Background

The game

As part of the MBA Managerial Negotiations class at CBS, students played a game of trust and deceit. Briefly, these are the rules of the game:
Students are randomly paired up. Each of them sees a card at random. That card can be either high (Jack, Queen, or King) or low (two, three, or four). Then, they tell their counterparts if they got a high card or a low card. Importantly, they do not have to tell the truth. Their counterparts know this, so they don’t necessarily believe them.
If A tells the truth and B believes them, they each get 10 points. If A lies and B believes them, A gets 20 points, and B loses 20 points. If B doesn’t believe them, they each get 0 points regardless of what A said.
Students rotate over 10 rounds of this. At the end, the student with the most points wins.

Our measures

We have three sets of measures: (1) Individual difference measures taken at the start of the semester; (2) individual difference measures taken before the game started; (3) individual difference measures taken after the game ended; and (4) round-by-round measures taken before and after every round.

The question

This is exploratory, so I can try all kinds of analyses. Generally, though, I’m interested in the collateral impact of generalized trust. Specifically, does encountering a distrusting person affect consequent cooperative-competitive behavior?

There are a few ways of doing this. A straightforward approach would be to simply look at the effect on an encounter at time t on behavior at time t + 1. But, we might need to adjust for behavior at time t - 1 in order to really isolate the effect of the encounter at time t. Of course, in each of these, I’ll add a fixed effect of Person to make sure I’m looking only at within person variability.

Limitations

One clear limitation is that we’re way underpowered. This was conducted only on two sections, in this data set, so we only have 69 students. But… we’ll treat this more as a proof concept right now. There are way more data untouched from previous iterations, and more coming in every semester.

Eligibility

There are some missing data. I took out participants who had more than 4 NA’s over the 10 rounds. That takes us from 69 participants to 68 participants.

Demographics

Gender

gender N Perc
Man 41 60.29
Woman 25 36.76
NA 2 2.94

Race

race N Perc
Asian/Asian-American/Pacific Islander 16 23.53
Asian/Asian-American/Pacific Islander,White/European American 1 1.47
Black/African American 7 10.29
Black/African American,White/European American 1 1.47
Hispanic/Latin American 9 13.24
White/European American 29 42.65
White/European American,Hispanic/Latin American 2 2.94
NA 3 4.41

US Nationality

us N Perc
0 29 42.65
1 37 54.41
NA 2 2.94

Parental income

parents_income N Perc
Below $30,000 9 13.24
$30,001 - $50,000 2 2.94
$50,001 - $70,000 2 2.94
$70,001 - $90,000 4 5.88
$90,001 - $110,000 3 4.41
$110,001 - $130,000 2 2.94
$130,001 - $150,000 2 2.94
$150,001 - $250,000 15 22.06
More than $250,000 18 26.47
NA 11 16.18

Measures

Pre-class measures

Comfortable lying

In negotiations, there is sometimes the potential for misrepresentation…
This can sometimes lead to possible gain, but it also possible loss if you get caught. Answer the following questions which respect to opportunities for misrepresentation, for possible gain.

How do you think you would feel about lying to someone in a negotiation? (1 = Extremely Uncomfortable / Bad to 7 = Extremely Comfortable / Good)

Efficacy of lying

What percent of the time do you think people would believe you if you tried to lie in a negotiation?

Efficacy of telling the truth

What percent of the time do you think your counterparts in a negotiation would believe you when you told the truth?

Acceptability of misrepresentation - by you

Sometimes in negotiations, misrepresenting things can get you ahead, and so misrepresentation for that reason is acceptable. What do YOU think? (1 = Not at All Acceptable to 7 = Extremely Acceptable)

Acceptability of misrepresentation - by others

Sometimes in negotiations, misrepresenting things can get you ahead, and so I think my COUNTERPARTS, on average, will find misrepresentation for that reason… (1 = Not at All Acceptable to 7 = Extremely Acceptable)

Trust

Mean score of the following items (1 = Strongly Disagree to 6 = Strongly Agree):

1. Most people can be counted on to do what they say they will do.
2. I tend to trust people, even those whom I have just met for the first time.
3. Unless you remain alert, someone will soon take advantage of you. (R)
4. Most people would tell a lie if they could gain by it. (R)
5. My typical approach is to be cautious with people until they have demonstrated their trustworthiness. (R)
6. I usually give acquaintances the benefit of the doubt if they do something that seems selfish.
7. Most people pretend to be more honest than they really are. (R)
8. I believe that most people are generally trustworthy.

Cronbach’s alpha = 0.84

Conflict styles

There are three questions. For each question, students are asked to rank five behaviors/characteristics from 1 (most natural/typical for me) to 5 (least natural/typical for me). Each of the five items represents a different conflict style (noted below in parentheses). We then take a sum score for each style (sum up the three items) and reverse-score it. So, scores range from 0 to 12.

When I have a conflict at work, I do the following…
1. I insist on getting my way (competing)
2. I examine ideas from both sides to find a mutually beneficial solution (collaborating)
3. I search for middle ground (compromising)
4. I give in to the wishes of the other party (accommodating)
5. I avoid direct confrontation or conflict whenever possible (avoiding)

1. I propose a compromise so that we both give something up (compromising)
2. I fight for a good outcome for myself (competing)
3. I work hard to satisfy the other party’s interests (accommodating)
4. I don’t like to create controversy or disagreement (avoiding)
5. I pursue both my own and the other person’s goals and interests (collaborating)

1. I go to great lengths to make the other party satisfied (accommodating)
2. I try to avoid or minimize any differences of opinion (avoiding)
3. I urge us both to give in a little (compromising)
4. I strive for a solution that suits both parties (collaborating)
5. I push to win or for an outcome that favors me (competing)

Pre-game measures

Predicted truth-tellers

How many people (out of 10 random interaction counterparts) do you expect will tell you the truth (versus lie to you) in this activity?

Predicted truth detection

If/when someone tells you the truth in this activity, what percent of time do you think you would accurately identify them as telling the truth?

Predicted lie detection

If/when someone lies to you in this activity, what percent of time do you think you would accurately identify them as lying?

Predicted truth-telling

How many times (out of 10 interactions with random counterparts) do you expect YOU will tell the truth in this activity?

Predicted truth detection of you

If/when you tell the truth in this activity, what percent of time do you think your counterparts would accurately identify you as telling the truth?

Predicted lie detection of you

If/when you LIE in this activity, what percent of time do you think your counterparts would accurately identify you as lying?

Predicted feeling: truth/lie

How do you think you would feel if the following happened during this activity (-5 = Extremely Negative to 5 = Extremely Positive):

1. youtru_cptrust: I told the truth … and they TRUSTED me
2. youtru_cpdistrust: I told the truth … and they DID NOT TRUST me
3. youlie_cptrust: I lied … and they TRUSTED me
4. youlie_cpdistrust: I lied … and they DID NOT TRUST me

Predicted feeling: trust/distrust

How do you think you would feel if the following happened during this activity (-5 = Extremely Negative to 5 = Extremely Positive):

1. youtrust_cptru: I trusted them … and they TOLD THE TRUTH
2. youtrust_cplie: I trusted them … and they LIED
3. youdistrust_cptru: I did NOT trust them … and they TOLD THE TRUTH
4. youdistrust_cplie: I did NOT trust them … and they LIED

Post-game measures

Open-end responses to the following questions:
1. What surprised you most in this activity?
2. What led people to TRUST you in this activity?
3. What led people to DISTRUST you in this activity?
4. What helped you accurately judge others’ trustworthiness in this activity?
5. What undermined or limited your ability to accurately judge others’ trustworthiness in this activity?

In-game measures

Intentions

Before talking further with your counterpart, decide whether you’re going to tell them the truth or lie.

If you tell them the truth, tell them whether your card is LOW (a 2, 3, or 4) or HIGH (a Jack, Queen, or King). You’re free to tell them what your specific card is. If you lie, you should say the opposite of whether your card is LOW or HIGH (for example, if it’s low, try to convince them that it’s high … you’re free to make up a specific card, like King of Diamonds or Jack of Clubs).

Recall that if you tell them the truth and they trust you, you gain 10 points. If you lie and they trust you, you gain 20 points. If they don’t trust you, you gain 0 points, regardless of whether you were telling the truth or lying.

You can say what you like to your counterpart but you can’t show them the image of your card.

1. I’m going to tell my counterpart the truth
2. I’m going to lie

Round-by-round

Cumulative

The percentage of times in which participants indicated that they will tell the truth.

Trust

Do you trust your counterpart? That is, do you think they’re telling you the truth about whether they have a low or high card? “Yes” means you trust that they’re telling the truth. “No” means you don’t trust them, you think they’re lying.

Recall that if you trust them and they’re telling you the truth, you gain 10 points. If you trust them and they’re lying, you lose 20 points. If you don’t trust them, you neither gain nor lose any points, regardless of whether they were telling the truth or lying.

Please indicate below: Do you trust your counterpart?

1. No
2. Yes

Round-by-round

Cumulative

The percentage of times in which participants indicated that they trusted their counterpart.

Truth

Did you tell your counterpart the truth? “Yes” means you told them the truth about whether you have a low or high card. “No” means you lied.

1. No
2. Yes

Round-by-round

Cumulative

The percentage of times in which participants indicated that they told the truth.

Outcomes by round per person

Points accumulated

Round-by-round

Cumulative

Feeling: Giving information

How did you feel about lying/truthing (piped from their answer) and your counterpart trusting/distrusting (piped from their answer)? (-5 = Extremely Negative to 5 = Extremely Positive)

Feeling: Receiving information

How did you feel about trusting/distrusting (piped from their answer) your counterpart’s truth/lie (piped from their answer)? (-5 = Extremely Negative to 5 = Extremely Positive)

Familiarity

How familiar were you with the person in this round before today’s game?? (1 = Not Familiar at All to 5 = Highly Familiar)

Round-by-round

Cumulative

Analysis

Correlation Matrix

Round 1 being trusted -> trusting/lying in rest of game

firstround_trusted n perc ptruth ptrust
distrusted 24 35.82 0.89 0.72
trusted 43 64.18 0.79 0.80

Round 1 being trusted/lying -> trusting/lying in rest of game

firstround_truth firstround_trusted n perc ptruth ptrust
lie distrusted 4 5.97 0.78 0.75
lie trusted 12 17.91 0.57 0.60
truth distrusted 20 29.85 0.91 0.71
truth trusted 31 46.27 0.87 0.87

Round 1 & 2 being trusted -> trusting/lying in rest of game

firstround_trusted secondround_trusted n perc ptruth ptrust
distrusted distrusted 4 6.25 0.97 0.72
distrusted trusted 19 29.69 0.88 0.75
trusted distrusted 20 31.25 0.69 0.76
trusted trusted 21 32.81 0.88 0.85

Being distrusted -> truthing at t+1

(#tab:unnamed-chunk-39)
Term \(\hat{\beta}\) 95% CI \(t\) \(\mathit{df}\) \(p\)
Intercept 0.79 [0.71, 0.86] 21.25 173.03 < .001
Ptrust cp 0.04 [-0.02, 0.11] 1.33 554.06 .183

Being distrusted + controlling for truthing at t -> truthing at t+1

(#tab:unnamed-chunk-40)
Term \(\hat{\beta}\) 95% CI \(t\) \(\mathit{df}\) \(p\)
Intercept 0.74 [0.65, 0.83] 16.05 273.17 < .001
Ptrust cp 0.04 [-0.03, 0.10] 1.08 545.30 .279
Ptruth 0.07 [-0.01, 0.15] 1.75 586.56 .080

Being distrusted X lying -> lying at t+1

(#tab:unnamed-chunk-41)
Term \(\hat{\beta}\) 95% CI \(t\) \(\mathit{df}\) \(p\)
Intercept 0.71 [0.60, 0.83] 12.50 434.91 < .001
Ptrust cp 0.07 [-0.06, 0.20] 1.09 540.45 .278
Ptruth 0.10 [-0.02, 0.23] 1.60 577.91 .110
Ptrust cp \(\times\) Ptruth -0.05 [-0.20, 0.10] -0.63 542.86 .530