Running Head: HORMONES AND ETHICS
Online Supplemental Materials
Title: Hormones and Ethics: Understanding the Biological Basis of Unethical Conduct
Authors: Jooa Julia Lee,1 Francesca Gino,1 Ellie Shuo Jin,2 Leslie Karen Rice2, Robert Alan Josephs2
Affiliations:
1
Harvard University
2
The University of Texas at Austin
* Correspondence to: bob.josephs@utexas.edu
Hormones and Ethics
2
Methods
Procedure to Measure Unethical Behavior
In this performance task, which we called “the Numbers Game,” participants were asked to solve 20 math problems. Each
problem consisted of a matrix of 12 sets of three-digit numbers. This task, which has been validated in previous behavioral ethics
research (Gino, Schweitzer, Mead, & Ariely, 2011), was selected because it provides participants with a simple and transparent means
of evaluating their arithmetic accuracy and thus does not afford participants an attributional “escape” (e.g., “The numbers I claimed
were correct didn’t match my actual performance because I wasn’t certain of my actual performance”).
To perform the math task, each participant was handed two sheets of papers. Participants were instructed to carefully read the
first sheet – that is, the collection slip, which included one practice example of the 20 matrixes that they would be solving on a second
sheet (the test sheet). For each matrix, the goal was to circle the two three-digit numbers embedded in a 12-number matrix that
summed to 10 (e.g., 4.19 and 5.81). Prior to completing the performance task, the participants completed a practice matrix to ensure
participants understood the task goal. Once they indicated that they understood the task, they were told to move on to the test sheet,
which contained the 20 matrices. Participants were told that they had exactly four minutes to solve as many of the 20 matrices as they
could. After four minutes had elapsed, participants were instructed to stop working, count the number of matrices on the test sheet
they had solved correctly, fold the test sheet in half, being told to make sure to remember how many matrices they had solved, and
place the test sheet through the slot of a recycling box. Participants were then asked by the experimenter to write down the total
number of correctly solved matrices on the collection slip. The experimenter looked at the collection slip and then paid participants
Hormones and Ethics
3
based on how many correct matrices they claimed to have solved (one dollar per matrix correctly solved for a maximum payment of
$20).
After the participants were debriefed and dismissed, research assistants who were blinded to the research hypothesis retrieved
their test sheets (identified, as mentioned earlier by the matching code on the collection slip), and calculated the actual number of
correctly solved matrices for each participant. This number was subtracted from the number of correct matrices claimed by the
participant. A positive discrepancy was used as our indicator of unethical behavior.
Hormones Data Collection, Assays, and Data Reduction
Potential participants were screened using a health questionnaire in order to exclude those who did not qualify due to their
conditions (pregnancy or suspected pregnancy) or because they were taking medications known to affect hormone activity. All
participants who passed the screening process were reminded that they should not: i) eat dairy products (e.g., milk, cream, or cheese)
or anything containing live bacterial cultures (e.g., yogurt, kefir); ii) consume caffeine or alcohol; iii) smoke cigarettes; iv) exercise; or
v) brush their teeth for at least one hour before coming to the laboratory. To increase compliance, we sent participants a follow-up
reminder email a day before their scheduled lab visit. Upon arrival at the lab, participants rinsed their mouths thoroughly with water to
facilitate procurement of a clean saliva sample.
After participants arrived in the lab, they read and signed consent forms, rinsed their mouths thoroughly, and then waited for
10 minutes prior to saliva collection. All saliva samples were collected between 1:30 p.m. and 6:00 p.m. to minimize diurnal variation
in testosterone and cortisol levels. We collected two saliva samples from each participant using the passive drool method, in which
Hormones and Ethics
4
participants drooled at least 1.5 mL of saliva into a sterile polypropylene microtubule using a straw. Once collected, samples were
immediately frozen and stored at –25°C. When data collection was completed, the samples from the pilot study were shipped on dry
ice to Salimetrics (State College, Pennsylvania) and the samples from the main study were shipped to the Clinical
Neuroendocrinology Lab at the University of Texas at Austin, where they were immuno-assayed in duplicate (by two of the authors –
LKR and ESJ) for testosterone and cortisol concentrations using Salimetrics Research Kits (State College, Pennsylvania).
All samples were assayed in the laboratory over the span of one week. All samples were thawed, vortexed, and centrifuged at
3,000 rpm for 15 minutes to collect saliva mucins at the bottom of the collection tube. Consistent with prior research (Mehta &
Josephs, 2010) intra-assay coefficients of variation (CV) and inter-assay CVs for cortisol and testosterone were each below 11%
across both studies, values which fit comfortably within the acceptable range.
Also consistent with previous research (Mehta, Jones, & Josephs, 2008; Popma et al., 2007), levels of pre-performance
testosterone and cortisol were positively correlated (combined sample: r=0.26, p=0.004; Males only: r=0.33, p=0.006; Females only:
r=0.53, p=0.001). Time of day was not correlated with either pre-performance testosterone or cortisol, ps>0.47. Further, neither pretask nor post-task cortisol levels were correlated with any self-reported emotions, trait anxiety, state anxiety, or state negative affect.
Testosterone. The assay used 25 μl of saliva per determination and has a lower limit of sensitivity of 1.0 pg/mL, a standard
curve ranged from 6.1 pg/mL to 600 pg/mL for the pilot study, and from 17.12 to 213.22 pg/mL for the main study. Mean and
standard deviations of the pre-performance testosterone are reported in Table S1 and S3, respectively.
As expected, there was a significant gender difference; males had higher testosterone levels (M=153.26 μg/dl, SD=8.36) than
Hormones and Ethics
5
females (M=72.95 μg/dl, SD=3.76), t(83)=9.67, p<0.001, d=2.12, for the pilot study. Similarly, there was a significant gender
difference in testosterone concentration; males had higher pre-performance testosterone levels (M=95.07, SD=43.54) than females
(M=49.75, SD=23.34), t(115)=6.67, p<0.001, d=1.24, for the main study. Averages for males and females fell within the normal
ranges, as reported by Salimetrics. To compare across sex, we z-scored testosterone levels separately for males and females (Josephs,
Sellers, Newman, & Mehta, 2006).
Cortisol The test used 25 μl of saliva per determination, had a lower limit of sensitivity of 0.003 μg/dl, a standard curve range
from 0.012 to 3.0 μg/dl for the pilot study, and from 0.01 to 0.58 pg/mL for the main study. When we looked at the samples by time
point in the main study, the mean of pre-performance cortisol concentration was 0.17 pg/mL (SD=0.11) with a range of 0.01 to 0.65
pg/mL, while the mean of the post-performance cortisol concentration was 0.18 pg/mL (SD=0.11) with a range of 0.00 to 0.77 pg/mL.
Cortisol levels were in the normal expected range reported by Salimetrics in both studies.
There was no significant difference between males (M=0.16 μg/dl, SD=0.02) and females (M=0.15 μg/dl, SD=0.01),
t(82)=0.65, n.s, d=0.14, in the pilot study. No sex difference in pre-performance cortisol levels was found (males: M=0.17, SD=0.09;
females: M=0.18, SD=0.13), t(115)=–0.89, p=0.37, d=–0.20, nor post-performance cortisol levels between males (M=0.18, SD=0.12)
and females (M=0.17, SD=0.08), t(115)=0.54, p=0.59, d=0.05 in the main study.
Cortisol typically generates a skewed distribution (e.g., Popma et al., 2007), and our data were no exception (skewness for preperformance cortisol=1.90 in the pilot study, pre-performance cortisol=1.72 and for post-performance cortisol=2.26 in the main
study). To normalize, cortisol values were log transformed and subsequently z-scored.
Hormones and Ethics
6
References
Gino, F., Schweitzer, M. E., Mead, N. L., & Ariely, D. (2011). Unable to resist temptation: How self-control depletion promotes
unethical behavior. Organizational Behavior and Human Decision Processes, 115(2), 191–203.
Josephs, R. A., Sellers, J. G., Newman, M. L., & Mehta, P. H. (2006). The mismatch effect: When testosterone and status are at odds.
Journal of Personality and Social Psychology, 90(6), 999–1013.
Mehta, P. H., & Josephs, R. A. (2010). Testosterone and cortisol jointly regulate dominance: Evidence for a dual-hormone hypothesis.
Hormones and Behavior, 58(5), 898–906.
Mehta, P. H., Jones, A. C., & Josephs, R. A. (2008). The social endocrinology of dominance: Basal testosterone predicts cortisol
changes and behavior following victory and defeat. Journal of Personality and Social Psychology, 94(6), 1078.
Popma, A., Vermeiren, R., Geluk, C. A. M. L., Rinne, T., van den Brink, W., Knol, D. L., et al. (2007). Cortisol moderates the
relationship between testosterone and aggression in delinquent male adolescents. Biological Psychiatry, 61(3), 405–411.
Hormones and Ethics
7
Table S1
Means and standard deviations for the key variables, and zero-order correlations between key variables, Pilot Study
Mean
SD
(1)
(2)
(3)
(4)
(5)
(1)
# Cheated
(2)
# Correctly Solved
8.12
3.87
–0.29**
(3)
Testosterone at T1
105.42
42.61
0.14
–0.01
(4)
Cortisol at T1
0.16
0.11
0.17
–0.09
0.45***
(5)
Time of Day
0.67
0.34
0.06
0.20
–0.03
0.08
(6)
Sex
0.59
0.49
–0.03
–0.08
–0.72***
–0.07
0.14
(7)
Age
21.90
2.78
–0.01
–0.07
0.07
–0.04
0.25*
2.24
(6)
3.44
–0.03
Note: ^p<0.10, *p<0.05, **p<0.01, ***p<0.001; Testosterone and Cortisol are raw scores (males and females combined) before transformation;
T1 = pre-performance hormone levels; sex was coded 0 if male, 1 if female.
Hormones and Ethics
8
Table S2
Hierarchical OLS regression models: Frequency of cheating as a dependent variable, Pilot Study
Frequency of Cheating
Predictor Variables
Step 1
Step 2
B
SE
95% CI
B
SE
95% CI
Actual Performance
–0.14
0.08
[–0.30, 0.03]
–0.15
0.08
[–0.31, 0.02]
Time of Day
7.49
9.86
[–12.10, 27.10]
10.3
10.1
[–9.73, 30.40]
Age
–0.01
0.12
[–0.24, 0.23]
–0.03
0.12
[–0.27, 0.20]
Pre-performance T
0.95
0.69
[–0.43, 2.32]
Pre-performance C
–0.29
0.71
[–1.69, 1.12]
Interaction: T x C
1.46
1.17
[–0.86, 3.80]
N
82
82
F
1
2.84
Prob > F
0.4
0.01
R-squared
0.04
0.18
Note: ^p<0.10, *p<0.05, **p<0.01, ***p<0.001.
Hormones and Ethics
9
Table S3
Means and standard deviations for the key variables, and zero-order correlations between key variables, Main Study
Mean
SD
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(1)
# Cheated
1.97
3.44
(2)
# Correctly Solved
8.02
3.87
–0.21*
(3)
Testosterone at T1
75.70
42.61
0.15
0.19*
(4)
Cortisol at T1
0.17
0.11
0.27**
0.14
0.26**
(5)
Cortisol at T2
0.18
0.11
0.03
0.16^
0.32***
0.67***
(6)
Negative Affect at T1
1.41
0.11
0.05
0.07
–0.10
0.02
-0.05
(7)
Negative Affect at T2
1.63
0.77
–0.14
–0.18*
–0.04
0.06
0.07
0.36***
(8)
State Anxiety at T1
1.46
0.67
0.07
0.04
–0.08
0.05
-0.07
0.88***
0.34***
(9)
State Anxiety at T2
1.58
0.68
–0.03
–0.05
–0.08
0.14
0.09
0.61***
0.80***
0.69***
(10)
Trait Anxiety
2.35
0.26
–0.07
0.15
0.01
0.09
0.12
0.01
0.07
0.06
0.17
(11)
Time of Day
0.64
0.05
0.07
0.01
0.05
0.07
0.02
-0.03
-0.08
0.02
-0.04
-0.12
(12)
(13)
Sex
Age
0.43
21.57
0.50
2.53
0.14
–0.06
-0.06
0.05
–0.53***
0.02
0.08
0.09
-0.06
0.09
-0.06
0.05
-0.09
-0.11
-0.02
0.08
-0.01
0.00
0.05
-0.06
Note: *p<0.05, **p<0.01, ***p<0.001; Testosterone and Cortisol are raw scores (males and females combined) before transformation; T1 = preperformance hormone levels, T2 = post-performance hormone levels; sex was coded 0 if male, 1 if female.
(11)
-0.08
0.14
(12)
0.01
Hormones and Ethics
10
Table S4
Hierarchical OLS regression models: Frequency of cheating as a dependent variable, Main Study
Frequency of Cheating
Predictor Variables
Step 1
Step 2
B
SE
95% CI
B
SE
95% CI
–0.18
0.09
[–0.35, –0.01]
–0.25**
0.08
[–0.42, –0.08]
5.9
7.02
[–8.02, 19.83]
2.97
6.7
[–10.31, 16.26]
Age
–0.10
0.16
[–0.41, 0.22]
–0.12
0.12
[–0.36, 0.12]
Trait Anxiety
–0.77
1.31
[–3.36, 1.82]
–1.07
1.28
[–3.61, 1.47]
0.5
0.5
[–0.49, 1.48]
0.81
0.46
[–0.12, 1.73]
1.24**
0.39
[0.47, 2.01]
Actual Performance
Time of Day
State Anxiety at Baseline
Pre-performance T
Pre-performance C
0.46*
0.2
[0.06, 0.86]
Interaction: T x C
0.48**
0.15
[0.19, 0.78]
N
110
F
1.45
4.14
Prob > F
0.21
>0.001
R-squared
0.06
0.27
Note: *p<0.05, **p<0.01, ***p<0.001.
110
Hormones and Ethics
11
Table S5
Hierarchical OLS regression models: Changes in Cortisol as a function of the frequency of cheating, Main Study
Predictor Variables
1
2
3
Changes in Cortisol Levels
Step 2
Step 1
Step 3
B
SE
95% CI
B
SE
95% CI
B
SE
95% CI
Actual Performance
–0.00
0.02
[–0.04, 0.03]
–0.02
0.02
[–0.05, 0.02]
–0.01
0.02
[–0.05, 0.02]
Time of Day
–1.88
1.51
[–4.87, 1.10]
–1.50
1.45
[–4.39, 1.38]
–0.35
1.34
[–3.01, 2.31]
Age
0.00
0.03
[–0.06, 0.06]
–0.00
0.03
[–0.06, 0.05]
–0.00
0.03
[–0.06, 0.05]
Trait Anxiety
Frequency of
Cheating
Pre-performance T
–0.26
0.3
[–0.85, 0.33]
–0.28
0.29
[–0.86, 0.28]
–0.19
0.26
[–0.72, 0.33]
–0.07**
0.02
[–0.11, –0.02]
–0.06**
0.02
[–0.11, –0.02]
0.19*
0.07
[0.04, 0.34]
Pre-performance C
–0.36***
0.07
[–0.51, –0.22]
Interaction: T x C
0.01
0.06
[–0.10, 0.12]
N
116
116
F
0.56
2.47
5.14
Prob > F
0.02
0.04
<0.001
R-squared
0.02
0.1
0.28
Note: *p<0.05, **p<0.01, ***p<0.001.
116