AI as a social skills equalizer
May 2024b: Analysis
Materials
Description
Participants were asked to read the following:
Imagine that you are a manager at a mid-size firm.
One of your employees, Alex, is leading a team on a high-stakes
project involving the development of a new product, crucial for the
company’s future growth. Alex has been with the company for five years
and is known for innovative problem-solving and leadership. Alex is
highly driven, always looking for the next challenge, and keen on moving
up the career ladder. Alex is respected and liked by team members for
being approachable and supportive but has shown signs of frustration
when under pressure.
Six months ago, Alex became a parent. This coincided with a critical
phase of the project, demanding an increased workload from Alex and
their team. Despite previous high performance, Alex’s recent outputs
have not met usual standards. Important deadlines have been missed, and
there have been a few instances where Alex’s work required significant
revisions. This has led to visible stress and a decrease in Alex’s usual
efficiency and creativity.
You want to motivate Alex to improve their work. Luckily, they value
transparency and constructive feedback, but unfortunately, they can also
take negative feedback personally if not delivered
thoughtfully.
To motivate Alex, you can send one of two messages. We will now
introduce those messages to you and ask you a few questions about each
one, before asking which one you would choose to send.
Messages
In random order, participants read the messages and answered questions about each. After that, they selected which message they would send as Alex’s manager. These are the messages:
Dominant message
Alex,
I want to address a critical issue regarding your recent
performance. As you know, the expectations for your role, especially
given our current high-stakes project, are substantially high. Over the
past few months, there has been a noticeable decline in your work output
and quality, which has led to missed deadlines and required
revisions.
It is imperative that we see immediate and sustained improvement in
your performance. Failure to meet these standards can have serious
consequences, potentially affecting your future with our firm. I would
like us to meet as soon as possible to discuss a clear and actionable
plan for you to turn this situation around. Please prepare for a
detailed discussion of your current projects and performance
metrics.
Regards,
[Your Name]
Non-dominant message
Dear Alex,
I’ve been reflecting on the challenges you’ve faced recently, both
personally and professionally. Becoming a parent is a significant
change, and I recognize that balancing this with an increased workload
has not been easy for you. I appreciate all the effort you are putting
in, and I want to make sure you have the support you need.
I believe in your abilities and potential, and it’s clear you are an
invaluable member of our team. Let’s schedule a time to discuss how we
can better support your needs, adjust your current workload, and set
achievable goals that take into account your current situation. I’m here
to help you navigate this period and ensure you continue to succeed and
grow with us.
Warm regards,
[Your Name]
Demographics
Race
df_aiss %>%
group_by(race) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| race | N | Perc |
|---|---|---|
| Asian | 5 | 5 |
| Black or African American | 15 | 15 |
| Hispanic, Latino, or Spanish origin | 7 | 7 |
| White | 67 | 67 |
| multiracial | 6 | 6 |
Gender
df_aiss %>%
mutate(gender = ifelse(is.na(gender) | gender == "","other",gender)) %>%
group_by(gender) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| gender | N | Perc |
|---|---|---|
| man | 53 | 53 |
| woman | 47 | 47 |
Age
df_aiss %>%
summarise(age_mean = round(mean(age,na.rm = T),2),
age_sd = round(sd(age,na.rm = T),2)) %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| age_mean | age_sd |
|---|---|
| 37.41 | 10.28 |
Education
df_aiss %>%
group_by(edu) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| edu | N | Perc |
|---|---|---|
| GED | 18 | 18 |
| 2yearColl | 14 | 14 |
| 4yearColl | 44 | 44 |
| MA | 16 | 16 |
| PHD | 8 | 8 |
Income
df_aiss %>%
ggplot(aes(x = income)) +
geom_bar() +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank(),
axis.title.y = element_blank()) +
coord_flip()
Employment
df_aiss %>%
group_by(employment) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| employment | N | Perc |
|---|---|---|
| Full-time | 82 | 82 |
| Part-time | 17 | 17 |
| Retired | 1 | 1 |
Work Experience
df_aiss %>%
group_by(workex) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| workex | N | Perc |
|---|---|---|
| Less than 5 years | 12 | 12 |
| 5 - 10 years | 27 | 27 |
| 10 - 20 years | 37 | 37 |
| More than 20 years | 24 | 24 |
Measures
Perspective-Taking
- I sometimes find it difficult to see things from the “other
person’s” point of view [R]
- I try to look at everybody’s side of a disagreement before I make a
decision
- I sometimes try to understand my friends better by imagining how
things look from their perspective
- If I’m sure I’m right about something, I don’t waste much time
listening to other people’s arguments [R]
- I believe that there are two sides to every question and try to look
at them both
- When I’m upset at someone, I usually try to “put myself in their
shoes” for a while
- Before criticizing somebody, I try to imagine how I would feel if I were in their place
Cronbach’s alpha = 0.74
df_aiss %>%
ggplot(aes(x = PT)) +
geom_histogram(binwidth = 0.3,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,5,1),
limits = c(0,6,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$PT,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Preference
If you had to choose one these messages to send to Alex, which one would you choose?
df_aiss %>%
group_by(pref) %>%
summarise(N = n()) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| pref | N |
|---|---|
| dom | 3 |
| nondom | 97 |
That’s… not good. We need much better dominant messages, or much worse non-dominant messages, to get some more variance. Maybe something like 30%-40% dominant message selection is realistic.
Dominant message
Likelihood
How likely would it be for you to send this message?
df_aiss %>%
ggplot(aes(x = dom_likely)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,7,1),
limits = c(0,8,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$dom_likely,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Empathic
How empathic is this message?
df_aiss %>%
ggplot(aes(x = dom_emp)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,5,1),
limits = c(0,6,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$dom_emp,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Threat
How threatening is this message?
df_aiss %>%
ggplot(aes(x = dom_threat)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,5,1),
limits = c(0,6,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$dom_threat,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Perceived impact
What will be the impact of this message on Alex’s work? (1 = Greatly Harm to 7 = Greatly Help)
df_aiss %>%
ggplot(aes(x = dom_eff)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,7,1),
limits = c(0,8,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$dom_eff,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Non-Dominant message
Likelihood
How likely would it be for you to send this message?
df_aiss %>%
ggplot(aes(x = nondom_likely)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,7,1),
limits = c(0,8,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$nondom_likely,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Empathic
How empathic is this message?
df_aiss %>%
ggplot(aes(x = nondom_emp)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,5,1),
limits = c(0,6,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$nondom_emp,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Threat
How threatening is this message?
df_aiss %>%
ggplot(aes(x = nondom_threat)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,5,1),
limits = c(0,6,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$nondom_threat,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Perceived impact
What will be the impact of this message on Alex’s work? (1 = Greatly Harm to 7 = Greatly Help)
df_aiss %>%
ggplot(aes(x = nondom_eff)) +
geom_histogram(binwidth = 1,
fill = "lightblue",
color = "black") +
scale_x_continuous(breaks = seq(1,7,1),
limits = c(0,8,1)) +
ylab("frequency") +
geom_vline(xintercept = mean(df_aiss$nondom_eff,na.rm = T),
color = "black",
size = 1.5,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_blank())
Correlation Matrix
PT: Perspective-taking
pref_isdom: selecting dominant message over
non-dominant message
dom_likely: Dominant message | likelihood
dom_emp: Dominant message | empathic
dom_threat: Dominant message | threat
dom_eff: Dominant message | perceived impact
nondom_likely: Non-Dominant message | likelihood
nondom_emp: Non-Dominant message | empathic
nondom_threat: Non-Dominant message | threat
nondom_eff: Non-Dominant message | perceived impact
df_aiss %>%
mutate(workex_num = as.numeric(workex),
edu_num = as.numeric(edu),
income_num = as.numeric(income),
pref_isdom = ifelse(pref == "dom",1,0)) %>%
dplyr::select(PT,pref_isdom,dom_likely:nondom_eff,age,workex_num:income_num) %>%
corPlot(upper = TRUE,stars = TRUE,xsrt = 270)