HR T1 - Assessment Centers

Arthur

Overview of AC process

Sorting and organizing observations; generation of initial dimension-level scores for the exercise

Assessors proceed to generateparticipants dimension-level scored

Assessors observe and record behaviors

Providing feedback to the participants

Participants go through exercises

AC =on one hand it reflects one's potential; it is a valid predictor. On the other hand we have doubts whether it measures what is supposed to measure

Steps in building an AC

Select/develop exercises to measure dimensions

Train assessors and administrators

Identify behavioral dimensions related to the KSAO's

Pilot the AC

Identify KSAO's / construct major underlying work ehaviors

Refine as warranted

Determine major wok behavior

Implement the AC

Job analysis

Assessors = should by psychologists = they are better at detecting behaviors

observe max 2 people

each game- different observation targets

use max 4-6 dimensions, not (too) related with each other

focus on representing the dimensions rather than workplace

SCORING:

within exercises

across exercises

score after each game

scored after completing the entire batter of tests

Highouse

beliefs

Variance i success of prediction

20% Technical

70% Unpredictable

10% Pschological

it is believeed that variance in measurement is due to bad measurement - in fact it is just our lack of knowledge

unstructured interview is the most popular measurement method, and it sucks = believe in horse-sense

there is belief that subjective measures>objective measures

myth of selection expertise - in fact there is no such effect

2 arguments in favor: Holisticity and Broken-leg problem

Expert ratings

lack insight into how they arrive at their decisions

poor inter judge agreement

rely on few pieces of information

more confident when irrelevant info is introduced

Ryan

Why HR practitioners have gaps in their knowledge?

Pressure form the companies to bring new candidates ASAP

Some stakeholder perceptions of methods is not aligned with practitioners

research is abstract and is lacking context(hard to apply)

HR professionals find legal requirements confusing

research not accessible and full of jargon

Methods of selection (in bold- most predictive). For racial biases check the paper (AND FOKIN DO IT)

Conscientiousness

Biographical info

Job knowledge tests

Sitiational judgment

WORK SAMPLES

Integrity test

Unstructures Interviews

AC

STRUCTURED INTERVIEWS

Reference checks

COGNITIVE TESTS

Referrals + Cognitive tests = best predictive combo BUT huge diversity issues

people from referrals 50% less likely to turnover relative to people from job ads

you can use different ways of presenting info (f.e. video) but then you affect content validity

THERE IS A GREAT LIST OF HR PRACTICES IN THE PAPER< READ IT

Best ways to reduce Adverse Impact of selection tools

Not using cognitive ability tests

Using tests focusing on citizenship behaviors/Teamwork in addition to task performance

Target recruitment ads at qualified minority groups

Use the tools with little adverse impact at the beginning and more adverse at the end of selection procedure

Ployhart

Situational judgment tests; present work-related situations and then ask how you should behave - usually multidimensional tests

types

"should do"

"would do"

maximum performance test = ability and knowledge

typical performance test

Development of SJT's

  1. Scoring
  1. Generate Situations

critical incidents at work / context specific situations

theory-driven situations

  1. Response Option generation

best if generated by job experts

4-6 responses per situation

Either choose best or put them on continuum from best to worst

Other Issues:

stem complexity = the more complex the situation the more racial bias, becasue of verba comprehension

branching stems

Stem fidelity=realism. the higher the better (situations can be put on videos)

would vs should (both question formats correlate with different constructs)

would do = personality constructs

should do= cognitive ability and knowledge = this one is more valid

Racial differences for SIT are as big as for Cognitive tests

SIT = moderate predictive ability

lowest bias goes for integrity tests and Personality (Conscientiousness)

you can lower the bias by reducing the verbalization = for example instead of providing stories on the paper in Situational Judgment tests, provide a short video

AC's vary in this matter, results vary from 0 to 0.5 in discrimination index (where 1 would be cognitive tests)

lowest bias = personality tests = conscientiousness and integrity. (with exception for biographical data where there is apparently stron white-black bias

Medium bias = work samples, interview, job knowledge, situational judgment, AC, work samples

Strong bias = cognitive tests and other similar. can be reduced when assessed non-verbally (but this is also a complicated issue)

You can also lower the adverse impact by introducing low-adverse-impact measurements at the beginning of the selection, and more adverse at the end of the selection