sociological perspectives and methods of research...

YOSH

IKA

ZU T

SUN

O/A

FP/G

etty

Imag

es

2 With Emphasis on Socially Assistive Robots

Sociological PerSPectiveS and MethodS of reSearch

46463_ch02_ptg01.indd 24 9/27/13 11:34 AM

NOT FOR SALE

Why Focus on a Socially Assistive Robots?

How do you think socially assistive robots (SARs) like the one shown in the photograph opening this chapter will redefine the way people work, play, learn, relax, and take care of each other? SARs are robots that are able to interact with humans and to meet cultural and social expectations associated with the roles they have been created to perform (International Journal of Social Robotics 2012). Those roles can involve assisting, coaching, or motivating people, whether it be by reminding them to take medicine, acting as a personal trainer, hand-ing tools to a surgeon, or teaching four-year-olds a foreign language. Sociology offers concep-tual and methodological tools to help us identify and sort through the issues accompanying this revolutionary technology and to formulate constructive response (Anderson, Mattin, and Webb 2012).

25

The integration of SARs into the home, school, and workplace is one of many global-scale trends predicted to revolutionize the way people live their lives. This trend is global in scale, because when we consider where the raw materials needed to manufacture robots come from, the locations of the work-forces that will manufacture them, the labor of the engineers and IT programmers who write the artificial intelligence (AI) al-gorithms to run them, and the populations that will eventually own them, there is no question that SARs are a phenomenon that is global in scale.

In this chapter, we showcase the three sociological perspec-tives and methods of sociological research as useful tools for analyzing and engaging with any trend or issue. We give special emphasis to one global-scale trend: the coming integration of SARs into society and our personal lives. More specifically these tools help us to study and think about a trend’s impact on work, family, and friendships and to evaluate and synthesize related information.

SARs can be viewed as the latest innovation in a progression of robots, beginning with industrial robots that do the heavy- duty repetitive manufacturing work in a fixed location. Industrial robots came on the scene in late 1960s, followed by the emergence of mobile or service robots in the mid- to late 1980s. Service robots can be professional or personal. Professional robots work for organizations devoted to defense, underwater explora- tion, agriculture (most notably milking cows), medicine, trans-portation, and rescue operations. Personal robots, on the other

hand, are produced for mass consumption to do things like vac-uum floors, mow lawns, entertain (toys, hobbies), provide care, and conduct surveillance. The International Federation of Robots (2012) estimates that 1.6 million industrial robots and 200,000 professional robots work in factories and other orga-nizational settings around the world. The federation also fore-casts that, between 2012 and 2015, people will buy 15.6 million personal robots.

SARs represent a major leap in innovation over industrial and basic service robots. For at least a decade, roboticists have been working to create SARs capable of assisting the growing numbers of elderly in need of varying degrees of care, and other popula-tions including children with autism. As a matter of policy, many governments anticipate that SARs will help meet their caregiving needs. Japan is a world leader in developing SAR technologies to care for its growing elderly population (Flandorfer 2012). The European Union has funded at least 156 projects, including the following titles: “Developing Robots’ Social Skills to Support Human-Robot Joint Action Execution,” “Computer Vision for Scene Understanding from a First-Person Perspective,” and “Intelligent Surgical Robotics” (European Commission 2013). Korea is a leader in developing robotic teaching assistants to be placed in more than 8,000 kindergartens (BBC Worldwide Radio 2013). The U.S. National Institutes of Health (2012) funds projects with the goal of “developing robots that can interact and work cooperatively with people and respond to changing environments.” (See “Sociological Imagination: Human Brains, Robotic Limbs.”)

d d j d d

46463_ch02_ptg01.indd 25 9/27/13 11:34 AM

NOT FOR SALE

26 Chapter 2 Sociological Perspectives and Methods of research

Some current news headlines offer a glimpse into a future that includes social robots:

● “Future Surgeons May Use Robotic Nurse, ‘Gesture Rec-ognition” (Science Daily 2011)

● “Robots in the Home: Will Older Adults Roll Out the Welcome Mat?” (Science Daily 2012)

● “Talking Robots Play Part in Therapeutic Treatment for People with Special Needs” (PBS Newshour 2013)

Sociologists draw on three broad perspectives—functionalist, conflict, and symbolic interaction—to guide analysis of any trend, issue, or situation. Each offers a cen-tral question to direct thinking and key concepts to orga-nize answers. We begin with a general overview of the functionalist perspective and then apply it to an analysis of SARs. Keep in mind that, taken alone, no single per-spective can offer a complete picture of a situation. But we can acquire a more complete picture by applying all three. Of course, as we will learn, each theory has its strengths and weaknesses.

Functionalist PerspectiveCORE CONCEPT 1 Functionalists focus on how the “parts” of society contribute in expected and unex-pected ways of maintaining an existing social order. They also focus on ways “parts” can disrupt that social order.

Functionalists define society as a system of interre-lated, interdependent parts. To illustrate this vision,

The sociological imagination alerts us to consider how time and place shape human biographies. People’s lives are products of three interrelated factors: chance (things over which they have no control), choice (decisions they make), and context (the larger social environment in which we live our lives). Innovations in robots are such that humans can use their brains to move robotic limbs, and the hope is that robotic limbs will eventually be capable of sending sensory feedback to the human brain. These developments change the context in which people who have been in accidents or were born without limbs (chance) now live their lives. Sociologists are also interested in questions of who will have access to robotic limbs.

Sociological Imagination Human Brains, Robotic Limbs

Vet

eran

s Af

fairs

and

DEK

A

26

Advances in SAR technologies are sure to create industrial robots that do more than stand in place 24/7, repeating the same motions. Technological advancements related to balance, flex-ibility, vision, and touch will allow robots, for example, to “zoom at the speed of the world’s fastest sprinters” and work in ware-houses where they “store, retrieve, and pack goods for shipment far more efficiently than people” (Markoff 2012).

Defe

nse

Adva

nced

Res

earc

h Pr

ojec

ts A

genc

y (D

ARP

A)

46463_ch02_ptg01.indd 26 9/21/13 2:32 PM

NOT FOR SALE

No Borders, No Boundaries Distribution of Industrial Robots by Region and Leading Countries

At the time of this writing, an estimated 1.4 million indus-trial robots were working in factories around the world, most notably in automotive and electronic sectors. World-wide, there are currently 55 industrial robots for every 10,000 manufacturing workers, and that ratio is expected to increase exponentially. The map highlights 15 countries in red that rank among the world leaders in industrial robots. The 23 countries highlighted in blue constitute Tier 2 countries. The map also shows the four countries with the greatest number of robots for every 10,000 car manufacturing workers (one segment of industrial robots).

The number of industrial robots is sure to increase dramatically. Simply consider that Taiwan-based Foxconn Technology Group, which makes electronic devices including Apple iPads, HP computers, and other elec-tronic devices, has begun the process of installing one million robots—called Foxbots—in Chinese factories (Kaiser 2012). While this map highlights the countries believed to be among the top 38 countries with indus-trial robots, we might predict that the countries high-lighted on this map will also be the Tier 1 and Tier 2 leaders in SARs.

functionalists use the human body as an analogy for society. The human body is composed of parts including bones, cartilage, ligaments, muscles, a brain, a spinal cord, hormones, blood, blood vessels, a heart, kidneys, and lungs. All of these body parts work together in impressive harmony. Each part functions in a unique way to maintain the entire body, but it cannot be sepa-rated from other body parts that it affects and that in turn help it function.

Society, like the human body, is made up of an incalcu-lable number of parts such as schools, cars, sports teams,

funeral rituals, holidays, religious rituals, laws, robots, and smartphones. Like the various body parts, each of society’s parts is interdependent and functions to main-tain a larger system. Functionalists define a function as the contribution a part makes to an existing social order.

27

FIguRE 2.1 Tier 1 and 2 Countries Leading the World in Use of Industrial RobotsSource of data: International Federation of Robotics (2012).

function The contribution a part of a society makes to an existing social order.

Germany1,176 robots per

10,000 auto workers

United States1,104 robots per

10,000 auto workers

Italy1,215 robots per

10,000 auto workers

Japan1,584 robots per

10,000 auto workers

Tier 1 CountriesTier 2 Countries

46463_ch02_ptg01.indd 27 9/21/13 2:32 PM

NOT FOR SALE


Social order refers to the way people have organized interaction and other activities to achieve some valued goal—whether it be to take care of the sick, to pass on

knowledge, to encourage interest in robots, and so on. A part’s function or contribution to the social order can be manifest or latent.

Manifest and Latent Functions

When a part’s effect on a social order is something that is expected, anticipated, or intended, that effect is a manifest function. When that part’s effect is unintended, not antic-ipated, or unexpected, that effect is a latent function. To grasp this distinction, consider the manifest and latent functions of sports teams—whether they be Little League, high school, college, or professional teams. A manifest function of a sports team is that it can unite fans who are often extremely different from one another economi-cally, culturally, linguistically, politically, religiously, and in countless numbers of other ways. Thus, sports teams function to transcend individual differences and foster a sense of belonging to a school, city, or country. Sports teams, if successful, can also have the unexpected or latent function of increasing enrollment at a school, espe-cially a university. Winning an NCAA national champi-onship in basketball or football, for example, is typically followed by an 8 percent increase in applications. Some universities such as St. Mary’s, Butler, and George Mason have reported 22 to 30 percent increases in applications after making successful NCAA tournament appearances (Crowell 2012).

In the most controversial form of this perspective, func-tionalists argue that all parts of society—even those that do not seem to serve a constructive purpose such as pov-erty, crime, and undocumented immigration—contribute in some way to maintaining some existing social order. In fact, functionalists argue that a part would cease to exist if it did not serve some function. Thus, functional-ists strive to identify how even seemingly problematic “parts” contribute to maintaining a social order. Consider

The U.S. government, through agencies like NASA, is making a concerted effort to support a social order that generates interest in robotics among school-age children. The Robotics Alliance Project was established to inspire high school students to pursue a college degree in robotics engineering (NASA 2013). In the United States, the most popular major is business management, with 8 percent of all majors declaring it (Johnson and Lubin 2011; U.S. Department of Education 2012). In India, engineering is among the most popular majors. In fact, in that country, 1.5 million seats are available to college students who want to become engineers (Anadd 2011).

Ella

Reh

man

U.S.

Air

Forc

e ph

oto/

Staf

f Ser

gean

t Mik

e M

eare

s

From a functionalist perspective, sports teams function to tran-scend differences among fans and foster a sense of belonging to an associated school, city, or country.

social order Refers to the way people have organized interaction and other activities to achieve some valued goal—to take care of the sick, to pass on knowledge, to encourage interest in robots, and so on.

manifest functions Intended or anticipated effects that a part has on the existing social order.

latent functions Unintended or unanticipated effects that a part has on the existing order.

46463_ch02_ptg01.indd 28 9/21/13 2:32 PM

NOT FOR SALE

manifest and latent functions and dysfunctions to orga-nize an answer to this key question. The purpose of the analysis that follows is not to generate an exhaustive list of functions and dysfunctions associated with this pending integration but to demonstrate how functionalists frame an analysis.

Manifest and Latent Functions of SARs

To identify the manifest functions (anticipated effects on social order and stability), we need to think about some of the most often-stated motives for creating SARs, the kind of robots capable of reading and responding to social cues, such as facial expressions or body movements, and interpreting facial expressions or body cues in cultural or social contexts. In addition, a truly social robot will be capable of responding with behavior, facial expressions, and gestures appropriate to the social and cultural context in which it is operating. At the time of this writ-ing, artificial intelligence (AI) researchers are working to perfect algorithms that allow robots to “learn and autono- mously develop sensory-motor and communicative skills” (Ramsey 2013). Because of dramatic improvements in supporting technologies such as microprocessors, sensors,

one function of poverty: Poor people often “volunteer” for over-the-counter and prescription drug trial tests. Most new drugs must eventually be tried on healthy human subjects to determine their potential side effects (for example, rashes, headaches, vomiting, constipation, and drowsiness). The chance to earn money motivates subjects to volunteer for these clinical trials. Because payment is relatively low, however, the tests attract a disproportion-ate share of low-income, unemployed, or underemployed people as subjects.

This function of poverty shows why a part of the soci-ety that everyone agrees is problematic and should be eliminated remains intact: It contributes to the stability of the pharmaceutical and medical systems. A functionalist would argue that without poverty, these systems would be seriously strained to find human subjects to test out new medial procedures and pharmaceutical products.

As you might imagine, this line of reasoning can lead to charges that functionalists defend the existing social order or status quo. Thus most functionalists also consider manifest and latent dysfunctions—concepts that help them to think, not just about a part’s contribution to maintain-ing a social order, but also its disruptive effects on order (Merton 1967).

Manifest and Latent Dysfunctions

Parts can have dysfunctions; that is, they can have disrup-tive consequences to the social order or to some segment within that social order. Like functions, dysfunctions can be either manifest or latent. Manifest dysfunctions are a part’s anticipated disruptions to a social order. While sports teams are known to foster fan loyalty to a school, community, or country, they also are known to foster loyalties so strong that in some cases fans have behaved in violent or disorderly ways after a loss, especially to a bitter rival. We can classify this response as a manifest dysfunction. Latent dysfunctions are unanticipated or unintended disruptions to a social order. Here we might argue that sports teams, in their quest to win champion-ships, can foster insecurities among players over being cut or losing playing time. In response, players may resort to playing with head and other injuries; using human growth hormone, testosterone, and other illegal substances; and engaging in unethical behavior such as deliberately trying to injure an opponent.

The Functionalist Perspective on Socially Assistive Robots

To see how the functionalist theory can be applied to a global trend or some other trend, we will consider how functionalists analyze the coming integration of SARs into society. Functionalists ask, “What are the anticipated and unintended consequences to social order of integrat-ing SARs into society?” Functionalists use the concepts of

functionalist Perspective 29

CB2/

ZOB/

Mac

hine

Per

cept

ion

Labo

rato

ry/W

E/N

ewsc

om

The research and development for this robotic boy, named Diego San, was funded by the National Science Foundation. Diego San has high-definition camera eyes that allow him to scan people’s gestures and facial expressions. AI algorithms allow the robot to mimic the processes by which children learn from observing others. It is important that Diego San be able to cor-rectly interpret facial expressions and gestures and respond with appropriate facial expressions and gestures if he is “to establish a relationship, and communicate intuitively to people” (Science Daily 2013).

dysfunctions Disruptive consequences of a part to the existing social order or some segment within that social order.

manifest dysfunctions A part’s anticipated disruptions to an existing social order.

latent dysfunctions Unintended, unanticipated disruptions to an existing social order.

46463_ch02_ptg01.indd 29 9/21/13 2:32 PM

NOT FOR SALE


and AI algorithms, robotics technology is approaching a “tipping point,” at which point SARs will take off (Kalil and Kota 2011).

Three of the most often-stated motives for developing social robots offer insights into expected or manifest func-tions of SARs: (1) to provide assistive care to people with disabilities, including children with autism and the grow-ing number of elderly in need of care; (2) to do jobs that are repetitive, boring, dangerous, and prone to human error; and (3) to reduce labor costs. These expectations qualify as manifest functions.

What might be some possible unexpected (latent) func-tions associated with social robots? If SARs realize the expected (manifest) functions, a number of unintended or latent functions to social order will surely follow. For one, if SARs are developed to the point where they can assume caregiving roles, governments will likely reduce the number of immigrants they admit as nurses aids, for example, to assist the growing number of elderly. By exten-sion, cultural misunderstandings and frictions that might have arisen between caregivers and clients with different national backgrounds are eliminated. In addition, risks of inappropriate sexual contact and abuse are eliminated. Finally, robots will not suffer from caregiver burnout or become impatient with their clients or students. Robots are also even-tempered, predictable, nonjudgmental, and always available and responsive.

Manifest and Latent Dysfunctions of SARs

One expected disruption (dysfunction) to the existing social order relates to employment. SARs are already being tested in prisons as guards. A South Korea prison tested the effectiveness of three robots acting as war-dens monitoring suspicious behavior exhibited by inmates (BBC Worldwide Radio 2012). At Carnegie Mel-lon, robotic programs are developing AI algorithms that allow SARs to assume roles as butlers, bartenders, and receptionists (Heater 2012). Robots at the University of California San Francisco Hospital have filled 350,000 prescriptions without error (Aquino 2012). Then there are SARs tied to transportation known as driverless cars. Will that innovation make taxicab drivers, package delivery drivers, and truckers irrelevant? Although some experts predict that other high-skilled, higher-paying jobs will be created to replace lost jobs in transportation, it is unlikely that those who lose jobs will be the ones to fill the newly created ones.

A latent or unexpected dysfunction relates to the emotional attachment humans may come to feel toward

SARs such that people may actually prefer the company of robots over that of humans. After all, there is “not so much to figure out” about a robot, “where there is a lot to figure out about a human being” (McGuire 2013). If this preference for robots becomes widespread, it might lead to the creation of a society like that described by sci-ence fiction writer Isaac Asimov, who warned “against the risks of dehumanization” that may accompany the “exces-sive and indiscriminate technical progress” on the planet Solaria:

Solaria was a planet inhabited by Spacer descendants. . . . Origi-nally, there were about 20,000 people living alone in vast estates. Solarians’ lives were marked by technology: Citizens never had to meet, save for sexual contact for reproductive purposes. All other contact was accomplished by sophisticated holographic viewing systems, with most Solarians exhibiting a strong phobia towards actual contact, or even being in the same room as another human. All work was done by robots: There were indeed thousands of robots for every Solarian. As centuries went by, Solaria became even more rigidly and obsessively isolationist. The planet cut off all contact with the rest of the Galaxy (although continuing to monitor hyperspatial communications). . . . At the final stage of Solarian civilization, the human inhabitants . . . had in fact withdrawn underground; their estates continued to be worked by millions of robots. (Antoci and Sodini 2011)

Conflict PerspectiveCORE CONCEPT 2 The conflict perspective focuses on conflict over scarce and valued resources and the strategies advantaged groups used to create and protect the social arrangements from which they benefit.

In contrast to functionalists, who focus on social order, conflict theorists focus on conflict as an inevitable fact of social life and as the most important agent for social change. Conflict can take many forms, including physical confrontations, exploitation, disagreement, tension, hos-tility, and direct competition. In any society, advantaged and less disadvantaged groups compete for scarce and val-ued resources (access to material wealth, education, health care, well-paying jobs, and so on). Those who gain control of and access to these resources strive to protect their own interests against the competing interests of others. Con-flict theorists ask this basic question: Who benefits from a particular social arrangement, and at whose expense? In answering this question, conflict theorists seek to describe the social arrangements that advantaged groups have established, consciously or unconsciously, to promote and protect their privileged position. Exposing these practices helps explain inequalities that exist in society.

Conflict theorists work to expose the facade of legitimacy—an explanation to justify the existing social

facade of legitimacy An explanation to justify the existing social arrangements that downplays or dismisses any possibility that the arrangement advantages some groups over others.

46463_ch02_ptg01.indd 30 9/21/13 2:32 PM

NOT FOR SALE

complex environments and to unpredictable situations (Scassellati, Admoni, and Matarić 2012). In other words, if SARs can be “taught” to work with autistic children, they can work with any population.

Conflict theorist would predict that SAR technologies will usher in a new divide separating advantaged popula-tions with the money to acquire SARs from those disad-vantaged populations without the financial resources to do so. The divide will created advantaged and disadvan-taged individuals, households, and communities within countries. That divide will also exist on a global scale, separating the richest countries with disproportionately greater access to SARs from the poorest countries with sig-nificantly less or no access.

Symbolic Interaction PerspectiveCORE CONCEPT 3 Symbolic interactionists focus on social interaction and related concepts of self-awareness/reflexive thinking, symbols, and negoti-ated order.

Sociologist Herbert Blumer coined the term symbolic inter-action and outlined its essential principles. Symbolic inter-actionists focus on social interaction, everyday encounters in which people communicate, interpret, and respond to one another’s words and actions. These theorists ask, when

arrangement that downplays or dismisses any possibil-ity that the arrangement advantages some groups over others. On close analysis, that justification is based on unsupported assertions and implausible premises but nevertheless is still presented as logical (Carver 1987). To illustrate, consider how some investors and CEOs of corporations, who depend on the minimum-wage labor of hundreds of thousands—if not millions of workers—to manufacture products or serve customers, justify their large salaries that are sometimes 500 times greater than that of those employees. These CEOs might argue that workers who are dissatisfied with their low wages are free to quit. Furthermore, CEOs further justify their advan-taged position when they claim that their leadership and decision-making skills are the reasons for a company’s financial successes, never acknowledging the contribu-tion of low-wage laborers. The popularity of the TV show Undercover Boss might be because it gives dignity to the millions of low-wage laborers who are the heart and soul of corporations such as Choice Hotels, Waste Manage-ment, Inc., Chiquita Brands International, and White Castle.

The Conflict Perspective on Socially Assistive Robots

In analyzing the integration of SARs in society, conflict theorists ask, “Who benefits from integration of these robots, and at whose expense?” In answering this ques-tion, they would point out an obvious fact: The creation of SARs is driven by a desire to create a “social arrangement” that maximizes profit by eliminating human labor and its associated costs (wages, benefits packages) in as many areas as possible, including product production (factories), distribution (packaging and transportation), and services (caregiving, teaching, rescue).

Conflict theorists view any altruistic motivations such as to assist the growing numbers of elderly or to work with children with autism spectrum disorder as a façade to cover up the profit motive that is really driving the creation of SARs (see “No Borders, No Boundaries: Five Coun-tries with Greatest Number and Percentage of People Age 65 and Over”). For example, although the U.S. government and some corporations fund SARs projects to help the esti-mated 1 in 88 children with autism-related disorders, there are likely other motives. Conflict theorists would argue that is it no accident that robot therapy for autism was “one of the first application domains in the field of socially assistive robotics” (Scassellati, Admoni, and Matarić 2012). This is a population of children who “do not behave consis-tently day to day or even hour to hour,” who are one minute engaged and the next can become distracted, angry, and uncooperative. In reality, these children are used as pawns to assist a much larger effort to create robust, flexible SARs with control architectures that can adapt to changing and

Symbolic interaction Perspective 31

Isse

i Kat

o/RE

UTE

RS/L

ando

v

From a conflict point of view, this SAR acting as a teacher’s assistant will eventually pay for itself many times over. Imagine that this robot costs the equivalent of $50,000 and that teach-ing assistants earn the equivalent of $25,000 per year plus an expensive benefit package. The robot would pay for itself in less than two years.

social interaction Everyday encounters in which people communicate, interpret, and respond to each other’s words and actions.

46463_ch02_ptg01.indd 31 9/21/13 2:32 PM

NOT FOR SALE

involved in interaction, how do people “take account of what each other is doing or is about to do” and then direct their own conduct accordingly (Blumer 1969)? The process depends on (1) self-awareness, (2) shared symbols, and (3) negotiated order.

Self-Awareness

Self-awareness occurs when a person is able to observe and evaluate the self from another’s viewpoint. People are self-aware when they imagine how others are view-ing, evaluating, and interpreting their words and actions. Through this imaginative process, people become objects to themselves; they come to recognize that others see them, for instance, “as being a man, young in age, a stu-dent, in debt, trying to become a doctor, coming from an undistinguished family, and so forth” (Blumer 1969, p. 172). In imagining others’ reactions, people respond and make adjustments (apologize, change facial expres-sions, lash out, and so on).

Shared Symbols

A symbol is any kind of object to which people assign a name, meaning, or value (Blumer 1969). Objects can be classified as physical (smartphones, cars, a color, a facial

expression), social (a friend, a parent, a celebrity, a bus driver), or as abstract (freedom, greed, justice, empathy). In the context of driving, for example, the color green has come to symbolize “go or proceed.” Objects can take on different meanings depending on audience and context: A tree can have different meanings to an urban dweller, a farmer, a poet, a home builder, an environmentalist,

symbol Any kind of physical phenomenon to which people assign a name, meaning, or value.

Robe

rt K

. Wal

lace

Look at the expression on these two children’s faces. Their blank facial expressions suggest that they have no idea what to do even with prompting from their mothers. That is because they lack self-awareness—the ability to think about how others see them and what others expect of them in a particular situation.

● Five Countries with Greatest Number of People Age 65 and Over

China 133.3 million India 67.1 million United States 42.5 million Japan 30.4 million Russia 18.5 million

● Five Countries with Greatest Percentage of Population Age 65 and Over

Japan 23.9% Russia 23.9% Germany 20.7% Italy 20.5% Greece 19.8%

There are 578.9 million people in the world who are age 65 and older (U.S. Central Intelligence Agency 2013).

About 43 percent live in five countries—China, India, United States, Japan and Russia. The five countries with the greatest number of people age 65 and over and the five countries with the greatest percentages of people age 65 and over are listed to the left. From a conflict point of view, which emphasizes profit making as the ultimate force driving innovation, the elderly represent a huge market to sell products such as SARs. One can easily envision the endless number of add-on modules or apps to be offered at extra cost beyond the basic SAR model—it might be an app for moving around cluttered indoor environments, an app for specific types of physical assistance like opening doors, turning water in showers on and off, carrying laundry, and so on. “Incrementally new” versions of SARs could be released every six months (Blackman 2013).

Source of data: U.S. Central Intelligence Agency (2013).

No Borders, No Boundaries Five Countries with Greatest Number and Percentage of People Age 65 and Over

32

46463_ch02_ptg01.indd 32 9/21/13 2:32 PM

NOT FOR SALE

or a lumberjack (Blumer 1969). People learn meanings that their culture attaches to objects. That is, they learn that a wave of the hand means good-bye, that letters of the alpha-bet can be selected and arranged to make countless words, and that dogs but not crickets are considered as pets.

Negotiated Order

When we enter into interaction with others—whether with a store clerk, a professor, friends, colleagues—we take for granted that a system of expected behaviors and shared meanings is already in place to guide the interaction. That is, we are generally aware that we should behave and talk in a certain way. Although expectations are in place, sym-bolic interactionists emphasize that established meanings and ways of behaving can guide the course of interaction, but that they can also be ignored, challenged, or changed (Blumer 1969). In most interactions, room for negotiation exists; that is, the parties involved have the option of nego-tiating other expectations and meanings. The negotiated order, then, is the sum of existing expectations and newly negotiated ones (Strauss 1978).

To illustrate, college students know, for example, that when they enter a classroom on the first day of class, they should not walk to the front of the room and give instructions to the class. Likewise, professors know that on the first day of class, students expect them to give an overview of the course and lay out expectations such as no texting during class. Already established expectations are in place, guiding interaction. Usually, however, some room for negotiation exists; that is, the parties involved have the option of negotiating a social order. So on the first day of class, a professor may negotiate with students, indi-cating it is okay to check text messages when the professor is passing out assignments but not during lecture or class discussions. However, professors know that they cannot “negotiate” a social order in which students pay money to receive a desired grade.

The Symbolic-Interactionist Perspective on Socially Assistive Robots

As we have learned, symbolic interactionists study people as they engage in social interaction. The symbolic interac-tion focus on self-awareness, shared symbols, and negoti-ated order are especially relevant to describing the skills SARs must possess if they are to interact successfully with humans. The robots must be guided by AI algorithms that enable them to step outside themselves and evaluate their performance from the point of view of those with whom they are interacting. For robots to be truly social and assis-tive, they must be able to interpret words, gestures, and facial expressions, and then respond accordingly. The con-cept negotiated order alerts us to a final AI hurdle: negotiat-ing social order. It would be relatively easy to create robots

to be caregivers if expectations for how people behave were fixed and unchanging—that is, if there were not so many exceptions to the rules depending on context and situations within contexts. The central question driving the writing of AI algorithms is one that is central to sym-bolic-interactionsist perspective. That question is: How do robots take account of what the people with whom they are interacting are doing or are about to do and then direct their own conduct accordingly?

Symbolic interaction Perspective 33

negotiated order The sum of existing expectations and newly negotiated ones.

TATR

CN

ASA

How does a robot’s appearance shape people’s beliefs about its ability to perform a role? If you were a soldier on a battlefield, would you be more confident if the robot rescuing you had a bearlike face or a helmeted face and bulging muscles? We might speculate people will have more confidence in robots’ abilities to do a task if they look like they fit the role to which they are assigned.

46463_ch02_ptg01.indd 33 9/21/13 2:33 PM

NOT FOR SALE


it leaves us wondering about a part’s overall effect on that order. So with regard to SARs, we are left asking: Do the manifest and latent functions of SARs outweigh the associ-ated manifest and latent dysfunctions?

One strength of the conflict perspective is that it forces us to look beyond popular justifications for why particu-lar social arrangements exist, and to ask questions about whose interests are being protected and promoted and at whose expense. A weakness of the conflict perspective is that it presents a simplistic profile of those who hold advantaged positions—that profile presents advantaged groups as driven only by profit and the desire to protect their interests. The most advantaged are portrayed as all-powerful and capable of imposing their will without resis-tance from disadvantaged groups, who are portrayed as exploited victims. But there are likely many people among the advantaged who think beyond the bottom line and self-interest. In addition, conflict theorists fail to recog-nize that the drive to make a profit is responsible for inno-vation. Without the profit motive, how much innovation would take place?

The strength of the symbolic-interactionist theory is that it focuses on up-close and personal factors that shape the course of interaction and relationships. A symbolic interactionist, however, can get caught up in interaction dynamics and lose sight of the larger structural issues in which that interaction is embedded, including the existing social order and a profit-driven economy.

Functionalist Conflict Symbolic Interactionist

Focus Order and stability Conflict over scarce and valued resources

Social interaction

Key Terms Function, dysfunction, manifest, latent

Conflict, facade of legitimacy Self-awareness, shared symbols, negotiated order

Vision of Society

System of interrelated parts Advantaged and less advantaged groups in conflict over scarce and valued resources

Web of social interactions

Central Question(s)

What are the expected/intended /anticipated and unexpected /unintended/unanticipated consequences of a part?

Who benefits from a particular pattern or arrangement, and at whose expense?

How do involved parties experience, interpret, influence, and respond to what they and others are doing in the course of interacting?

Strength Balanced analysis of positive and negative effects

Encourages analysis beyond popular explanations that justify existing social arrangements

Encourages direct, firsthand, and in-depth analysis

Weakness Defends existing socialarrangements; difficult to determine a part’s overall effect

Presents simplistic view of advantaged versus disadvantaged groups; fails to acknowledge the role of profit making in advancing major innovations

Focuses on interaction dynamics and may lose sight of the larger social context in which interaction is embedded

FIguRE 2.2 Overview of the Three Theoretical Perspectives

Because the success of SARs is dependent on their ability to interact successfully with people, the symbolic-interactionist perspective is especially useful for think-ing about factors that shape the success of human–robot interaction. To what extent does successful interaction depend on the robot’s appearance; for example, does it matter if the robot is anthropomorphic (humanlike), zoomorphic (animallike), or machinelike in appearance? To be even more specific, should a robot charged with babysitting appear adultlike in appearance so that par-ents perceive the robot as responsible? Should it appear gender-neutral or have traits that evoke associations of masculinity or femininity? Apart from appearance, there are questions about how a robot should convey disagree-ment. Should the robot express disagreement in explicit ways (“I think this choice is not correct”) or in implicit ways (“Are you sure?”)? Research suggests that German clients and Chinese clients tend to be more comfortable with explicit styles, and Americans with implicit (Li, Rau, and Li 2010).

Critique of Three Sociological Theories

Each perspective has its strengths and weaknesses (see Figure 2.2). One strength of the functionalist perspec-tive is that it gives a balanced overview by considering a part’s intended and unintended consequences to the exist-ing social order. One weakness of that perspective is that

© C

enga

ge L

earn

ing

2015

46463_ch02_ptg01.indd 34 9/21/13 2:33 PM

NOT FOR SALE

● identifying variables and specifying hypotheses, ● collecting and analyzing the data, and ● drawing conclusions.

It is important to know that researchers do not always follow the six steps in sequence. For example, they may decide on a research question after they have familiar-ized themselves with the existing literature. Although the six steps need not be followed in any particular order, all must eventually be completed to ensure the quality of the project.

In the sections that follow, we will examine each stage. Along the way, we will feature a research project of soci-ological significance—a National Science Foundation-funded study of a socially assistive robot integrated into a workplace—conducted by Min Kyung Lee, Sara Kiesler, Jodi Forlizzi, and Paul Rybski (2012a, 2012b) at the Human-Computer Interaction Institute and the Robotics Institute, and hereafter referred to as Lee.

Establishing a Research Question/Reviewing the Literature

It is impossible to list all the research topics that are con-sidered sociological because almost any subject involv-ing human activity qualifies. Sociology is distinguished from other disciplines, not by the specific topics it inves-tigates, but by the questions that guide inquiry. The soci-ological significance of the social robot study lies with its focus on meanings employees assign to a social robot and the effect the robot’s presence has on the workplace. Specifically, the questions driving Lee’s research were: How will employees relate to and interact with a social robot? How will the robot affect the workplace? Might the findings of this study have implications for designing social robots and for successfully integrating them into work environments?

When doing research, it is important to review the existing literature to take into account what experts have already published on the chosen topic, if only to avoid reinventing the wheel. Reading the relevant literature can also generate insights about a chosen topic that may not been have considered. Even if researchers believe that they

To this point, we have seen how each of the three socio-logical perspectives organizes analysis through a central question and key concepts (see Figure 2). Before we turn to methods of social research, we will consider how the sociological perspectives and the theories we will cover later in this text relate to the research process.

The Methods of Social ResearchResearch methods are the various strategies sociologists and other scientists use to formulate and answer meaning-ful research questions and to collect, analyze, and interpret data gathered. Here, the term data is used in the broadest sense; it applies to observations recorded, responses to sur-vey and interview questions, and much more. The three perspectives inspire the questions guiding research and offer a framework for interpreting the findings. In turn, research findings can offer support to or challenge the the-oretical perspective(s).

CORE CONCEPT 4 Sociologists adhere to the scien-tific method; that is, they acquire data through obser-vation and invite others to critique and replicate the research.

Sociologists adhere to the scientific method, a care-fully planned data-gathering and data-analysis process that researchers open to outside critique and replication. Replication is “the heart of good research” (Dye 1995, p. D5). No finding can be taken seriously unless other researchers can repeat the research study and obtain the same results. When researchers know that others are cri-tiquing and replicating their work, the process serves to reinforce careful, thoughtful, honest, and conscientious behavior. Moreover, this “checking” encourages research-ers to maintain objectivity; that is, it encourages them not to let their personal, or subjective, views about the topic influence their observations or the outcome of the research.

Ideally, the research process and any critique of that process should be guided by the core values of honesty, skepticism, fairness, collegiality, and openness (National Academy of Sciences 1995). In practice, though, some research is dismissed as unimportant or unworthy of examination simply because the topic or the researcher is controversial or because the findings depart from main-stream thinking. Moreover, some researchers fabricate data to support a personal, an economic, or a political agenda.

The research process involves at least six interdependent steps:

● establishing a research question, ● reviewing the literature, ● choosing a research design,

the Methods of Social research 35

research methods Various strategies that sociologists and other scientists use to formulate or answer meaningful research questions and to collect, analyze, and interpret data gathered.

scientific method A carefully planned data-gathering and data-analysis process that researchers open to outside critique and replication.

objectivity A stance in which researchers’ personal, or subjective, views do not influence their observations or the outcomes of their research.

46463_ch02_ptg01.indd 35 9/21/13 2:33 PM

NOT FOR SALE

Researchers can choose from a variety of data-gathering methods, including self-administered questionnaires, interviews, observation, and secondary sources.

Self-Administered Questionnaire A self-administered questionnaire, one of the most common methods of data collection, is a set of questions given to respon-dents, who read the instructions and fill in the answers on their own. The questions may ask respondents to write out answers (open ended) or to select one answer from a list of choices (forced choice).

This method of data collection offers a number of advantages: No interviewers are needed to ask respon-dents the questions, and the questionnaires can be distributed to large numbers of people at one time. Also, an interviewer’s facial expressions or body language can-not influence respondents, so the respondents feel freer than they otherwise might to give unpopular or contro-versial responses. Self-administered questionnaires pose some problems, too. First, respondents can choose not to participate. Second they can misunderstand or skip over questions. Finally, the quality of the information collected depends not only on respondents’ decisions to answer questions conscientiously but also on the quality of the questions.

Interviews Compared with self-administered question-naires, interviews are more personal. In these face-to-face, phone, or electronically mediated conversations

between an interviewer (which may be a robot) and a respondent, the interviewer asks questions and records the respondents’ answers. As respondents give answers, interviewers (whether human or robotic) must avoid pauses, expressions of surprise, or body language that reflect value judgments. Refraining from such conduct helps respondents feel comfortable and encourages them to give honest answers.

Interviews can be structured or unstructured, or some combination of the two. In a structured interview, the wording and sequence of questions are set in advance and cannot be altered during the course of the interview. In contrast to the structured interview, an unstructured interview is flexible and open ended. The questions are not in a set order or worded in advance; rather, the question-and-answer sequence is spontaneous and resembles a normal conversation. The interviewer allows respondents to take the conversation in directions they present as crucial. The interviewer’s role is to give focus to the interview, ask for further explanation or clari-fication, and probe and follow up on interesting ideas expressed by respondents. In other words, interviewers appraise the meaning of respondents’ and then decide what to ask next.

Observation As the term implies, observation involves watching, listening to, and recording behavior and conversations as they happen. This research technique

Methods of Data Collection

have a new idea, they must still review the literature to establish how their research verifies, advances, or corrects the existing literature.

In the write-up of this research, Lee et al. included about 40 scholarly books and articles as references. Ide-ally, scholarly writings are written by experts who care-fully document the sources of their ideas and who have actually conducted research on the subject of interest. Scholarly writings are typically reviewed by experts who assess the quality of the arguments made, offer sugges-tions for improvement, and recommend whether the work deserves to be published.

In choosing the literature to review, researchers cer-tainly read the most current materials, but the publication date should not be the only criterion. Classic and ground-breaking articles written decades or even centuries ago can be very important resources. In reviewing the literature,

self-administered questionnaire A set of questions given to respondents who read the instructions and fill in the answers themselves.

interviews Face-to-face, telephone, or electronically mediated conversations between an interviewer and a respondent, in which the interviewer asks questions and records the respondent’s answers.

structured interview An interview in which the wording and sequence of questions are set in advance and cannot be changed during the interview.

unstructured interview An interview in which the question-and-answer sequence is spontaneous, open-ended, and flexible.

observation A research technique in which the researcher watches, listens to, and records behavior and conversations as they happen.

36

46463_ch02_ptg01.indd 36 9/21/13 2:33 PM

NOT FOR SALE

may sound easy, but it entails more than just watch-ing and listening. The challenge of observation lies in knowing what to look for while remaining open to other considerations. “It is a crucial choice, often determining the success or failure of months of work, often differ-entiating the brilliant observer from the . . . plodder” (Gregg 1989, p. 53). Good observation techniques are developed through practice; observers learn to recognize what is “worth” observing, to become alert to pat-terns, to take detailed notes, and to make connections between observed behaviors.

Observational techniques are especially useful for three purposes: (1) studying behavior as it occurs, (2) learning information that survey questions cannot easily capture, and (3) appreciating the experiences and viewpoints of those under observation. Observation can take two forms: participant and nonparticipant. Non-participant observation consists of detached watching and listening; that is, researchers merely observe but do not interact with the study subjects or become emotion-ally involved in their daily life. In contrast, researchers engage in participant observation when they join a group and assume the role of a group member, interact directly with individuals whom they are studying, or live in a community under study. In other words, observa-tion research qualifies as participant observation when researchers became involved in the daily life of the people studied, even actively participating.

In both participant and nonparticipant observation, researchers must decide whether to hide their identity and purpose or to announce them. One major reason for choosing concealment is to avoid the Hawthorne effect, a phenomenon in which research subjects alter their behavior when they learn they are being observed. If researchers choose to announce their identity and pur-pose, they must give participants adequate time to adjust to their presence. Usually, if researchers are present for a long enough time, their subjects will eventually display natural, uninhibited behaviors.

Secondary Sources (Archival Data) Another data-gather-ing strategy relies on secondary sources, data that have been collected by other researchers for some other pur-pose. Researchers draw on secondary data, for example, when they use data governments routinely collect, includ-ing records of births, deaths, marriages, divorces, crime, educational attainment, travel, and trade. Any researcher who takes some pre-existing data set and then uses it to address their research question is using a secondary source (Horan 1995). Another kind of secondary data source consists of materials that people have written, recorded, or created for reasons other than research (Singleton, Straits, and Straits 1993). Examples include television commercials and other advertisements, letters, diaries, home videos, poems, photographs, artwork, graf-fiti, movies, and song lyrics.

researchers learn not only what is known about the topic of interest but also learn about gaps in the literature or things not yet known. By identifying gaps, researchers establish how their research can advance knowledge on the chosen topic.

After reviewing the existing literature, Lee learned that people are most likely to accept, trust, and engage with SARs when the robots are capable of making small talk and displaying signs of empathy. Lee noticed that much of the research on how humans respond to SARs had been conducted in labs or in public settings involv-ing one-time or otherwise very short encounters. To her knowledge, “no studies have followed the same employ-ees over an extended period.” In other words, there seemed to be no research about how socially interac-tive robots fit into the “culture of a real workplace.” Lee realized that the literature could benefit from a research

study that focused on human responses to social robots over extended and repeated encounters to offset the novelty effect.

37

nonparticipant observation A research technique in which the researcher observes study participants without interacting with them.

participant observation A research technique in which the researcher observes study participants while directly interacting with them.

Hawthorne effect A phenomenon in which research subjects alter their behavior when they learn they are being observed.

secondary sources (archival data) Data that have been collected by other researchers for some other purpose.

46463_ch02_ptg01.indd 37 9/21/13 2:33 PM

NOT FOR SALE


Choosing a Research Design

After determining the research question, researchers typ-ically decide upon a research design. That involves decid-ing who or what to study and the method of gathering data. Lee’s research design involved observing employ-ees as they interacted with a robot, doing face-to-face interviews with employees about their experiences, and distributing surveys to employees to learn related infor-mation. For this study, Lee and her colleagues designed Snackbot, a robot capable of delivering snacks, and decided to try it out on people who worked in a uni-versity. Before settling on a snack delivery service, Lee surveyed employees who worked on one floor of the com-puter science building to find out what kind of service they would like to see offered. The survey results showed that employees were open to snack delivery, especially to deliveries of fresh, healthy snacks. Lee hung flyers, sent out postcards, and used snowball sampling (refer-rals from those who had already signed up) to recruit potential customers from among those who worked on that floor. In the end, 21 people signed up—eight women and 13 men ranging in age from 22 to 51. Eleven of the customers were graduate students, eight were staff, and two were faculty. Only one customer had some experi-ence with robots. Customers ordered desired snacks from a website, indicated a preferred delivery day, and gave their office number.

Snackbot, a 4.5-foot-tall anthropomorphic robot with a male voice, made the deliveries on Monday, Wednes-day, and Friday between 2:30 and 4:00 p.m. Because it was impossible for Snackbot to deliver 21 customer orders during the 4.5 hours the service was made available, half of the customers had access to the service during the first two months of the study and the other half had access in the last two months. Snackbot, who looked something like the robot pictured on this page, could smile, frown, or show a neutral expression. Snackbot was fitted with a web camera and microphone to record his interactions with customers. Later, Lee and her colleagues reviewed these visual and audio recordings in search of meaning-ful patterns.

If customers were out of their office when Snackbot arrived, the snack was hung in a paper bag on their office doorknob. Altogether, the researchers recorded 346 inter-actions over a 4-month period. The snacks were free and given as compensation to those participating in the study who agreed to do three things: allow their interactions

with Snackbot to be recorded, complete a satisfaction sur-vey, and do a 30- to 60-minute face-to-face interview with Lee when the study ended.

Specifying Variables and Operational Definitions

A variable is any behavior or characteristic that consists of more than one category. Age is a variable, as age cat-egories can range from seconds old (at birth) to 100-plus years old. Gender is a variable, as people can identify as male, female, or transgender. Time elapsed is a variable, as time can range from less than a second passed to any number of years passed. All variables used in a research study must be operationalized or given an operational definition;

CORE CONCEPT 5 If findings are to matter, research-ers must create operational definitions that are reliable and valid. Researchers also test hypotheses.

Operational definitions are clear, precise instructions about how the variables are observed or measured. Note that the words observed and measured have the same meaning. If gender is a variable, researchers could measure it by asking people “what is your gender?” and offering set responses from which people must choose. Researchers could also measure or observe gender by looking at a person and deciding what gender the person appears to be.

For Lee’s study, four of the variables important to the study were (1) presence or absence of personalized conver-sation with the Snackbot, (2) whether or not agreed-upon rules emerged for how to treat and interact with Snackbot,

U.S.

Nav

y ph

oto

by J

ohn

F. W

illia

ms

Snackbot’s speech, head, and mouth were “controlled remotely with a laptop connected to the robot through a wireless network.” A “Wizard of Oz”—or “behind-the-curtain-style” setup allowed operators working in a room off-site to see and hear customer interactions with Snackbot using video/audio feeds.

variable Any behavior or characteristic that consists of more than one category.

Operational definitions Clear, precise definitions and instructions about how to observe and/or measure the variables under study.

46463_ch02_ptg01.indd 38 9/21/13 2:33 PM

NOT FOR SALE

for those assigned to the second group, Snackbot engaged in personalized conversation.

Norm formation refers to rules for behavior that emerge as a group adjusts to some change in routine until enough times passes that the “change” even-tually becomes part of the routine. With respect to this study, the change to the office was a robot-guided snack delivery service. When Snackbot first appeared on the scene, the group had obviously established no norms about how to relate to the robot and make it part of the work environment. Over time, norms about how to treat Snackbot and make the robot part of the work environment emerged.

Cooperation is action taken to assist or meet a request someone makes. To measure cooperation with the robot, Lee recorded whether or not customers (1) gave suggestions to Snackbot when it asked for names of places on campus that might be of interest to tourists, (2) complied when the robot suggested to a customer to take a break from work and join it in doing a “neck stretch,” and (3) agreed to try a special healthy mys-tery snack Snackbot recommended. Cooperation is a variable because customers can cooperate on all three requests, two requests, one request, or none of the requests.

(3) whether or not customers cooperated with Snackbot’s requests, and (4) whether or not customers assigned or did not assign humanlike qualities to Snackbot. The opera-tional definition of each follows:

Personalized conversation involves dialogue in which one party makes reference to things specific to the other party. Conversation between Snackbot and customers was impersonal when the com-ments could apply to anyone, such as “It is a nice day today. I am glad to see you and hope you are doing well.” Conversation was personalized when it contained statements like, “It seems as though you really like apples. This is the fifth time you have ordered that,” or “You were out of the office last time I brought you an apple. I missed seeing you.” Personalized conversation is a variable because Snackbot followed either an impersonal script or a personalized one. For the first four deliveries, Snackbot stuck to an impersonal script. After that, customers were randomly assigned to one of two groups. For those assigned to one group, Snackbot continued to engage in impersonal conversation;


NAS

A

Lee and her colleagues were interested in the how people respond to a robot integrated into workplace settings. What factors, for example, might make some people more likely to cooperate with robots and assign them humanlike qualities and motivations?

Taking Action

Chris

Cal

deira

In many research studies, gender is treated as a variable with only two categories—male and female. The assumption is that each category captures shared experiences associ-ated with being or appearing male or female that shape identity, behavior, and thinking. In the case of Lee’s study, she wanted to learn whether being male and female shaped the way customers interacted with Snackbot. The LBGT movement, however, has prompted researchers to question the two-category scheme. For example, how do people who identify as transgender fit in the two-category system? What about customers who feel like or are perceived as “masculine females” or as “feminine males.”

46463_ch02_ptg01.indd 39 9/21/13 2:33 PM

NOT FOR SALE


Anthropomorphize is the act of ascribing human char-acteristics to something that is not considered human. This is a variable in the study because customers either assigned Snackbot human characteristics such as feel-ings and motivations or they did not.

Operational definitions must be reliable and valid. An operational definition has reliability if everyone who repeats that operational definition obtains essentially the same results. Consider the reliability of Lee’s measure of cooperation. If customers answered Snackbot’s ques-tion about good locations in the building to take tour-ists, took a break from work to join the robot in doing a “neck stretch,” and agreed to try the mystery snack, it seems likely that anyone would classify those custom-ers as fully cooperative. So in this sense, Lee’s measure is reliable. Issues of reliability arise, however, when it is not clear whether someone cooperated with Snackbot. What if someone did the neck stretch but only half-heartedly stretched while still working? What if the person rolled her eyes while doing the neck stretch? Two people trying to classify the customers’ reactions as cooperative or not may not reach the same conclusion, raising questions about the reliability of the operational definition of coopera-tion with Snackbot. Unless clear instructions guide how to classify “half-hearted” efforts or eye rolling, the measure for cooperation cannot be reliable.

An operational definition must also be assessed for its validity or the extent to which the described measure actu-ally measures what it claims to measure. Does Lee’s mea-sure of cooperation really measure willingness to cooperate with a robot? One might argue that agreeing to do a neck stretch or answering a question about the location of an event are trivial tests of cooperation and therefore not valid measures. Perhaps a stronger measure of cooperation with a robot might involve giving the robot a different, more

serious role in the workplace such as a health coach pro-moting employee fitness. As health coach, the robot could request customers eat a healthier snack or to take a brisk walk on their lunch break. If customers cooperated by doing something that really mattered, then we have a valid measure of cooperation with a robot in workplace settings.

Hypotheses

Researchers strive to test hypotheses, predictions about how variables are related. The behavior to be predicted is the dependent variable. The variable that is believed to predict the behavior of interest is the independent vari-able. In Lee’s study the researchers sought to determine whether customers’ exposure to personalized conversation with Snackbot (the independent variable) increased the likelihood they would cooperate with the robot’s requests (the dependent variable). Specifically, if the hypothesis is supported by the data, then researchers can claim that if they know the value of an independent variable, then they can better predict the dependent variables. Three of the hypotheses Lee tested are:

Hypothesis 1: Customers who experience personal-ized conversations with Snackbot are more likely to cooperate with Snackbot’s requests than those who experience impersonal scripted conversation.

Hypothesis 2: Customers who engage in personal-ized conversations with Snackbot are more likely to anthropomorphize Snackbot than those who engage in impersonal scripted conversation.

Hypothesis 3: As time passes and Snackbot a fixture in the workplace, norms about how to treat him will emerge.

In addition to identifying independent and dependent variables, researchers also identify control variables, variables other than the independent variables that are associated with both the dependent and independent vari-ables and that researchers hold constant so they can focus on just the relationship between the independent variable and the dependent variable. Think about it this way—there are any number of variables that both relate to how someone might respond to personalized-impersonalized conversation and a decision to cooperate. One such vari-able might be the customer’s gender. Perhaps female and transgender customers are more likely to value personal-ized conversation than are males, and for that reason, per-sonalized conversation is a significant factor in females’ and transgenders’ decisions to cooperate than it is for males. How do researchers hold variables such as gender constant? They compare the males’ level of cooperation in Group 1 (impersonalized) with the males’ level of coop-eration in Group 2 (personalized). If males’ levels of coop-eration are the same in both groups, we can say that type of conversation had no bearing on decisions to cooperate.

reliability The extent to which an operational definition gives consistent results.

validity The degree to which the described measure actually measures what it claims to measure.

hypothesis A trial explanation put forward as the focus of research; it predicts how independent and dependent variables are related and how a dependent variable will change when an independent variable changes.

dependent variable The variable to be explained or predicted.

independent variable The variable that explains or predicts the dependent variable.

control variables Variables other than the independent variables that are associated with both the dependent and independent variables and that researchers hold constant so they can focus on just the relationship between the independent variable and the dependent variable.

46463_ch02_ptg01.indd 40 9/21/13 2:33 PM

NOT FOR SALE


FIguRE 2.3 The figures summarize data collected on the percentage of customers who cooperated with Snackbot’s request to do a neck stretch and the percentage that agreed to try a recommended snack. We can see from the charts that personalized conversation dramatically increased the likelihood that customers cooperate with Snackbot’s requests. Source of data: Lee et al. (2012a, 2012b).

a) Percentage of Customers in Personalized versus Impersonalized Conversation Groups Who Cooperated with Snackbot’s Request to Do Neck Stretches

92%

61%

Personalized Impersonalized0

100

50

Perc

ent W

ho D

id N

eck

Stre

tch

b) Percentage of Customers in Personalized versus Impersonalized Conversation Groups Who Cooperated with Snackbot’s Recommendation to Try a Mystery Snack

95%

58%

Perc

ent W

ho T

ried

Mys

tery

Sna

ck


100

50

Likewise, if we compare females’ levels of cooperation who are in Group 1 (impersonalized) and Group 2 (personal-ized) and we find that females who have engaged in per-sonalized conversation are more cooperative than those who have not, then we can claim that personalized con-versation increases female propensity to cooperate. We do the same comparisons for transgender customers. Con-trolling for gender allows researchers to be more precise about under what circumstances personalized conversa-tion matters or does not matter in securing cooperation. If, however, the level of cooperation is no different across the three gender categories—male, female, transgender—then we can say that gender does not play a role in explaining cooperation. We could control for other customer charac-teristics—age, race, and status in workplace—to be clearer about the conditions that yield cooperation.

Collecting and Analyzing the Data

CORE CONCEPT 6 Researchers collect data that they then analyze to see if there is support for their hypotheses.

When researchers analyze the collected data, they search for common themes and meaningful patterns. In presenting their findings, researchers may use graphs, tables, photos, statistical data, quotes from interviews, and so on. Lee and her colleagues presented data in the form of bar charts. Two examples follow (see Figures 2.3 and 2.4).

1.59

0.75


2

1.5

1

0.5

Num

ber o

f Loc

atio

ns N

amed

FIguRE 2.4 Average Number of Locations Customers in Personalized versus Impersonalized Conversation Groups Named in Response to the Snackbot’s Request for Good Places to Take Tourists This bar chart shows the average number of locations customers named in the personalized conversation group versus in the impersonalized conversation group. On average, customers who received personalized attention offered 1.59 locations that tourists to campus might like to see versus .75 locations by those customers who received impersonal attention.Source of data: Lee et al. (2012a, 2012b).

46463_ch02_ptg01.indd 41 9/21/13 2:33 PM

NOT FOR SALE


Among other things, Lee analyzed the transcripts of conversations customers had with Snackbot, looking for patterns about if and when customers anthropomor-phized the robot. She found that those in the personal-ized group were more likely to treat Snackbot like a person than those in the impersonalized group. Lee offered the following examples to illustrate. When Snackbot asked one customer assigned to the impersonalized conversa-tion group to recommend locations to take tourists, that person answered: “Snackbot, let’s not be ridiculous, can I take my snack?” In contrast, a customer assigned to the personalized group responded, “Let’s see. You could visit the [exhibit name] on the first floor or the third floor. The second floor has a lot of cool other robotic stuff that you could check out or show people, I think they would like that.”

Drawing Conclusions

CORE CONCEPT 7 Researchers conclude by discuss-ing the broader implications of their research study.

For a research study to be significant, it must have implications that extend beyond the people or setting in the study. Lee maintained that her findings (not all of which are described in this chapter) have important implications for a future in which SARs will inevitably become part of the workplace. Her research confirmed that under the right conditions, people can, and do, form collegial (even social) relationships with work-place robots.

We can also use the three perspectives to help clarify implications. From a functionalist’s perspective, Lee’s study speaks to manifest and latent functions and dys-functions. Robots can reduce human labor needed to do routine, repetitive tasks like snack delivery (a manifest function). People can become attached to robots (latent function). Emotions for SARs can fuel workplace conflicts and jealousies if people compete for the attention and favor of their robot “colleague”(latent dysfunction).

With regard to the conflict theory, Lee’s findings sug-gest that one way to ease the transition of robots into the workplace is to make them likeable and even preferable to humans. Those emotions can distract people’s atten-tion away from the real purpose of integrating them into workplaces—to eliminate human labor and the associated wages and benefits. Finally, from a symbolic-interactionist point of view, Lee’s research suggests that the success of SARs depends on a robot’s ability to interact with human workers, most notably on their ability to engage in person-alized conversation.

generalizability The extent to which findings can be applied to the larger population from which a sample is drawn.

Taking Action

Jos

h El

lings

on. C

ourt

esy

of C

hris

Cal

deira

Lee’s research focused on human–robot interaction in the workplace and examined the process by which norms emerged to guide how customers treat robots. As such, it fits especially well with the symbolic-interactionist perspec-tive. This is because Lee’s research offers important insights about what factors contribute to successful human–robot interaction. The most important factor identified in Lee’s research was the robot’s ability to engage in personalized small talk.

A research study also concludes with a discussion about generalizability, the extent to which findings can be applied beyond the random or nonrandom sample studied. One might question the generalizability of Lee’s study because she only studied one workplace setting, a floor in a computer science building on a college campus. In addition, that workplace setting involved computer sci-ence employees who may be predisposed favorably toward social robots. Finally, the customers she studied were not randomly chosen from all the employees who worked on

46463_ch02_ptg01.indd 42 9/21/13 2:33 PM

NOT FOR SALE

Summary of core concepts 43

that floor. Rather, the people she studied were self-selected. That is, they chose to sign up for a snack delivery service. We do not know if, and in what ways, the people who did not sign up were different from those who did. Perhaps they are less sociable, for example.

Lee acknowledged these potential shortcomings but she countered with the argument that the chosen setting might, in fact, prove more difficult to integrate a robot into because computer science types may be less likely to think of robots as social beings (to humanize them, for example). If that is in fact the case, then other audiences may prove even more susceptible to cooperating with, humanizing, and accepting robots into the work environment. Despite doubts over generalizability, Lee’s research is still intrigu-ing and offers valuable insights about human –robot inter-actions in workplace settings.

In evaluating the shortcomings of Lee’s study, keep in mind that researchers cannot possibly study everyone. Instead, they study samples, or a subset of cases from a larger population of interest. Ideally, samples should be

random, a situation in which every person or case has an equal chance of being selected. When a sample is ran-domly chosen, it increases the chances that it is repre-sentative of the population from which it is selected. We should note that there are compelling reasons to study nonrandom samples. For example, there has been no research about the process of integrating social robots into workplace settings. So Lee’s research is useful for it fills a void about how any employees in any workplace setting react to SARs.

samples Portions of the cases from a larger population.

random sample A type of sample in which every case in the population has an equal chance of being selected.

representative sample A type of sample in which those selected for study have the same distribution of characteristics as the population from which it is selected.

Summary of CORE COnCEpTS

In this chapter, we showed how the three sociological perspectives and methods of sociological research are useful tools for analyzing and engaging with trends and issues. We gave special emphasis to one trend: the coming integration of SARs into the workplace and our personal

lives. More specifically, these tools help us to identify signs that a trend is in the making; to study a trend’s impact on work, family, and friendships; and to evaluate and synthesize related information.

Corp

oral

Pat

rick

Fl

eisc

hman

With regard to SARs, functionalists focus on anticipated (manifest) and unintended (latent) consequences to the existing social order. Manifest functions of SARs include reducing labor costs, providing assistive care to people with disabilities and doing jobs that are repetitive, bor-ing, dangerous, and prone to human error. SARs’ latent functions include eliminating the risk of abusive behavior

and inappropriate sexual contact and caring for clients 24/7 in an even-tempered, predictable, and nonjudgmen-tal way. One manifest dysfunction is that SARs will create massive unemployment. A latent dysfunction is that people may come to prefer the company of robots over that of fellow humans, creating a society where people can and do avoid each other.

CORE CONCEPT 1 Functionalists focus on how the “parts” of society contribute in expected and unexpected ways to maintaining an existing social order. They also focus on ways “parts” can disrupt that social order.

CORE CONCEPT 2 The conflict perspective focuses on conflict over scarce and valued resources and the strategies advantaged groups use to create and protect the social arrangements from which they benefit.

Conflict theorists ask this basic question: Who benefits from a particular social pattern or arrangement? In the case of the SARs, conflict theorists argue that an honest analysis of the situation would reveal that advantaged groups will be those with the resources to produce and own SARs. Employers able to replace human labor with

SARs will save money, as they will not have to pay sala-ries and benefits. Unlike human employees, SARs will work 24/7 and never complain. Populations that need care and other assistance represent a market to whom SARs can be sold, including endless upgrades and various applications.

46463_ch02_ptg01.indd 43 9/21/13 2:33 PM

NOT FOR SALE


When doing research, sociologists explain why their research topic is important, tie their research in with existing research, and specify the core concepts guid-ing analysis. Sociologists decide on a plan for gathering data, identifying whom or what they will study and how they will select (sample) subjects for study. Sociolo-gists use a variety of data-collection methods, including

self-administered questionnaires, interviews, observation, and secondary sources. In addition, they propose and test hypotheses that specify the relationship between inde-pendent and dependent variables. In their study of human cooperation with robots, Lee and her colleagues followed the scientific method and used a research design that involved observation, face-to-face interviews, and surveys.

CORE CONCEPT 5 If findings are to matter, researchers must create operational definitions that are reliable and valid. Researchers also test hypotheses.

Operational definitions are clear, precise instructions about how to observe or measure the variables. Good operational definitions are both reliable and valid. Lee and her colleagues created operational definitions for personalized conversation, norm formation, coopera-tion, and anthropomorphizing. Researchers strive to

test hypotheses that predict how variables relate to each other. One hypotheses Lee tested that was supported by the data was that customers who experience personalized conversations with Snackbot are more likely to cooper-ate with the robot’s requests than those who experience impersonal scripted conversation.

CORE CONCEPT 6 Researchers collect data that they then analyze to see if there is support for their hypotheses.

When researchers analyze the collected data, they search for common themes and meaningful patterns. In present-ing their findings, researchers may use graphs, tables, photos, statistical data, quotes from interviews, and so

on. After systematically reviewing transcripts of conversa-tions customers had with Snackbot, Lee found support for all her hypotheses.

CORE CONCEPT 7 Researchers conclude by discussing the broader implications of their research study.

Lee’s findings have important implications for a future in which SARs will inevitably become part of the work-place. Her research showed that people can and do form collegial, and even social, relationships with robots.

Despite doubts about generalizability, Lee’s research is still intriguing and offers valuable insights about human–robot relationships.

control variables 40dependent variable 40dysfunctions 29

facade of legitimacy 30function 27generalizability 42

hawthorne effect 37hypothesis 40independent variable 40

Key Terms

CORE CONCEPT 4 Sociologists adhere to the scientific method; that is, they acquire data through observation and invite others to critique and replicate the research.

CORE CONCEPT 3 Symbolic interactionists focus on social interaction and related concepts of self-awareness/reflexive thinking, symbols, and negotiated order.

Symbolic interactionists focus on social interaction and ask, How do involved parties experience, interpret, influence, and respond to what they and others are doing while interacting? They draw upon the follow-ing concepts: (1) reflexive thinking, (2) symbols, and (3) negotiated order to think about any social situation.

Because the success of SARs depends on their ability to interact successfully with humans, the symbolic-interactionist perspective is especially useful for thinking about factors that increase likelihood of successful inter-actions. Those factors are the ability to think reflexively, interpret symbols, and negotiate interaction.

46463_ch02_ptg01.indd 44 9/21/13 2:33 PM

NOT FOR SALE

Key terms 45

interviews 36latent dysfunctions 29latent functions 28manifest dysfunctions 29manifest functions 28negotiated order 33nonparticipant observation 37objectivity 35observation 36

operational definitions 38participant observation 37random sample 43reliability 40research methods 35samples 42scientific method 35secondary sources (archival data) 37self-administered questionnaire 36

social interaction 31social order 28structured interview 36symbol 32unstructured interview 36validity 40variable 38

46463_ch02_ptg01.indd 45 9/21/13 2:33 PM

NOT FOR SALE

sociological perspectives and methods of research...

Documents