ORM Week 2: Causality
I. Student presentation
II. Overview lecture/discussion
III. Discussion of Moyle paper
Moyle: The role of negative affectivity in the stress process: Tests of alternative models
Purpose
1. Setup: Purpose to examine role of NA in relations between stressors and strains.
2. Approach: Comparison of competing mechanisms/hypotheses.
3. Gaps in literature: Help to support/refute various suggested mechanisms
Hypotheses
1. Five hypotheses derived from suggested mechanisms
2. Mechanisms imply causal connections although no causal tests
3. How many variables? Triplets: stressors, strains, NA
Workload, workload fluctuation, control, social support, time orientation
Job satisfaction, well-being (GHQ)
4. Basis for hypotheses: Mechanisms with supporting literature
Method
1. Cross-sectional survey, single source
2. n = 143, heterogeneous in UK
3. Measures reliable? All but workload fluctuation (alpha = .61)
4. Established measures, yes
5. Not much detail on how data collected
6. Limits to sample generalizability? [limited occupations]
7. Limits to measurement generalizability [limited stressors/strains]
8. Ethical issues? Anonymous?
Results
1. Data analytic approach [Correlation/regression. Patterns to support hypotheses.]
Correlation, mediation, moderation
2. Logical sense. Yes, logical progression from simple to complex
3. Different patterns for job satisfaction and well-being
4. Not overly complex
Discussion
1. What was concluded? [Support for all mechanisms but more for WB than JS
2. Conclusions logical and not stretched.
[Assumes reduction from partialing due to bias and might not be]
3. Are there plausible alternative explanations (more than given here?)
4. Limitations discussed? [Yes, limits to sample generalizabililty and causal conclusions]
General
1. What conclusions do you personally reach from this—what convinced you?
2. Paper well-written and communicate clearly?
3. How to improve the study?
4. Where to go from here—how to follow-up?
Week 3: Validity and Threats To Validity
1. Student presentations of articles
2. Student presentation of hypothesis/variables
Next step: Choose measures
Overlap in measures
3. Lecture/discussion
4. Discussion of Frese paper
I. Purpose: To rule out plausible alternatives to the stressorsstrain idea
II. Approach
Specify alternative mechanisms/possibilities
Introduce design features to address alternatives
“Objective measures”
Longitudinal design
III. Method
Design
Two cross-sectional, 1 longitudinal
Cross-sectional: Self, observer, group (mean of incumbents in same job)
Longitudinal
Measures
Psychological stressors (RA, RC, constraints, noise, danger)
Physical stressors (physical demands and workload)
Psychosomatic complaints
Third variables: Perceived employment alternatives, Leisure time stressors,
spouse lack of support, political exaggeration (union activities), SES
IV. Results
Test for third variables accounting for stressorstrain
Objective stressors relate to strains
Good support for psychological stressor (Table 3)
Less support for physical stressor (but more than claimed)
Note halo effect for observers
Argue superiority of group rating since incumbents have more information
Third variables account for stressorstrain
Partialled specific variables from stressorstrain correlations
Little impact of partialling on correlations (Table 4)
Note that other third variables might yield different results
Reverse causation
Longitudinal data with cross-lagged panel
Comparison of cross-lags
More support for stressorstrain than opposite
Results not entirely consistent
Individual differences as moderators
Tendency to over- or under-report compared to colleagues in same job
Not much evidence for moderator
V. Conclusions
Effect size about r = .20
Argues this is important and expected: Inus condition—other factors and health
Argues for validity of self-report symptoms
Relate to physical illness and mortality
Tend to be under-reports rather than over-reports
Support that stressorsstrain
Some differences found between psychological and physical stressors
Inability of plausible alternatives to rule out causal conclusion
VI. Overall reaction to this paper
What do you conclude?
Frese’s arguments convincing?
What could be done to improve study?
Could his methodology/approach be used for different questions?
Week 4: Construct and External Validity
1. Student presentation of articles
2. Student presentation of scales/discussion of project
3. Lecture/discussion
4. Van Dyne article
A. Purpose
Refine construct of organizational psychological ownership
Develop a scale
B. Introduction
Heavy on conceptual/theoretical development
Are hypotheses nomological net or test of ownership theory?
C. Method
Three samples
n = 186, self-reports
n = 409, self-reports
n = 227, 2 peers OCB, supervisor OCB, performance, test-retest
Ownership scale
7-item
3 not supported by CFA but used anyway
Good alpha, good test-retest (.72)
High correlations with commitment (.68, .70, .40)
D. Results
Supported some but not all hypotheses
Related to commit, job satis., organization-based self-esteem
Related to perform (.15), and OCB (.26, .30)
Incremental validity over attitudes for OCB but not job performance
E. Controlled for demographics
“…because maturity, gender, and tenure can influence performance” p. 450
Some weak relations became nonsignificant after controls
No test for significant reduction
Ownership-performance r = .15 (p < .05), regression with controls b = .15 (p < .06)
F. Discussion/conclusions
Convincing evidence for validity of scale?
Convincing evidence for viability of construct?
Weaknesses of study
What could be improved?
Strengths of study
What should be done next?
Week 5: Quasi-experimental Designs
1. Student presentations of articles
2. Project: Should bring copies of scales to coordinate
3. Lecture/discussion of method variance
4. Discuss Shadish and Stone-Romero
a. Reactions/questions on readings.
b. One group post-test: Any cases useful in I/O?
c. Case control design: Where useful in I/O?
d. Stone suggests lab is as generalizeable as field. Do you agree?
e. Is I/O field biased against lab as Stone suggests? If so why a third lab studies?
5. Discuss Barling, Weber & Kelloway
A: Purpose
To demonstrate that charisma can be trained.
To show impact of leadership change on unit performance
B. Introduction
Short
Makes case for importance of transformational leadership
Notes gap in literature showing trainability and effectiveness
C. Method
20 branches of a national bank in Canada
Multifactor Leadership Questionnaire completed by subordinates
Organizational commitment of subordinates
Financial outcomes for branch
Two group (treated vs. controls) Pretest-posttest design
Treatment was 1-day group training session and 4 individual boosters
D. Results
Treated group seen as more transformational
Treated group increased commitment
Unit performance improved?
Based on comparison of posttests
Table 2 shows credit card sales going down pretest-posttest for trained
E. Discussion
What do authors conclude?
Issue of small sample (9 trained vs. 11 control managers)
Levels issue with commitment: Mean commitment per group
Optimal time frame for posttest (5 months for this study)
F. Conclusions
Strengths of study
What could be improved?
Convinced transformational leadership can be trained?
Convinced it improves performance?
Is it the training content or just that there was training?
Week 6: Quasi-experimental Designs 2
1. Student presentation of articles.
2. Project: Time for students to discuss with their groups
3. Lecture/discussion
4. Student questions
5. Discuss Ludwig
A. Purpose: Determine effectiveness of goal setting on driver safety
B. Background
Problem
Pizza deliverers have 3x national average accident rate.
$millions due to injuries/fatalities
Factors in accidents
Young inexperienced drivers 18-24 years old
Pressure for fast delivery
Commissions
Solution
Goal setting
Literature suggests it is effective
Response generalization
Impact of intervention generalizes beyond target behavior
C: Method
Subjects
n = 324 deliverers
College students, mean age 21
3 stores, 1 per condition
Observations
Unobtrusive windows of nearby businesses
Observed seatbelt use, turn signals, kind of stop (full, rolling, etc.)
Design
Nonequivalent control group with multiple baseline (6 weeks)
Participative: Discussion-based meeting, group feedback
Assigned group: Lecture-based meeting, group feedback
Control: Nothing
D. Results
Manipulation check: Post meeting questionnaire
Participative reported more participation
Compared baseline, intervention, withdrawal of intervention, follow-up
n = 40 with complete data
Complete stops
Goal setting groups improved but returned to baseline
Turn signal
Participative increased and maintained
Control, assigned no increase
Safety belt
Participative increased and maintained
Control, assigned no increase
Note initial group differences
Participative almost double controls, assigned in middle
E. Discussion
Goal setting effective
Participation and assigned equal in target behavior
Participation worked on other nontarget behaviors
Assigned didn’t work and even declined
Major contribution
Looking at more than target behavior
Argues intrinsic vs. extrinsic motivation
Cites Deci-Ryan but not counter evidence
F. Limitations
What could improve the study
Was it goal setting or feedback?
What are plausible alternative explanations
Threats with nonequivalent designs
Pre-test differences
Possible mechanisms for response generalization
Compliance vs. acceptance
Week 7: Longitudinal Designs
1. Student presentations
2. Projects. Questionnaires should be done and ready to copy/distribute
3. Lecture/discussion
4. Student questions
5. Discuss Laschinger
A. Purpose: Determine if empowerment leads to job satisfaction
B. Background
Distinguishes structural vs. psychological empowerment
Structural: Perceptions of conditions that provide power
Psychological: Reactions to structural
[Distinction unclear—both are psychological states]
Empowerment leads to job satisfaction
Nursing in Canada heavily downsized
Empowerment needed to maintain job satisfaction
C. Method
Subjects: 185 nurses
Two-wave, 3 year study
Scales
Measure of both powers and work satisfaction
Structural: 6 areas of empowerment—rate how much
[Not clear on what dimensions are precisely]
Psychological: 4 areas—doesn’t say how rated
[Overlap with JDS: meaningful work, autonomy, impact]
D: Results
Doesn’t report zero-order correlations
Notes little relation of demographics with variables of interest
SEM test of empowermentjob satisfaction
Supports structuraljob satisfaction
Mediation of psychological not supported
E: Discussion
Concludes evidence found for causal effect
Little evidence for mediation
Found for cross-sectional time 1 data
Suggests method variance but not how
Perhaps relations just smaller over time
F: Limitations
Nature of empowerment not precise
Distinction of both empowerment types not clear
Aren’t both psychological?
Empowerment scales overlap with job characteristics
What is really measured?
How does empowerment fit with other measures of control/power?
Not very thoughtful discussion of longitudinal aspect
Why 3 years?
Reason to expect change in job satisfaction?
Rule out plausible alternatives?
Method variance oversimplified in limitations
Week 8: Experimental Design
1. Student presentations.
2. Projects. Data collection should be underway.
3. Lecture/discussion.
4. Discuss Bedeian.
1. What’s his major point: Articles don’t reflect what was really done.
Introduction distortion of researcher’s thinking.
Forced structure of article obscures real story.
Worst case is hypothesis as if it drives study.
Little role for serendipity.
2. Do you share Bedeian’s concern?
3. What would be an alternative?
4. Would the alternative help?
5. Don’t we all know what’s really going on, so what does it matter?
5. Discuss Wainer: Nonsampling error (statistical reasoning).
1. What is nonsampling error?
2. What is the problem with the longevity study?
Sample of death certificates that included occupation.
Why do students die youngest?
3. Model of nonresponse.
Bullet holes in returning aircraft.
Assume places there weren’t holes reflected vulnerable spots.
Those not returning shot there.
4. Study of unemployment.
Collect sample of unemployed.
How do we know they weren’t different to begin with?
How to fix the design?
Pretests to compare those fired from those not.
Week 9: Survey Methods
1. Student presentations
2. Projects. Data collection should be underway
3. Lecture/discussion of papers
Week 10: Theory
Discussion of Sutton-Staw
Very strong bias toward necessity for theory
APA task force argued against premature theory
1. Do you agree that theory is more important than data?
2. What is a theory?
Distinguish from model or framework
Jerry Greenberg—CWB model a framework
Fails to explain how/why things relate
How does emotion lead to CWB
3. Can social science have the same sort of theory as natural science?
Can we workout mechanisms the same way?
4. Is content of Journal of Applied Psychology dustbowl empiricism?
Strawman—they suggest most journals/work devoid of theory
5. p. 381, makes Bedeian’s point that what’s reported isn’t exactly what’s done
6. p. 381, Assumes good theory well tested will be instant classic. What is good?
7. Argues for value of strong conceptual arguments in theory.
Doesn’t acknowledge importance of data and empirical testing.
By this standard is Goleman’s EI better than other research?
8. p. 383, mentions Freud and Marx as what we should aspire to
Notes current journals wouldn’t have published them.
Are Freud-Marx really science?
Are they testable?
S&S arguing against science?
Is there a distinction between scholarship and science?
Locke
1. Philosophy of science
Where do deductive theories come from?
Is there really a deductive theory?
Qualitative vs. quantitative basis
Does science really advance entirely by falsification?
Where are all the falsified theories?
2. Are major scientific advances based on inductive methods?
Newton as induction led to Newtonian mechanics
Decartes as deduction led to wrong ideas
E.g., soul meets the body in the pineal gland
3. “Unfortunately, social scientists learned the wrong thing from the hard sciences…rejection of introspection as a legitimate method. Its rejection robbed psychology of a key source of data for inductive theorizing.”
4. “…as the inductive method came increasingly under attack…researchers often had to pretend that they had theories before they had a firm basis for any. This method makes for quick and often short-lived theories.”
5. Recommendations
a. Develop a substantial body of observations using many methods
b. Make use of introspection as data.
c. Look for evidence of causality
d. Identify boundary conditions
e. Be thorough and patient
Hambrick
1. How do you like the writing style? Is it effective?
Refers to field’s orientation to theory as
Fetish, obsession
Uses religious metaphors “holy water” p. 1347.
2. Do you agree that pure empirical fact finding studies should be published?
3. How do you know if findings are theoretically interesting?
4. Can you argue the other side with Hambrick?
Week 11: Levels
1. Student presentations
2. Data should be collected for projects
3. Lecture on levels
4. Discussion of Hofmann
Types of group constructs p. 251
Global: Can’t be divided, e.g., profit
Shared: Combined individual, e.g., climate
Configural: Arises from interactions, football team
Says gender breakdown is configurational. Is it?
Forms of composition
Additive: Sum of individual scores
Direct consensus: Agreement, e.g., norms
Referent-shift: Asking about group rather than self (how do people in your group?)
Dispersion: Level of agreement, variance on variable within group
Process: Maps individual output to group output. Unclear.
Data Analysis
ICC(1) and HLM most popular
WABA never really caught on
6. Discussion of Pirola-Merlo & Mann
a. Purpose: Explore individual and team creativity
b. Background
Creativity studied at both individual and team level
Lack of integration between levels
Creativity based on aggregate of individual beliefs/perceptions
Is team creativity additive (sum of individuals) or disjunctive (best individual)?
c. Theoretical suggestion
Individual creativity raw material of novel ideas
Team process determines how raw material is adopted/used
Team can facilitate or inhibit creative behavior
d. Method
Subjects: 56 R&D teams with 319 individuals
Questionnaire study
Team climate
Organizational encouragement of innovation
Ratings of individual/team creativity (self and team leader)
e. Results
rwg low (<.70) for some teams/variables, but others were high.
Some relations of climate with creativity
Organizational encouragement strongest r
HLM supported some links with climate
Various creativity measures intercorrelated
f. Discussion
Climate is shared based on agreement
Climate affects both individual and team creativity
Team creativity interaction of individual creativity and climate
Team creativity largely aggregate of individuals
Week 13: Reviews and Meta-analysis
1. Student presentations
2. Projects—written version due in two weeks at next class
3. Lecture on reviews
4. Discussion of Judge, Heller, & Mount
a. Purpose: Explore FFM and job satisfaction
Mean rs of each FFM dimension and job satisfaction
Tested 3 moderators
Cross-sectional vs. longitudinal
Direct FFM measures vs. indirect
Type of job satisfaction measure used
b: Background
Cites Spector in noting literature mostly descriptive
Notes theory development impeded by lack of framework of personality structure
Criticizes focus on NA-PA and job satisfaction
NA-PA independence debate
Other traits may also be important
States consensus has emerged on FFM
Explains why each FFM dimension should relate to job satisfaction.
Do you buy the notion that the problem is lack of framework?
Has consensus emerged?
Should focus be on FFM rather than more specific variables like NA-PA?
Do proposed links make sense?
c. Method
Thorough procedure to locate studies
k = 163
Personality variables placed into FFM using Barrick & Mount procedure
Normal employed adult samples
Individual level not group
d. Analyses
S-H approach
Adjusted for unreliability using observed or estimated values
e. Results
Mean rs
Neuroticism-.24
Extraversion.19
Openness.01
Agreeableness.13
Conscientiousness.20
Moderators
Job satisfaction measure, especially on neuroticism (Hoppock largest r)
Little impact of direct/indirect measure
Little impact of cross-sectional vs. longitudinal
f. Conclusions
Neuroticism strongest correlate—linked with extraversion happy personality
Conscientiousness 2nd strongest—explains away negative rs as sampling error
Suggests link through performance but fails to note job type moderator
Agreeableness low and inconsistent findings—unexplained
Openness half studies positive and half negative rs—unexplained
Links results to life satisfaction showing similar correlations with FFM
Suggests PA is extraversion—but why are rs different from Connolly-Vish?
Argues NA measure is quasi-dispositional and less stable and confounded with life satis