skills are similar to jobcostly to administerwork best with mechanical, rather than people-oriented tasksAssessment CentersRealistic tasks done in groupsAssessed by multiple of raters rating multiple domainsMultiple methodsin basket group exerciseleaderless group exerciseStrong criterion validity (e.g., teachers, police)overall scores predict job performanceMeasurement issuescostly to administerdifferent ratings on a task too highly correlateddimension ratings not correlated strongly across tasksfix? focus on behavior checklists and rater trainingDrug Testingopinion?People are more accepting of it if job involves risks to others (Paronto, et al., 2002)Measurement issuesreliability is very high, but not perfectValidity?Normands, Salyards, & Mahoney (1990)over 5000 postal service applicantsthose who tested positive had 59% higher absenteeism, 47% more likely to be firedno differences in injury or accidentsLetters of Recommendationever written a letter of recommendation for someone?worst criterion validity of all commonly used assessment toolssome use for screening extremely bad candidatesMeasurement issuesrestriction of rangewriter bias/investmentPsychological Test CharacteristicsGroup vs. individualObjective vs. open-endedPaper & pencil vs. performancePower vs. speedPsychological Test TypesAbility TestsCognitive abilityPsychomotor abilityKnowledge and skill or achievementIntegrityPersonalityEmotional IntelligenceVocational interestIntegrity TestsDesigned to predict whether employee will engage in counterproductive work behavior (CWB)overt vs. personality (covert)Better at predicting general CWB and performance than theft (r = .30 -.40)Measurement issuesdifficult to measure criteria!proprietary issueslegal and privacy issuesfakingPersonality Testsmeasures predispositions toward particular feelings and behaviorsnot all tests are based on past researchmany have shown incremental validitye.g., predict when controlling for IQMeasurement issuesjob relevancenot easily/often faked or a problem if faked (e.g., job faking too)The Big Five InventoryOpenness Highs: imaginative, creative, and to seek out cultural and educational experiences. Lows: more down-to-earth, less interest in art & more practical. Conscientiousness Highs: methodical, well organized and dutiful. Lows: less careful, less focused & more likely to be distractedExtraversionHighs: energetic and seek out the company of others. Lows (introverts): tend to be more quiet and reserved. AgreeablenessHighs: tend to be trusting, friendly and cooperative. Lows: tend to be more aggressive and less cooperative NeuroticismHighs: prone to insecurity and emotional distress. Lows: more relaxed, less emotional and less prone to distress.Intelligence TestsHave greatest validityOften very easy and inexpensive to useWonderlic Personnel Test50 items12 minute time limitSample questionsInterpreting scores?Scores vary as a function of race and ethnicityEthical issues?Face validity?Determining “Test” UtilityGoal of testing is to make decisions about individuals on the basis of the amount of a given trait they possess.A test should give us a “true” picture of a person’s traitsTest Score = True Score + ErrorReliability and ValidityReliabilityTest-retestParallel (Alternate) formsInternal ConsistencyValidityFaceContentCriterion-relatedConstruct-relatedFactors Influencing Selection QualityThree factors influence selection qualityPredictor validitySelection ratioBase rateSelection DecisionsRejectAcceptPredictor ScoreCriterion PerformanceSuccessfulPerformanceUnsuccessfulPerformancer =.60ABCDbr = .50sr = .50TrueNegative sFalsePositivesFalseNegative sr =1.00TruePositivessr = .25Effect of Selection Ratio on Predictor UtilityRejectAcceptPredictor ScoreCriterion Performancer =.40Selection Cutoff Scoresr =.50 sr =.10sr =.95Effect of Predictor Validityon Predictor UtilityPredictor ScoreCriterion Performancer =.40Selection Cutoff Scoresr =.50r =.80r =.00Selection Strategies3 Basic StrategiesMultiple RegressionAssumes relationships between predictors and criterion are linearAssumes having a lot of one attribute compensates for having little of anotherMultiple CutoffApplicants must achieve a set, minimum score on all predictorsMultiple HurdleApplicants must achieve satisfactory scores on a number of predictors that are administered over