- Slides: 33
Personnel selection Week 6
What is personnel selection? • Traditional definition • “identifying individuals from a pool of applicants who are suitable to work in a target job position” • But technically, it can also be about selecting who to admit as students to a university
Why do we care? • Work takes up ± 1/3 of your lives • School takes up ± 2/3 of your present lives (I hope) • How you work affects others around you – they are stuck with you for ± 1/3 of their lives
What was personnel selection like in the past?
Sidetrack: Sociology/History of work • CVs didn’t exist • People get hired based on connections, wordof-mouth • Based on walk-up interviews
Why do you think people were ‘unsophisticated’ in personnel selection? Workplace • Your job candidate Neighborhoo d Districts pool was restricted, largely because of transport limitations
The modern pseudoscience of personnel selection
Lots out there 1. MBTI 2. MMPI (valid for clinical use, but not for job selection) 3. Graphology aka. handwriting analysis (our focus today)
Semantic associations in graphology • Example: If a person has consistent handwriting, he/she has consistent personality • These associations could be true, but are not. • Puzzle: Why do practitioners think the associations are true?
Why are ideas resistant to extinction? 1. They are true. 2. They are false, but they tap on some fundamental aspects of human nature
Illusory correlations • A fundamental human tendency to see patterns among random events • Could illusory correlations explain why graphology is a zombie that you can’t kill? King, R. & Koehler, D. (2002). Illusory correlations in graphological evidence. J Exp Psy.
Evidence for illusory correlation in graphology • Subjects read… 40 handwriting samples randomly paired with 40 personality profiles • For each sample, they had to judge the personality of the writer • Objective correlation should be zero • Obtained correlations, r =. 65 (large correlation) King, R. & Koehler, D. (2002). Illusory correlations in graphological evidence. J Exp Psy.
Assume MBTI & graphology are valid • Criterion validity: the extent to which two measures that purportedly measure the same constructs correlate Extraversion If: MBTI & If: Graphology Extraversion • Then MBTI’s extraversion score would correlate with graphology’s extraversion score, right? • But they don’t. Either one or both are invalid. Bayne & O’Neil. (1988). Handwriting and personality: A test of some expert graphologists' judgments. Guide
The modern science of personnel selection There is hope in humanity
Job-work analysis • What is the nature of the job? • What’s the nature of the work? • Is there a difference?
Work in your teams: What are the skills needed for each position? • • • Team 1: Mc. Donald’s restaurant manager Team 2: Psychology professor Team 3: Prime Minister of India Team 4: Sales manager Team 5: [Your ideal job] Think of the ‘soft’ and ‘hard’ skills
Measurable success • Sometimes you can call it “KPI” (key performance indicators) • But KPI neglects small day-to-day bits of work skills
Measurable hard skills • Intelligence…but what about rationality (Week 10) • Competence • Efficiency • Hands-on
Measurable soft skills • • Leadership (what’s that? ) Followership Big 5 Procrastination Interpersonal Empathy Organizational citizenship
So far… ‘Hard skills’ ‘Softskills’ Job performance
Measurement problems unique to personnel selection • Typically valid questionnaires do a good job of measuring what they are suppose to measure • But when one’s career is on the line… • People may have ‘self-presentation’ concerns (social desirability)
Measurement problems unique to personnel selection • It is difficult to remove self-presentation bias. • Some people try to remove self-presentation bias by using measures that tap into the unconscious (Rorsarch, TAT, etc. ) • We have seen in Week 4 how wrong these could go. (In Week 10, we will see how the unconscious can be properly measured. )
Measurement problems unique to personnel selection • Some companies use ‘round robin evaluation’ techniques • Also known as “ 360 -degree feedback”, “multi-rater feedback” • Subjects within the same pre-assigned group can evaluate each other. • Lots of software available to automate the calculations.
Recall: Reliability • There are many forms of reliability: test-retest, internal consistency, inter-rater • Inter-rater reliability is a concern especially in a round-robin setup
Suppose you want to predict job success. What can you do? • Because we can measure hard skills, soft skills, and job performance, we can technically do what is technically called a regression analysis. • Don’t worry if you don’t know this term; just treat it as a prediction analysis.
Example of a regression statistical output All you need to know is that there are statistical models aimed at predicting outcomes from a set of variables. DV: Worker output Model 1 c y 1 y 2 y 3 y 4 y 5 Unstandard ized 46. 3 4 0. 58 6 2. 47 0 1. 77 2 3. 27 7 2. 47 2 Standardiz Error ed 5. 94 8 1. 57 0. 07 6 0 0. 87 0. 44 2 7 1. 00 0. 26 7 7 1. 19 0. 50 7 4 1. 38 0. 29 6 6 t-value 7. 79 1 0. 37 2 2. 83 4 1. 76 0 2. 73 7 1. 78 3 p . 001 0. 71 1 0. 00 6 0. 08 3 0. 00 8 0. 07 9
Problem: Individual vs Sample vs Population • Most statistical tests are designed to generalize from sample to population Populatio n Sample • They are not designed to say something about a particular individual Individual • The problem is, the individual is often who we are concerned about
Ecological fallacy • A logical fallacy in the interpretation of statistical data where inferences about the nature of individuals are deduced from inference for the group to which those individuals belong. • Formal mathematical proof: • Don’t worry, I don’t understand the above formula either. Robinson (1950). Ecological correlations and the behavior of individuals. Am Sociol Rev.
Example: Ecological fallacy • You measured the math scores of a particular classroom and found that they had the highest average score in the district. • Later (probably at the mall) you run into one of the kids from that class and you think to yourself “she must be a math whiz. ”
Visual example Large prediction error r = + 1. 00 r = +. 70 r = +. 30 Don’t worry if you don’t know what r is. Just treat it as “perfect”, “good”, “poor correlation”, respectively.
Not all tests are useless • Remember in Week 4 we discussed cut-off scores for clinical diagnosis? • The same logic applies. If an individual does not meet a cut-off for a test (e. g. , eyesight test for pilots), then there is no debate about ecological validity.
Now, let’s suppose the ideal situation • You are a recruiter. You find a perfect match between a potential candidate and the job fit. • One problem: Job scopes changes over time • This change is especially detrimental for nontransferable ‘hard skills’ (this is why you should always seek to learn new skills even when you’re employed)
Take home messages • Selecting the right person for the job has been the aim of many organizations • Graphology is dubious, or just plain silly. • Modern scientific methods exists, but there are still limitations