Network resources, contact resources, and status attainment

20

Click here to load reader

Transcript of Network resources, contact resources, and status attainment

Page 1: Network resources, contact resources, and status attainment

Ž .Social Networks 20 1998 159–178

Network resources, contact resources, and statusattainment 1

Gina Lai a,), Nan Lin b, Shu-Yin Leung c

a Department of Sociology, Hong Kong Baptist UniÕersity, Kowloon Tong, Hong Kong, Chinab Department of Sociology, Duke UniÕersity, Durham, NC, USA

c Department of Sociology, State UniÕersity of New York at Albany, Albany, NY, USA

Abstract

This paper is a continuation of a research program designed to examine the role of social resources in thestatus attainment process. Unlike previous studies that conceptualized and operationalized social resources aseither network resources or contact resources, it proposes a complementary approach that links the potentialpool of the social resources embedded in a person’s network to the actual mobilization of social resources in a

Ž . Žspecific event. The causal link between access to for example, network resources and use of for example,.contact resources social resources is empirically tested in a series of models in which the effects of personal

resources and social resources on status outcome are assessed. The analysis is based on a representative sampleof employed males in a metropolitan area in upstate New York. Findings support the conceptualization ofsocial resources as having two components. The analysis allows a further re-evaluation of the relative effects,both direct and indirect, of the ascribed versus achieved statuses on status attainment. Results suggest thatascribed factors may exert greater influence on status attainment than previously assumed when mediatedeffects of achieved factors on status outcome via access to and use of social resources are taken intoconsideration. q 1998 Elsevier Science B.V.

1. Introduction

A major theme in recent studies of status attainment focuses on the methods andconsequences of job search. Particularly, the emphasis on job search using informalmethods and subsequent effect on status gains has stimulated an ongoing program of

Žresearch revolving around the notions of social ties and social resources for example,

) Corresponding author. Fax: q86-2339-7893.1 The paper has been presented at the annual meeting of the American Sociological Association in

Washington, DC, August 1990. The authors contributed equally to the paper.

0378-8733r98r$ 19.00 q 1998 Elsevier Science B.V. All rights reserved.Ž .PII S0378-8733 97 00012-9

Page 2: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178160

see the works of Flap and De Graaf, 1986; Granovetter, 1973, 1974, 1982; Lin, 1982,.1983; Marsden et al., 1984, 1988 . Two issues are central to the development of this

research program. First, how is the notion of social resources conceptualized andoperationalized? Second, what is the role of social resources in the process of statusattainment? For example, how do social resources mediate the impacts of ascribed andachieved attributes on the status outcome?

This paper begins with a brief review of the social resources theory and discusses twoapproaches of conceptualization and operationalization of social resources, namely, the‘contact-as-resources’ and the ‘network-as-resources’ perspectives. Then, we will pro-pose how the two approaches can be integrated into a single framework. Finally, we willdemonstrate how this integrated model can further specify the relative roles of ascriptionand achievement in determining the status outcome.

2. The theory of social resources

Resources can be defined as valued goods in a society, the possession of whichŽmaintains and promotes a person’s self-interest for survival and preservation Lin,

. Ž1982 . Examples of societal resources are wealth, status, and power Lin, 1982; Boxman.and Flap, 1990 . Social resources are valued goods embedded in one’s social network

Žand accessible through one’s direct and indirect ties Lin et al., 1981a,b; Lin and Dumin,.1986 . Thus, social resources are not possessed by the individual. They can be

considered second-order resources belonging to the alters in one’s social networkŽ .Boissevain, 1974 . In a specific action, for instance, looking for a job, an ego maymobilize hisrher social resources by reaching out to alters in hisrher personal networkfor information andror influence, which, in turn, would help improve the chance ofhaving a successful outcome. Therefore, the general hypothesis, known as the socialresources proposition, is that access to and use of social resources would lead to better

Ž .job search outcome Lin, 1982 .The social resources theory further hypothesizes that access to and use of social

Žresources are influenced by two factors: the initial status of an individual the strength of.position proposition and the strength of the relationship between job-seeker and contact

Ž .person when a contact is used in the search process the strength-of-tie proposition .ŽSpecifically, the theory argues that better initial positions for example, parental

.socioeconomic resources or initially attained socioeconomic statuses promote thelikelihood of reaching and using better social resources. These initial positions offeropportunities to establish ties with those who have better personal and social resourcesthrough the so-called homophily or like-me principle. Homophily ties are ties that linkindividuals with similar characteristics, for example, positions and ascribed traits. The

Žtheory, informed by Granovetter’s weak-tie argument Granovetter, 1973; Granovetter,.1982 , also argues that given similar initial positions, those who use weaker ties in job

search may have access to better social resources because weaker ties tend to possessdiverse and, therefore, useful information and influence, which would increase thechance of getting better jobs.

Page 3: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 161

While a significant body of literature has substantiated the major propositions of thetheory, 2 an important issue has emerged concerning how social resources should beconceptualized and measured. The debate focuses on whether social resources should be

Ž .measured relative to the specific contact used in job search contact resources or in theŽ .resource characteristics of one’s routine social network network resources .

3. Contact resources versus network resources

As argued in the social resources theory, both access to and use of social resourcesplay an important role in affecting job-search outcomes. Earlier studies have focusedalmost exclusively on the use of social resources. Inquiries have sought to identify acontact who successfully leads to a specific job and to determine how the contact’s

Žresource characteristics for example, occupational status, authority position, core versus. Žperipheral sector affect the outcome variables for example, the individual’s occupa-

.tional prestige, wage, authority, firm size . This approach of conceptualization andmeasurement of social resources is termed as contact resources. According to thisapproach, the effect of social resources on instrumental outcomes represents the directresult of reaching a resource-rich contact person in one’s social network.

The contact resources approach has been applied in empirical studies by Lin et al.Ž . Ž . Ž .1981a,b , Marsden and Hurlbert 1988 , De Graaf and Flap 1988 , Sun and HsungŽ . Ž .1988 , and Bian and Ang 1997 . Findings generally show that higher socioeconomicposition of the contact is related to better search outcomes because the higher the statusof the contact is, the more information the contact possesses and the more influence thecontact can exert on behalf of the individual seeking help.

Recent attention has been given to the access to social resources, that is, socialresources embedded in the routine social networks to which an ego belongs. Known asthe network resources approach, the argument focuses on the networks an ego routinely

Žmaintains and the configuration of the members’ resource characteristics for example,.diversity and range of resources , which are indicative of the extent of available

resources an ego can draw from in a specific instrumental action. 3

Ž .Campbell et al. 1986 have offered some research evidence for the viability of thisŽargument. They show that socioeconomic statuses are related to network range size,

2 ŽStudies have strongly confirmed two of the three propositions the social resources and strength of.position hypotheses . The strength-of-tie proposition has not received systematic confirmation. Further

Ž .conceptualization Lin, 1990b; Montgomery, 1992 is needed to further refine the hypothesis.3 Different approaches have been used to generate information about the characteristics of personal

networks. These approaches impose different constraints on the number of ties, content of ties, intimacy withŽties, geographical location, and time frame in which ties are evoked Campbell and Lee, 1991; Fischer, 1982;

.Wellman, 1979; Wellman and Wortley, 1990 . Regardless of the approaches used, variations in networkcomposition and range, in terms of age, education, race, and sex, are found to be fairly similar across studiesŽ .Campbell and Lee, 1991 .

Page 4: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178162

4. Ždiversity, and density of the network and network composition mean and maximal.education, job prestige, and family income of alters . If network range and network

composition are indicators of social resources, these results lend support to the strength-of-position hypothesis as espoused in the social resources theory.

Ž .The work of Lin and Dumin 1986 also focuses on network resources, which areŽ .operationalized as the range of occupational statuses network range and the highest

Ž .occupational status reached network composition in the network. Their results showthat higher initial socioeconomic status and weak ties allow access to better networkresources. In addition, weak ties enable greater access to better resources when theinitial position of the individual is low.

Ž .Informed by the network resources approach, the study by Boxman et al. 1991Ž .shows that managers with larger social capital resources , which is measured by the

frequency of contacts with people in other organizations and the number of member-ships of elite clubs, are more likely to use informal channels to find their jobs.Moreover, access to social capital has a substantial independent effect on income, net ofhuman capital and levels of managerial function.

Ž .Erickson 1996 applies the notion of network resources to the study of culturalresources. Defining network resources in terms of the number of distinct classes

Ž .accessed through a network network variety , she finds that network variety is moreŽimportant than class position in influencing the extent of cultural resources knowledge

.in books, art, magazines, and sports .

4. An integrated approach

While empirical evidence supports both approaches of measuring social resources, noeffort has yet determined the relationship between network resources and contactresources. In this paper, we argue that network resources and contact resources representtwo causally-related aspects of social resources. One may view the routine socialnetwork as a resource opportunity structure with diverse ties and rich resources, towhich an ego has access. When an ego needs to evoke a tie as a contact in a particularaction, a resource-rich network would offer a greater likelihood of finding a resource-richcontact. In other words, the richer the resources and the more diverse the ties that areembedded in the network, the more likely an ego may locate and find a resource-rich tieas a contact to accomplish an instrumental action, such as finding a job. If thisconception is correct, network resources should causally enhance contact resources. 5

4 Diversity is measured by the proportions of respondent’s alters who are of different religion, ethnicity,sex, or who are engaged in a different kind of work. Density is measured as the average closeness, frequency

Žof contacts, duration of friendship, content multiplexity mean number of contents found in ties between.respondent and alters , and role multiplexity.

5 Ž .In an instrumental action for example, searching for jobs , an ego may choose not to evoke any tie as acontact, and rely, rather, on formal channels or direct application. It is not clear at this point how such choiceis made. Empirical evidence is also equivocal as to what factors influence such choices. Nor is there any

Ž .evidence that evoking a contact s for job search is dysfunctional and detrimental to the search outcomes.Nevertheless, when a contact is activated for a particular action, we may hypothesize that resources in thenetworks should affect the extent of resources associated with that contact.

Page 5: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 163

Moreover, the actual mobilization of resources, rather than the availability of resources,should play a more critical role in affecting the outcome.

The above contention, however, does not imply that network resources is the onlyfactor affecting the reach to a specific contact. Actions, in principle, should be dictated

Ž .by both personal and social resources Lin, 1990a . As stated above, network resourcesas the morphological structure of social resources provide an opportunity structure thatone can activate a resource-rich contact for an instrumental action. Personal resourcesreflect the attributes of one’s structural position. These personal resources may include

Ž .both ascribed resources for example, parental socioeconomic characteristics andŽ .achieved resources for example, education and attained socioeconomic characteristics .

The principle of homophily states that individuals at higher structural positions with richresources would tend to interact with alters of similar resource characteristics. Thuspersonal resources may enrich network resources. Personal resources may also benefitthe mobilization of social resources, for example, seeking help from a contact person injob search. While the principle of homophily predicts the reach of individuals at highinitial positions to resource-rich alters, it may also govern the helping behavior of thelatter in relation to the former. A resource-rich contact person might be more willing tohelp an individual who are of similar resource characteristics than one who has less

Ž .resources that is, from a lower position . Thus personal resources would allowindividuals to more effectively mobilize resources acquired through social ties.

The above deliberations lead us to the following hypotheses. A systematic presenta-tion of these hypotheses can be found in Fig. 1.

Ž . Ž1 Network resources configuration of network members’ socioeconomic character-. Žistics would affect contact resources socioeconomic characteristics of a successfully-

Fig. 1. The conceptual model.

Page 6: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178164

.activated tie causally. Empirically, there should be a significant and positive relation-ship between the two variables.

Ž .2 Contact resources would affect status attainment directly whereas the effect ofnetwork resources is mostly indirect. Empirically, the standardized path coefficient fromcontact resources to status outcome should be greater than that from network resourcesto the outcome variable.

Ž .3 Contact resources are both directly and indirectly affected by personal resources.ŽEmpirically, there should be significant and positive paths from ascribed parental

. Ž .characteristics and achieved attained education and socioeconomic characteristicsvariables to contact resources, both directly and indirectly via network resources.

5. Further theoretical implications

The specified model has implications for two important conceptual issues in theresearch literature: the role of the strength of ties and the relative importance ofascription and achievement in the status attainment process. The strength-of-tie argu-ment looks at the nature of dyadic relationship between ego and contact in the job searchprocess. It contrasts with the principle of homophily, which focuses on the similarity inindividual attributes of the two parties in the process of resources flow. Extending

Ž . Ž .Granovetter’s weak-tie argument Granovetter, 1973; Granovetter, 1982 , Lin 1982clarifies the link of weak ties to status outcome by emphasizing the role of contactresources in the transmission of weak-tie effect on status attainment. That is, weak ties

Ž .would lead to better contacts, who in turn, bring successful outcomes. Burt 1992highlights the properties of social ties that bring successful outcomes of certain actions.He examines the strategic location of social ties in networks and contends that networkbridges, which connect two or more otherwise disconnected social clusters, oftenfacilitate the flow of information across these social clusters. Access to network bridgesmight thus bring information benefits, which would increase the likelihood of havingsuccessful outcomes for a certain action. Further, while network bridges can be strong orweak ties to ego, weak ties are more likely than strong ties to be bridges. Thus anetwork of weak ties may offer better resources than a network of strong ties. Empirical

Ž . Ž .evidence Lin and Dumin, 1986 shows that weaker ties friends and acquaintancesŽ .have higher mean socioeconomic statuses than stronger ties relatives .

Following the above line of reasoning, it could be hypothesized that a resource-richnetwork contains more weak ties than a resource-poor network. If this hypothesis issupported, there should be a positive relationship between richness and diversity of the

Žego’s network, and the likelihood for the successful contact that is, contact that leads to.a certain job to be a weak tie. As shown in Fig. 1, such a relationship is postulated from

network resources to strength of tie.In addition, this model offers an opportunity to further assess the relative contribu-

tions of ascribed and achieved statuses in the status attainment process. In the earlymodels of status attainment, education, an indicator of the achieved status, was seen as

Ž .the mediating factor between ascribed status that is, parental status and the outcome

Page 7: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 165

Ž .variable. Analyses on the American data Blau and Duncan, 1967 showed that the totalŽ .effect of the ascribed status, direct and indirect via education , on attained status was

relatively small compared to that of education. These findings have led to the conclusionthat in the American society, the achieved status is more important than the ascribedstatus for status attainment. This conclusion holds true when both education and first-jobstatus are considered indicators of achieved status.

Ž .Previous studies incorporating social resources variables mainly contact resourceshave argued that when the mediated effects of parental status through social resourcesare taken into account, ascribed status might be just as important, if not more so, than

Ž .achieved status Lin et al., 1981a,b . The present approach incorporating both networkresources and contact resources affords a more refined model to specify how the

Ž .ascribed status parental status might operate through both the ego’s routine networkand the specific contact mobilized in the job search process in accounting for asuccessful outcome.

In Fig. 1, the effect of the ascribed status can be identified by assessing the directeffect of parental resources on status attainment as well as its indirect effects viaachieved resources, network resources, strength of tie, and contact resources. Such ananalysis should yield more refined estimates of the relative contributions of the ascribedand achieved statuses on status attainment than those provided by previous efforts.

Section 6 describes the data and measures used to estimate the proposed model.Whenever our data set permits, multiple indicators are used for each concept. The model

Žis estimated by the structural equation modeling technique Joreskog and Sorbom,.1993 .

6. Data and method

The data were collected in a job search survey conducted in the Albany–Schenectady–Troy, New York area in 1975. The data set consists of information onjob-seeking activities for a random sample of 399 males, who were aged 20–64.

6 Since network resources are measured as the occupational characteristics of the network at the time of theinterview rather than at the time the respondents found their jobs, it is conceivable for one to be concernedabout the possibility that one’s status outcome may actually affect the extent of one’s network resources. Thismay be especially true for those respondents who found their current jobs long before the study. Unfortunately,the demonstration of the stability or instability of network resources is beyond the ability of our data.However, we are interested in one’s routine network, in which social ties tend to be built in a gradual processover time. Although the average length of time for a social tie to become routinized in a network is yet to beknown, it is reasonable to assume that one’s routinized network ties should remain stable for a certain periodof time. Our data showed that 52.7% of the respondents found their current jobs within the last 5 years.Therefore we may assume with some confidence that network resources of at least half of our sample are fairlystable. Nevertheless, about 30% of the sample found their current jobs 10 or more years ago. Networkresources of these individuals, as measured at the time of survey, might thus be different from networkresources at the time of job search. Readers are urged to take note of the potentially confounding relationshipbetween network resources and status outcome.

Page 8: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178166

The key variables in the hypothesized model, presented in Fig. 1, are measured asfollows. Parental resources is operationalized as the resources that are obtained throughone’s family background and measured by the respondent’s father’s occupational statusand educational level. Attained resources are indicated by respondent’s educational leveland first job status. These variables have long been recognized as crucial determinants of

Ž .status outcome Blau and Duncan, 1967 .Network resources are measured in terms of the extent of accessibility to a variety of

occupational positions through an individual’s social ties. During the interview, eachrespondent was presented with a list of 20 occupations and asked to indicate whetherany of his relatives, friends, or acquaintances held such occupations. 6 Network diversity

Ž .is indicated by 1 the difference between the lowest and the highest occupationalŽ . Ž .statuses accessed, 2 the highest occupational status accessed, and 3 the number of

Ž . 7different occupations accessed Lin and Dumin, 1986 . The number of occupationalstatuses as an indicator of network resources is self-explanatory. The variable of thehighest occupational status reached requires some clarification. According to the social

Ž .resources theory Lin, 1982 , one tends to take advantage of the ‘better’ resourcesembedded in one’s ties or network in an instrumental action. Thus, the upper reach ofone’s social network to resources, such as occupational status, should be a meaningfulindicator of the ‘best’ potential social resources. The range of occupational statusessuggests the vertical depth of social resources and indicates the diversity and novelty of

Ž .resources in one’s network Campbell et al., 1986 . It has been a typical networkmeasure in the literature. However, as an indicator of network resources in the presenttheoretical formulation for status attainment, it is expected to be a secondary butcomplementary indicator to the highest occupational status in predicting the actual use

Ž .of social resources in the job search process contact resources .Contact resources and the strength of ties are measured only for individuals who used

personal contacts in finding their jobs. Respondents in the survey were asked how theyfound their current jobs. The methods reported are categorized into three major types:personal contacts, direct application, and formal channels. Personal contacts includecontacts with a relative, friend, acquaintance, neighbor, person one used to work with,person one was working with at the time, former employer, and new employer. Thecontact person may provide the respondent with the information, advise about the job, orinfluence the hiring of the respondent. 8 Direct application refers to job change throughnewspaper advertisements, intra-company promotion, and company’s sales catalogue.

7 In the subsequent modeling process, two of the indicators, range of occupational statuses and highestoccupational status, are highly correlated. However, we decided to keep all three indicators in the model asmeasures of network resources, as more indicators tend to increase the reliability of the estimates of themeasurement error, and, thus, to increase the reliability of the estimates of the structural effects.

8 It is beyond the scope of the present paper to test what specific functions these contacts provide. PastŽ .literature Lin et al., 1981a suggests that they serve both information and influence functions. Teasing these

functions out would require more detailed measures and designs, which are not available in the current dataset.

Page 9: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 167

Formal channel includes such sources as union, vocational counsellor at school, privateemployment agency, public employment agency, and professional and vocational groups.

Excluding the respondents who did not report the job search methods that led to theircurrent jobs, 56.3% of the 318 respondents found their current jobs through successful

Žmobilization of a particular personal tie as opposed to a vast majority of the sample.obtained job information via personal contacts; see also footnote 9 , 27.4% applied

directly to an employer, and 16.3% used formal channels. For those who utilizedpersonal contacts to obtain current jobs, the relationship between the contact and therespondent and the occupational status of the contact are recorded. 9

Contact resources are measured by the occupational status of the contact. Thestrength of tie is indexed by the relationship between the respondent and the contact as

Ž . Ž .being either weak acquaintances and indirect ties or strong friends and relativesŽ . 10Granovetter, 1973; Lin et al., 1981a .

The ultimate dependent variable is status outcome, measured by the occupationalstatus of the current job. Univariate statistics and the zero-order correlations amongthese variables can be found in Appendix A. Structural equation modeling is adopted to

Ževaluate the causal relationships among the variables, and LISREL program Version. 118.14 is used to estimate the parameters in the equations.

9 While 97% of 383 respondents obtained information about current jobs from at least one personal contact,not all personal contacts led to successful outcomes. In such cases, resources of unsuccessful contacts were notmeasured. Therefore, contact resources, as operationalized in the present paper, represents the successfullyactivated resources. In addition, network resources of the individuals who used other job-search methods werealso not measured. As a result, one may be concerned about sample selectivity. This issue is dealt with in thebeginning of the section ‘Hypothesis Testing’.

10 Friends and relatives are sometimes analyzed separately for the purpose of ascertaining their independentŽ .effects Lin and Dumin, 1986 , with the strength of tie with friends being ranked between acquaintances and

Ž . .relatives. Studies by Wellman 1990 ; Wellman and Wortley, 1989, 1990 show that extended kin, such ascousins and aunts, are much weaker ties than friends and that only parentsradult children are stronger ties thanfriends. Therefore, he argues against the mix of extended kin with immediate kin. Further, means comparisonsshow that users of friend ties exhibit closer resemblance to users of relative ties than those of acquaintanceties, in terms of parental characteristics, attained resources, social resources, and status outcomes.Nevertheless,

Ž .Marsden and Campbell 1984 argue that intimacy may be a better indicator of strength of tie than rolerelationship. However, our data set does not have the information on intimacy either. While we acknowledgethe limitation in the operationalization of strength of tie in the present paper, we also urge that future research

Ž .should incorporate other dimensions of tie strength Granovetter, 1973 .11 Ž .The fit of the model with the data is assessed by Goodness-of-Fit Index GFI , Adjusted Goodness-of-Fit

Ž . Ž . Ž .Index AGFI , and Comparative Fit Index CFI . All three indexes have values ranging from 0 worst fit to 1Ž . Ž .best fit Joreskog and Sorbom, 1993 . If the constructed model does not fit with the observed data matrix,modifications will be made. Guided by the modification indices and theoretical formulations, the modificationsinvolve either constraining or freeing parameters. The modification procedure follows a successive hierarchy;each new run would involve a single modification. Thus the significance of each modification can be assessedin relation to the preceding and subsequent modifications. The selection of the ‘best’ model relies on threecriteria. First, the model achieves a reasonable degree of goodness of fit, as reflected by the GFI, the AGFI,and the CFI. Second, the model shows significant improvement over the previous models. This is evaluated byan increase in the values of the GFI, the AGFI, and the CFI, and a nonsignificant increase in chi-square valuesrelative to a corresponding increase in degrees of freedom. Third, the model offers the most meaningfulinterpretations and the structural parameters are stable in a series of successive runs.

Page 10: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178168

7. Hypothesis testing

Since the proposed model examines social resources in terms of two causally-relatedcomponents—network resources and contact resources, it is applicable only to individu-

Žals who activated personal contact in obtaining current jobs 56% of the total valid.sample, or 179 respondents . This raises the issue of sample selectivity, which has wide

Ž .implications involving many research topics and disciplines Heckman, 1980 . ToŽ .resolve this issue, a dichotomous variable 1scontact-users; 0sotherwise is com-

puted to indicate whether a contact was successfully activated to lead to the current job,and regressed on parental resources, attained resources, and network resources.

Ž .Results from logistic regression analysis Table 1 show that none of the independentvariables have significant effect on the likelihood of using personal contact to obtaincurrent jobs. Therefore, it can be concluded that the present sample does not suffer fromselectivity bias.

The estimation of the proposed model follows a sequential process. Results aredisplayed in Table 2. Due to limitations of space, only the estimates for the final modelare presented. However, the goodness-of-fit measures for the intermediate models areprovided for reference.

The fully-recursive baseline model, Model 1, contains all possible relationshipsŽamong the latent variables, such that some paths e.g., from parental resources and

.achieved resources to strength of tie , though theoretically non-existent, are alsoincluded. It serves as a benchmark to which other alternative models are compared. The

Ž . Ž .Goodness-of-Fit Index GFI , the Adjusted Goodness-of-Fit Index AGFI , and theŽ .Comparative-Fit-Index CFI suggest that the baseline model has an acceptable, though

Ž .not desirable, fit with the data GFIs0.86; AGFIs0.70; CFIs0.88 . An examinationof the structural parameters shows that the following paths are not significantly at the

Table 1Logistic regression of use of personal contact on parental resources, attained resources, and network resourcesa

Independent variables Use of personal contact

Parental resourcesFather’s occupational status y0.004Father’s education 0.02

Attained resourcesEducation y0.01First job status y0.01

Network resourcesRange of occupational statuses 0.004Highest occupational status y0.02Number of occupations 0.04

2y2 Log likelihood x 457.39Degrees of freedom 333Number of valid cases 341

a None of the coefficient is statistically significant at p-0.05.

Page 11: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 169

Ž . Ž .probability level of 0.05: 1 from parental resources to network resources, 2 fromŽ . Ž .parental resources to strength of tie, 3 from strength of tie to status outcome, 4 fromŽ . Ž .network resources to status outcome, 5 network resources to strength of tie, and 6

Table 2The model estimation process and parameter estimates for the final model

A. Model estimation process

Goodness-of-fit measures Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 72x rdf 144.87r26 145.40r27 146.40r28 148.75r29 152.07r30 155.31r31 157.54r32

GFI 0.86 0.86 0.85 0.85 0.85 0.85 0.84AGFI 0.70 0.71 0.72 0.72 0.73 0.73 0.73CFI 0.88 0.88 0.88 0.87 0.87 0.87 0.87

Ž .Effective sample size after listwise deletion s153

Ž .B. Parameter estimates for the final model Model 7

Measurement model — factor loadings of indicators on latent Õariables

Indicators Parental Attained Network Strength Contact Statusresources resources resources of tie resources outcome

Father’s education 0.64 y y y y yFather’s job status 1.00 y y y y yEducation y 0.69 y y y yFirst job status y 1.00 y y y yRange of job statuses y y 1.00 y y yHighest job status y y 0.85 y y yNumber of occupations y y 0.82 y y yRole relationship y y y 1.00 y yContact’s job status y y y y 1.00 yCurrent job status y y y y y 1.00

Structural model — statistically significant standardized effects of independent latent ÕariablesIndependent variables Attained Network Strength Contact Status

resources resources of tie resources outcome

Parental resources 0.54 ns ns 21 nsAttained resources y 0.34 0.38 0.31 0.38Network resources y y ns 0.22 nsStrength of tie y y y 0.22 nsContact resources y y y y 0.47

2R 0.29 0.12 0.14 0.40 0.57

GFI: Goodness-of-fit Index.AGFI: Adjusted Goodness-of-fit Index.CFI: Comparative-Fit-Index.Model 1: Fully recursive model.Model 2: Model 1 without the path from parental resources to network resources.Model 3: Model 2 without the path from parental resources to strength of tie.Model 4: Model 3 without the path from strength of tie to status outcome.Model 5: Model 4 without the path from network resources to status outcome.Model 6: Model 5 without the path from network resources to strength of tie.Model 7: Model 6 without the path from parental resources to status outcome.ns: Parameters not statistically significant and thus removed from the equation.y: Parameters not included in the equation.

Page 12: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178170

from parental resources to status outcome. Each path is thus eliminated in eachsubsequent step.

Ž .After deleting the insignificant paths, the model Model 7 still indicates a less thandesirable fit with data. Modification indices and first-order derivatives suggest alterna-tives for model improvement but the alternatives are not consistent with our conceptualformulation. Therefore we consider Model 7 the best model and its parameter estimatesare used to validate our hypotheses. The lack of desirable fit of the model may seemnotable because of our small sample size.

The first hypothesis states that network resources are positively related to contactresources. Supporting evidence is found for the hypothesis. Individuals with access to aresource-rich network are more likely to locate a resource-rich contact person during job

Ž .search betas0.15 . The indirect effect of network resources on contact resources viastrength of tie is not statistically significant. Network resources may not facilitate thereach to a successful contact through weak tie. Weak ties are found to facilitate the reachto a contact person with higher occupational status, who eventually leads to a better jobŽ . 12 Žbetas0.22 . This finding replicates previous results Lin et al., 1981a,b; Marsdenand Hurlbert, 1988; De Graaf and Flap, 1988; Sun and Hsung, 1988; and Bian and Ang,

.1997 .Ž .Burt 1992 argues that the extent of resources is related to one’s location in a

collectivity of networks rather than weak ties. The most strategic or most resource-filledlocation is the central node, which connects various disconnected isolated networks. Therelationship between this central node and connected alters might be strong or weak ties.In other words, resource-rich networks may not necessarily consist of a majority of weakties. Therefore it is reasonable to find that network resources and the likelihood ofactivating weak ties for instrumental action are not related.

The second hypothesis postulates that contact resources have a stronger direct impacton status outcome than network resources and that the effect of network resources ismostly indirect. This hypothesis is supported by the findings. Consistent with the

Ž .hypothesis, contact resources have a significant direct effect on status outcome 0.47whereas the effect of network resources on status outcome is mainly indirect. Theseresults suggest that use of social resources is relatively more important than access tosocial resources in affecting the outcome of an instrumental action.

The third hypothesis that contact resources is influenced by personal resources alsoreceives support. Both parental resources and attained resources have a positive relation-

Ž .ship with contact resources 0.21 and 0.31 respectively . In other words, positional

12 ŽRecent literature suggests that strength of tie may have an interaction effect with social origins Wegener,. Ž .1990 and with contact resources on status attainment Lin et al., 1981a; Wegener, 1990 . In addition, a

reviewer suggests a negative interaction effect between attained resources and network resources on contactŽ .resources. To test the validity of these interaction effects, four sets of interaction terms are computed: 1

Ž . Ž .strength of tie and parental resources, 2 strength of tie and attained resources, 3 strength of tie and contactŽ . Ž .resources, and 4 attained resources and network resources. The method of Smith and Sasaki 1979 is used to

reduce the problem of multicollinearity between the product variable and the component terms. An examina-tion of the zero-order correlation coefficients shows that none of the interaction terms is significantly related tothe dependent variables of interest. This finding is further confirmed using multiple regression analysis.

Page 13: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 171

advantages, in terms of better family background, own education, and first job status,facilitate one’s ability to reach a contact with rich resources, who would help bringing abetter job outcome.

In sum, the present findings support our three hypotheses proposed earlier. Amongindividuals who use personal contact in getting current job, the extent of mobilizedsocial resources plays a significant role in generating high-status jobs. The availability ofsocial resources only provides a necessary but not sufficient condition for status

Ž .outcomes. The pool of social resources network resources , however, would facilitatethe extent of mobilization. Better contact resources can also be reached via weak ties.Further, both access to and use of social resources are facilitated by personal resourcesŽ .parental and attained resources .

8. Ascription versus achievement

The results from the final model allow us to examine a critical issue in the statusattainment and social network literature; namely, the relative contributions of ascriptionand achievement in explaining status attainment in the United States. An exercise in

Ž . Ž .effect decompositions Alwin and Hauser, 1975 Table 3 shows that the total effects ofparental resources on status outcome is 0.47 whereas the total effects of attainedresources on status outcome is 0.53. The ratio of contributions of ascription versusachievement factors is about 1:1.3. Although ascription does not have a direct effect onstatus attainment, it still remains an important factor in affecting status outcome.

A closer examination of the results suggests that the role of achievement in statusattainment may need reinterpretation. Conventionally, effects of achievement are thoughtto correspond with merit, qualifications, ability, and so on. In the present study, 36% ofthe total effect of attained resources on status outcome is found to be mediated by socialresources. The mediating effect via social resources represents one’s effort in the courseof an instrumental action, that might not necessarily be related to heightening abilities. Aperson with good education and a prestigious first job might be placed in a resource-richnetwork because of structural opportunities, and a resource-rich network opens addi-tional opportunities for a better outcome. Further, part of the social resources areembedded in ascribed ties—relatives. Therefore the total effects of achievement, aspresented above, might partly be indicative of forces other than meritocracy. To obtain amore accurate estimate of the total effects of achievement as ability and merit, we haveto exclude the mediating effect via social resources. As a result, we only consider thedirect effect, which is 0.38. The ratio of ascription effects to achievement effectsbecomes 1.11:1, suggesting that ascription may play a more important role thanachievement in affecting status outcome.

The role of social resources is also highlighted by its contribution to mediating theeffect of parental resources on status attainment. Social resources transmit 50% of thetotal effect of parental resources on status outcome. In sum, the above findings not onlydemonstrate the importance of ascription, but also highlight the role of social resourcesin the status attainment process.

Page 14: Network resources, contact resources, and status attainment

()

G.L

aietal.r

SocialNetw

orks20

1998159

–178

172Table 3Decomposition of the total effects of parental resources and attained resources on status attainment

A. Decomposition of the total effect of parental resourcesPaths Effects Percentage of the total effect

Direct effect 0.09 19%

Indirect effectDirectly from attained resources 0.16s0.54=0.30 34%Via social resources 0.22 47%

Ž .Via network resources 0.01s 0.07=0.10Ž .q 0.07=0.14= 0.09Ž .q 0.07=0.14=0.21=0.40Ž .q 0.07=0.15=0.40

Via the effect of attained resourcesŽ .On network resources 0.01s 0.54=0.30=0.14=0.09Ž .q 0.54=0.30=0.14=0.21=0.40Ž .q 0.54=0.30=0.15=0.40

Ž .On strength of tie 0.03s 0.54=0.29=0.09Ž .q 0.54=0.29=0.21=0.40

On contact resources 0.07s0.54=0.31=0.40Total effect 0.47 100%

B. Decomposition of the total effect of attained resourcesPaths Effects Percentage of the total effect

Direct effect 0.30 57%Ž .Indirect effect via social resources 0.23 43%

Ž .Ž .Via network resources 0.06s 0.30=0.10 0.30=0.14=0.09Ž .q 0.30=0.14=0.21=0.40Ž .q 0.30=0.15=0.40

Ž .Via strength of tie 0.05s 0.29=0.09Ž .q 0.29=0.21=0.40

Via contact resources 0.12 s0.31=0.40Total effect 0.53 100%

Page 15: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 173

9. Further discussion

Several issues remain to be discussed. The present analysis includes only individualswho use personal contact in getting current jobs. The social resources of thoseindividuals who have found their jobs without activating, or successfully activating,informal channels, are thus not measured. While parental resources, attained resources,and network resources do not significantly predict the likelihood of the successful use of

Ž .personal contact that is, no selectivity bias , one may question if the impacts of personalŽ .resources and social network resources on status attainment differ between the

individuals who use and those do not use personal contact.To resolve this issue, parental resources, attained resources, and network resources

Žare used to predict status outcome for the two groups of respondents that is, contact-users.and non-contact-users . Multi-group analysis in structural equation modeling is em-

ployed to estimate the parameters in the equation. Particular attention is given towhether the two models have the same structural forms and parameter estimates. 13

The test of equal structural forms fails to support the hypothesis that the structuralrelationships among the variables are the same for contact-users and non-contact-usersŽ 2 .x s201.53; dfs42 . Further examination of the statistical significance levels of thestructural relationships for the two groups reveals that some paths are significant in onegroup but not the other. As a result, the nonsignificant paths in each group are removedaccordingly. The common paths are then tested for equality. Panels A and B of Table 4display the final estimates for contact-users and non-contact-users respectively.

The results show that the parameter estimates for the paths from parental resources toattained resources, from attained resources to network resources, and from attainedresources to status outcome are statistically equal in the two groups. Compared to themodel without the equality constraints, the inclusion of the constraints does notsignificantly increase the x 2 value. However, the structural models for contact-usersand non-contact-users are different regarding the effect of network resources on statusoutcome. While network resources are enhanced by parental resources and attainedresources for both contact-users and non-contact-users, only the former would enjoy thebenefits of network resources in a job search. This finding suggests that althoughnetwork resources do not predict the likelihood of using social ties to accomplish aninstrumental action, network resources tend to yield a better outcome when a tie issuccessfully activated.

One possible explanation for this finding is that the composition of network resourcesmay differ for contact-users and non-contact-users, thus leading to different outcomes.

Ž .However, T-Test not shown indicates no difference in the extent of network resources

13 The test for equal structural forms and parameters is taken in two steps. First, a model is set up to imposeequal structural forms on the two groups. An insignificant chi-square statistic suggests equal structural forms.Second, if the two groups have equal structural forms, further constraint will be imposed on the values of thestructural parameters. A nonsignificant chi-square statistic indicates equal structural parameters. If thehypothesis of equal structural forms is not supported, modifications will be made to the structural relationshipsfor the two groups and the common structural parameters will be subject to equality test. The selection of the‘best’ model for each group follows the same criteria used for the proposed model, as stated in 11.

Page 16: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178174

Table 4Comparison of the structural effects of parental resources, attained resources, and network resources on statusattainment for 153a contact-users and 155a non-contact-users

A. Contact-users

Measurement model — factor loadings of indicators on latent Õariables

Indicators Parental resources Attained resources Network resources Status outcome)Father’s education 0.64 y y y)Father’s job status 1.00 y y y

)Education y 0.69 y y)First job status y 1.00 y y

)Range of job statuses y y 1.00 y)Highest job status y y 0.85 y)Number of occupations y y 0.81 y

)Current job status y y y 1.00

Structural model — standardized effects of independent ÕariablesIndependent variables Attained resources Network resources Status outcome

)Parental resources 0.55 ns ns) )Attained resources y 0.28 0.66

)Network resources y y 0.152 ) ) )R 0.31 0.08 0.51

B. Non-contact-usersMeasurement model — factor loadings of indicators on latent ÕariablesIndicators Parental resources Attained resources Network resources Status outcome

)Father’s education 0.61 y y y)Father’s job status 1.00 y y y

Education y 0.58) y y)First job status y 1.00 y y

)Range of job statuses y y 1.00 y)Highest job status y y 0.84 y)Number of occupations y y 0.75 y

)Current job status y y y 1.00

Structural model — standardized effects of independent ÕariablesIndependent variables Attained resources Network resources Status outcome

)Parental resources 0.55 ns ns) )Attained resources y 0.27 0.65

Network resources y y ns2 ) ) )R 0.30 0.07 0.422x rdf 203.25r44

GFI 0.89CFI 0.89

) p-0.05.ns: Parameters not statistically significant and thus removed from the equation.y: Parameters not included in the equation.a Effective sample size after listwise deletion.

Page 17: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 175

between the two groups. Another possible explanation deals with the nature of tiesestablished in the networks. The presence of the effect of parental resources on networkresources for non-contact-users but not for contact-users suggests different processes inwhich network resources are acquired. Contact-users accumulate network resourceslargely through their own attained resources whereas non-contact-users build theirnetwork resources from both parental resources and attained resources. This leads to thespeculation that network resources built in part from parental resources may not be aseffective for instrumental purpose as network resources built only from one’s ownattained resources. It is because social ties established through family connection might

Ž . Žbe more likely to be strong ties that is, relatives than weak ties that is, friends and.acquaintances , and weak ties tend to be more useful than strong ties for instrumental

actions. However, no difference is found between contact-users and non-contact-users inthe proportion of relatives in the network nor the extent of network resources acquiredthrough relatives. Therefore a satisfactory explanation for this puzzling result awaitsfuture research.

Another issue is concerned about the restriction of the analyses to men only. Previousstudies have suggested that parental resources and social resources are less effective on

Ž .women’s status outcome Lin and Bian, 1991 . Therefore the relative effects ofascription and achievement should differ between women and men. We hypothesize thatwomen are less likely to benefit from parental and social resources but more likely torely on attained resources. While there may be a smaller ascriptive effect on statusoutcome, women may also experience greater structural constraint from their ownsocioeconomic positions. Clarification of these conceptual and empirical issues wouldsignificantly enrich our understanding of the opportunity structure for mobility inparticular and social stratification in general.

9.1. Concluding remarks

This study has made two contributions to theory. In the context of the socialresources theory, it clarifies the mechanism through which social resources, embeddedin the routine network and mobilized in a job search, affect status attainment. Specifi-cally, network resources and contact resources form a complementary and causalrelationship in influencing the status outcome. In other words, the ‘normative’ networkresources provide a context in which social resources are mobilized for a particularinstrumental action, such as searching for jobs.

The established relationship between access to and use of social resources may shedsome light on the distinction between two related terms, namely social resources and

Ž .social capital. Coleman 1988, 1990 defines social capital as resources embedded in aŽ .social network and mobilized for a purposive action. Flap et al. 1986, 1988 also refer

social capital to social ties activated during job search. Social capital, thus conceived,represents the use of social resources. To refine the concept of social resources anddelineate the conceptual boundaries of social resources and social capital, we mayconsider social resources a collection of resources located in the structure of relationsamong individuals, that is, a pool of available resources that could be mobilized forcertain purposes, whereas social capital is the portion of resources used to achieve a

Page 18: Network resources, contact resources, and status attainment

()

G.L

aietal.r

SocialNetw

orks20

1998159

–178

176

Appendix A. Zero-order correlation coefficients, means, and standard deviations ofstudy variables for 153 respondents who used personal contacts to find jobsa.

Education First job Range of job Highest job Number Strength Contact’s Status Father’s job Father’sstatus statuses status of occ. of tie job status outcome education

Education 1.00First job status 0.69 1.00Range of job statuses 0.36 0.34 1.00Highest job status 0.52 0.45 0.85 1.00Number of occupations 0.40 0.35 0.82 0.66 1.00Strength of tie 0.21 0.30 0.19 0.22 0.24 1.00Contact’s job status 0.51 0.55 0.35 0.46 0.34 0.32 1.00Status outcome 0.71 0.64 0.39 0.54 0.46 0.33 0.68 1.00Father’s job status 0.56 0.54 0.23 0.36 0.25 0.21 0.47 0.49 1.00Father’s education 0.55 0.46 0.23 0.32 0.23 0.14 0.44 0.45 0.64 1.00Mean 13.03 38.64 60.38 80.16 9.37 0.35 53.04 46.74 38.34 9.76Standard deviation 3.03 24.84 21.86 17.61 5.51 0.48 21.84 25.11 24.57 3.84

aSample size after listwise deletion.

Page 19: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178 177

specific end. The present study shows that the extent of mobilized resources would becontingent upon the richness of the resource pool. However, the circumstances underwhich social resources are translated in social capital are yet to be studied.

This study also advances our understanding of the process by which ascriptive andachieved factors exert their influence on status attainment. By demonstrating howascriptive and achieved statuses affect the resources pool of one’s social network andhow such network resources can be translated into an advantage by reaching bettercontacts to get better job outcomes, this study calls for a serious reconsideration of thethesis that the achieved factors largely account for status attainment in the United States.Particularly, meritocracy associated with achieved factors might have been over-stated inprevious studies. More than one third of achieved effects is found to be transmitted viaone’s social network, which may not be indicative of hisrher skills and ability. Thisstudy, therefore, poses a challenge to the ‘functional’ argument of industrial develop-

Ž .ment and mobility Treiman, 1970 that the ascribed factors are more influential in theless developed or more traditional societies whereas the achieved factors are moreinfluential in the more developed or more ‘modern’ states. More importantly, itchallenges the conventional understanding of ‘human capital’. Beyond its economicconnotation of investment on the part of an individual, human capital may moresubstantively reflect the inheritance afforded by one’s family background. In sum, thisstudy highlights the critical role of social resources in theories and studies in stratifica-tion and mobility.

Acknowledgements

The study was in part supported by a grant from the National Science FoundationŽ .INT-90-12727 to Nan Lin.

References

Alwin, D.F., Hauser, R.M., 1975. The decomposition of effects in path analysis. Am. Sociol. Rev. 40, 37–47.Bian, Y., Ang, S., 1997. Guanxi networks and job mobility in China and Singapore. Social Forces 75,

981–1005.Blau, P.M., Duncan, O.D., 1967. The American Occupational Structure. Wiley, New York.Boissevain, J.F., 1974. Friends of Friends. Basil Blackwell, Oxford.Boxman, E.A.W., De Graaf, P.M., Flap, H.D., 1991. The impact of social and human capital on the income

attainment of Dutch managers. Social Networks 13, 51–73.Boxman, E.A.W., Flap, H.D., 1990. Social capital and occupational chances. Paper presented at the

International Sociological Association, XIIth World Congress of Sociology, July, Madrid.Burt, R.S., 1992. Structural Holes: The Social Structure of Competition. Harvard Univ. Press, Cambridge.Campbell, K., Lee, B., 1991. Name generators in surveys of personal networks. Social Networks 13, 203–222.Campbell, K.E., Marsden, P.V., Hurlbert, J.S., 1986. Social resources and socioeconomic status. Social

Networks 8, 97–117.Ž .Coleman, J.S., 1988. Social capital in the creation of human capital. Am. J. Sociol. 94 Supplement ,

S95–S120.Coleman, J.S., 1990. Foundations of Social Theory. Harvard Univ. Press, Cambridge.

Page 20: Network resources, contact resources, and status attainment

( )G. Lai et al.rSocial Networks 20 1998 159–178178

De Graaf, N.D., Flap, H.D., 1988. With a little help from my friends. Social Forces 67, 452–472.Fischer, C., 1982. To Dwell Among Friends. University of California Press, Berkeley.Flap, H.D., De Graaf, N.D., 1986. Social capital and attained occupational status. Neth. J. Sociol. 22,

145–161.Flap et al., 1986.Flap et al., 1988.Granovetter, M., 1973. The strength of weak ties. Am. J. Sociol. 78, 1360–1380.Granovetter, M., 1974. Getting a Job: A Study of Contacts and Careers. Harvard Univ. Press, Cambridge.Granovetter, M., 1982. The strength of weak ties: a network theory revisited. In: Marsden, P.V., Lin, N.

Ž .Eds. , Social Structure and Network Analysis. Sage, Beverly Hills, pp. 105–130.Erickson, B.H., 1996. Culture, class, and connections. Am. J. Sociol. 102, 217–251.Heckman, J.L., 1980. Sample selection bias as a specification error with an application to the estimation of

Ž .labor supply functions. In: Smith, J.P. Ed. , Female Labor Supply: Theory and Estimation. PrincetonUniv. Press, Princeton, pp. 206–248.

Joreskog, K., Sorbom, D., 1993. LISREL 8. Scientific Software International, Chicago.Ž .Lin, N., 1982. Social resources and instrumental action. In: Marsden, P.V., Lin, N. Eds. , Social Structure and

Network Analysis. Sage, Beverly Hills, pp. 131–145.Lin, N., 1983. Social resources and social actions: a progress report. Connections 6, 10–16.Lin, N., 1990a. Social structure and social mobility: a structural theory of status attainment. In: Breiger, R.

Ž .Ed. , Social Mobility and Social Structure. Cambridge Univ. Press, Cambridge.Lin, N., 1990b. Social resources and the emergence of social structure. Paper presented at the XII World

Congress of Sociology, July, Madrid, Spain.Ž .Lin, N., Bian, Y., 1991. Social connections Guanxi and status attainment in urban China. Paper presented at

the Second European Conference on Social Network Analysis, June 10–22, Sorbonne, Paris.Lin, N., Dumin, M., 1986. Access to occupations through social ties. Social Networks 8, 365–385.

Ž .Lin, N., Ensel, W.M., Vaughn, J.C. Eds. , 1981. Social resources and strength of ties: structural factors inŽ .occupational status attainment. Am. Sociol. Rev. 46 August : 393-405.

Lin, N., Vaughn, J.C., Ensel, W.M., 1981b. Social resources and occupational status attainment. Social Forces9, 1163–1181.

Marsden, P.V., Campbell, K.E., 1984. Measuring tie strength. Social Forces 63, 482–501.Marsden, P.V., Hurlbert, J.S., 1988. Social resources and mobility outcomes: a replication and extension.

Social Forces 66, 1038–1059.Marsden et al., 1984.Marsden et al., 1988.Montgomery, J.D., 1992. Job search and network composition: implications of the strength-of-weak-ties

hypothesis. Am. Sociol. Rev. 57, 586–596.Smith, K.W., Sasaki, M.S., 1979. Decreasing multicollinearity: a method for models with multiplicative

functions. Sociol. Meth. Res. 8, 35–56.Sun, C.-S., Hsung, R.-M., 1988. Social Resources and Social Mobility. Tunghai University, Taiwan.Treiman, 1970.Wegener, B., 1990. Mobility and the structure of ties: interactions between social resources and human capital

in status attainment. Paper presented at the International Sociological Association, XIIth World Congressof Sociology, July, Madrid.

Wellman, B., 1979. The community question. Am. J. Sociol. 84, 1201–1231.Wellman, B., 1990. The place of kinfolk in community networks. Marriage and Family Rev. 15, 195–228.Wellman, B., Wortley, S., 1989. Brother’s keepers: situating kinship relations in broader networks of social

support. Sociol. Perspectives 32, 273–306.Wellman, B., Wortley, S., 1990. Different strokes from different folks: community ties and social support.

Am. J. Sociol. 96, 558–588.