blinded by the sun? celebrity-tie bias effects on nfl ... · • if we used network of all nfl...
TRANSCRIPT
Blinded by the Sun? Celebrity-Tie
Bias Effects on NFL Coaches’ Careers
Martin Kilduff, University College London
Craig Crossland, University of Texas at Austin
Wenpin Tsai, Penn State
Matthew Bowers, University of Texas at Austin
Why do people get ahead?
Network ties as lenses that bias decision-makers’
assessments of candidates observable
qualifications
Appropriate experience and record of achievement (universalism)
The right social networkconnections (particularism)
Social Network Theory
• Hard-to-achieve network connections can
signal underlying quality to observers.
• High-status network connections function as
prisms through which the quality of the
individual is assessed (cf. Podolny, 2001).
Prior cognitive social network
research
• Cognitive social network research has been
limited to the use of data concerning
perceived networks (Brands, 2013)
• The sociology of social networks uses
cognitive theory applied more generally to
understand how actual social network
connections affect appraisals (Podolny, 2001)
Research Questions
• How do celebrity ties distort decision-makers’
views of applicants’ qualifications?
• How do celebrity ties affect employees’ career
outcomes over time?
– Scapegoating
– Demotion
Contributions to Theory
• Celebrity-tie bias concept
– Industry stars bring beneficial gains to colleagues
• Social ties as lenses
– distort perceptions of other attributes –
contribution to networks as prisms
• Balanced perspective on social capital
– the debate over benefits and drawbacks of social
connections
Celebrity Ties: a connection between a star manager, whose
name is known and renowned to everyone in the profession, and
a subordinate member of the star manager's team
• weak ties (distant and sporadic connections – Granovetter,
1973).
• interorganizational latent ties (ties between organizations
that are currently inactive – Mariotti and Delbridge, 2012).
• interpersonal dormant ties (former social ties that have
lapsed – Levin, Walter, and Murnighan, 2011).
• career imprinting (the stamp of a distinctive organizational
culture on individuals' careers – Higgins, 2005).
Celebrity-Tie Bias
People:
• see others through the halo of perceived
high-status friends
• use cognitive reference points (such as
high status people) to make sense of
complex environments
• exaggerate the importance of high-status
others
Celebrity-Tie Bias
Affiliation with high status actors is a positive predictor of career
advancement (Halgin, 2006).
Hypothesis 1: Possessing a workplace tie to a
celebrated manager is positively associated
with the likelihood of being promoted.
How is celebrity-tie bias magnified or
corrected?
People get ahead because others distort their
value based on celebrity ties?
People get punished if the great expectations
aroused by celebrity ties are not fulfilled?
Celebrity-tied promoted people reach their level
of incompetence?
Magnification: The celebrity-tie positive effect
on promotions is enhanced for individuals…
…with less rather than more industry experience.
…with a record of affiliation with successful
organizations.
Celebrity tie weakens the importance of
career experience (H2)
Biases affect judgment under uncertainty (T & K,
1982).
Celebrity ties are likely to prove most beneficial
for individuals with little relevant work
experience.
Celebrity ties confer cognitive social capital that
substitutes for experience.
Celebrity Ties Makes Good
Performance Count More (H3)
Performance of junior coaches hard to assess.
Confirmation bias will tend to attribute team
success more to coaches with celebrity ties.
Celebrity coaches as cognitive reference points.
Easier to answer: “Is this person associated with
someone whose performance is easy to
assess?”
Correction of celebrity-tie bias
• Scapegoating: Celebrity tie magnifies the
negative aspects of bad performance
• Peter Principle: People rise to their level of
incompetence
Organization’s leader is fired � celebrity-tied
subordinates exit the industry (H4)
Violation of positive expectations leads to
disappointment and unwillingness to work
with the “great expectations” individual (Rink &
Ellemers, 2012)
Individuals, not teams, are blamed for failure
(Naquin & Tynan, 2003).
High profile individuals “sacrificed” for the
collective good (Boeker, 1992).
Peter-Principle Effect• People rise to their level of incompetence due
to mismatches between job requirements and
skills
• Observers are prompted to correct their
celebrity-tie bias by penalizing the celebrity-
tied individuals.
• Celebrity-tied promoted individuals are more
likely to receive demotions relative to non-
celebrity-tied promoted individuals (H4).
Competition for new
knowledge
•Sign stealing
•Payment for injury
• NFL draft
• Revenue sharing
• Salary cap
• Schedule adjustments
• Free agency restricted
•Head coach like CEO
of 3 divisions
•NFL valued at $33B
28-32 team National Football League1980-2010, 1298 coaches, 10,269 coach-team-years
Methods
Sample
• 1298 coaches entered the NFL between 1980-
2010: our sample
• 10,269 coach-team-year observations
• Data from NFL Record and Fact Book
• College and/or non-NFL coaching and playing
Name Position
Jim Caldwell Head Coach
Clyde Christensen Offensive Coordinator
Larry Coyer Defensive Coordinator
Jim Bob Cooter Offensive Assistant
Richard Howell Assistant Strength and
Conditioning Coach
Gene Huey Running Backs
Pete Metzelaars Offensive Line
Tom Moore Senior Offensive Assistant
Mike Murphy Linebackers
Rod Perry Special Assistant to the
Defense
Ron Prince Assistant, Offensive Line
Frank Reich Quarterbacks
Ray Rychelski Special Teams
Bill Teerlinck Defensive Assistant
John Teerlinck Defensive Line
Ricky Thomas Tight Ends
Jon Torine Strength and Conditioning
Ron Turner Wide Receivers
Alan Williams Defensive Backs
2010 Indianapolis Colts Coaching Roster
Coaching Positions for 1298 coaches in Sample
1. Head Coach (86)
2. Assistant head coach (80) Senior
3. Coordinator (140)
4. Position coach 565)
5. Assistant position coach / other (427)
1st head coach = 1 in year Smith attains level 1.
Promotion = 1 in year when Smith moves up.
Celebrity tie1. Winning 1 or more superbowls (23).
2. Appearing in 2 or more superbowls (18).
3. Winning 5 or more playoff games (30).
4. Appearing in 10 or more playoff games (24).
5. Winning 100 or more regular season games
(24).
6. Being named Coach of the Year in 2 or more
years by AP or Sporting News (13).
celebrity-tie inter-item correlation = .67, α = .92
Career Performance, etc prior to
start of focal year
• Individual’s Career performance = # NFL wins
prior to focal year / total number of games
coached
• Career experience = # of NFL years coaching
• Senior coaching tenure = # of years coaching
at levels 1, 2, 3
• Head coaching tenure =total # of years at level
1
Head coach dismissal (180 left)
• the dismissal of a head coach will lead to
assistant coach’s exit? (11 others on the
roster)
• 2 independent coders, (ICC .91), 142 =
dismissals, 38 = voluntary
• Tested on sub-sample of all individuals (not
head coaches) on rosters where head coach
left during or just after the season
• NFL exit = 1 if assistant coach never coached
again
Post-Promotion Outcome• We coded the outcome of every promotion in our data (N =
630).
• An individual could conclude a promoted role in one of four
ways: 1) a further promotion with the same or another team
(N = 86); 2) a lateral move to another team (N = 133); 3) a
demotion with the same or another team (N = 204); or 4) exit
from our sample (N = 83).
• We therefore coded outcomes into positive category including
a further promotion or a lateral move, and negative category
including demotion.
• In addition, 114 coaches were still in their original promoted
roles (i.e., with the same team and at the same coaching
level) in 2010, at the end of our sample frame.
1. Celebrity tie
2. Promotion
3. First senior
4. First HC
5. Career performance
6. Career experience
7. Senior tenure
8. HC tenure
9. Team performance
10. Exit
11. Team HC dismissal
12. Age of coach
13. Non-NFL experience
14. NFL playing tenure
15. NFL playing success (AP All-Pro
Team
16. QB college
17. QB NFL
18. Roster size = # coaches
19. Team total (# diff. teams)
20. NFC (vs. AFC)
21. Team age (# years in NFL)
22. Team past performance
23. Team super bowls
24. Super bowl recency
25. Previous HC co-working
26. IMR (inverse Mills Ratio)
27. Calendar year
Endogeneity
• Coaches seeking employment are likely to
perceive all 32 teams as potentially successful.
• Probit model to regress the binary celebrity tie
variable on variables likely to affect coach hire.
• Includes predictors not included in second
stage model: QB NFL, QB College
• Inverse Mills ratio (IMR) used in our second-
stage model.
DV: Celebrity tie
Constant -2.96** (0.18)
Age -0.14** (0.01)
Playing experience 0.00 (0.00)
Playing success 0.05 (0.10)
QB college 0.15** (0.05)
QB NFL 0.22* (0.11)
Career experience 0.09** (0.01)
Non-NFL experience -0.02 (0.02)
Career performance 2.19** (0.13)
Roster size 0.04** (0.01)
Team total 0.23** (0.01)
Log likelihood -4392.09
Chi-squared 3785.18**
First-stage Probit Model for Selection
Bias Correction
Analyses
• Binary DVS, so we used random-effects logistic
regression, fixed effects logit models after
omitting all coach-level variables (that don’t
change year to year.
• interaction effects: we created interaction
terms by mean-centering and multiplying,
respectively, career experience and career
performance with celebrity tie.
Results
• The longer an individual stays in the NFL, the
greater the chance of attaining a celebrity tie
(r = .53, p < .01).
• Positive team performance protects people
from industry exit (r = -17, p < .01)
Celebrity Ties and Career Success (H1)
Possessing a celebrity tie to a successful head
coach is positively associated with:
- being promoted (β = 0.39, p < .01);
- attaining first senior coaching position
(β = 0.55, p < .01);
- attaining first head coaching position
(β = 1.11, p < .01).
Did Celebrity Tie Facilitate Promotion?
• Overall promotion probability increased 45%.
– mean probability of being promoted in any given
coach-year was 3.81% for individuals without
celebrity ties and 5.54% for those with celebrity
ties.
• probability of receiving a first senior
promotion increased by 73%
• probability of receiving a first head coaching
position increased by 200%
Did a celebrity tie weaken the importance
of career experience? (H2)?
–being promoted (β = -0.02, ns)
–1st senior coaching position (β = -0.09, p < .01);
–1st head coaching position (β = -0.09, ns).
Did Celebrity Tie Make Good
Performance Count More? (H3)
Having a celebrity tie boosted how career performance affected:
–being promoted (β = 2.18, p < .01);
–1st senior coaching position (β=2.47, p < .05);
–1st head coaching position (β=5.55, p < 05).
Did celebrity tie enhance the negative
aspects of bad performance? (H4)
The dismissal of a head coach predicted
assistant coach’s exit (β=0.54, p < 05).
• Probability of exiting the NFL was 59% higher
for those with celebrity ties
• probability 15.4% vs. 9.7%
Peter Principle? Celebrity-tied promotions
more likely to end in demotion? (H5)
Coaches with
celebrity ties
Coaches without
celebrity ties
Total
Promotion or
lateral move71 (46.1%) 158 (56.6%) 229
Demotion 83 (53.9) 121 (43.4%) 204
Total 154 279 433
Chi-square test (χ2 = 4.413, df = 1; p < .05), Fisher’s exact test (p < .05).At coach-year level, pairwise correlation between celebrity-tied promotionand positive outcome was negative and significant (r = -.07, p < .01).
Robustness ChecksResults unchanged :
• if all individual-level control variables omitted and fixed effects used
instead of random effects.
• If we used network of all NFL coaches from 1980 onward (N = 1565) rather
than just those that first entered the NFL during or after 1980 (N = 1298).
• If we replace binary celebrity tie measure with an ordinal measure. For
each coach-year where celebrity tie was coded as zero, we made no
changes. For coach-years where the celebrity tie was coded as one, we
replaced this with the number of years that the individual had worked
with the sender of the celebrity tie (i.e., the relevant successful head
coach). If an individual had worked with multiple senders, we used the
sender that the individual had spent the most time working with. This
alternative measure provides an indication of the strength of the celebrity
tie.
Robustness of celebrity tie coding
• If the general celebrity tie measure replaced with, in turn,
each of the six component measures, results were consistent
for five of these measures (Super Bowl wins, Super Bowl
appearances, playoff wins, playoff appearances, and 100
wins), but were not robust to the use of the celebrity tie
(coach of the year) measure on its own.
• Instead of including all 36 coaches that had achieved a single
milestone (Table 3), we restricted our sample to only those 22
coaches that had achieved at least four of the six milestones.
Our findings were again unchanged.
Did celebrity ties represent knowledge transfer?
• Assuming rational observers perceive greater knowledge
transfer when there was a knowledge-match between the two
parties.
• A celebrity tie in a given coach-year was classified as a
“knowledge match” if the celebrity tie-sending coach’s
historical experience was in the same domain as the tie-
receiving coach. If not, the tie was categorized as a “no-
knowledge match.”
• both knowledge match ties and no-knowledge match ties
separately predicted promotions, first senior promotions, and
first head coaching appointments.
• regression coefficients for the two different types of celebrity
tie were not significantly different for any of the three models
(χ2 = 1.20; χ2 = 1.02; χ2 = 0.01; all ns).
Did celebrity coaches continue to pass on expertise to
ex-members of their coaching staff for years after they
leave?
• If so, celebrity-tie effects persist for many years.
• We re-coded data to record the length of time (in
years) that had passed since an individual coach had
first received their most recent celebrity tie.
• Working under a successful coach has no
incremental impact on promotion likelihood after the
first year in which the subordinate coach is out on his
own across all promotions, first senior promotions,
and first head coach promotions.
Summary of Results
• A celebrity tie to a successful head coach helped individuals
earn promotions in the NFL, including to senior positions and
head coach.
• For those individuals with little career experience (relative to
individuals with lots of experience), celebrity ties increased
the chances of achieving first senior appointments.
• The successes of celebrity-tied individuals counted for more in
promotion tournaments.
• Dark side of such celebrity connections: coaches with
celebrity ties were more likely to exit from the NFL following
the dismissal of the head coaches under whom they worked.
• Celebrity-tied coaches tended to get promoted beyond their
abilities relative to promoted coaches without the benefit of
celebrity ties.
Discussion
• What is new: celebrity tie as a biasing lens
that magnifies the importance of human
capital and exposes the individual to
scapegoating and demotion
• Does it pay to have connections to people
who are celebrated industry leaders (cf. Burt,
2010)?
Contribution
• A new theoretical direction social network
research: ties as lenses that distort individuals’
qualities both beneficially and detrimentally
• Different from cognitive social structure
research that focuses on misperceptions of
network ties
Contributions
• to the social capital approach that has
emphasized the effects of social network ties
on job mobility and achievement (e.g., Lin,
Cook, & Burt, 2001).
• To executive succession research: an
executive’s prior associations with celebrity
managers may help determine whether he or
she will rise to the top of a firm.
Practical Implications
• celebrity ties may have a “dark side” for both
organizations and individuals.
• social connections rather than just skills and
abilities enable people to move into positions
such as head coach.
• lessons learned concerning recruitment of
players have failed to be applied to hiring and
promoting coaching staff.
Limitations
• Special nature of the NFL
• Absence of specific coach performance
metrics
• No data on cognitive biases
Conclusion
• Affiliation with a successful manager can
facilitate or damage career progress
• Surprising given that competitive markets
reduce social network effects
• In making momentous decisions – such as
appointing a senior executive – judgments
concerning human capital can be swayed by
celebrity affiliations