Peter V. Marsden

Handbook of survey research

With chapters on: sampling; measurement; questionnaire construction and question writing; survey implementation and management; survey data analysis; special types of surveys; and integrating surveys with other data collection methods, this title includes topics such as measurement models, the role of cognitive psychology, and surveying networks.

Social structure and network analysis

Network analysis is being increasingly looked to as a means of understanding social structure. It can shed light on how individual actions create social structure, how social structure constrains the individual, and how attitudes and behaviour are determined by social structure. Articles by leading proponents of network analysis and structuralism examine how these methodological techniques and this theoretical approach can be applied to a variety of social phenomena. Written by some of the leading proponents of network analysis, this book will be welcomed by professionals in sociology and their students.

Network studies of social influence

Network analysts interested in social influence examine the social foundations for influence—the social relations that provide a basis for the alteration of an attitude or behavior by one network actor in response to another. This article contrasts two empirical accounts of social influence (structural cohesion and equivalence) and describes the social processes (e.g., identification, competition, and authority) presumed to undergird them. It then reviews mathematical models of influence processes involving networks and related statistical models used in data analysis. Particular attention is given to the “network effects” model. A number of empirical studies of social influence are reviewed. The article concludes by identifying several problems of specification, research design, and measurement and suggesting some research that would help to resolve these problems.

Homogeneity in confiding relations

Abstract Patterns of inbreeding (or homophily) and social distance in data from the 1985 General Social Survey on dyads discussing important matters are examined. Stratifying variables include age, education, race/ethnicity, religion, and sex. Discussion relations are most constrained by race/ etnicity, and least by sex and education. Inbreeding effects are present for all five stratifying variables, and account for virtually all structure in dyads classified by race/ethnicity and religions Appreciable social distance biases in the formation of these strong ties are found for age and education, but not for other stratifying variables. The analysis illustrates the use of loginear and log-multiplicative association models in the analysis of cross-classifications of attributes of alters and respondents.

SOCIAL RESOURCES AND SOCIOECONOMIC STATUS

We address two questions central to the “network as resources” argument, using network data from two mass surveys. First, how is range best measured? We identify six dimensions of range: one each reflecting network size and complexity, and two each representing density and diversity. Second, what is the nature of the relationship between SES and social resources? Evidence here supports the proposition that network range and composition are positively related to an actors socioeconomic status.

Egocentric and sociocentric measures of network centrality

Abstract Egocentric centrality measures (for data on a node’s first-order zone) parallel to Freeman’s [Social Networks 1 (1979) 215] centrality measures for complete (sociocentric) network data are considered. Degree-based centrality is in principle identical for egocentric and sociocentric network data. A closeness measure is uninformative for egocentric data, since all geodesic distances from ego to other nodes in the first-order zone are 1 by definition. The extent to which egocentric and sociocentric versions of Freeman’s betweenness centrality measure correspond is explored empirically. Across seventeen diverse networks, that correspondence is found to be relatively close—though variations in egocentric network composition do lead to some notable differences in egocentric and sociocentric betweennness. The findings suggest that research design has a relatively modest impact on assessing the relative betweenness of nodes, and that a betweenness measure based on egocentric network data could be a reliable substitute for Freeman’s betweenness measure when it is not practical to collect complete network data. However, differences in the research methods used in sociocentric and egocentric studies could lead to additional differences in the respective betweenness centrality measures.

Social Networks, Job Changes, and Recruitment

This chapter reviews scholarship on how the matching of people to jobs is influenced by networks of interpersonal ties. By all accounts, that role is substantial on both the individual’s side and the employer’s side of the labor market. The mediation of job change and recruitment/selection processes by networks illustrates the embeddedness of labor market processes in ongoing structures of social relations (Granovetter 1985) with special clarity.

Gender Differences in Organizational Commitment Influences of Work Positions and Family Roles

Data obtained from the 1991 “Work Organizations” module of the General Social Survey (GSS) reveal a small but significant tendency for employed men to display higher organizational commitment (OC) than employed women do. This article examines the gender differences and factors that arguably heighten or dampen it. The authors consider both job models highlighting gender differences on job attributes such as autonomy or rewards, and gender models that stress socialization, family ties, and differential labor market opportunities. They find that the primary explanation for the gender difference is that men are more likely than women to hold jobs with commitment-enhancing features. Gender differences in family ties do little to affect male-female OC difference. When job attributes, career variables, and family ties are simultaneously controlled, the authors find that, if anything, women tend to exhibit slightly greater OC. Contrary to implications of some gender models, the correlates of OC do not appear to be appreciably different for men and women.

Effects of a Quality Improvement Collaborative on the Outcome of Care of Patients with HIV Infection: The EQHIV Study

Context Multi-institutional quality improvement collaboratives are popular, but are they effective? Contribution This controlled study evaluated an HIV care Breakthrough Series program that emphasized provider teams, sessions on quality improvement theory and techniques, and report backs about implementing quality improvement. A review of the medical records of 9986 HIV-infected patients showed no important differences in quality of care (viral load suppression, pneumocystitis prophylaxis, and screening for tuberculosis and hepatitis) between the 44 intervention clinics and the 25 control clinics. Cautions Patient adherence and satisfaction were not measured, and some control clinics may have used quality improvement techniques similar to those recommended by the collaborative program. The Editors In the pastdecade, tremendous improvements have occurred in measuring and monitoring the quality of medical care in the United States. Despite these advances, striking problems with quality persist (1, 2). The quality of care for patients with HIV infection is of particular concern. Substantial evidence shows that obtaining medical services and treatment for patients with HIV infection may lead to longer survival and better quality of life (3, 4), yet serious quality-of-care problems and striking disparities in quality by race and social class have been documented (4-6). In the 1980s, continuous quality improvement techniques were introduced into health care (7, 8). These strategies emphasize that most quality problems are a result of system failings rather than problems with individual practitioners (9). In 1995, the Institute for Healthcare Improvement introduced the concept of the Breakthrough Series, which brings together health care organizations dedicated to improving the quality of care in particular clinical areas through the application of continuous quality improvement techniques (10). These techniques (known as Plan/Do/Study/Act or PDSA cycles; Figure 1) first identify deficiencies in quality, next repeatedly implement small-scale interventions and measure changes, and then refine and expand interventions to improve processes of care (11, 12). Typically, each Breakthrough Series collaborative is composed of 20 to 40 participating health care organizations and a faculty with expertise in the clinical area and quality improvement methods (13). To date, the Institute for Healthcare Improvement has conducted collaboratives with more than 700 teams working on 23 clinical conditions or treatment processes, including improving asthma care and reducing medication errors. Although some evaluations of quality before and after a collaborative support the validity of this approach, only a few limited controlled trials have been conducted (14, 15). Figure 1. Theoretical construct of continuous quality improvement. An important source of funding for HIV care is the Ryan White Comprehensive AIDS Resources Emergency (CARE) Act, which is administered by the HIV/AIDS Bureau of the Health Resources and Services Administration. Title III of the CARE Act supports comprehensive primary health care for HIV-infected individuals and currently supports primary care services for more than 150000 patients receiving care in more than 200 community health centers, hospital-based clinics, and city or county health services (16). In 1999, the Health Resources and Services Administration required all clinical sites that were newly awarded funding under Title III of the CARE Act to participate in a quality improvement collaborative conducted by the Institute for Healthcare Improvement. Other sites already receiving Title III funding were also invited to participate. This study evaluates the impact of the collaborative by examining pre- and postimplementation quality-of-care information on samples of patients from both participating and matched nonparticipating clinics. Methods Study Site Selection and Controls Of the 200 Title III sites in the continental United States in May 2000, we excluded 16 that reported caseloads of fewer than 100 cases per year, 12 that were initially slated to participate in the Breakthrough collaborative but elected not to do so, and 1 that lost its CARE Act funding shortly before the collaborative began. Of the remaining 171 sites, 62 participated in the collaborative. Among these sites, 54 agreed to participate in the study and 44 (including 11 mandatory participants and 33 voluntary participants; 71% of collaborative participants) provided chart review data. Of the 109 nonparticipating sites eligible to be selected as control sites, 65 provided information needed for matching. The potential control sites were matched with intervention sites on the basis of the type of site (community health center, community-based organization, health department, hospital, or university medical center), location (rural or urban), number of locations delivering care, region, and number of patients with active HIV infection. When these criteria were used, 40 sites were selected as potential controls and 37 of them (93%) agreed to participate in the study. Of these, 25 (63% of potential control sites) participated in the chart review portion of the study. The Committee on Human Studies of Harvard Medical School approved the study protocol. Quality Improvement Intervention Each participating clinic selected a team, usually consisting of at least one administrator and one or more clinicians, and a population of focus on which the teams interventions would be tested. Usually, the population of focus consisted of all HIV-infected patients in a particular site, but participants sometimes chose to focus on a subset of patients, such as those under the care of a particular group of clinicians. Originally, the design of the collaborative extended for 12 months and included a kickoff meeting and 2 subsequent 2-day meetings called learning sessions. The kickoff learning session included instruction in the theory and practice of quality improvement by identifying problems in HIV care and then introducing the techniques of continuously implementing, measuring, and refining changes (the Plan/Do/Study/Act cycles) (11, 12) to improve the care of HIV-infected patients. Each learning session included additional instruction in quality improvement techniques and breakout sessions that focused on improving specific aspects of care, developing an information infrastructure to track progress, and specific aspects of quality improvement theory. In addition, teams exchanged ideas and presented storyboards of their progress to date. At each session, teams reported on activities, methods, and results. Toward the end of the 12-month period, the Health Resources and Services Administration decided to extend the collaborative by 4 months and add a third learning session. Between the sessions (action periods), team members implemented concepts and ideas. Each site had access to a collaborative listserve, participated in monthly conference calls with the collaborative faculty, and submitted monthly reports of its improvements, which included charts that tracked the sites improvements to date in the required key quality measures described in the next section. Detailed descriptions of the Breakthrough Series collaboratives are available elsewhere (10, 17-19). Quality-of-Care Monitors We selected quality-of-care measures (Table 1) to coincide with required and optional quality measures selected by the collaborative faculty as areas for improvement. These measures were selected by the faculty after reviewing the literature to identify areas of quality deficiency in the delivery of HIV care, particularly for underserved populations targeted by the CARE Act. Because of the paramount importance of antiretroviral therapy to the treatment of HIV infection, the faculty focused on measures related to antiretroviral treatment, including the percentage of patients receiving highly active antiretroviral therapy, the percentage of patients with a controlled viral load, and the percentage of patients who received adherence counseling, as required key measures for the collaborative. Measures were then developed on the basis of consensus guidelines appropriate for the period of care (20). Our primary measures were rates of highly active antiretroviral therapy use and control of HIV viral load for appropriate patients. Patients eligible for highly active antiretroviral therapy included those with CD4+cell counts less than 0.350 109 cells/L, those with CD4+counts between 0.350 and 0.500 109 cells/L and a viral load greater than 5000 copies/mL, all patients with a viral load greater than 30000 copies/mL, and patients already receiving highly active antiretroviral therapy, as per the guidelines. We also assessed the use of highly active antiretroviral therapy for those with CD4+counts less than 0.350 109 cells/L to reflect recommendations that were published after the end of the collaborative (21). Because of the variability in viral load assays available at the time, viral load was considered controlled if it was undetectable or if the total viral load was less than 400 copies/mL. We also assessed the use of screening and prophylaxis, as well as access to care. The only key measure followed by the collaborative that we could not assess was related to adherence counseling because this information is not reliably available from medical records. Table 1. Quality of Care Indicators* Quality-of-Care Data Collection To identify pre- and postintervention samples of patients, we requested lists of all HIV-infected patients in care at each of the sites during the 2 time periods (Figure 2). For the first sample, sites were asked to provide encrypted lists of all HIV-infected patients age 18 years or older as of June 2000 seen at the site between 1 January and 30 June 2000. For the second sample, sites were asked to provide a similar list of active patients age 18 years or older as of Dec

Externalizing employment: Flexible staffing arrangements in US organizations

Abstract Flexible staffing arrangements (such as temporary, contract, and part-time work) enable organizations to externalize administrative control or limit the duration of employment. We examine the prevalence and correlates of such arrangements using a recent large, representative survey of US establishments. We first develop a typology of flexible staffing arrangements and discuss reasons why organizations may adopt them. We then present measures of these flexible staffing arrangements and describe their distribution among US establishments. Finally, we examine hypotheses about the types of employers that are more or less likely to use the various types of flexible staffing arrangements, finding support for both cost-reduction and resource dependence perspectives. The use of flexible arrangements is more common in large establishments, in seasonal industries, and in establishments with highly female workforces.