Problem 2
EVALUATION SELECTION METHODS
Criterion: what do we want to predict? job performance
Predictor: with what do we want to predict the criterion? unstructured interview,
psychological tests
DISCRIMINATION
1. Measurement procedures provide clear discrimination between candidates,
discrimination should be based on job-relevant criteria, not irrelevant factors
like gender or ethnicity.
2. Job selection: most likely to perform best on the job (suitable vs unsuitable)
3. Discriminatory power: distinguish between candidates based on their
expected future performance.
4. Selection methods : first come first serve, random selection, weighted
random selection, selection procedure focused on selecting the best.
LEGALITY & FAIRNESS
1. Measures must not discriminate unfairly against members of any specific
subgroup of the population (ethnic minorities).
2. Intentional or direct discrimination: on relevant/irrelevant features that are
especially common in one type of race or gender.
3. Adverse impact/unintentional or indirect discrimination: same rule applies to
everyone, but it affects some groups more than others a use 80% rule.
VALIDITY AND RELIABILITY: test can be reliable but not valid but can’t be the other
way around (not reliable=not valid)
1. Adequate technical qualities of measurement procedures
2. Validation process
o Privotal stage: selection decision is taken; candidate is offered or not the job
o Job analysis alone cannot prove that candidate characteristics will do better
o Validation: finding out whether parts of selection process were effective
o Always do ‘cross-validation’ study on second sample of people to cross-check the
results obtained, especially when having many predictors.
Criterion-related validity:
CONCURRENT: predictor and criterion are collected at the same time in current
employees. The test scores are correlated with recent performance.
(+) less time consuming (+) Not having to hire people with ‘bad’ scores
(-) Different population than actual applicants
(-) These people may already have acquired skills on the job
PREDICTIVE: predictor measured before criterion. All in the sample are hired. After
time has passed, criteria are assessed. (+) Vigorous (-) Practical issues (money, time)
(-) Hard to ignore predictor data during selection
, Other forms of validity
- Incremental validity: whether adding other predictor adds more validity to the test
- Face validity: when something looks right/ logical link between test & job
performance. ADVANTAGE: positive impact on acceptability= otherwise applicant
may take legal action.
- Content validity: no validity studies, validity is based on a logical basis (car-driving
ability expected to include all essential activities for competent driver like emergency
stop and reversing)
- Construct validity: whether measures what was intended
o Identify psychological characteristics that underlie successful performance of task
Reliability: extent to which instrument measures consistently under varying
conditions (done many times and have same outcome)
o High reliability then relatively free from measurement errors (score obtained is
close to ‘true’ score)
How to use the predictor in selection procedures
-The multiple hurdle approach: For each predictor, a passing score has to be met=
hurdle is passed. Hurdles are expensive = the company spends most money on those
with high potential.
- Regression approach: use an equation to combine all predictors to provide a
numerical forecast of the criterion.
Low scores = overcome by combining the 2 approaches.
Personal selection & assessment methods
1. Interviews: - Long been the most popular method - Used by every organization
- Either 1 interviewer/ panel interview - Structured > unstructured
Unstructured: Primacy/recency effect/ Social dynamics influence conclusions
Sensitive to bias/ Focus on negative aspects= look for proof of negative aspects
Structured: preplanned structure/high proportion of job -related questions
Take care of design/execution = valid & reliable
Used by nearly every organization and for selecting employees at all levels
Panel interview: when several interviewers are involved
Most important features: extent to which pre-planned structure is followed &
proportion of questions that are directly related to the job
Criticized for being subjective, unreliable and vulnerable to bias.
o Structure and interviewee trained to enhance validity
Forms of structured interviews
o Behavioral interview (measures typical performance): asking interviewees to
describe pervious behavior in past situations that are relevant to the job
= looking for evidence that applicants are capable of handling similar job situations
Rate responses to such questions with behaviorally anchored rating scale (BARS)
- Ensure consistency - Anchors are based on requirements - Little bias
EVALUATION SELECTION METHODS
Criterion: what do we want to predict? job performance
Predictor: with what do we want to predict the criterion? unstructured interview,
psychological tests
DISCRIMINATION
1. Measurement procedures provide clear discrimination between candidates,
discrimination should be based on job-relevant criteria, not irrelevant factors
like gender or ethnicity.
2. Job selection: most likely to perform best on the job (suitable vs unsuitable)
3. Discriminatory power: distinguish between candidates based on their
expected future performance.
4. Selection methods : first come first serve, random selection, weighted
random selection, selection procedure focused on selecting the best.
LEGALITY & FAIRNESS
1. Measures must not discriminate unfairly against members of any specific
subgroup of the population (ethnic minorities).
2. Intentional or direct discrimination: on relevant/irrelevant features that are
especially common in one type of race or gender.
3. Adverse impact/unintentional or indirect discrimination: same rule applies to
everyone, but it affects some groups more than others a use 80% rule.
VALIDITY AND RELIABILITY: test can be reliable but not valid but can’t be the other
way around (not reliable=not valid)
1. Adequate technical qualities of measurement procedures
2. Validation process
o Privotal stage: selection decision is taken; candidate is offered or not the job
o Job analysis alone cannot prove that candidate characteristics will do better
o Validation: finding out whether parts of selection process were effective
o Always do ‘cross-validation’ study on second sample of people to cross-check the
results obtained, especially when having many predictors.
Criterion-related validity:
CONCURRENT: predictor and criterion are collected at the same time in current
employees. The test scores are correlated with recent performance.
(+) less time consuming (+) Not having to hire people with ‘bad’ scores
(-) Different population than actual applicants
(-) These people may already have acquired skills on the job
PREDICTIVE: predictor measured before criterion. All in the sample are hired. After
time has passed, criteria are assessed. (+) Vigorous (-) Practical issues (money, time)
(-) Hard to ignore predictor data during selection
, Other forms of validity
- Incremental validity: whether adding other predictor adds more validity to the test
- Face validity: when something looks right/ logical link between test & job
performance. ADVANTAGE: positive impact on acceptability= otherwise applicant
may take legal action.
- Content validity: no validity studies, validity is based on a logical basis (car-driving
ability expected to include all essential activities for competent driver like emergency
stop and reversing)
- Construct validity: whether measures what was intended
o Identify psychological characteristics that underlie successful performance of task
Reliability: extent to which instrument measures consistently under varying
conditions (done many times and have same outcome)
o High reliability then relatively free from measurement errors (score obtained is
close to ‘true’ score)
How to use the predictor in selection procedures
-The multiple hurdle approach: For each predictor, a passing score has to be met=
hurdle is passed. Hurdles are expensive = the company spends most money on those
with high potential.
- Regression approach: use an equation to combine all predictors to provide a
numerical forecast of the criterion.
Low scores = overcome by combining the 2 approaches.
Personal selection & assessment methods
1. Interviews: - Long been the most popular method - Used by every organization
- Either 1 interviewer/ panel interview - Structured > unstructured
Unstructured: Primacy/recency effect/ Social dynamics influence conclusions
Sensitive to bias/ Focus on negative aspects= look for proof of negative aspects
Structured: preplanned structure/high proportion of job -related questions
Take care of design/execution = valid & reliable
Used by nearly every organization and for selecting employees at all levels
Panel interview: when several interviewers are involved
Most important features: extent to which pre-planned structure is followed &
proportion of questions that are directly related to the job
Criticized for being subjective, unreliable and vulnerable to bias.
o Structure and interviewee trained to enhance validity
Forms of structured interviews
o Behavioral interview (measures typical performance): asking interviewees to
describe pervious behavior in past situations that are relevant to the job
= looking for evidence that applicants are capable of handling similar job situations
Rate responses to such questions with behaviorally anchored rating scale (BARS)
- Ensure consistency - Anchors are based on requirements - Little bias
