The Internet is a magnifying glass

Sorin A. Matei

Studying the Studying the social effects of social effects of

the Internet the Internet with a with a

“magnifying “magnifying glass”glass”

Is the Internet:

A Bridging tool? Once introduced it might reduce the distance

between individuals and groups Emphasis on access to technology

(Hiltz-Turoff – Network nation, Rheingold, Wired)

A chasm creator? The technology creates new inequalities – microserfs, symbolic manipulator masters and “the great digitally unwashed” (Barbrook and Cameron)

The “social effects of the Internet” seen in The “social effects of the Internet” seen in relationship with the digital dividerelationship with the digital divide

ProblemProblem

Questions are

Technology-centric (Wellman)Focused primarily on access

Re-framing the questionsRe-framing the questions• We need to broaden the work in this arena by looking at:

– What happens after access is not a problem anymore?– What happens to the social content / substance of Internet use?

• Assumption: The Internet does not singlehandedly create social gaps, or can reduce them

• The Internet is a catalyst, is the “yeast” in the social mix

• It favors specific behaviors (especially those with socially consequential effects) if these behaviors are already present

Two pronged approachTwo pronged approach

• Individual – what social behaviors are enhanced by the Internet?

• Social/group level – is social capital increased / diminished by Internet connections?

The individual approachThe individual approach

• A number of studies have noticed “magnification effects”

• Those socially active are more likely to• Adopt the Internet• Or to use it for social goals

Empirical evidence for Empirical evidence for magnification effectsmagnification effects

• Metamorphosis (Ball-Rokeach and Matei)

• GSS 2000 (Robinson and Neustadtl)

• Pew Internet Polls (Rainie, Jones and Howard)

• Syntopia (Katz, Rice and Aspden)

• Cyberville study in Toronto (Wellman and Hampton)

Belonging

Mass Communication

On-line connections

ChineseGreater Monterey Park

MexicanEast LA

CaucasianSouth Pasadena

KoreanGreater Koreatown

CaucasianWestside

African-AmericanGreater Crenshaw

Central AmericanPico Union

Bilingual Telephone Interviews1812 Households

Metamorphosis research Metamorphosis research strategystrategy

Some empirical evidence – Some empirical evidence – Metamorphosis Metamorphosis

• People who are more solidly anchored to their neighborhoods are more likely to make a friend on-line

• 7% increase in likelihood of making a friend on-line for each increase in a

“belonging index” score

• Married people are more likely to make a friend on-line than singles when they know someone in the neighborhood

• Singles are less likely to make a friend on-line when they do not know someone in the neighborhood

Community level effects in Community level effects in Metamorphosis studyMetamorphosis study

• In white neighborhoods the Internet indirectly contributes to social integration

• In Asian and Latino neighborhoods the Internet does not contribute -- directly or indirectly -- to social integration

Group effects at national levelGroup effects at national level

• Study of the 48 contiguous union states– states with the highest amount of social

capital are more likely to produce virtual groups – Yahoo! clubs – and to produce the most active groups

GSS 2000 Internet moduleGSS 2000 Internet module

• Contains questions about:– Reasons for using the Internet social reasons

included– Time spent with people off-line (distinguishes

between family and friends)

– Some findings:• Users are more likely to spend more time with neighbors and

friends• Those who use the Internet for social reasons are also more

likely to spend more time with friends – although not with family members

Further research questionsFurther research questions

• Individual level

• If high “socializers” in Real Life are high “socializers” in Virtual Space, how sustainable is this in the long run?– Will there be a tipping-off point, which will lead to a “reversal of

fortunes”? – Will, in the long run, on-line ties replace off-line ties? – Will this affect especially the virtual class, those living the digital

life, isolated in their “nerdistans” (Kotkin)?– This question, although asked many times and allegedly

answered, is still to be addressed. It requires longitudinal, national or large scale representative studies.

Further research questionsFurther research questions

• Group level

– How should we confront the failure of hardware dissemination to alleviate the problems of the poorest, less vital communities?

– Maybe we should address the issue of presence/absence of social capital first, before assuming that technology will create it

– Internet connections revitalize pre-existing community resources, cannot invent them from scratch

Social capital theorySocial capital theory• In trying to explain when and how the Internet might

contribute to / detract from social interaction I rely on Social Capital Theory (Coleman, Putnam)

• SCT main goal to explain social action

Social capital = social networks and informal / semi-formal organizations

SCT identifies the resources and motivations that explain social involvement and collective action

Generalized reciprocity: main resource for Generalized reciprocity: main resource for generating social capitalgenerating social capital

Social capital is more likely to be produced where social attitudes, expectations and obligations are directed by the principle of “generalized reciprocity”:

“I’ll do this for you without expecting anything specific back from you, in the confident expectation that someone else will do something for me down the road” (Putnam, 2000, Bowling Alone, p. 21).

Where generalized reciprocity is strong, there is are important “unintended consequence:” groups are easier to form and transactions (social, economic, political) to be negotiated more social capital

Extending social capital theory to studying the social Extending social capital theory to studying the social impact of communication technologyimpact of communication technology

• Generalized reciprocity is “trained” and developed in geographically situated communities: families, neighborhoods, schools, circles of friends and associates, political units

– shared cultures and values facilitate trust– a sense of obligation is stronger to those closer to us

(dark side)

• A sense of “generalized reciprocity” once acquired, becomes portable and extensible to other realms. We can take it with us wherever we go.

SCT implications for studying the InternetSCT implications for studying the Internet

• Generalized reciprocity and the habits of the heart associated with it will inflect our use of communication technology

– When colonizing the Internet we take our capacity to generate social capital with us

– We will generate social ties in a proportion commensurate with our general ability to produce social capital

The broad research questionThe broad research question

• Does off-line propensity for sociability influence on-line social interactions?

The need for a dual level of The need for a dual level of analysis strategyanalysis strategy

• Generalized reciprocity is a form of “positive externality:” neighborhood watch groups benefit even those who do not participate

• IS NOT an entirely individual phenomenon

• It is BOTH an individual and group process

• Not only individuals that present high propensity for generating social capital will be more likely to generate on-line social ties

• Social groups with potential for high social capital will manifest the same tendencies

More specific research questionsMore specific research questions

• Do individuals who have a higher propensity for generating social capital (GR), also have a higher propensity for involvement on-line?

• Do social environments with higher potential for generating social capital (GR) produce more on-line activity?

Individual-level side of the questionIndividual-level side of the question

• Explored in Los Angeles using a geographically focused sample

• Results reported in American Behavioral Scientist (2001) and in the Journal of Communication (in press)

Belonging IndexBelonging Indexcaptures level of social capital at individual level – assumes trust and captures level of social capital at individual level – assumes trust and

generalized reciprocitygeneralized reciprocity• Objective

– Number of neighbors known to:

• Talk about a personal problem

• Ask for a ride

• Watch over your home

• Assist with a repair

• Subjective

– Agree/disagree:

• It is easy to make friends with your neighbors

• You enjoy talking with your neighbors

• Your neighbors borrow things from you

• You are interested in knowing what your neighbors are like

Cronbach alpha .8

The question, again, is…The question, again, is…

• Are people with higher level of belonging (social capital) more likely to establish bonds on-line?

Analysis: logistic regressionAnalysis: logistic regression

• Depedent variable is binary– “Yes” / “No” answers to the question: “Have

you ever met someone you consider a personal friend?”

• How much does a predictor variable increase the odds of choosing one of two categories of the binary variable, controlling for other variables

Logistic regression resultsLogistic regression results

Those who “belong” more are 7% more likely to make a friend on-line for each “belonging index unit increase”

Controlling for gender, ethnicity, income, education, age, immigration history

Midway conclusionsMidway conclusions

• Belonging is positively associated with making friends on-line

• Social capital might be involved in generating sociability on-line

Switching levels of analysisSwitching levels of analysis

• Do individuals who have a higher propensity for generating social capital (GR) also have a higher propensity for involvement on-line?

• Do social environments with higher potential for generating social capital (GR) produce more on-line sociability?

Study started as part of an undergraduate research methods class

Paper presented in Maastricht, at the 3rd Conference of Internet researchers, under review at the Journal of Broadcasting and Electronic Media

MethodologyUsing states as units of analysis:

Do states with higher capacity for producing social capital generate more on-line sociability?

MethodologyUsing states as units of analysis:

Do states with higher capacity for producing social capital generate more on-line sociability?

Operationalizing “on-line sociability”Operationalizing “on-line sociability”

• Number of Yahoo! groups associated with a specific state of the union

• Yahoo! groups: Web-based electronic spaces where people interested in a specific location (state or smaller locations) can meet and communicate

• Bulletin board, chat, file sharing, photo uploads, community databases

• 4,597 Groups (M=95 / state)

• 170,050 Members

• 340,789 Messages

• Group size range: 1 - 2,239 members

AnalysisAnalysis

• Multiple (OLS) regression:

– Predict number of groups per 100,000 using capacity for generating social capital trust level (proxy for generalized reciprocity)

– Controlling for population homogeneity and density.

Clubs per 100,000 inhabitantsClubs per 100,000 inhabitantsdependent variabledependent variable

Main predictor variableMain predictor variable

• State-level of trust (capacity to generate social capital)

– % of those who answered “Yes” to the GSS question – “Most people can be trusted”

– Rough indicator of “generalized reciprocity”

• Used by Putnam in his “social capital” index.

% Yes “Most people can be % Yes “Most people can be trusted”trusted”

Co-variates (controls)Co-variates (controls)

• Population density• Population homogeneity (% foreign born)

Variables dropped after exploratory analysis• Internet penetration• Gross state product

– Highly correlated between them (r=0.6).– Highly correlated with level of trust (r=0.6).

ResultsResults  Unstandardized

CoefficientsStandardized Coefficients

t p

B Std. Error β    

% Yes: Most people

can be trusted 0.03 .014 .327 2.285 .02

Percent population

foreign born -.08 .032 -.420 -2.546 .01

Adjusted R2=.18

The higher the social capital, the more numerous the groups

The more homogeneous the population, the more numerous the groups

Follow-up analysisFollow-up analysis

• Is the consequence of high potential for social capital – higher social involvement – connected to on-line sociability?

• Can on-line sociability directly be predicted by off-line sociability?

• Predict number of Yahoo! groups using number of NGOs (501c3) / 1000 people

Follow up analysis resultsFollow up analysis results

Dependent variable: Yahoo! groups per 100,000 Adjusted R-square = .12

The more numerous the non-profit organizations, the more numerous the on-line groups

The more homogeneous the population, the more numerous the groups -- ns

ConclusionsConclusions

• Individual level:– On-line sociability probably has a “magnifying

glass effect” – helps those who have high level of belonging to extend their relationships on-line

• State level:– High social capital states generate more on-

line groups – sociability on-line reflects sociability off-line

Practical implicationsPractical implications

For designing on-line venues:• Sociability builds on sociability

• Sticky sites and groups are made of:a) sticky individuals who share at least some proximity

b) sticky technologies

• Seed the group with high social capital opinion leaders and motivators

Practical implicationsPractical implications

For the community activist / policy practitioner

• Hardware alone does not revitalize community or democracy

• Is the level of social capital sufficient to expect a specific pay-off from implementing the technology?

• IF NOT, energize first the social networks in the community

• Make them the anchors of the new computer network

The road from here…The road from here…

• Study of Lexington modeled after the Los Angeles study

– how does the specific spatial location of each respondent in a specific geographic location influence their social ties on and off-line?

• Yahoo! study follow-up:

– longitudinal analysis

How does the Internet/media How does the Internet/media interact with our social contexts?interact with our social contexts?

• The Internet as other media plays an important role in the process of social integration

• It facilitates emergence of “ties the bind”

• It serves as a “magnifying glass” – strengthens pre-existing propensities for social action

• Do you talk with other people about your neighborhood? (1-10 scale, median split)

• Are you a member of any community organization? • Do you primarily use community media or• Mainstream media for

– community information, entertainment or shopping – TV, newspapers, radio

• Do you have Internet access from home, work or anywhere else?

• Have you ever met someone on-line you consider a personal friend?

– Coleman: multiplexity and strength of ties between social actors (manifested as obligations and expectations), social norms related to trust and access/density of information channels

– Putnam: those features of social life—social

networks, norms and trust—that enable collective action

Belonging and new/old media Belonging and new/old media connectedness: a communication connectedness: a communication

infrastructure modelinfrastructure model

Connections toCommunity

Organizations

Local/Community Media

Connections

Participation in Interpersonal Storytelling BELONGING Internet

connection

Mainstream Mass Media

Connections1.8

1.7

5.6

1.4

1.4

1.6

Metamorphosis study: English-speaking samples

On-line sociability predicted by belongingOn-line sociability predicted by belonging((dv: “Have you ever met someone on-line you consider a personal friend?”)dv: “Have you ever met someone on-line you consider a personal friend?”)

Variable B S.E. Wald Sig Exp(B)

BELONGING .0639 .0296 4.6612 .0309 1.0660

GENDER .5391 .3013 3.2019 .0736 1.7144

AGE -.0095 .0143 .4344 .5098 .9906

EDUC .1811 .1156 2.4549 .1172 1.1985

INCOME -.1030 .0863 1.4238 .2328 .9021

IMMIG.GEN. -.1362 .1296 1.1045 .2933 .8727

KOREATOWN 3.2065 1.3314 5.7996 .0160 24.6915

KOREAN/BELONG.-.1231 .0702 3.0737 .0796 .8842

CRENSHAW .2197 .5752 .1459 .7025 1.2457

ELA -1.2143 .8942 1.8441 .1745 .2969

MONTEREY PARK .5827 .5824 1.0010 .3171 1.7908

WESTSIDE .1334 .5405 .0609 .8051 1.1427

PICO UNION -.5566 .8220 .4586 .4983 .5731

Operationalizing “Intensity of on-line activity”Operationalizing “Intensity of on-line activity”

Starting Point: How much activity is generated by a typical club in any given state?

– First instinct: average number of messages/member for each club, then average the averages

– Problem: ignores the fact that some clubs are larger or older, had more chances to facilitate activity

– Solution: “Adjusted” measure of “average number of messages” per club

• MEASURE CONCEPTUALLY: What would the number of messages sent to a typical club in any given state be if the influence of number of members and club age would be constant (the same)?

• OLS-procedure: DV: Number of Messages; IVs: # of members; club age in months; R-square = .65 Predict number of messages for each club using group size (# of members) and club age (longevity in months)

• MEASURE OPERATIONALLY: Average predicted number of messages for each state: SUM of predicted number of messages / Number of clubs in each specific state

ResultsResults

  Unstandardized Coefficients

Standardized Coefficients

t Sig.

  B Std. Error Beta    

Population density -0.0003 .000 -.315 -1.905 .065

% Answered Yes “Most people can be

trusted”0.05 .023 2.277 2.292 .028

% Answered Yes “Most people can be

trusted” squared-.0007 .000 -2.634 -2.637 .012

Adjusted R2=.21

Curvilinear relationship between trust and weighted average group activity

Sparser populated states generate more active clubs

Curvilinear relationship between Curvilinear relationship between trust and on-line activitytrust and on-line activity

GSS: "Most people can be trusted" % agree

70605040302010

Me

an

we

igh

ted

clu

b a

ctiv

ity

2.8

2.6

2.4

2.2

2.0

1.8

1.6

1.4

Geographic Region

West Coast

Southwest

Mountain

Plain States

Midwest

South

Mid Atlantic

North East

Total Population

WYWI

WV

WA

VA

VT

UT

TX

TN

SC

RI

PA

OR

OK

OH

ND

NC

NY

NJ

NHMT

MS

MO

MN

MI

MA

MD

LA

KY

KS

IA

IN

IL

GA

FL

CT

COCA

AR

AZ

AL

Curvilinearity of relationship between trust and activity Curvilinearity of relationship between trust and activity mirrored by that between off-line involvement and on-line mirrored by that between off-line involvement and on-line

involvementinvolvement

DDB Needham dataset: Club meetings attended last year

1110987654

Mea

n w

eigh

ted

club

act

ivity

2.8

2.6

2.4

2.2

2.0

1.8

1.6

1.4

Geographic Region

West Coast

Southwest

Mountain

Plain States

Midwest

South

Mid Atlantic

North East

Total Population

WV

WA

VA

VT

UT

TX

TN

SDSC

PA

OR

OK

OH

ND

NC

NY

NJ

NHNV

NE

MT

MS

MO

MN

MI

MA

LA

KY

KS

IA

IN

IL

IDGA

COCA

AR

AZ

AL