PSY 3711 1st Edition Lecture 7Outline of Last Lecture I. RecruitmentII. 3 key dimensions of employer informationIII. Personnel selectionIV. Correct vs incorrect hiring decisionsOutline of Current Lecture I. Psychological measurementII. How are constructs related to measures?Current LectureI. Psychological measurementa. Construct: hypothetical, non-measureable attribute of person (e.g. happiness, job performance)b. Theory: hypothetical relationship between constructsc. Operational definition: the “real world” way of measuring a construct. Predictor and criterion measure fit hered. Hypothesis: the “real world” way of testing theorye. Use circles to indicate construct, use box to indicate what is used to measure the constructII. How are constructs related to measures?a. Reliability: how much of the differences in scores reflect true differences vs. error?i. Whenever we take a measurement there will be error ii. Any measurement (data point) is some combination of a “true score” and an error componentiii. Observed score= true score + erroriv. Reliability provides an index of how much an observation is the “true”part expressed as a correlationv. How consistently is the operational definition measuredvi. Different kinds of error  different times, different items, different scoresvii. Different kinds of reliability measure different kinds of errorviii. Depending on what kind of measures we have and the reason we’re measures  we should different kinds of reliabilityThese notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.ix. Error1. Transient error: inconsistent situational effects (e.g. mood, sickness, noise, temperature, randomness)  test-retest reliability 2. Item sample: effects of the particular items that were chosen  equivalent/parallel forms reliability (test taker takes two forms of the same test) or internal consistency reliability (how much do items on the same form relate to the others; cronbach alpha3. Rater sampling: see different behaviors, interpret behaviors differently, idiosyncratic preferences and biases interrater reliability (correlation between the scores given to the same person by two raters) a. Interrater reliability  people are really bad at making ratingsb. Most interrater reliabilities are very low  job performance .51, extraversion .41, emotional stability .34c. Reliablity is higher with more familiarity, observing same behaviors, and very specific rating rulesx. If reliability is too low  more items, more raters, aggregate more info about the construct we want to measure (MORE) or training for raters (BETTER)xi. Types of reliability  test-retest, alternate forms, internal consistency,inter-raterb. Validity: does the measure assess the thing we want to assess? i. Construct validity: do all the operational definitions represent the constructs they should?1. Convergent  does this test correlate with other measures of the same thing2. Divergent does it not measure the things it shouldn’t be measuring3. Assessed using the correlations between measuresii. Content validity: does the substance of the operational definition match the construct1. Are all important parts of the construct measured2. Assessed using content analysisiii. Criterion related validity (most important in I/O): does the measure predict important criteria?1. Assessed using the correlation between the predictor and criterion measures2. Do relationships between predictors and criteria reflect theory?3. Predictive validity studies  take test, hire, then assess test later4. Concurrent validity studies  people already in the