perspective-taking, information processing and workplace ... · 4/18/2018  · person. in one...

37
Perspective-taking, information processing and workplace verbal abuse: a Bangladesh factory manager field experiment Laura Babbitt Drusilla Brown 18 April 2018

Upload: others

Post on 11-Jul-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

Perspective-taking, information processing and

workplace verbal abuse: a Bangladesh factory

manager field experiment

Laura Babbitt

Drusilla Brown

18 April 2018

Page 2: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

2

PERSPECTIVE-TAKING, INFORMATION PROCESSING AND WORKPLACE VERBAL ABUSE: A BANGLADESH FACTORY MANAGER FIELD EXPERIMENT

By LAURA BABBITT AND DRUSILLA BROWN*

Workplace verbal abuse is common despite its adverse effects on workers and firm

performance. Dehumanization of workers in the minds of managers may contribute

to the use of inefficient managerial practices if it affects processing of information

on the relationship between abuse and work effort. Existing research demonstrates

that when subordinates are dehumanized, their superiors are less likely to process

information concerning the ineffectiveness of abusive motivational techniques.

Managers of dehumanized subordinates are more likely to employ abuse and less

likely to process information related to the techniques’ ineffectiveness than

managers who perceive subordinates in humanized terms. Perspective-taking has

been shown to promote rehumanization, suggesting that it may also improve

manager information processing. A randomized controlled trial analyzing the impact

of a perspective-taking exercise on processing information concerning verbal abuse

was conducted with factory managers in 16 Bangladesh apparel factories. Managers’

modal perception of the prevalence of verbal abuse was inconsistent with worker

reports, indicating managers’ resistance to acknowledging verbal abuse. The

perspective-taking exercise improved information processing but the response was

heterogeneous. Among managers who engaged with the exercise or viewed verbal

abuse as inappropriate, perspective-taking did not affect the reported perception of

the prevalence of verbal abuse, but treatment did increase interest in and willingness

to make changes based on the data. Perspective-taking increased acknowledgement

of the prevalence of verbal abuse only in those managers for whom there was little

dissonance between personal beliefs and the data. These managers were also more

surprised but less willing to make changes.

* Tufts University, Department of Economics, 8 Upper Campus Road, Medford, MA 02155 (e-mail: [email protected]). Financial and

logistical support was provided by Good World Solutions. The research was conducted under SBER Protocol 1607004.

Page 3: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

3

I. Introduction

Verbal abuse such as yelling or derogatory remarks is a common workplace

motivational technique, yet workers who experience verbal abuse are less productive and

require a compensating pay differential (Rourke, 2014). So why does this inefficient

practice continue? Two likely factors are a lack of managerial attention and data

availability. Processing information on the relationship between management practices

and firm performance requires that managers first choose which dimensions of the

production process should receive their attention and then that they receive data in a

useful form (Hanna, Mullainathan, and Schwartzstein, 2014). Firm managers, particularly

in developing countries, often fail to adopt straightforward changes like keeping factory

floors clear and organizing inventory because they do not realize that these changes would

affect firm productivity (Bloom et al., 2013). Verbal abuse could be tolerated simply

because managers do not realize it is harmful to firm performance or do not realize how

pervasive it is.

Simply providing information about verbal abuse may not reduce its use, however.

Information processing related to subordinates is also affected by the extent to which

subordinates are dehumanized in the minds of managers—not unlikely in a factory context.

Powerful people are more willing to take action to achieve their goals, even if that action

has negative consequences for others, and they justify this behavior by dehumanizing those

they have harmed (Lammers and Stapel, 2011). For example, managers might dehumanize

workers to downplay the negative consequences of requiring overtime.

Just as harm can lead to dehumanization, harm is more easily inflicted when

workers are already dehumanized. In laboratory studies in which subordinates are

dehumanized, supervisors administer more severe punishments and are less likely to

process information concerning the ineffectiveness of abusive motivational techniques

(Bandura, Underwood and Fromson, 1975). When subordinates made more mistakes

following abusive punishment, indicating that abuse does not improve performance,

supervisors who had heard a humanizing description of their subordinates, or no

Page 4: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

4

description at all, generally realized that more intense punishment did not help their

subordinates improve, and plateaued or decreased the punishment intensity. However,

supervisors with dehumanized subordinates did not process the negative relationship

between harsh motivational techniques and subordinate performance, and actually

increased the intensity of punishment.

However, fMRI and field experiment evidence indicate that dehumanization can be

reversed through the simple act of imagining the preferences or experiences of another

person. In one study, researchers showed photos of various types of people to participants

while measuring their brain activity and found that dehumanized people (i.e., homeless

people or drug addicts) were processed in the brain as objects rather than humans (Harris

and Fiske, 2006). Yet simply asking participants to imagine whether the person in the

photo liked a certain kind of vegetable was enough to activate processing in the social areas

of the brain. Trying to determine someone else’s preferences made them human again

(Harris and Fiske, 2007). Another study found that when participants were given a high-

power role and asked to think about the experiences of another person, they felt more

responsibility and concern for others’ well-being. Though, the effect disappeared when

participants were explicitly asked to take the perspective of the other person (Scholl, et al.,

2017). A face-to-face analogic perspective-taking exercise has been shown to reduce

transphobia (Broockman and Kalla, 2016). Participants in the treatment condition first

considered a time when they were judged negatively for being different, then were

encouraged to generalize their experiences to a disadvantaged group. The intervention

reduced transphobia in one of ten participants.

The literature on suboptimal labor management practices and dehumanization

suggests that factory managers may be more receptive to information about verbal abuse

after a perspective-taking exercise. Specifically, managers who see graphs illustrating

worker reports about working conditions might be more likely to process that information

after having made the effort to imagine their workers’ preferences and daily routines. This

question is especially important given the key role managers play in receiving and acting

on worker complaints (Townsend, 2014).

Page 5: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

5

Perspective-taking, however, is not without risks. There is some possibility of

creating defensiveness in managers when asking them to focus on workers’ preferences

and then providing negative information about working conditions for those workers.

Negative information about one’s behavior (or the behavior of a group one belongs to)

threatens the image of oneself as a good person (Sherman and Cohen, 2006). For example,

White American participants who strongly identify with being White are more likely to

react to being reminded of their unearned privileges by denying that racial discrimination

exists (Branscombe, Schmitt, and Schiffhauer, 2007). Similarly, feeling that one’s status is

threatened can lead to oppressive behavior in an effort to retain power (Townsend, 2014),

as may happen when managers are alerted that they will be expected to make changes.

To test whether perspective-taking improves information processing concerning the

prevalence of verbal abuse, we conducted a randomized controlled trial with factory

managers in apparel factories in Bangladesh. Managers were randomly assigned to one of

two conditions. Managers in the treatment condition first completed a perspective-taking

exercise, reflecting on the thoughts, preferences and aspirations of workers in their factory.

Those in the control condition answered questions about their own thoughts, preferences,

and aspirations.

Both groups were then shown two pieces of information taken from surveys of

workers in factories in and around Dhaka, Bangladesh. One piece of information was

positive: Most workers are satisfied with their supervisors. The other was negative: More

than half of workers report being verbally abused at work. Both groups of managers were

then asked questions about the data they had just been shown.

Manager perceptions of worker satisfaction with their supervisor match the data as

reported by workers. However, remarkably, in spite of having just been shown data

indicating that over half of workers report verbal abuse at work, nearly 65 percent of

managers reported that verbal abuse rarely or never occurs in their factory. Anticipating

findings, perspective-taking does alter information processing but the effect is

heterogeneous. The more attention the managers in the treatment condition devoted to

the perspective-taking exercise, the less happy but more interested they were in the data,

Page 6: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

6

and the more inclined they were to make changes based on the data. However,

perspective-taking increased willingness to acknowledge the incidence of verbal abuse

itself only among mangers who exhibited low indicators of humanization of subordinates.

The experimental design and data are described in section II. Findings are reported in

section III and conclusions follow.

II. Experimental Design and Data

Participants were 334 managers in 16 factories in and around Dhaka, Bangladesh.

Managers with incomplete data were excluded, leaving a final sample of 277. There were

10 factories in the first group (October 2016) and six factories in the second group (March

2017). Prior to participation, managers in the first group were informed at the time of the

experiment that a social dialogue program would be introduced in the factory in the

coming months.

In each factory, participants were invited to a managerial meeting to hear data

about their factory. All participants were instructed to call a particular number using their

cell phone. After the informed consent process, they were asked to use their phone to

complete an automated survey.

The survey software randomly assigned each manager to complete one of two 10-

question surveys. In the treatment condition, managers were asked to imagine their

workers’ preferences, daily routines and life aspirations (e.g., “What do you think workers

in your factory prefer to have for breakfast?”). Managers in the control condition answered

the same questions about themselves (e.g., “What do you prefer to have for breakfast?”).1

Next, managers were shown two graphs of worker-reported survey data,

summarized across several factories.2 A positive worker perception was paired with a

1 Survey text is available in the Appendix in Table A1. 2 Pilot testing indicated that managers became upset and defensive after hearing that workers

reported verbal abuse, so they were told the data came from “factories like yours” instead of

“your factory.”

Page 7: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

7

negative worker perception. The positive finding is that 78 percent of workers report

being satisfied with their supervisor. The negative finding is that 53 percent of workers

report verbal abuse by their supervisor.

After seeing graphs depicting the two findings, managers called a given number to

complete a second survey. The second survey captured their reactions to the data

presented and their attitudes about verbal abuse (e.g., “How common is verbal abuse in this

factory?” “How appropriate do you think it is to yell or use harsh language with workers in

this factory?”). In each factory, all managers participated at the same time, though they

were asked not to discuss the surveys or data with each other until everyone had

completed the second survey.

Summary statistics for all variables are reported in Table 1. Manager reactions to

the data were overall positive. Managers reported that they found the data useful (mean of

3.71 on a 4-point scale) and important (mean of 3.87 on a 4-point scale). There was a

ceiling effect for the item “How important are these results to you?” with 90 percent of

participants selecting “very important,” so this item was not included in the analysis below.

A preponderance of managers also planned to use the results to make changes (mean of

3.67 on a 4-point scale).

{Table 1 about here}

However, in spite of their apparent receptivity to the data, managers did not

generally process the central finding. Managers were shown data indicating that over half

of workers report verbal abuse. Yet when asked how common verbal abuse is, 42.9 percent

of the control group and 39.2 percent of the treatment group responded, “not at all

common.” Only 22.8 percent of the treatment group and 21.8 percent of the control group

responded that verbal abuse is “somewhat” or “very common.” A summary of responses is

provided in Figure 1.

{Figure 1 about here}

Treatment effects are analyzed with four regression models. In the simple model,

Model 1, we analyze whether treatment affected responses on the second survey.

Page 8: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

8

𝑀𝑜𝑑𝑒𝑙 1 𝑋𝑖 = 𝛽𝑜 + 𝛽1𝑇𝑟𝑒𝑎𝑡𝑒𝑑 + 𝜀𝑖

where 𝑋𝑖 is an outcome variable for participant i and Treated is a binary variable, taking on

the value 1 for participants in the treatment condition and 0 otherwise. Outcomes are (1)

duration of the pre-presentation perspective-taking or control survey (call_duration), (2)

appropriateness of verbal abuse (VA_app), (3) whether the participant was surprised by

the data (surprised), (4) whether the participant found the data understandable

(understandable), (5) whether the participant found the data interesting (Interested), (6)

whether the data made the participant happy (Happy), (7) whether the participant believed

the data were personally accurate (accurate_you), (8) whether the participant believed the

data were accurate about their factory (accurate_factory), (9) whether the participant

personally believes that verbal abuse is effective (VA_effective), (10) whether the

participant believed that workers in the factory were satisfied with their supervisor

(sup_sat_VA), (11) whether the data were useful (useful), (12) whether the participant

would make changes based on the data (make_changes) and (13) whether the data indicate

a need for additional training (Training).

We then turn to consider variables that may mediate the treatment effect. With

Model 2, we consider the possibility that the more attention a participant devoted to the

perspective-taking exercise, the more the exercise affected information processing.

Attention is measured by the time spent on the perspective-taking exercise.

𝑀𝑜𝑑𝑒𝑙 2 𝑋𝑖 = 𝛽𝑜 + 𝛽1𝑇𝑟𝑒𝑎𝑡𝑒𝑑 + 𝛽2𝐶𝑎𝑙𝑙_𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛 + 𝛽3𝑇𝑟𝑒𝑎𝑡𝑒𝑑 ∗ 𝐶𝑎𝑙𝑙_𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛 + 𝜀𝑖

Call_duration is measured in seconds.

Conversely, participants may be less receptive to information about verbal abuse if

it is in conflict with other beliefs they hold about themselves or about proper factory

management. Participants may resist information about the prevalence of verbal abuse if

they see workers in humanized terms or believe that verbal abuse is ineffective and

inappropriate. Participants may more readily process information if they already believe

that yelling at workers is an effective and appropriate business practice. A participant’s

beliefs concerning whether verbal abuse is appropriate can be an indicator both of the

Page 9: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

9

extent to which workers are seen in dehumanized terms as well as an indicator of

acceptance of verbal abuse as a business practice. With Model 3, we capture the interaction

between treatment and beliefs concerning the appropriateness of verbal abuse.

𝑀𝑜𝑑𝑒𝑙 3 𝑋𝑖 = 𝛽𝑜 + 𝛽1𝑇𝑟𝑒𝑎𝑡𝑒𝑑 + 𝛽4𝑉𝐴_𝑎𝑝𝑝𝑟𝑜𝑝𝑟𝑖𝑎𝑡𝑒 + 𝛽5𝑇𝑟𝑒𝑎𝑡𝑒𝑑 ∗ 𝑉𝐴_𝑎𝑝𝑝𝑟𝑜𝑝𝑟𝑖𝑎𝑡𝑒 + 𝜀𝑖

𝑉𝐴_𝑎𝑝𝑝𝑟𝑜𝑝𝑟𝑖𝑎𝑡𝑒 is measured on a 4-point scale.

Participants may have also been affected by framing. Prior to the experiment,

participants in Group 1 were told that changes in the factory related to the management of

workers would be implemented. No framing was presented to managers in Group 2.

Model 4 explores whether framing affects the effectiveness of perspective-taking on data

processing, though it is important to note that framing may not be the most important

characteristic that differentiates the two factory groups.

𝑀𝑜𝑑𝑒𝑙 4 𝑋𝑖 = 𝛽𝑜 + 𝛽1𝑇𝑟𝑒𝑎𝑡𝑒𝑑 + 𝛽6𝐺𝑟𝑜𝑢𝑝2 + 𝛽7𝑇𝑟𝑒𝑎𝑡𝑒𝑑 ∗ 𝐺𝑟𝑜𝑢𝑝2 + 𝜀𝑖

Group2 is a binary variable taking on the value 0 for factories that received framing and 1

for factories for which there was no data framing.

III. Empirical Results

Models 1, 2 and 3 are estimated with OLS with standard errors clustered by factory.

A. Base Model

Estimates of Model 1 are reported in Table 2. The only significant treatment effect

was on call duration during the perspective-taking exercise, reported in column 1. Call

duration ranged from 24 to 838 seconds, with an average of 296 seconds. Managers who

received the worker-focused survey spent an average 56 seconds longer responding than

those who answered questions about themselves, indicating that managers put some time

into imagining their workers’ preferences and routines.

{Table 2 about here}

Page 10: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

10

B. Participant Attention

The amount of time participants spent on the perspective-taking survey could be an

indicator of how seriously they took it and the extent to which they rehumanized workers.

In the second model, reported in Table 3, the effect of perspective-taking, call duration, and

the interaction between the two are considered. That is, did the effect of treatment on the

outcome variables depend on how long managers spent taking the perspective-taking

survey?

{Table 3 about here}

Treated managers who spent longer thinking about their workers’ preferences were

more likely to report being interested in the results, as can be seen in column 4 (Table 3).

Figure 2 depicts the predicted treatment effect at various levels of perspective-taking call

duration with 90% confidence intervals. Managers who spent at least 400 seconds on the

perspective-taking exercise were more likely to report being interested in the results if

they were in the treatment condition rather than the control condition. Managers who

spent less time did not show a significant treatment effect on whether they reported being

interested.

{Figure 2 about here}

Overall, treated managers were more likely to report being happy with the data as

reported in column 5 (Table 3), but those who spent longer on the survey were less likely to

feel happy. The predicted marginal treatment effects at various levels of time spent

perspective-taking are depicted in Figure 3. Participants who spent less than 50 seconds

on the perspective-taking exercise were more happy about the data than participants in the

control condition. Participants who spent more than 400 seconds on the perspective

taking exercise were less happy than those in the control condition.

{Figure 3 about here}

Similarly, treated managers were overall less likely to report an intention to make

changes based on the results, unless they spent longer on the survey, as reported in column

Page 11: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

11

13 (Table 3). The more time managers invested in the perspective-taking exercise the

more likely they were to want to make changes, as depicted in Figure 4.

{Figure 4 about here}

Thus, managers in the treatment group who engaged with the perspective-taking

exercise were more likely to be interested, less likely to be happy and were more willing to

make changes based on the data than managers who were assigned to the control

condition. The result that participants in the engaged group were also less happy about the

data indicates increased concern.

By contrast, managers assigned to the perspective-taking treatment who did not

engage with the exercise were more likely than managers in the control group to be happy

with the data and less likely to intend to make changes based on the data. That is,

treatment among managers who resisted the perspective-taking exercise actually

increased the resistance to implications of the data.

Taken together, these results suggest that the treatment may have made some

managers less willing to engage with the data (particularly the negative aspect of the data).

After being asked to imagine workers’ preferences, managers who invested less effort in

the perspective-taking exercise reported being happier about the data and less likely to

make changes. Those who spent longer on the survey—and therefore presumably

rehumanized their workers more—were more willing to engage with the negative aspect of

the data and make changes. However, interestingly, while investment in the perspective-

taking exercise increased interest in negative aspects of the data and intention to use the

data to make changes, perspective-taking did not lead managers to directly acknowledge

the extent of verbal abuse in their factories.

C. Beliefs

Model III considers the contribution of conflicts between the data and personal

beliefs on the receptivity to the data. Results are reported in Table 4. Personal beliefs are

captured by the item “How appropriate do you think it is to yell or use harsh language with

workers in this factory?”

Page 12: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

12

{Table 4 about here}

As can be seen in column 9 (Table 4), managers who believe that verbal abuse is

appropriate and, possibly, see their workers in dehumanized terms are more likely to

acknowledge the pervasiveness of verbal abuse. The group most likely to accurately

process the data concerning the prevalence of verbal abuse are those who were assigned to

perspective-taking and think that verbal abuse is appropriate. The marginal treatment

effects are depicted in Figure 5. Participants who engaged in perspective-taking and

acknowledge any level of appropriateness of verbal abuse (2=somewhat appropriate,

3=appropriate and 4=very appropriate) are more likely to acknowledge the prevalence of

verbal abuse than participants in the control condition.

{Figure 5 about here}

This result suggests that treatment improves receptivity to the data if there is not a

conflict between the implications of the data and the participant’s beliefs and values.

Perspective-taking increases the willingness to acknowledge the prevalence of verbal

abuse if the person also believes that verbal abuse is appropriate. More to the point, if a

belief that verbal abuse is appropriate can be taken as an indicator of dehumanization, the

results indicate that participants will resist the data if there is too much conflict between

the implications of the data and their perception of workers as human beings.

Treated managers who believe verbal abuse is appropriate were also more likely to

be surprised by the findings (see Figure 6). It is not clear what aspect of the results

surprised these managers. It is possible they are surprised by the fact that workers are

generally satisfied with their supervisors even though workers also report that verbal

abuse is common. By contrast, treated managers who believe verbal abuse is inappropriate

were more likely to be interested in the results (see Figure 7) and less likely to find them

understandable. One way of interpreting the “understandable” item is that lower ratings

indicate more thoughtfulness. The two pieces of data the managers viewed were

somewhat contradictory. Thus, managers who were paying attention to both pieces of

information would find the results confusing.

Page 13: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

13

{Figures 6 and 7 about here}

Model 3 indicates, then, that managers who were assigned to the treatment group

but do not see their workers in humanized terms improved their processing of the most

direct information concerning the prevalence of verbal abuse. By contrast, managers

assigned to the treatment group who see their workers in humanized terms were less likely

to acknowledge the prevalence of verbal abuse but did seem to appreciate the implications

of the data for their behavior in the factory.

D. Framing

Turning to framing, managers in Group 1 knew that they would be expected to make

changes; managers in Group 2 did not. As reported in Table 5, treated managers who were

in Group 2 were more likely to find the results understandable (column 3), more likely to

think that the results accurately reflected their relationship with their workers (column 6),

and less likely to think that verbal abuse is effective (column 8). Not being put on guard

about upcoming changes could make managers more receptive to the information

presented, though perhaps less engaged with the findings.

{Table 5 about here}

III. Conclusions

Lab experiments find that dehumanization increases the abuse of subordinates and

interferes with the rational processing of the negative relationship between abuse and

subordinate performance. Perspective-taking has been shown to promote rehumanization.

Information processing on the negative relationship between abuse and performance, then,

may be improved by perspective-taking.

Analyzing data from a field experiment with Bangladesh factory managers, we find

that perspective-taking affected information processing concerning verbal abuse but that

treatment is endogenous and has a heterogeneous effect. Those managers who spent more

time on the perspective-taking exercise were more likely to find the data on the high

Page 14: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

14

prevalence of verbal abuse interesting, less likely to be happy about the results and more

likely to be willing to make changes than untreated managers. Treated managers who

spent less time on the perspective-taking exercise were more likely to report being happy

with the data and less likely to use the data to make changes than untreated managers.

These findings indicate that managers who spent more time on the perspective-

taking exercise were better able to process the negative implications of the data and more

willing to make changes based on the data. Although these managers were not necessarily

willing to acknowledge the prevalence of verbal abuse, their readiness to make changes is

arguably equally important. By contrast, for managers who resisted the perspective-taking

exercise, treatment made them less likely to engage with negative aspects of the data. They

focused attention on the positive findings about work satisfaction and did not plan to make

changes as a consequence.

The impact of perspective-taking on information processing was mediated by the

consonance or dissonance between a participant’s beliefs and the data. The perception of

verbal abuse as appropriate is an indicator of beliefs concerning acceptable motivational

techniques and, possibly, the extent to which workers are dehumanized in the eyes of

managers.

For managers who believe that verbal abuse is not appropriate, treatment increased

the probability that they would find the results interesting and made them more thoughtful

about the results. Although perspective-taking did not lead these managers to acknowledge

the prevalence of verbal abuse, it improved processing of the meaning of the data and

promoted more humane treatment of workers.

By contrast, managers who think that verbal abuse is appropriate were more likely

to believe that verbal abuse is effective and acknowledge that verbal abuse is common. For

these managers, treatment increased the probability of being surprised by the results and

acknowledging the prevalence of verbal abuse. Perspective-taking clearly improved

information processing for this group, but not in a way that promotes more humane

treatment of their subordinates.

Page 15: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

15

Forewarning managers of future changes may have increased resistance to the data.

Treated managers who did not know that the information would be used to introduce

changes in the factory were more likely to find the results understandable and an accurate

reflection of their relationship with their subordinates. These treated managers were also

more likely to think that verbal abuse is not effective.

There is some risk of creating defensiveness in managers when asking them to focus on

workers’ preferences. A defensive reaction could explain the positive results among

managers who were not led to expect change, and the tendency of treated managers to

report being happier about the information they saw, and less likely to make changes.

However, managers who engaged with the treatment were more interested in the data and

more willing to make changes—indicating that perspective-taking does have the potential

to reduce verbal abuse in factories.

Page 16: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

16

.

REFERENCES

Bandura, Albert, Bill Underwood, and Michael E. Fromson. 1975. Disinhibition of

aggression through diffusion of responsibility and dehumanization of victims. Journal of

Research in Personality 9:253–269.

Bloom, Nicholas, Benn Eifert, Aprjit Mahajan, David McKenzie, and John Roberts. 2013. “Does

management matter? Evidence from India,” Quarterly Journal of Economics, 128(1), 1-51.

doi.org/10.1093/qje/qjs044

Branscombe, Nyla R., Michael T. Schmitt, and Kristin Schiffhauer. 2007. “Racial attitudes in

response to thoughts of White privilege,” European Journal of Social Psychology, 36, 203–

215.

- -. 2007. “=Social groups that elicit disgust are differentially processed in mPFC. Social

Cognitive and Affective Neuroscience 2:45–51.

Broockman, David and Joshua Kalla. 2016. Durably reducing transphobia: A field

experiment on door-to-door canvassing. Science 352 (6282):220-224. doi:

10.1126/science.aad9713.

Hanna, Rema, Sendhil Mullainathan, and Joshua Schwartzstein. 2014. Learning through

noticing: Theory and experimental evidence in farming. Quarterly Journal of Economics

129(3):1311-1353. https://doi.org/10.1093/qje/qju015

Harris, Lasana S. and Susan T. Fiske. 2006. Dehumanizing the lowest of the low:

Neuroimaging responses to extreme out-groups. Psychological Science 17:847–53.

doi:10.1111/j.1467-9280.2006.01793.x

Lammers, Joris and Diederik A. Stapel. 2011. Power increases dehumanization. Group

Processes and Intergroup Relations 14:113–126. doi:10.1177/1368430210370042

Page 17: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

17

Rourke, Emily L. 2014. “Is there a business case against verbal abuse? Incentive structure,

verbal abuse, productivity and profits in garment factories,” Better Work Discussion Paper

No. 15.

Scholl, Anika, Kai Sassenberg, Daan Scheepers, Naomi Ellemers, and Frank de Wit. 2017. A

matter of focus: Power-holders feel more responsible after adopting a cognitive other-

focus, rather than a self-focus. British Journal of Social Psychology 56:89–102.

doi:10.1111/bjso.12177

Townsend, Keith 2014. The role of line managers in employee voice systems. In Handbook

of Research on Employee Voice, A. eds. Wilkinson, J. Donaghey, T. Dundon and R. Freeman.

Glouchestershire: Edward Elgar.

Page 18: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

18

Table 1

Summary Statistics

Variable Treatment Control Overall

Obs Mean Std. Dev. Min Max Obs Mean Std. Dev. Min Max Obs Mean Std. Dev. Min Max

call_duration 158 348.31 86.55 24.42 656.48 119 290.18 108.40 34.96 837.64 277 323.33 100.58 24.42 837.64

understandable 147 3.47 0.71 1 4 111 3.60 0.65 1 4 258 3.53 0.68 1 4

Interested 154 0.25 0.44 0 1 115 0.19 0.40 0 1 269 0.23 0.42 0 1

Happy 154 0.49 0.50 0 1 115 0.50 0.50 0 1 269 0.50 0.50 0 1

accurate_you 156 2.49 0.69 1 3 113 2.49 0.68 1 3 269 2.49 0.68 1 3

accurate_factory 154 2.30 0.74 1 3 112 2.33 0.75 1 3 266 2.31 0.75 1 3

VA_not_appropriate 147 3.48 0.85 1 4 106 3.42 0.97 1 4 253 3.46 0.90 1 4

VA_appropriate 147 1.52 0.85 1 4 106 1.58 0.97 1 4 253 1.54 0.90 1 4

VA_effective 143 1.95 1.10 1 4 109 2.00 1.11 1 4 252 1.97 1.11 1 4

VA_common 133 1.94 1.07 1 4 99 1.86 1.05 1 4 232 1.91 1.06 1 4

sup_satisfaction_VA 133 1.93 0.76 1 3 94 1.94 0.77 1 3 227 1.93 0.76 1 3

sup_sat_VA_NR 133 0.42 0.50 0 1 94 0.40 0.49 0 1 227 0.41 0.49 0 1

useful 146 3.75 0.54 1 4 110 3.67 0.67 1 4 256 3.71 0.59 1 4

make_changes 139 3.67 0.63 1 4 107 3.68 0.68 1 4 246 3.67 0.65 1 4

Training 148 0.59 0.49 0 1 113 0.55 0.50 0 1 261 0.57 0.50 0 1

Page 19: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

19

results_important 149 3.89 0.38 2 4 113 3.85 0.52 1 4 262 3.87 0.44 1 4

surprised 132 2.82 1.05 1 4 106 2.64 1.18 1 4 238 2.74 1.11 1 4

Page 20: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

Table 2

Model 1: Basic Treatment Effect

VARIABLES

call_duration

(1)

VA_app

(2)

Surprised

(3)

Understandable

(4)

Interested

(5)

Happy

(6)

accurate_you

(7)

accurate_factory

(8)

Treated 56.13*** -0.0653 0.192 -0.124 0.0625 -0.0155 0.00779 -0.0434

(8.59e-05) (0.575) (0.284) (0.185) (0.216) (0.806) (0.909) (0.618)

Constant 295.9*** 1.583*** 2.631*** 3.593*** 0.187*** 0.509*** 2.473*** 2.339***

(0) (4.48e-10) (0) (0) (0.000603) (2.82e-07) (0) (0)

Observations 272 248 233 253 264 264 264 261

VARIABLES

VA_effective

(9)

VA_common

(10)

sup_sat_VA

(11)

sup_sat_VA_NR

(12)

Useful

(13)

make_changes

(14)

Training

(15)

Treated -0.0379 0.0771 -0.00296 0.0254 0.0702 -0.00154 0.0413

(0.848) (0.590) (0.980) (0.689) (0.465) (0.984) (0.400)

Constant 1.981*** 1.854*** 1.935*** 0.391*** 3.673*** 3.673*** 0.555***

(3.85e-09) (-3.85e-09) (0) (2.23e-06) (0) (0) (6.72e-08)

Observations 247 227 224 224 251 241 256

*** p<0.01, ** p<0.05, * p<0.10

Page 21: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

21

Table 3

Model 2: Treatment and Call Duration

VARIABLES

VA_app

(1)

Surprised

(2)

Understandable

(3)

Interested

(4)

Happy

(5)

accurate_you

(6)

accurate_factory

(7)

Treated 0.382 0.374 -0.0210 -0.313 0.411* -0.134 0.111

(0.453) (0.497) (0.966) (0.146) (0.0884) (0.613) (0.719)

Call_duration -0.000585 -0.000289 -0.00101 -0.000207 0.000181 -0.000398 0.000164

(0.301) (0.638) (0.249) (0.277) (0.718) (0.475) (0.782)

TreatedXCall_duration -0.00120 -0.000489 -0.000144 0.00110* -0.00125* 0.000463 -0.000461

(0.387) (0.764) (0.918) (0.0779) (0.0547) (0.570) (0.609)

Constant 1.758*** 2.717*** 3.893*** 0.249*** 0.455** 2.592*** 2.290***

(1.04e-06) (4.55e-10) (1.83e-10) (0.0100) (0.0265) (5.73e-10) (5.75e-09)

Observations 248 233 253 264 264 264 261

VARIABLES

VA_effective

(8)

VA_common

(9)

sup_sat_VA

(10)

sup_sat_VA_NR

(11)

Useful

(12)

make_changes

(13)

Training

(14)

Treated -0.309 0.0349 0.245 0.0516 -0.0117 -0.443* 0.194

(0.627) (0.933) (0.639) (0.861) (0.974) (0.0705) (0.472)

Call_duration -0.00247** -0.00153** -0.000263 0.000888* 9.53e-05 -0.000958* 0.000893*

(0.0431) (0.0165) (0.812) (0.110) (0.860) (0.0972) (0.0894)

Page 22: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

22

TreatedXCall_duration 0.00110 0.000349 -0.000655 -0.000248 0.000221 0.00141* -0.000569

(0.524) (0.769) (0.671) (0.776) (0.811) (0.0785) (0.491)

Constant 2.728*** 2.310*** 2.009*** 0.140 3.645*** 3.960*** 0.289

(6.64e-06) (3.34e-09) (1.04e-06) (1.68e-05) (0.406) (0) (0)

Observations 247 227 224 224 251 241 256

*** p<0.01, ** p<0.05, * p<0.10

Page 23: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

23

Table 4

Model 3: Treatment and Verbal Abuse Attitudes

VARIABLES

Surprised

(1)

Understandable

(2)

Interested

(3)

Happy

(4)

accurate_you

(5)

accurate_factory

(6)

Treated -0.128 -0.229 0.161* -0.0858 -0.0529 0.0693

(0.634) (0.216) (0.0964) (0.359) (0.563) (0.766)

VA_appropriate -0.0589 -0.106 0.0345 -0.0236 0.0494 0.0704

(0.512) (0.243) (0.457) (0.730) (0.350) (0.388)

TreatedXVA_appropriate 0.171 0.0654 -0.0646 0.0501 0.0495 -0.0836

(0.137) (0.590) (0.274) (0.458) (0.215) (0.532)

Constant 2.755*** 3.754*** 0.140* 0.542*** 2.379*** 2.232***

(8.05e-11) (0) (0.0754) (8.50e-05) (0) (2.09e-10)

Observations 223 237 248 248 246 242

VARIABLES

VA_effective

(7)

VA_common

(8)

sup_sat_VA

(9)

sup_sat_VA_NR

(10)

Useful

(11)

make_changes

(12)

Training

(13)

Treated 0.152 -0.395* 0.0681 -0.00610 0.0884 -0.0260 0.0307

(0.579) (0.100) (0.750) (0.965) (0.569) (0.759) (0.831)

VA_appropriate 0.561*** 0.272*** 0.0675 -0.0740** -0.0748 -0.00159 -0.100**

(7.81e-05) (0.00465) (0.419) (0.0282) (0.347) (0.977) (0.0481)

Page 24: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

24

TreatedXVA_appropriate -0.120 0.324** -0.0563 0.0258 -0.00866 0.0123 0.0152

(0.427) (0.0110) (0.621) (0.705) (0.935) (0.761) (0.860)

Constant 1.090*** 1.384*** 1.825*** 0.487*** 3.773*** 3.666*** 0.705***

(2.73e-05) (1.26e-07) (2.75e-09) (1.21e-05) (0) (0) (2.10e-07)

Observations 236 218 209 209 240 229 243

*** p<0.01, ** p<0.05, * p<0.10

Page 25: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

25

Table 5

Model 4: Treatment and Factory Group

VARIABLES

VA_app

(1)

Surprised

(2)

Understandable

(3)

Interested

(4)

Happy

(5)

accurate_you

(6)

accurate_factory

(7)

Treated -0.0207 0.257 -0.299** 0.102 -0.0206 -0.129 -0.0564

(0.885) (0.352) (0.0176) (0.169) (0.824) (0.105) (0.683)

Group2 0.158 -0.177 -0.231** 0.0381 -0.0707 -0.187 0.00102

(0.522) (0.487) (0.0244) (0.669) (0.546) (0.136) (0.993)

TreatedXGroup2 -0.0866 -0.193 0.398** -0.0931 -0.00106 0.314*** 0.0330

(0.712) (0.577) (0.0150) (0.319) (0.993) (0.00489) (0.845)

Constant 1.509*** 2.717*** 3.702*** 0.169*** 0.542*** 2.559*** 2.339***

(0) (3.95e-09) (0) (0.00943) (6.07e-06) (0) (0)

Observations 248 233 253 264 264 264 261

VARIABLES

VA_effective

(8)

VA_common

(9)

sup_sat_VA

(10)

sup_sat_VA_NR

(11)

Useful

(12)

make_changes

(13)

Training

(14)

Treated 0.323* 0.200 -0.0281 -0.0397 0.0823 0.0894 0.00366

(0.0834) (0.345) (0.868) (0.658) (0.583) (0.454) (0.960)

Group2 0.568 0.113 -0.0492 -0.140 -0.0618 0.0133 -0.193*

(0.108) (0.594) (0.677) (0.141) (0.641) (0.897) (0.0909)

Page 26: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

26

TreatedXGroup2 -0.826** -0.280 0.0538 0.135 -0.0438 -0.224 0.0626

(0.0328) (0.311) (0.819) (0.278) (0.808) (0.128) (0.502)

Constant 1.724*** 1.800*** 1.958*** 0.458*** 3.702*** 3.667*** 0.644***

(1.06e-09) (4.67e-08) (0) (6.34e-05) (0) (0) (3.33e-08)

Observations 247 227 224 224 251 241 256

*** p<0.01, ** p<0.05, * p<0.10

Page 27: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

Figure 1. How common is verbal abuse? Percent responses by treatment and control groups.

3.45.1

9.2

11.412.6

11.4

18.5

22.2

42.9

39.2

13.4

10.8

01

02

03

04

0

perc

ent

NR very somewhat not very not at all DK/DW

C T C T C T C T C T C T

Verbal Abuse Common

Page 28: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

28

Figure 2. How do you feel about these results? [Interested] Marginal treatment effects by seconds to complete perspective-taking exercise.

Page 29: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

29

Figure 3. How do you feel about these results? [Happy] Marginal treatment effects by seconds to complete perspective-taking exercise.

Page 30: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

30

Figure 4. How likely are you to use these results to make changes? Marginal treatment effects by seconds to complete perspective-taking exercise.

Page 31: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

31

Figure 5. How common do you think it is to yell or use harsh language with workers in this factory? Marginal treatment effects by belief that verbal abuse is appropriate.

Page 32: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

32

Figure 6. How surprising are these results to you? Marginal treatment effects by belief that verbal abuse is appropriate.

Figure 7. How do you feel about these results? [Interested] Marginal treatment effects by belief that verbal abuse is appropriate.

Page 33: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

33

Appendix

Table A1 Treatment, Control and Follow-up Survey

Perspective Taking Survey What do you think workers in your factory prefer to have for breakfast?

1) Steamed rice with vegetables 2) Roti/bread 3) Steamed rice with water and chili 4) Other 5) Don't know/Don't want to answer

What do you think workers in your factory think about on their way to work?

1) What they will do at work that day or production target achievements 2) What their children or other family members will do that day 3) Future plans for their children 4) Other 5) Don't know/Don't want to answer

What do you think workers in your factory think about on their way home from work?

1) What they did at work that day or any special thing at the factory 2) Getting promoted to a better job within the factory or getting a production bonus 3) Their plans for the week 4) Other 5) Don't know/Don't want to answer

What do you think workers in your factory talk about with their children or other family members in the morning?

1) Getting to work/school on time 2) Doing a good job at school or work 3) Future plans or minimizing family expenses 4) Other 5) Don't know/Don't want to answer

What do you think workers in your factory talk about with their children or other family members in the evening?

1) any special things that happened at work or school that day 2) Upcoming celebrations 3) Preparing for a better future (through job promotions, more schooling, etc.) 4) Other 5) Don't know/Don't want to answer

What do you think workers in your factory prefer to eat when they get home from work?

1) Steamed rice with vegetables 2) Steamed rice with fish 3) Steamed rice with chicken/beef

Page 34: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

34

4) Other 5) Don't know/Don't want to answer

What concerns do you think workers in your factory have for their family?

1) Financial solvency or assets for the future 2) Children's education or future marriage 3) Happy family life 4) Other 5) Don't know/Don't want to answer

What hopes do you think workers in your factory have for their family?

1) Continuing to improve their quality of life 2) Children continuing their education 3) Minimizing their family expenses 4) Other 5) Don't know/Don't want to answer

What do you think workers’ goals are? 1) Build a better life of financial

solvency 2) educate their children or arrange good marriage(s) 3) Advance in their jobs 4) Other 5) Don't know/Don't want to answer

Control Survey What do you prefer to have for breakfast?

1) Steamed rice with vegetables 2) Roti/bread 3) Steamed rice with water and chili 4) Other 5) Don't know/Don't want to answer

What do you think about on your way to work?

1) What you will do at work that day or production target achievements 2) What your children or other family members will do that day 3) Future plans for your children 4) Other 5) Don't know/Don't want to answer

What do you think about on your way home from work?

1) What you did at work that day or any special thing at the factory 2) Getting promoted to a better job within the factory or getting a production bonus 3) Your plans for the week 4) Other 5) Don't know/Don't want to answer

Page 35: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

35

What do you talk about with your children or other family members in the morning?

1) Getting to work/school on time 2) Doing a good job at school or work 3) Future plans or minimizing family expenses 4) Other 5) Don't know/Don't want to answer

What do you talk about with your children or other family members in the evening?

1) any special things that happened at work or school that day 2) Upcoming celebrations 3) Preparing for a better future (through job promotions, more schooling, etc.) 4) Other 5) Don't know/Don't want to answer

What do you prefer to eat when you get home from work?

1) Steamed rice with vegetables 2) Steamed rice with fish 3) Steamed rice with chicken/beef 4) Other 5) Don't know/Don't want to answer

What concerns do you have for your family?

1) Financial solvency or assets for the future 2) Children's education or future marriage 3) Happy family life 4) Other 5) Don't know/Don't want to answer

What hopes do you have for your family?

1) Continuing to improve your quality of life 2) Children continuing their education 3) Minimizing your family expenses 4) Other 5) Don't know/Don't want to answer

Follow-up Survey Do you think these results accurately reflect what happens in this factory?

1) Yes, these results very accurately reflect the relationship between you and your workers 2) Yes, these results somewhat accurately reflect the relationship between you and your workers 3) No, these results do not reflect the relationship between you and your workers

Page 36: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

36

Do you think these results accurately reflect the relationship between you and your workers?

1) Yes, these results very accurately reflect the relationship between you and your workers 2) Yes, these results somewhat accurately reflect the relationship between you and your workers 3) No, these results do not reflect the relationship between you and your workers

How easy were these results to understand?

1) Very easy 2) Somewhat easy 3) Not very easy 4) Not at all easy

How do you feel about these results? 1) Happy 2) Upset 3) Interested 4) Neutral

How important is it for workers to be satisfied with the relationship with their supervisor?

1) Very important 2) Somewhat important 3) Not very important 4) Not at all important

How important are these results to you?

1) Very important 2) Somewhat important 3) Not very important 4) Not at all important

How surprising are these results to you?

1) Very surprising 2) Somewhat surprising 3) Not very surprising 4) Not at all surprising

How useful are these results to you? 1) Very useful 2) Somewhat useful 3) Not very useful 4) Not at all useful

How likely are you to use these results to make changes?

1) Very likely 2) Somewhat likely 3) Not very likely 4) Not at all likely

[If very likely or somewhat likely] What kind of changes are you most likely to make?

1) Strengthen HR 2) Increase training 3) Change factory policy

Page 37: Perspective-taking, information processing and workplace ... · 4/18/2018  · person. In one study, researchers showed photos of various types of people to participants while measuring

37

4) Other

How effective do you think it is to yell or use harsh language with workers in this factory?

1) Very effective 2) Somewhat effective 3) Not very effective 4) Not at all effective

How appropriate do you think it is to yell or use harsh language with workers in this factory?

1) Very appropriate 2) Somewhat appropriate 3) Not very appropriate 4) Not at all appropriate

How common do you think it is to yell or use harsh language with workers in this factory?

1) Very common 2) Somewhat common 3) Not very common 4) Not at all common