Vous êtes sur la page 1sur 11

Social Networks 35 (2013) 593–603

Contents lists available at ScienceDirect

Social Networks
journal homepage: www.elsevier.com/locate/socnet

Ask and you shall receive: Social network contacts’ provision of help
during the job search
Lindsey Trimble O’Connor ∗
California State University Channel Islands, United States

a r t i c l e i n f o a b s t r a c t

Article history: Social network contacts—the people who are asked to help with others’ job searches—are key players
Received 21 January 2013 in the job networking process. Before job seekers can become employed with the help social networks,
Received in revised form 27 May 2013 contacts must first be able and willing to share the social resources job seekers need for their search.
Accepted 25 July 2013
Little is known about the factors that affect contacts’ ability and willingness to do this. Analyses of a
unique dataset of contacts show that they typically have access to resources and help job seekers by
Keywords:
sharing them. Still, contacts are better able to help when they are male, employed, and better educated
Social capital access/mobilization
than job seekers. They are more willing to help when they perceive job seekers to be “good” workers.
Job searching
Inequality
In identifying the conditions in which contacts provide social resources, this study illustrates how social
networks are a productive job search strategy for some, but not all, job seekers.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction yet, it is safe to assume that some job seekers do not receive help
from contacts when they ask for it. For example, some contacts
Researchers have long known that a primary way people may be unable to help job seekers with their search and some may
search for and find work is by asking the members of their social choose not to (Marin, 2007, 2012, 2013; Lin, 2001; Newman, 1999;
network—their friends, family members, neighbors, coworkers, Royster, 2003; Smith, 2005, 2007, 2010). Job seekers who activate
acquaintances—for help with their job searches (Granovetter, 1995; social capital through their networks are likely to have more pro-
Ioannides and Loury, 2004; Marsden and Gorman, 2001). Social net- ductive searches than job seekers who are unable to—they may
work connections provide job seekers with beneficial resources, spend less time, energy, and resources searching and becoming
or social capital, making “networked” job seekers more competi- employed than job seekers whose networks are unable or unwilling
tive job applicants than job seekers who hear about jobs through to help with their searches. In short, contacts control access to ben-
formal methods (e.g., newspaper advertisements, public or private eficial social capital and have the “power” to help job seekers have
employment agencies, job advertisements on websites, etc.) or who a successful search, making them an important topic of empirical
directly apply for jobs (Lin, 1999; Portes, 1998; but see Mouw, investigation (Trimble et al., 2013).
2003). Scholars know little about the part contacts play in the job
What is less clear is how job seekers activate resources for their search because of data collection and analysis methods which over-
job searches. Before job seekers become employed with the help of look or simplify their role. Most research on social networks and
their social network contacts, they first “access” social capital and the job search focuses on workers either after they have been
“mobilize” it for their searches (Lin, 1999, 2001). In other words, job referred by contacts to jobs (for example, see Fernandez et al.,
seekers must seek out contacts who possess the kinds of resources 2000; Fernandez and Fernandez-Mateo, 2006; Fernandez and Sosa,
they need for their job search and convince them to share those 2005; Petersen et al., 2000), or after they found their jobs with
resources.1 Relatively little is known about how job seekers do this, the help of social network contacts (for example, see Drentea,
1998; Elliott, 1999, 2001; Green et al., 1999; Hanson and Pratt,
1991; Kmec and Trimble, 2009; Leicht and Marx, 1997; Mencken
∗ Corresponding author. Tel.: +1 (850) 437 3873. and Winfield, 2000; Smith, 2000; Stainback, 2008; Straits, 1998).
E-mail address: lindsey.oconnor@csuci.edu These approaches pose two significant and inter-related problems
1
Many job seekers are informally recruited into new jobs by social network con- for understanding contacts’ role in the job search. First, by focus-
tacts (for example see, McDonald, 2011; McDonald et al., 2009). Referred to as ing on job seekers who have already received assistance from
“non-searching,” this process occurs when people find work through the receipt
social network contacts, researchers know little about the rate at
of unsolicited job-related help from contacts. In this paper, I focus solely on “active
searching”—situations in which job seekers solicit help from their social network which contacts are asked to help job seekers and the factors that
contacts. affect whether contacts actually help (Trimble et al., 2013). This is

0378-8733/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.socnet.2013.07.005
594 L. Trimble O’Connor / Social Networks 35 (2013) 593–603

exacerbated by researchers’ tendency to study job seekers rather most contacts help with job searches when asked. Still, I find that
than contacts directly (but see Marin, 2007, 2012, 2013; Smith, male contacts are better able to help than female, employed con-
2005, 2007, 2010) because job seekers may not be aware of the tacts are better able to help than unemployed, and contacts with
reasons contacts do or do not help them. Second, by focusing on more education than job seekers are better able to help than con-
“successful” job seekers—those who find jobs with the help of tacts with less. I also find that contacts are better able to help when
contacts—scholars have little understanding of the job seekers who they are familiar with job seekers’ work qualifications than when
are unable to access and mobilize social capital and the extent to they are unfamiliar, and more willing to help when they perceive
which an inability to access and mobilize social capital is a prob- job seekers to be “good” workers (i.e., having a good work ethic and
lem that some groups (i.e., racial/ethnic minorities, women, or being trustworthy) than when contacts perceive job seekers to be
the unemployed) may confront more than others (Trimble et al., “bad” (having a poor work ethic and being untrustworthy). I con-
2013). clude with a discussion of the implications of my findings for the
In this paper, I take a unique approach to studying social capital social capital, job search, and labor market inequality literatures
access and mobilization for a job search. Rather than examine social and provide suggestions for future research.
network contact use from the perspective of job seekers, I study
social network contacts directly to understand the factors that
1.1. Social capital access and mobilization
affect their provision of social capital for others’ job searches. This
paper builds on the growing body of literature on social network
About half of all jobs are found with the help of a social network
contacts’ provision of help (Marin, 2007, 2012, 2013; Newman,
connection (Granovetter, 1995; Marsden and Gorman, 2001), mak-
1999; Royster, 2003; Smith, 2005, 2007, 2010) in three important
ing contact use one of the most common job search strategies. Social
ways. First, most studies of contacts’ provision of help with a job
network contacts are thought to be beneficial for the job search
search use qualitative methods to examine social capital activation
because they share useful information about job opportunities and
among low-income, urban, or minority contacts (Newman, 1999;
employers, use their work-related influence to sway hiring deci-
Royster, 2003; Smith, 2005, 2007, 2010) or entry-level white-collar
sions, and use their own reputations to “signal” to employers that
workers (Marin, 2007, 2012, 2013). These job seekers are looking
job seekers will be good hires (Bian, 1997; Fernandez and Weinberg,
for work in unique labor markets and their contacts may use a
1997; Lin, 1999, 2001; Marsden and Gorman, 2001; Portes, 1998;
different set of criteria than the general population when decid-
Seibert et al., 2001; Yakubovich, 2005).
ing if and how to help. In this paper, I draw on quantitative data
Yet, in order for network use to result in employment, social cap-
from a large, representative sample of contacts living in Washing-
ital has to first be “accessed and/or mobilized in purposive action”
ton State. This research is the first to my knowledge to identify
by job seekers (Lin, 2001, p. 29). Job seekers gain access to social cap-
the extent to which a representative sample of contacts help job
ital by being involved in particular networks, so that the resources
seekers with their searches and the conditions in which they
embedded in those networks are that individual’s accessed capital
do so.
(Lin, 1999, 2001). Social capital is mobilized when job seekers make
Second, previous research on the factors affecting contact help
use of contacts’ social resources for instrumental action.
during a job search focused on their provision of unsolicited help
Job seekers’ access and mobilization of social capital is depend-
to “non-searchers” (Marin, 2007, 2012, 2013) or do not discern
ent on the extent and diversity of contacts’ resources and their
between the provision of solicited and unsolicited help (Newman,
provision of those resources. Some contacts may not have the
1999; Smith, 2005, 2007, 2010). In contrast, this paper focuses
resources job seekers need for their searches because their pos-
exclusively on active job seekers—those who directly seek out
itions within social structures and institutions limit their access to
the assistance of their social network contacts and who consti-
social capital (Lin, 2001, p.100). Other contacts may have access to
tute about 40 percent of workers who find their jobs with the
resources but decide not to share them (e.g., a contact may decline
help of network contacts (Granovetter, 1995). Contacts have dis-
to help a job seeker because he does not feel she would be a good
tinct roles in the active and non-search processes (Elliott, 2000)
“fit” for a job) (Lin, 1999, 2001). In other words, being embedded
and job seekers are likely to face unique challenges in activating
in social networks with rich social capital does not equate to being
social capital in each situation. In an active search, for example, job
able to use that capital for a job search because contacts actively
seekers approach contacts with requests for help and so they may
decide with whom and under what conditions they will share their
encounter contacts who are unable or unwilling to help them. In
resources (Lin, 1999, 2001; Marin, 2007, 2012; Smith, 2005, 2007,
the non-search process, contacts initiate the sharing of resources
2010). For job seekers to access and mobilize social capital for a
and are unlikely to reach out to a job seeker unless they are able and
job search, they must seek out the help of contacts who are both
willing to help. Contacts’ roles in each process warrant an explo-
able and willing to share their social resources. In the following sec-
ration of the ways in which the factors that affect provision of help
tion, I describe three sets of factors that affect contacts’ ability and
might differ in each situation.
willingness to help job seekers with their searches.
Finally, most studies of social network contacts focus on social
capital mobilization—that is, whether contacts choose to share
their resources with job seekers (Marin, 2007, 2012, 2013; Smith, 1.2. Ability to help with a job search
2005, 2007, 2010). However, job seekers’ success with a job search
depends on accessing and mobilizing contacts’ social capital (Lin, 1.2.1. Contact status
1999, 2001), and so I distinguish between the factors that affect Contacts’ ability to help with a job search depends, in part, on
each outcome. This distinction is particularly important for under- their status. Status is directly linked to access to social resources.
standing how social network contact use perpetuates labor market Social resource theory (SRT) suggests that contacts who are high
inequality because some groups of job seekers (e.g., minorities or in status have larger and more diverse social networks than con-
women) may face distinct barriers in accessing and mobilizing tacts who are low in status (Campbell et al., 1986; Marsden, 1987;
social capital for their searches. Marsden and Hurlbert, 1988; McPherson et al., 2001, 2006; Lin,
To study contacts’ provision of help, I use unique data I col- 1999, 2000), meaning that high status contacts not only have more
lected from a random sample of approximately 350 contacts living social resources which they can share with job seekers but a greater
in Washington State about the last time someone sought their help variety of them compared to low status contacts (Lin, 1999, 2000,
with a job search, hereafter the “job seeker.” My analyses show that 2001; Marsden and Hurlbert, 1988). SRT implies that high status
L. Trimble O’Connor / Social Networks 35 (2013) 593–603 595

contacts are in a better position to know how to help job seekers Second, contacts may hesitate to help weakly tied job seek-
than low status contacts (Lin, 1999; Lin et al., 1981; Marin, 2013). ers because they do not want to provide unwanted or ill-matched
Indeed, research on job seekers’ use of social networks sug- resources. Marin’s (2007, 2012) research on the factors that affect
gests that high status contacts may be better able to help job whether contacts share unsolicited job leads showed that contacts
seekers with their searches than low status contacts. For exam- were reluctant to pass along job information for fear of seem-
ple, Wilson (1996) showed that economically disadvantaged, ing intrusive. As a result, contacts tended to share job leads with
urban black men—whose networks consisted of other low-income strong ties—people whose qualifications and interests were well-
blacks—lacked the necessary connections to resourceful contacts known and for whom they knew the job information would be
who could provide information about job opportunities or influ- welcome (Marin, 2007, 2012). Contacts may feel similarly when
ence over hiring decisions. Other empirical work has found that approached by job seekers with whom they share a weak connec-
racial/ethnic minorities and women have fewer high status net- tion, and opt to convey an inability to help. I use four measures
work members compared with white men, and that this deficit of tie strength—contacts’ familiarity with job seekers’ work qua-
helps explain why social network use during the job search is less lifications, contacts’ closeness to job seekers, the length of their
beneficial for minorities and women than white men (Kasinitz and relationship, and history of exchanging favors—to test the following
Rosenberg, 1996; Marsden and Gorman, 2001; McGuire, 2000; Lin, hypothesize:
2001; Rankin and Quane, 2000; Royster, 2003; Tigges et al., 1998,
but see Fernandez and Fernandez-Mateo, 2006; Newman, 1999). H2 . Contacts with strong ties to job seekers are better able to help
Because racial/ethnic minorities and women tend to rely on other than contacts with weak ties.
minorities and women for help with their searches (Braddock and
Contacts who are socially similar to job seekers may also be more
McPartland, 1987; Falcón and Melendez, 2001; Leicht and Marx,
likely to know how to help than contacts who are socially dissim-
1997; McDonald, 2011; Mencken and Winfield, 2000; Stainback,
ilar. This is because socially similar contacts likely have access to
2008; Straits, 1998), this research suggests that minorities and
the particular kinds of resources (i.e., job information and influ-
women are less likely to have the ability to help with a job search
ence) job seekers need for their search by virtue of their shared
than white men.
social structural location. Differences in the labor force participa-
Other indicators of status, including employment, education,
tion rates of particular groups, sex and race/ethnic segregation in
and age, are also likely to affect contacts’ ability to help with a job
the labor market, and the experience and education requirements
search. Employed contacts may be more likely than unemployed
of jobs may limit the extent of resources contacts have access to,
contacts to maintain work-related relationships (e.g., coworkers,
in turn, limiting their ability to help with socially dissimilar job
supervisors) from whom they can learn about job opportunities,
seekers’ searches. In other words, the more socially dissimilar two
and to come across beneficial resources through routine inter-
individuals are, the less likely they are to have the resources needed
actions at work. Contacts with high levels of education typically
to help one another with a job search. In this analysis, I examine the
have larger and more diverse networks (Marsden, 1987; McPherson
degree of social similarity between contacts and job seekers using
et al., 2001, 2006) and have greater levels of social capital than con-
four characteristics: race/ethnicity, gender, age, and education. I
tacts with low levels of education (Lin, 2001), putting them in good
hypothesize:
position to help with others’ job searches. Finally, older individuals
tend to have more social capital than younger individuals, partic- H3 . Contacts who are socially similar to job seekers are better able
ularly social capital from work-related ties (McDonald and Mair, to help than contacts who are socially dissimilar.
2010). For this reason, older individuals may be better able to help
with a job search. I use these five markers of a contact’s status—race,
gender, employment status, education level, and age—to test the 1.3. Willingness to help with a job search
following hypothesis:
1.3.1. Contact-job seeker relationship
H1 . High status contacts are better able to help with job seekers’
Contacts’ relationships with job seekers also affect whether con-
searches than low status contacts.
tacts are willing to help with a job search. Because contacts must
devote their own time and effort to help job seekers (Fernandez
1.2.2. Contact-job seeker relationship and Castilla, 2001; Lin, 2001; Royster, 2003; Smith, 2005, 2007),
Contacts’ ability to provide the resources job seekers need also interpersonal relationships are an important pre-determinant of
depends on the relationships they share with job seekers, specif- contacts’ willingness to help. Contacts are unlikely to exert a
ically the strength of their connections and their degree of social great deal of effort helping job seekers with whom they are
similarity with job seekers. Strong ties might be better able to help only weakly tied (Lin, 2001), or to whom they do not feel obli-
when asked than weak ties for two reasons. First, contacts who gated.
are strongly tied to job seekers will better know how to help than In addition to the time and effort contacts exert on behalf of
weak ties because they are likely to be familiar with job seekers’ job seekers, tie strength affects whether contacts will want to help.
work qualifications. As a result, strong ties will have a better under- Contacts who are familiar with and feel close to job seekers may be
standing of job seekers’ skills and the types of jobs at which they more motivated or obliged to help, or feel more invested in ensuring
would be successful than weak ties (Marin, 2007, 2012). For exam- they find work (Bian, 1997; Granovetter, 1982; Yakubovich, 2005).
ple, imagine a contact is approached by a job seeker who formerly Contacts may also be more willing to assist when job seekers have
worked as a librarian. Knowing this job seeker’s work history as assisted them in the past, or when they expect that job seekers
a librarian might trigger the contact’s memory of a job advertise- will assist them in the future because of feelings of expected or
ment she recently saw for a position at a bookstore. Had the contact real mutual obligation, a characteristic of strong ties (Bian, 1994,
not known about this job seekers’ work experience, she might not 1997; Granovetter, 1985; Smith, 2007). Typically, close ties have a
have thought to share information about this job opening. In short, history of cooperation which fosters trust and further cooperation
knowing something about job seekers is likely to help contacts (Burt, 2001). As such, I hypothesize:
remember specific, relevant resources which they can then share
with the job seeker, and to help them decide whether they possess H4 . Contacts with strong ties to job seekers are more willing to
resources that might be of use to job seekers. help than contacts with weak ties.
596 L. Trimble O’Connor / Social Networks 35 (2013) 593–603

Contacts might also be more willing to help if they are socially economy, perceptions about the possibility of finding work, and
similar to job seekers. Homophily creates feelings of obligation whether respondents had been asked to help with someone else’s
(Fiske, 1998; Reskin, 2000), and so contacts may feel more com- job search in the three years preceding the survey. If respondents
pelled to help socially similar job seekers than dissimilar job seekers were recently asked for help, I asked them to describe the last time
because of the affinity they feel toward them. Moreover, because someone sought their help, elaborating on the type(s) of help they
individuals feel “at ease” with socially similar others (Fiske, 1998; were asked for, whether they in fact provided the help, their rela-
Reskin, 2000), they may feel more comfortable providing their tionship with the person who last asked for help, and that person’s
assistance to job seekers who are socially similar than dissimilar demographic characteristics.3 I also asked respondents to provide
(Marin, 2012). Empirical evidence suggests that one type of social their demographic information, both at the time of survey comple-
similarity is an important determinant for whether contacts “put tion and, in the case of non-constant attributes (e.g., age), when they
in a good word for job seekers”—contacts are more likely to vouch were last asked for help. I refer to the data I collected from these
for job seekers if they are the same gender than if they are different questionnaires as the Social Networks and Employment Assistance
(McDonald, 2011). I expect: Survey (SNEAS).
To select a sample, I contracted a private vendor who drew
H5 . Contacts who are socially similar to job seekers are more
a list of addresses from the United States Postal Service (USPS)
willing to help than contacts who are socially dissimilar.
delivery sequence file (DSF), an electronic list of virtually all res-
idential and commercial addresses to which the USPS delivers mail
1.3.2. Job seekers’ status as a worker
(Dillman et al., 2009; Link et al., 2008).4 To ensure a random sam-
Finally, contacts’ assessments of job seekers’ work-related
ple of individuals within households were selected to complete
behaviors also affect their willingness to help (Smith, 2005, 2007).
the questionnaire, I instructed the adult (age 18 years or older)
Contacts’ perceptions of a job seeker affect their judgment of the
in the household with the most recent birthday to complete the
worker, the kinds of jobs that he or she could potentially work,
questionnaire.
whether the job seeker is worth the investment, and subsequently
I administered the SNEAS questionnaire using the tailored
whether the contact is willing to help. For example, Smith (2005,
design method (TDM) (Dillman et al., 2009). I contacted each
2007) found that urban, black contacts were unwilling to help job
household with four separate mailings in order to increase sur-
seekers with their searches because they were afraid of hurting
vey response (Dillman et al., 2009). The first mailing consisted
their own reputations by helping job seekers who exhibited “poor
of a pre-notice letter which informed households that they had
worker” qualities (e.g., a lack of motivation, irresponsibility, or a
been selected to take part in a survey about how people look for
tendency toward delinquency). Smith (2005, 2007, 2010) suggests
work. The second mailing included a cover letter, the question-
that contacts consider their perception of job seekers’ work ethic
naire, a $2.00 cash incentive, and a business reply envelope. One
and trustworthiness when deciding to help.
week after mailing the questionnaire packet, I mailed a postcard
Similarly, contacts might be more willing to help job seek-
to all households selected for the sample. I used the postcard to
ers who are employed when they ask for help because holding
thank households that had already completed the questionnaire
down a job indicates a certain level of trustworthiness (Smith,
and to remind non-responders to do so. Nearly one month after
2005). In other words, contacts know that job seekers were “good
I mailed the first questionnaire packet, I mailed non-responders
enough” workers to get and keep their current jobs. Contacts
a replacement questionnaire packet. The replacement question-
might also be more willing to help white, male job seekers than
naire packet was similar in content to the first; it included a
minority or female job seekers because of stereotypes suggesting
cover letter, a replacement questionnaire, and a business reply
that white men are committed and competent workers (Correll
and Ridgeway, 2006; Ridgeway and Correll, 2004). In sum, these
characteristics—work ethic, trustworthiness, employment status,
race/ethnicity, and gender—can signal to contacts that job seek-
3
ers are serious, committed to work, and deserving of the help. Some requests for help are more memorable than others (e.g., a request for an
introduction to a hiring manager versus a request to review a résumé), and contacts
As a result, contacts’ investments are less risky because they might be more likely to remember situations in which they were able to help or
can be more certain that job seekers will not tarnish their rep- actually provided help to job seekers than when they could not or did not provide
utations and that they will not be wasting their investment on help. To ensure respondents recalled the last person who asked for their help, I
an “undeserving” job seeker (Sherman, 2009; Smith, 2005). I asked respondents whether anyone sought nine different types of common job-
related help from them in the three years preceding the survey. These nine types
hypothesize:
of help included situations in which someone asked respondents to: (1) tell him or
H6 . Contacts who perceive job seekers as “good” workers are more her about available job opportunities, (2) talk with an employer on his or her behalf,
(3) share information about a specific job or employer, (4) give advice about how to
willing to help than contacts who perceive job seekers as “bad” look for or apply for a job, (5) hire him or her, (6) be a reference or provide a formal
workers. recommendation, (7) give advice on how to dress or how to act around an employer,
(8) introduce him or her to someone who could help with their job search, and (9)
2. Data and measures pick up or drop off a job application for him or her. By listing several types of job-
related help, I hoped to “jog” respondents’ memory so that they remembered the last
person who asked for their help with a job search—regardless of how significant the
2.1. Data request for help might have been, whether they were able to or had actually provided
the help, or whether the job seeker received an offer as a result of their request for
To test these hypotheses, I draw on data I collected from help. Results from 15 cognitive interviews I conducted to pre-test the questionnaire
suggested that respondents had no trouble remembering the last person who sought
approximately 350 social network contacts. In 2010, I mailed a
their help.
questionnaire to a stratified, random sample of 1850 Washing- 4
I stratified the sampling frame by urbanicity in order to make meaningful
ton State households.2 In the questionnaire, I asked respondents comparisons between urban and rural areas in Washington State, although these
to share their views on a variety of topics including the state of the comparisons are not the focus of this paper. Half of the addresses were selected
from census blocks considered “urban” in the year 2000 (n = 925), and half were
selected from “rural” blocks (n = 925). The Census considers a block to be “urban” if
it has a population density of over 1000 people per square mile or if it is adjacent to
2
The questionnaire was an 11-page (bi-folded on three sheets of 8.5 × 14 paper), a block with a population density of 1000 people per square mile and has a density
62-item paper questionnaire titled “How do people look for work? An effort to under- of 500 people per square mile. All census blocks that do not fit these criteria are
stand how Washington residents help people find jobs.” considered rural (United States Bureau of the Census, 2000).
L. Trimble O’Connor / Social Networks 35 (2013) 593–603 597

envelope.5 Using this method, I received 577 completed question- from contacts: (1) information about job opportunities/openings,
naires and achieved a response rate of 34.4 percent.6 (2) information about a job or employer, (3) providing an intro-
I compared the gender, age, race/ethnic composition, and duction, or (4) providing a reference or letter of recommendation.
employment status of the SNEAS sample with the Washington State Contacts who knew how to help or were asked to provide a refer-
population to establish data representativeness (see Appendix A). ence are coded “1” if contacts provided all of the help job seekers
The SNEAS sample has significantly more women, older people, sought. I coded contacts who were asked for help, but did not
and whites than the State of Washington. To ensure that my sam- provide it as “0.” Those contacts who were unable to help with
ple approximates the Washington State population, I created a job seekers’ searches are coded as missing for this variable. In other
weight that adjusts for respondents’ gender, age, and urban/rural words, only contacts who knew how to provide the help job seekers
status using the “raking” procedure in Stata.7 The raking procedure, needed have values for this variable.10,11
also called “sample balancing,” is ideal when population totals are
unavailable for the variables one wants to weight by (Battaglia et al.,
2004). For example, to adjust the sample for respondents’ gender, 2.3. Independent variables
age, and urban/rural status, one would need to know the total popu-
lations for each cell (e.g., the total population of rural women age 65 2.3.1. Contact’s status
and older). The raking procedure adjusts the sample one variable The first set of independent variables measure contacts’ sta-
at a time to match the populations’ proportions (Battaglia et al., tus, a proxy for their access to social resources. Models include
2004).8 measures of contacts’ race/ethnicity and gender because ascrip-
tive labor market inequality limits minorities and women’s access
2.2. Dependent variables to resources at work which could affect their ability to share
job-related resources (England, 2010; Reskin, 1993; Smith, 2002;
2.2.1. Ability to help with a job search Tomaskovic-Devey, 1993). Contacts’ race/ethnicity is a dichotomous
The first outcome of interest measures contacts’ ability to help variable coded “1” when contacts were racial/ethnic minorities and
with job seekers’ searches. I created this variable by drawing on “0” when white. Due to the small sample of racial/ethnic minori-
the responses to three questionnaire items. The questionnaire first ties in the SNEAS dataset, I coded respondents as minorities if they
asked contacts whether job seekers requested three types of job- described themselves as American Indian/Alaska Native, Asian,
related help: (1) information about job opportunities/openings, (2) black or African American, Latino/Hispanic, or other (non-white). I
information about a job or employer (e.g., information about a also coded respondents as racial/ethnic minorities if they selected
work organization’s culture/environment), and (3) an introduction two or more race categories (including white in conjunction with
to someone who could help with their job search. After establishing any other category). Contacts’ gender is measured using a dichoto-
whether job seekers sought these three types of help, the question- mous variable where “1” is female and “0” is male. Models also
naire then asked contacts if they knew how to provide the help (e.g., include a measure of contacts’ employment status coded “1” when
did contacts know of any job opportunities they could share with contacts were employed and “0” if they were unemployed, retired,
job seekers?). Contacts were coded “1” if they knew how to provide or a non-employed student, and an ordinal measure of contacts’
at least one type of help that the job seeker needed (e.g., they knew education (1, less than a high school diploma, high school graduate,
of at least one job opportunity that they could share with the job or having a GED; 2, some college; 3, 2-year college degree; 4, 4-year
seeker). I coded contacts who were asked for a particular type of college degree; 5, graduate or professional degree) at the time the
help, but did not know how to provide it as “0.”9 job seeker sought their help with the job search. Finally, contact’s
age in years is an ordinal measure of their age at the time they were
2.2.2. Willingness to help with a search asked for help (1, 18–34 years; 2, 35–44 years; 3, 45–54 years; 4,
The second outcome of interest, contacts’ willingness to help, 55–64 years; and 5, 65 years and older).
measures whether contacts actually provided the help that job
seekers sought. To create this outcome variable, I drew on a series
of questions about whether job seekers sought four types of help
10
Unlike the variable measuring contacts’ ability to help, I code contacts as being
willing to help when contacts provided all the help for which job seekers asked, and
5
Due to budget limitations, 500 of the 1850 households received two mailings as unwilling if they did not provide at least one of the types of help for which job
(the questionnaire mailing including a $1.00 incentive instead of $2.00, and a post- seekers asked. In other words, this variable measures whether contacts show any
card) instead of four. hesitation to help job seekers.
6 11
I calculated this response rate using AAPOR’s Standard Definitions (American Some contacts were asked to help job seekers in more than one way (e.g.,
Association for Public Opinion Research, 2010) response rate formula #6. they were asked to share job information and provide an introduction), and many
7
I opted to not weight the data for respondents’ race/ethnicity because of the contacts were able (n = 51) and willing (n = 12) to provide multiple types of help.
small number of racial/ethnic minority contacts in the SNEAS dataset. With so few Dichotomous outcome variables do not account for this variation and may mask
minorities, and with adjusting for respondents’ sex, age, and urban/rural status, I did the effects of contact status, contact-job seeker relationship strength and similar-
not have enough respondents in each cell to weight for respondents’ race/ethnicity ity, and job seekers’ status on the extent to which contacts are able and willing to
as well. help. In analysis not shown, I explore an alternative way of coding the outcome
8
Recall, that I instructed household members with the most recent birthday to variables to account for the fact that some contacts are asked to help more than
complete the questionnaire. Typically, researchers who use a within-household others, and some are able and willing to provide more help than others. To do this, I
sampling strategy estimate weights that adjust for the probability of selection at created two alternate outcome variables. The first measures the proportion of types
the within-household level (see Link et al., 2008). Link et al. (2008), for example, of help contacts were able to provide relative to the types of help they were asked
calculate these weights using the total number of adults per household. Since I did to provide. The second measures the proportion of types of help contacts were will-
not collect this information in my questionnaire, I do not adjust the SNEAS data in ing to provide relative to the types of help they were able and asked to provide.
this way. Findings from the analyses using the proportion outcome variables are not substan-
9
Contacts were only coded as unable to help if they did not know how to provide tially different from the analyses using the dichotomous outcome variables. With
any of the help that job seekers wanted. For example, if a job seeker asked the the exception of three coefficients—respondents’ gender, employment status, and
respondent for information about a job opportunity and an introduction to someone respondents’ education—becoming marginally statistically significant in the model
who could help with their search and the respondent was able to help in at least estimating contacts’ ability to help, the models are very similar. Because the mod-
one of these ways, I coded him or her as knowing how to help with the job search. els using dichotomous outcome variables are more parsimonious, I present findings
If the job seeker did not know how to provide either of these types of help, then she using these outcomes. Findings from this alternative analysis are available upon
or he is coded as being unable to help. request.
598 L. Trimble O’Connor / Social Networks 35 (2013) 593–603

2.3.2. Contact-job seeker relationship Table 1


Weighted descriptive statistics by contacts’ ability to help, n = 342.
I use two variables to capture the relationship between con-
tacts and job seekers: tie strength and social similarity. I measure Able to help (78%) Unable to help (22%)
tie strength with a series of four questionnaire items. Contacts’ Contact’s status
familiarity and closeness to job seekers are ordinal-level variables Minority 0.25 (—) 0.19 (—)
measuring the extent to which contacts are familiar with job seek- Female 0.44** (—) 0.64 (—)
ers’ work qualifications and the degree to which they feel close to Employed 0.80 (—) 0.68 (—)
Education 3.28 (1.43) 3.59 (1.49)
job seekers. Higher values represent more familiarity and greater
Age 3.29 (1.29) 3.31 (1.28)
levels of closeness. Years known is an ordinal-level measure of
Strength of relationship
the number of years contacts knew job seekers when they sought
Familiar 2.20* (0.80) 1.98 (0.80)
help (1, less than one year; 2, one to two years; 3, more than Close 1.98 (0.82) 2.01 (0.82)
2 years, but less than five; and 4, five or more years). Finally, Years known 3.00 (1.16) 3.02 (1.16)
past favor is a dichotomous variable coded “1” when job seek- Past favor 0.67 (—) 0.58 (—)
ers had done contacts a favor in the past and “0” when they had Social similarity 0.55 (—) 0.56 (—)

not. Control variable


Social similarity is a measure of the degree to which contacts and Urban/rural status 0.83 (—) 0.76 (—)
job seekers are socially similar in terms of race/ethnicity, gender, Source. Social Networks and Employment Assistance Survey. Analyses are weighted
age, and education. Values range from 0–1, with 1 indicating that for contacts’ urban/rural status, gender, and age. Standard deviations are in paren-
contacts and job seekers are similar across all four characteristics theses. Differences between able and unable contacts significant at: †p < .10,*p < .05,
**p < .01, ***p < .001 (two-tailed).
and 0 indicating that they are dissimilar across the four character-
istics (values of .75, .5, and .25 signify that contacts and job seekers
are similar along 3, 2, or 1 characteristics, respectively). Table 2
Weighted descriptive statistics by contacts’ willingness to help, n = 344.

2.3.3. Job seekers’ status as a worker Willing to help (92%) Unwilling to help (8%)
The last set of independent variables measure job seekers’ Strength of relationship
characteristics, in particular characteristics signaling pro-work Familiar 2.27 (0.79) 2.21 (0.79)
behaviors. Good worker is an index measuring contacts’ percep- Close 2.03** (0.80) 1.52 (0.91)
Years known 3.06 (1.12) 3.07 (1.12)
tions of a job seeker’s work ethic and trustworthiness (alpha = .62).
Past favor 0.68 (—) 0.50 (—)
Responses range from 1–3 with higher values indicating that the Social similarity 0.55 (—) 0.52 (—)
job seeker was a better worker in the eyes of the respondent. Job
Job seeker’s status as a worker
seeker’s employment status measures whether a job seeker was
Good worker 2.54*** (0.56) 1.86 (1.11)
working when they asked the contact for help (1, employed and 0, Employed 0.28 (—) 0.25 (—)
unemployed). I also include a measure of job seekers’ race/ethnicity Minority 0.20† (—) 0.32 (—)
(1, minority and 0, white) and gender (1, female and 0, male) to tap Female 0.52 (—) 0.56 (—)
into the effects of race and gender stereotypes about workers on Control variable
contacts’ willingness to help.12 Urban/rural status 0.83 (—) 0.76 (—)

Source. Social Networks and Employment Assistance Survey. Analyses are weighted
2.4. Control variables for contacts’ urban/rural status, gender, and age. Standard deviations are in
parentheses. Differences between willing and unwilling contacts significant at:
†p < .10,*p < .05, **p < .01, ***p < .001 (two-tailed).
Because I oversampled rural Washington State residents, and
because the different economic opportunities in urban and rural
environments may affect contacts’ ability and willingness to help were asked to provide a reference or letter of recommendation
with job searches (Vias, 2012; Vias and Peter, 2006), I control for (n = 344).
contacts’ place of residence at the time of the survey. Urban equals I weight to account for the oversampling of rural residents and
one when contacts were residing in urban areas at the time they for the overrepresentation of women and older respondents in the
completed the survey (0, contacts were residing in rural areas). SNEAS sample. I estimate the models using listwise deletion for
cases that are missing values for the dependent variables and mul-
3. Analytic sample and missing data tiple imputation (MI) for missing data on the independent variables
and control variable. When data are missing at random, MI is pre-
In the first analysis (testing hypotheses 1–3), I restrict the ana- ferred to listwise deletion or other methods of imputation as MI is
lytic sample to contacts who were asked by a job seeker to share least likely to introduce bias into the models, violate missing com-
information about a job opportunity/opening, to share information pletely at random assumptions or produce standard error estimates
about a job or employer or were asked to provide an introduc- that are too low (Allison, 2001).13
tion to someone who could help with their search (n = 342). Of
the 577 original SNEAS respondents, 121 (21 percent) had not 4. Results
been asked for help recently, 109 (19 percent) were not asked
to provide the three types of help I examine in the first analysis, Tables 1 and 2 show the weighted descriptive statistics for the
and five were missing data for the variables I used to construct variables used in the following analyses, by contacts’ ability and
the analytic weights. The second analysis (testing hypotheses 4–6) willingness to help with a job search.
includes respondents who knew how to provide at least one of The majority of SNEAS contacts—78 percent—knew how to
the three types of help I investigate in the first analysis, or who provide the kinds of help job seekers needed (see Table 1). Means

12 13
I instructed contacts to answer these questions while thinking about the time In analyses not shown, I estimate the analyses using list-wise deletion. Findings
when the job seeker asked for help rather than when the contacts were completing are not substantially different from the results using multiple imputation (results
the survey. available upon request).
L. Trimble O’Connor / Social Networks 35 (2013) 593–603 599

Table 3 Table 4
Logistic regression estimates for contacts’ ability to help with a job search (n = 342). Logistic regression estimates for contacts’ willingness to help with a job search
(n = 344).
ß expß
ß expß
Contact’s status
Minority 0.21 (.45) 1.23 Strength of relationship
Female −0.92* (.34) 0.40 Familiar −0.26 (.33) 0.77
Employed 0.76* (.38) 2.14 Close 0.72 (.48) 2.06
Education −0.25* (.12) 0.78 Years known −0.38 (.27) 0.68
Age 0.03 (.12) 1.03 Past favor −0.11 (.60) 0.90
Social similarity −0.45 (.95) 0.64
Strength of relationship
Familiar 0.56* (.26) 1.74 Job seekers’ status as a worker
Close −0.15 (.29) 0.86 Good worker 1.66*** (.48) 5.24
Years known −0.18 (.17) 0.84 Employed −0.32 (.80) 0.72
Past favor 0.36 (.38) 1.43 Minority −0.91 (.70) 0.40
Social similarity −0.39 (.68) 0.68 Female −0.22 (.61) 0.80
Constant 1.09 Constant −0.64

Source. Social Networks and Employment Assistance Survey. Standard errors in Source. Social Networks and Employment Assistance Survey. Standard errors in
parentheses. Analyses are weighted for contacts’ urban/rural status, gender, and parentheses. Analyses are weighted for contacts’ urban/rural status, gender, and
age. Model controls for contacts’ residence in an urban or rural census block. age. Model controls for contacts’ residence in an urban or rural Census block.
†p < .10,*p < .05, **p < .01, ***p < .001; two-tailed test. †p < .10,*p < .05, **p < .01, ***p < .001; two-tailed test.

comparisons tests revealed that able contacts, on average, are more dissimilar contacts are just as likely to help job seekers as socially
likely to be male than female and more likely to be familiar with similar contacts.
job seekers work qualifications than unable contacts. Next, I examine the factors that affect social capital mobilization.
Almost all contacts—92 percent—were willing to provide the Table 4 shows the results of a logistic regression model predicting
help job seekers needed for their search (see Table 2). contacts’ willingness to help with a job search.14 This model tests
Willing contacts differ from unwilling contacts in three impor- hypothesis 4–6.
tant ways. First, willing contacts are more likely, on average, to Although means comparisons tests showed that willing con-
report feeling close to job seekers than unwilling contacts. Will- tacts, on average, feel closer to job seekers than unwilling ones,
ing contacts also think job seekers are better workers, on average, after controlling for job seekers’ characteristics and contacts’ place
than contacts who are unwilling to help. Finally, willing con- of residence, I find no support for hypothesis 4 in the multivari-
tacts are less likely to be racial/ethnic minorities than unwilling ate analysis (Contacts with strong ties to job seekers are more willing
contacts, although this difference is only marginally statistically to help than contacts with weak ties). I also find no support for
significant. hypothesis 5 (Contacts who are socially similar to job seekers are more
Table 3 presents findings from a model regressing contacts’ abil- willing to help with searches than contacts who are socially dissimi-
ity to help with a job search on contacts’ status, tie strength, and lar). Socially similar contacts are no more willing to help job seekers
social similarity. This model tests hypotheses 1–3. than socially dissimilar ones.
Three variables measuring contact status significantly predict Finally, one measure of job seekers’ status significantly predicts
whether contacts are able to help: contacts’ gender, employment contacts’ willingness to help: whether contacts perceive job seekers
status, and level of education. The odds that women know how to to be good workers. A one unit increase in job seekers’ perceptions
help with a job search are 60 percent (ß = −0.92; expß = 0.40) the of contacts’ work ethic and trustworthiness yields a 424% increase
odds of men, net of all variables in the model. Respondents who in the predicted odds that contacts are willing to help, net of con-
were employed at the time they were asked for help are more trol variables (ß = 1.66; expß = 5.24). Thus, I find strong evidence
likely to know how to help job seekers than respondents who that contacts’ perceptions of job seekers as workers matter for their
were unemployed. The predicted net odds that employed respon- willingness to help with a job search. No other measures of job seek-
dents know how to help are approximately 2 times the odds for ers’ status affects contacts’ willingness to help, and so I find limited
unemployed respondents (ß = 0.76; expß = 2.14). Surprisingly, abil- support for hypothesis 6 (Contacts who perceive job seekers as good
ity to help with a job search decreases with respondents’ level workers are more willing to help with job searches than contacts who
of education. A one-unit increase in level of education results in perceive job seekers as bad workers).
a 22 percent decrease in ability to help (ß = −0.25; expß = 0.78).
Respondents’ race/ethnicity and age are not statistically signif-
icant predictors of being able to help with a job search. Thus, 5. Discussion
I find mixed support for hypothesis 1 (High status contacts are
better able to help with job seekers’ searches than low status con- Using a unique approach to studying social capital and the job
tacts). search—one that focuses on social network contacts rather than
I find moderate support for hypothesis 2 (Contacts with strong job seekers—this paper sought to identify the extent to which con-
ties to job seekers are better able to help than contacts with weak ties). tacts help with other’s job searches when asked, and the factors
Contacts’ familiarity with job seekers’ work qualifications predicts that affect their provision of help. Before job seekers can become
contacts’ ability to help. A one-unit increase in the strength of the employed with the help of contacts, they must first gain access
relationship between contacts and job seekers is associated with a to and make use of their contacts’ social capital. In other words,
74 percent (ß = 0.56; expß = 1.74) increase in the predicted odds of job seekers must seek out the help of contacts who possess the
knowing how to help, net of all other variables in the model. None resources they need for their search and then make use of those
of the other measures of tie strength significantly predict ability to
help with a job search.
Finally, I find no evidence to support hypothesis 3 (Contacts 14
Because so few contacts were unwilling to help with job seekers’ searches (8
who are socially similar to job seekers are better able to help with percent), these findings should be considered exploratory and future researchers
job searches than contacts who are socially dissimilar). Socially should reexamine this issue using a larger sample of contacts.
600 L. Trimble O’Connor / Social Networks 35 (2013) 593–603

resources to become employed. Studying the conditions in which of job leads they possess (Marin, 2007, 2012, 2013). In addi-
job seekers are able to do this provides key insights into why con- tion to explaining the advantages employed contacts have in
tact use can be a productive job search strategy for some, but not providing help, examining other indicators of social resources
all job seekers. (e.g., access to “insider information” or influence) could help
Analyses show that most contacts—78 percent—are able to help researchers pinpoint the kinds of resources that are most bene-
with a job search when asked. That is, more often than not, contacts ficial for job seekers, the contacts who are most likely to possess
typically possess the kinds of resources job seekers need for their them, and the kinds of job seekers who know resourceful con-
searches. Still, I find that some factors increase (or decrease) con- tacts.
tacts’ ability to help. Consistent with SRT (Lin, 1999, 2000, 2001), My analysis revealed that contacts’ ability to help with others’
I find that male and employed contacts are more likely to know job searches declines with their level of education. This finding is
how to help with job searches than female and unemployed con- surprising because, according to SRT (Lin, 1999, 2000, 2001), indi-
tacts. One likely reason that men are better able to help than viduals’ network size not only increases with education, but also
women is that discrimination and occupational sex segregation diversifies (Marsden, 1987; McPherson et al., 2001, 2006), sug-
limits women’s access to jobs with authority, power and pres- gesting that highly educated contacts should be particularly well
tige (England, 2010; Reskin, 1993; Tomaskovic-Devey, 1993) which positioned to help with others’ job searches. One possible expla-
restricts their ability to help with a job search. Indeed, women are nation for this unexpected finding is that contacts’ knowledge
less likely than men to have access to high status connections at of job-related resources narrows when contacts reach a certain
work because they are concentrated in lower ranking occupations level of specialization within a particular field (e.g., when they
than men (McGuire, 2000). As a result, women may be less likely possess an advanced professional or graduate degree). A study
than men to possess job-related resources and less likely to be able which not only examines how knowledge of job-related resources
to act as an intermediary for a job seeker’s search. More refined varies by level of education, but also by field of specialty would
measures of contacts’ employment situations (e.g., occupation or help uncover whether contacts who are highly educated in some
job duties, working in a job with hiring authority, support networks fields (i.e., MBA) learn of more diverse job resources than con-
at work, or occupational sex composition) would help to clarify the tacts who are highly educated in other fields (i.e., veterinary
mechanisms at play here. medicine).
An alternate possibility is that women have access to simi- An alternative hypothesis is that contacts are better able to help
lar resources as men, but that women “devalue” the resources with a job search when they have higher levels of education than job
they possess. Status expectations theory suggests that women are seekers. In other words, contacts’ level of education is less impor-
expected to be less competent and committed at work than men tant for predicting their ability to help than is their relative level
(Correll and Ridgeway, 2006; Ridgeway and Correll, 2004), and as a of education. To test this possibility, I examined the effects of a
result are judged and stereotyped more harshly than men (Heilman, series of dichotomous variables measuring contacts’ relative edu-
2001). Evidence suggests that women, like men, “buy into” these cation on their ability to help. This supplemental analysis revealed
stereotypes (Hirsh and Cha, 2008) which could affect their sense of that relative education matters. Specifically, the odds that higher
the value of their resources for job seekers. For example, women educated contacts are able to help are about 3.5 times the odds of
may not view their job-related information as something that job lower educated contacts. Contacts with the same level of education
seekers would really want or that would be helpful, so they err on as job seekers are just as likely to be able to help as contacts with
the side of caution and say they do not know how to help when less education.15 Future research should continue to tease out the
asked. Men, on the other hand, who enjoy a privileged status in effects of contact and job seekers’ education on contacts’ ability to
the workplace (Ridgeway and Correll, 2004), may feel extra confi- help with a job search.
dent in their ability to help or feel like they are “expected” to know My analyses also show that the strength of the relationship
how to help with work-related matters. A qualitative study that between contacts and job seekers matters for contacts’ ability to
fleshes out the decisions women and men make about whether help. I find that contacts who are strongly tied to job seekers are
they are able to help and their impressions of the kinds of help that better able to help than weakly tied contacts. This finding supports
would benefit particular job seekers would help shed light on this previous research on contact-job seeker tie strength and contacts’
finding. provision of help with job searches which showed that contacts
No matter the mechanisms at play, this research suggests hesitate to help weak ties with their job searches because contacts
that job seekers who network with female contacts will have were uncertain whether the help was wanted (Marin, 2007, 2012).
less productive job searches than job seekers who network with This finding has implications for the “strength of weak ties” argu-
male contacts. Because job seekers tend to find jobs with the ment, which suggests that weakly-tied contacts are more beneficial
help of same gender contacts (Leicht and Marx, 1997; Mencken for a job search than strongly tied because they provide access to
and Winfield, 2000; Straits, 1998), women may be particularly new, non-redundant job information and resources (Burt, 1992;
disadvantaged by female contacts’ inability to help with job Granovetter, 1995). While weak ties may provide a job seeker with
searches. novel resources, this study and Marin’s (2007, 2012) work suggest
I also find that employed contacts are more likely to know that job seekers are less likely to access weak ties’ social capital
how to help with job seekers’ searches than unemployed con- than strong ties’.
tacts. By virtue of being connected to a workplace, employed Another possible explanation is that job seekers are better able
contacts likely have better access to job-related resources than to gauge the resources that strong ties possess compared with weak
unemployed contacts (e.g., work-related networks who key them ties and only target those likely to have the resources job seekers
into upcoming job openings or to whom they can introduce job need. In other words, because job seekers know more about strong
seekers when they need help). These job-related resources put ties than weak, they are less likely to ask a strong tie than a weak
employed contacts in the position to know how to help when tie for help he or she is unable to provide. Until this possibility is
asked. Future research examining the types of useful resources examined, job seekers should take extra care to make sure weakly
that employed and unemployed contacts possess would help
test this possibility. For example, future studies might ask study
participants to list job openings they know of, and then com-
pare how employed and unemployed contacts differ in the extent 15
Results from this supplemental analysis are available upon request.
L. Trimble O’Connor / Social Networks 35 (2013) 593–603 601

tied contacts are familiar with their qualifications so that contacts These factors did not significantly predict contacts willingness to
know how to help with their job searches. help in my multivariate analysis, however future research should
My analyses show that few contacts are unwilling to help to reexamine their effects using a larger sample of willing and unwill-
others’ job searches. Only eight percent of contacts showed some ing contacts.
hesitation in helping job seekers—that is, they were unwilling to Finally, future research should consider alternative ways of col-
provide at least one of the types of help for which job seekers lecting data from contacts on their ability and willingness to help
asked. That so few contacts were unwilling to help is surprising, others with their job searches. It is possible that my analysis over-
especially in light of Sandra Smith’s (2005, 2007, 2010) qualita- estimates contacts’ ability and willingness if contacts remembered
tive research on social capital mobilization. Smith (2005, 2007) the last person they were able to help with a job search, or willing
found that most (over 80 percent) of the low-income black workers to help, rather than the last person who asked for help. In other
she studied showed hesitation in helping with others’ job searches words, contacts may be forgetting instances in which they were
which stemmed from contacts’ fear of tarnishing their own work- unable or unwilling to help. There is no way of knowing whether
place reputations by making “bad” referrals. this affects my findings in significant ways; however I did take sev-
Why are my findings so different from Smith’s (2005, 2007)? eral precautions when designing the survey to ensure respondents
One possible explanation is that I rely on a different sample of recalled the last person who asked for help with a job search. For
respondents for my analysis. Perhaps, high levels of distrust and example, in an effort to “jog” respondents’ memory about the last
unwillingness to help others with job searches is limited to low- time they were asked to help, I instructed them to read through
income, urban black contacts. Employers perceive low-income a list of nine common types of help that job seekers might have
black workers as unreliable which results in tenuous employment asked them for before identifying the last person who asked for
situations (Kirschenman and Neckerman, 1991; Moss and Tilly, help. I also conducted 15 cognitive interviews to pretest the SNEAS
2001; Wilson, 1996), and workers with extremely precarious labor questionnaire. During the interviews, I probed respondents about
market positions are understandably more likely to be guarded whether they were in fact remembering the last person who asked
about making “bad” referrals than workers with stable positions for their help with a job search. Findings from these interviews sug-
(Smith, 2005, 2007). In addition, Smith (2005, 2007) found that gest that contacts do typically remember the last person who asked
hesitancy to help with a job search was more prevalent in neigh- them for help, usually because the request happened recently (e.g.,
borhoods with concentrated disadvantage (i.e., extreme poverty, within the last month in many cases). Still, future research should
high rates of welfare dependence, high crime rates) because these consider this a possibility and think creatively about ways to study
environmental conditions foster pervasive distrust and impede col- social capital activation and mobilization from the perspective of
lective efficacy. The SNEAS sample is representative of the State contacts.
of Washington, which has few low-income urban blacks living in
areas of concentrated disadvantage relative to other states. For this
reason, my sample may have more willing contacts than studies of 6. Conclusion
low-income, urban, black populations.
Another possible explanation for the difference in findings has to The saying, “ask and you shall receive,” applies to many job
do with the way in which I measure contacts’ willingness to help. seekers who turn to their social network contacts for help with
I considered contacts as being willing to help job seekers if they a job search. Most job seekers have contacts who help with their
provided the help for which job seekers asked. Yet, it could be prob- searches, however those who rely on female, unemployed, less
lematic to conflate providing help with willingness to help. With educated, or unfamiliar contacts, or whose contacts think they are
the measure of contact willingness I use, I am unable to determine “poor” workers do not get help. This research raises many ques-
whether contacts were at all hesitant to help job seekers. More- tions for network and labor market scholars to explore through
over, since I did not ask contacts to rate the “quality” of help they future research. How do contacts make decisions about the types
provided, I am unable to tell whether contacts hesitated to help of resources they share? Do gender and race/ethnic biases affect
some job seekers but offered them “low quality” help as a conso- contacts’ assessment of the types of job-related resources they
lation or a way to “save face.” Future research that examines the think job seekers want? Do contacts provide “good” quality help to
quality of help contacts provide to job seekers could help explain some job seekers, but not to others? How do contacts decide who
the discrepancy between my findings and Smith’s (2005, 2007, they will “go the extra mile” for? Future research that attempts to
2010). address these sorts of questions will shed light on the mechanisms
Despite a high percentage of contacts being willing to help, driving social network contact use during the job search, contacts’
one factor emerged as an important predictor of contacts’ will- critical role in the job search, and the ways in which networks
ingness: their perceptions of job seekers’ work quality. Analyses continue to perpetuate labor market inequality.
show that contacts were more willing to help job seekers
whom they perceived to be “good” workers than job seek- Acknowledgments
ers whom they perceived to be “bad.” This finding supports
Smith’s (2005, 2007) research demonstrating that contacts con- I would like to thank Julie Kmec, Don Dillman, Amy Wharton,
sider job seekers’ reputations before agreeing to help with their the anonymous reviewers and editors of Social Networks for pro-
job searches. Although I find that contacts are more willing to viding insightful comments on this paper. An earlier version of this
help with others’ job searches than Smith’s (2005, 2007, 2010) article was presented at the 2011 Annual Meeting of the American
work shows, it appears that my respondents used the same Sociological Association, Las Vegas, Nevada. This research was sup-
criteria as Smith’s (2005, 2007, 2010) when deciding who to ported in part by NSF Doctoral Dissertation Improvement Grant,
help. SES 1003692, and by the Washington State University Department
Descriptive analysis showed that willing and unwilling contacts of Sociology Dissertation Grant. I would also like to thank the
differ in other interesting ways. Means comparisons tests revealed Michelle R. Clayman Institute for Gender Research at Stanford Uni-
that willing contacts, on average, feel closer to job seekers and are versity for the financial support they provided while writing this
less likely to be racial/ethnic minorities than unwilling contacts. article.
602 L. Trimble O’Connor / Social Networks 35 (2013) 593–603

Appendix A. Dillman, D.A., Smyth, J.D., Christian, L.M., 2009. Internet, Mail and Mixed-Mode
Surveys: The Tailored Design Method. John Wiley & Sons Inc., Hoboken, New
Jersey.
Comparison of social network and employment assistance sur- Drentea, P., 1998. Consequences of women’s formal and informal job search methods
vey respondents with Washington State population (adults ages 18 for employment in female-dominated jobs. Gender Soc. 12, 321–338.
and over)a Elliott, J.R., 1999. Social isolation and labor market insulation: network and
neighborhood effects on less-educated urban workers. Soc. Quart. 40,
Washington Total SNEAS 199–216.
State sample Elliott, J.R., 2000. Class, race and job matching in contemporary urban labor markets.
Soc. Sci. Quart. 81, 1036–1052.
Gender
Elliott, J.R., 2001. Referral hiring and ethnically homogeneous jobs: how prevalent
Male .50 .44†
is the connection and for whom? Soc. Sci. Res. 30, 401–425.
Female .50 .56*
England, P., 2010. The gender revolution: uneven and stalled. Gender Soc. 24,
Age 149–166.
18–34 years .31 .22* Falcón, L.M., Melendez, E., 2001. The social context of job searching for racial groups
in urban centers. In: ’Connor, A., Tilly, C., Bobo, L. (Eds.), Urban Inequality: Evi-
35–44 years .18 .13
dence from Four Cities. Russell Sage, New York, pp. 341–371.
45–54 years .19 .27*
Fernandez, R.M., Castilla, E.J., 2001. How much is that network worth? social capital
55–64 years .16 .21
in employee referral networks. In: Lin, N., Cook, K.S., Burt, R.S. (Eds.), Social
65 years and older .16 .18 Capital: Theory and Research. Transaction Publishers, New Brunswick, NJ, pp.
Race/ethnicity 85–104.
Fernandez, R.M., Fernandez-Mateo, I., 2006. Networks, race, and hiring. Am. Sociol.
White alone .74b .84***
Rev. 71, 42–71.
Minority .27 .16*
Fernandez, R.M., Sosa, M.L., 2005. Gendering the job: networks and recruitment at
Black/African American alone .03 .02
a call center. Am. J. Sociol. 111, 859–904.
Asian only .07c .07 Fernandez, R.M., Weinberg, N., 1997. Sifting and sorting: personal contacts and
American Indian/Alaska Native alone .01 .01 hiring in a retail bank. Am. Sociol. Rev. 62, 883–902.
Latino/Hispanic alone .10 .03 Fernandez, R.M., Castilla, E.J., Moore, P., 2000. Social capital at work: networks and
Other alone .02 – employment at a phone center. Am. J. Sociol. 105, 1288–1356.
Two or more races .04 .03 Fiske, S.T., 1998. Stereotyping, prejudice, and discrimination. In: Gilbert, D.T., Fiske,
S.T., Lindzey, G. (Eds.), The Handbook of Social Psycholog, 4th ed. McGraw Hill,
Employment status New York, pp. 357–411.
Employed .62d .65 Granovetter, M., 1982. The strength of weak ties: a network theory revisited. In:
Unemployed, but looking for a job .06d .09 Marsden, P.V., Lin, N. (Eds.), Social Structure and Network Analysis. Sage, Beverly
Unemployed, not looking for a job .32d .26 Hills, CA, pp. 105–130.
n 5,093,008 577 Granovetter, M., 1985. Economic action and social structure the problem of embed-
Source. Social networks and employment assistance survey (SNEAS) and 2009 dedness. Am. J. Sociol. 91, 481–510.
American Community Survey (2009) analyses are weighted using a design weight Granovetter, M., 1995. Getting a Job: A Study of Contacts and Careers, 2nd ed. Uni-
versity of Chicago Press, Chicago.
to account for the overrepresentation of rural Washington State residents in the
Green, G.P., Tigges, L.M., Diaz, D., 1999. Racial and ethnic differences in job-search
sample.
strategies in Atlanta, Boston, and Los Angeles. Soc. Sci. Quart. 80, 263–278.
Differences between the SNEAS sample and WA state population significant at: Hanson, S., Pratt, G., 1991. Job search and the occupational segregation of women.
†p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests). Ann. Assoc. Am. Geog. 81, 229–253.
a
Author calculations using data from the 2009 American Community Survey Heilman, M.E., 2001. Description and prescription: how gender stereotypes prevent
(2009). women’s ascent up the organizational ladder. J. Soc. Issues 57, 657–674.
b Hirsh, C.E., Cha, Y., 2008. Understanding employment discrimination: a multilevel
The ACS does not provide information on the share of non-Latino/Hispanic racial
minorities over the age of 18. Therefore, I calculated the racial/ethnic composi- approach. Sociol. Compass, 1989–2007.
tion for the entire WA State population (n = 6,664,195) instead of the racial/ethnic Ioannides, Y.M., Loury, L.D., 2004. Job information networks, neighborhood effects,
composition of those 18 years and older. and inequality. J. Econ. Lit. 42, 1056–1093.
c Kasinitz, P., Rosenberg, J., 1996. Missing the connection: social isolation and employ-
Category includes Native Hawaiian and other Pacific Islander.
d
ment on the Brooklyn Waterfront. Soc. Probl. 43, 180–195.
Civilian population, 20 years or older. Kirschenman, J., Neckerman, K., 1991. We’d love to hire them, but. . .: the meaning
of race for employers. In: Jencks, C., Peterson, P.E. (Eds.), The Urban Underclass.
Brookings Institute Press, Washington D.C., pp. 203–232.
References Kmec, Julie A., Lindsey B. Trimble. 2009. “Does it Pay to have a Network Contact?
Social Network Ties, Workplace Racial Context, and Pay Outcomes.” Social Sci-
Allison, S., 2001. Missing Data. Sage Publications, Thousand Oaks, CA. ence Research 38: 266-278.
American Association for Public Opinion Research (AAPOR), 2010. Standard Defini- Leicht, K.T., Marx, J., 1997. The consequences of informal job finding for men and
tions: Final Dispositions of Case Codes and Outcome Rates for Surveys, Available women. Acad. Manage. J. 40, 967–987.
at http://www.aapor.org/Standard Definitions1.htm. Retrieved November 15, Lin, N., 1999. Social networks and status attainment. Ann. Rev. Sociol. 25, 467–487.
2010. Lin, N., 2000. Inequality in social capital. Contemp. Sociol. 29, 785–795.
American Community Survey (ACS), 2009. 2009 American Community Survey 1 Lin, N., 2001. Social Capital: A Theory of Social Structure and Action. University Press,
Year Estimates: Washington: Detailed Tables. Generated by AUTHOR using Cambridge.
American Factfinder, U.S. Census Bureau, U.S. Dept. of Commerce, Available Lin, N., Ensel, W.M., Vaugh, J.C., 1981. Social resources and strength of ties: structural
at http://factfinder2.census.gov/faces/nav/jsf/pages/index.xhtml. Retrieved factors in occupational status attainment. Am. Sociol. Rev. 46, 393–405.
December 3, 2010. Link, M.W., Battaglia, M.P., Frankel, M.R., Osborn, L., Mokdad, A.H., 2008. A com-
Battaglia, M.P., Izrael, D., Hoaglin, D.C., Frankel, M.R., 2004. Tips and Tricks parison of address-based sampling (abs) versus random digit dialing (rdd) for
for Raking Survey Data (a.k.a. Sample Balancing). American Association for general population surveys. Public Opin. Quart. 72, 6–27.
Public Opinion Research, Available at http://www.amstat.org/sections/srms/ Marin, A., 2007. Social Capital as Process: The Network Sources of Latent, Available,
Proceedings/y2004/files/Jsm2004-000074.pdf. Retrieved August 15, 2011. and Accessed Job Information. Department of Sociology, Harvard University,
Bian, Y., 1994. Guanxi and the allocation of urban jobs in China. China Quart. 140, Cambridge, MA (PhD dissertation).
971–999. Marin, A., 2012. Don’t mention it: why people don’t share job information, when
Bian, Y., 1997. Bringing strong ties back in: indirect ties, network bridges, and job they do, and why it matters. Soc. Networks 34, 181–192.
searches in China. Am. Sociol. Rev. 62, 366–385. Marin, A., 2013. Who can tell? network diversity, within-industry networks and
Burt, R., 1992. Structural Holes: The Social Structure of Competition. Harvard Uni- opportunities to share job information. Sociol. Forum (forthcoming).
versity Press, Cambridge, Mass. Marsden, P.V., 1987. Core discussion networks of Americans. Am. Sociol. Rev. 52,
Burt, R., 2001. Bandwidth and echo: trust information, and gossip in social networks. 122–131.
In: James, E.R., Alessandra, C. (Eds.), Networks and Markets. Russell Sage Foun- Marsden, P.V., Gorman, E.H., 2001. Social networks, job changes, and recruitment.
dation, New York, pp. 30–74. In: Berg, I., Kalleberg, A. (Eds.), Sourcebook of Labor Markets: Evolving Structures
Braddock, J.H., McPartland, J.M., 1987. How minorities continue to be excluded from and Processes. Kluwer Academic, New York, pp. 467–502.
equal employment outcomes: research on labor market and institutional barri- Marsden, P.V., Hurlbert, J.S., 1988. Social resources and mobility outcomes: a repli-
ers. J. Soc. Issues 43, 5–39. cation and extension. Soc. Forces 66, 1038–1059.
Campbell, K.E., Marsden, P.V., Hurlbert, J.S., 1986. Social resources and socioeco- McDonald, S., 2011. What’s in the old boys network? accessing social capital in
nomic status. Soc. Networks 8, 97–117. gendered and racialized networks. Soc. Networks 33, 317–330.
Correll, S.J., Ridgeway, C.L., 2006. Expectation states theory. In: Delamator, J. (Ed.), McDonald, S., Mair, C.A., 2010. Social capital across the life course: age and gendered
Handbook of Social Psychology. Springer, New York, pp. 29–51. patterns of network resources. Sociol. Forum 25, 335–359.
L. Trimble O’Connor / Social Networks 35 (2013) 593–603 603

McDonald, S., Lin, N., Ao, D., 2009. Network of opportunity: gender, race, and job Smith, R.A., 2002. Race, gender, and authority in the workplace: theory and research.
leads. Soc. Probl. 56, 385–402. Ann. Rev. Sociol. 28, 509–542.
McGuire, G., 2000. Gender, race, ethnicity, and networks: the factors affecting the Smith, S.S., 2000. Mobilizing social resources: race, ethnic, and gender differences
status of employees’ network members. Work Occup. 27, 501–523. in social capital and persisting wage inequalities. Sociol. Quart. 41, 509–537.
McPherson, M.J., Smith-Lovin, L., Brashears, M., 2006. Social isolation in America: Smith, S.S., 2005. Don’t put my name on it: social capital activation and job-finding
changes in core discussion networks over two decades. Am. Sociol. Rev. 71, assistance among the back urban poor. Am. J. Sociol. 111, 1–57.
353–375. Smith, S.S., 2007. Lone Pursuit: Distrust and Defensive Individualism Among the
McPherson, M.J., Smith-Lovin, L., Cook, J.M., 2001. Birds of a feather: homophily in Black Poor. Russell Sage Foundation, New York.
social networks. Ann. Rev. Sociol. 27, 415–444. Smith, S.S., 2010. A test of sincerity: how black and latino service workers make
Mencken, F.C., Winfield, I., 2000. Job search and sex segregation: does sex of social decisions about making referrals. Ann. Am. Acad. Polit. Soc. Sci. 629, 30–52.
contact matter? Sex Roles 42, 847–864. Stainback, K., 2008. Social contacts and race/ethnic job matching. Soc. Forces 87,
Moss, P., Tilly, C., 2001. Stories Employers Tell: Race, Skill and Hiring in America. 857–886.
Russell Sage Foundation, New York. Straits, B.C., 1998. Occupational sex segregation: the role of personal ties. J. Vocat.
Mouw, T., 2003. Social capital and finding a job: do contacts matter? Am. Sociol. Rev. Behav. 52, 191–207.
68, 868–898. Tigges, L.M., Browne, I., Green, G.P., 1998. Social isolation of the urban poor. Sociol.
Newman, K., 1999. No Shame in My Game: The Working Poor in the Inner City. Knopf Quart. 39, 53–77.
and the Russell Sage Foundation, New York. Tomaskovic-Devey, D., 1993. Gender and Racial Inequality at Work: The Sources and
Petersen, T., Saporta, I., Seidel, M.L., 2000. Offering a Job: meritocracy and social Consequences of Job Segregation. ILR Press, Ithaca, NY.
networks. Am. J. Sociol. 106, 763–816. Trimble Lindsey, B., Julie Kmec, A., Steve McDonald, 2013. Social Networks and the
Portes, A., 1998. Social capital: its origins and applications in modern sociology. Ann. Job Search: Focusing on People Asked to Provide Job Assistance. Nova Science
Rev. Sociol. 24, 1–24. Publishers (forthcoming).
Rankin, B., Quane, J.M., 2000. Neighborhood poverty and the social isolation of inner- United States Bureau of the Census, 2000. Urban and Rural Classification, Avail-
city African–American families. Soc. Forces 79, 139–164. able at http://www.census.gov/geo/www/ua/ua 2k.html. Retrieved December
Reskin, B.F., 1993. Sex segregation in the workplace. Ann. Rev. Sociol. 19, 241–270. 19, 2011.
Reskin, B.F., 2000. The proximate causes of employment discrimination. Contemp. Vias, A.C., 2012. Perspectives on U.S. rural labor markets in the first decade
Sociol. 29, 319–328. of the twenty-first century. In: Kulscár, L.J., Curtis, K.J. (Eds.), International
Ridgeway, C., Correll, S., 2004. Unpacking the gender system: a theoretical perspec- Handbook of Rural Demograph, 3rd ed. Springer, Dordercht, Netherlands,
tive on gender beliefs and social relations. Gender Soc. 18, 510–531. pp. 273–291.
Royster, D.A., 2003. Race and the Invisible Hand: How White Networks Exclude Black Vias, A.C., Peter, N.B., 2006. Changing livelihoods in rural America. In: Kandel, W.A.,
Men from Blue-Collar Jobs. University of California Press, Berkeley, CA. Brown, D.L. (Eds.), Population Change and Rural Society. Springer, Dordercht,
Seibert, Scott, E., Maria, L., Kraimer, Robert, C., Liden, 2001. A social capital theory of Netherlands, pp. 75–102.
careers. Acad. Manag. J. 44, 219–237. Wilson, W.J., 1996. When Work Disappears. Knopf, New York.
Sherman, J., 2009. Those Who Work, Those Who Don’t: Poverty, Morality, and Family Yakubovich, V., 2005. Weak ties, information, and influence: how workers find jobs
in Rural America. University of Minnesota Press, Minneapolis, Minnesota. in a local Russian labor market. Am. Sociol. Rev. 70, 408–421.

Vous aimerez peut-être aussi