Complex climate and network effects on internal migration in South Africa revealed by a network model

Climate variability and climate change influence human migration both directly and indirectly through a variety of channels that are controlled by individual and household socioeconomic, cultural, and psychological processes as well as public policies and network effects. Characterizing and predicting migration flows are thus extremely complex and challenging. Among the quantitative methods available for predicting such flows is the widely used gravity model that ignores the network autocorrelation among flows and thus may lead to biased estimation of the climate effects of interest. In this study, we use a network model, the additive and multiplicative effects model for network (AMEN), to investigate the effects of climate variability, migrant networks, and their interactions on South African internal migration. Our results indicate that prior migrant networks have a significant influence on migration and can modify the association between climate variability and migration flows. We also reveal an otherwise obscure difference in responses to these effects between migrants moving to urban and non-urban destinations. With different metrics, we discover diverse drought effects on these migrants; for example, the negative standardized precipitation index (SPI) with a timescale of 12 months affects the non-urban-oriented migrants’ destination choices more than the rainy season rainfall deficit or soil moisture do. Moreover, we find that socioeconomic factors such as the unemployment rate are more significant to urban-oriented migrants, while some unobserved factors, possibly including the abolition of apartheid policies, appear to be more important to non-urban-oriented migrants.

Market ranking and network structure: pathway to dominance

PurposeThe relationship between interlocking directorates and firm performance has been increasingly debated, with a focus on whether firm's centrality in interlock networks is associated with performance. The purpose of this study is to examine not only how a firm's position in this network is associated with performance but also how the performance of network partners can impact a firm's performance. This study examines how firms effectively utilise the interlock network to achieve the goal of higher market capitalisation – termed market capitalisation rank (MCR).Design/methodology/approachThe premise of the study is the UK FTSE 350 firms from 2014 to 2018. The paper makes use of a temporal network autocorrelation model to examine how firm characteristics, the structural position in the interlock network and the performance of network partners affect MCR over time.FindingsThe analysis indicates that firms with ties (via the interlock network) to firms with high market capitalisation are more likely to enhance their own MCR, highlighting network partners have the opportunity to play a critical role in a firm's dominance strategy to optimise firm value.Originality/valueThe value of this research is that it does not only look at the impact of a firm's position in the network on performance, but the impact of the performance of network partners on a firm's market performance as well.

Network Analysis of the Social Environment Relative to Preference for and Tolerance of Exercise Intensity in CrossFit Gyms

Known for its ability to improve fitness and health, high-intensity functional training (HIFT) focuses on functional movements completed at high intensities, often yielding outcomes superior to repetitive aerobic workouts. Preference for and tolerance of high-intensity exercise are associated with enjoyment of and adherence to HIFT. Similarly, the social environment present within CrossFit, a popular group-based HIFT modality, is important to the enjoyment of and adherence to HIFT. This study aimed to test whether preference and tolerance were related to social connections within CrossFit networks. Linear network autocorrelation models (LNAMs) and exponential random graph models (ERGMs) were computed on sociometric and attribute data from members of three CrossFit networks (n = 197). LNAMs showed the preference and tolerance scores of someone’s social connections were associated with their own in all three gyms, and ERGMs demonstrated preference and tolerance scores were associated with the presence of social ties within all networks. This study is the first to provide evidence for a relationship between social connections and preference and tolerance. Future longitudinal research is needed to determine if the social environment may influence and optimize a person’s preference of and tolerance for HIFT.

Network Autocorrelation of Perceived Physical Activity Skill Competence Among Adolescents at a Summer Care Program: A Pilot Study

Purpose: This study investigates the possible association between adolescent friendship networks and perceived physical activity skill competence in a summer care program. Design: Adolescents participated in researcher-administered surveys at the start (T1) and end (T2) of summer. Setting: Adolescents at a Boys & Girls Club were sampled. Sample: Adolescents (age 8-12) completed researcher-administered surveys at T1 (n = 100; µ age = 9.9 years; 47% male; 55% Black) and T2 (n = 77; µ age = 9.8 years; 51% male; 49% Black). Measures: Perceived skill competence was measured by asking adolescents to rate how good they felt they were at physical activity at the club. Adolescents were also asked to provide names of up to 5 peers whom they hung around with, talked to, and did things with the most while at the club. Analysis: Linear network autocorrelation models were used to determine network effects or clustering of perceived physical activity skill competence within the club. Results: There were significant network effects for adolescent perceived skill competency scores at T1 (β = 0.05, p < 0.01) and T2 (β = 0.05, p = 0.02), indicating adolescent perceived skill competence scores were associated with those of their friends. Conclusions: Practitioners may wish to encourage the use of group or collaborative skill competency improvement activities as well as possibly pairing adolescents with differing skill competencies to foster improvement and possible diffusion of perceived skill competency.

Network Autocorrelation Modeling: Bayesian Techniques for Estimating and Testing Multiple Network Autocorrelations

The network autocorrelation model has been the workhorse for estimating and testing the strength of theories of social influence in a network. In many network studies, different types of social influence are present simultaneously and can be modeled using various connectivity matrices. Often, researchers have expectations about the order of strength of these different influence mechanisms. However, currently available methods cannot be applied to test a specific order of social influence in a network. In this article, the authors first present flexible Bayesian techniques for estimating network autocorrelation models with multiple network autocorrelation parameters. Second, they develop new Bayes factors that allow researchers to test hypotheses with order constraints on the network autocorrelation parameters in a direct manner. Concomitantly, the authors give efficient algorithms for sampling from the posterior distributions and for computing the Bayes factors. Simulation results suggest that frequentist properties of Bayesian estimators on the basis of noninformative priors for the network autocorrelation parameters are overall slightly superior to those based on maximum likelihood estimation. Furthermore, when testing statistical hypotheses, the Bayes factors show consistent behavior with evidence for a true data-generating hypothesis increasing with the sample size. Finally, the authors illustrate their methods using a data set from economic growth theory.

Using Social Network Analysis to Understand Trust, Reciprocity, and Togetherness in Wildlife Tourism Microentrepreneurship

High levels of trust, reciprocity, and togetherness embedded within entrepreneurial networks are believed to facilitate cooperation that enables success among individual business owners. This study examines the effects of social influence, network characteristics, and entrepreneurial motivations on trust, reciprocity, and togetherness in a network of wildlife tourism microentrepreneurs. Thirty-seven wildlife tourism microentrepreneurs from North Carolina’s Pamlico Sound Region were recruited for in-person structured interviews. Data were analyzed using social network analysis, specifically a series of linear network autocorrelation models in conjunction with supportive qualitative assessment. Microentrepreneurs expressing a high level of trust were connected with microentrepreneurs expressing a low level of trust in their peers. Conversely, microentrepreneurs with strong feelings of reciprocity were connected with microentrepreneurs having similar feelings. These findings illustrate that the presence of equally reciprocal relationships is not an indication of equally trusting relationships. The findings also suggest that higher numbers of business ties tend to diminish the levels of trust, reciprocity, and togetherness toward connected peers.

Estimating the Attributable Disease Burden and Effects of Interhospital Patient Sharing on Clostridium difficile Infections

Objective:To estimate the burden of Clostridium difficile infections (CDIs) due to interfacility patient sharing at regional and hospital levels.Design:Retrospective observational study.Methods:We used data from the Healthcare Cost and Utilization Project California State Inpatient Database (2005–2011) to identify 26,878,498 admissions and 532,925 patient transfers. We constructed a weighted, directed network among the hospitals by defining an edge between 2 hospitals to be the monthly average number of patients discharged from one hospital and admitted to another on the same day. We then used a network autocorrelation model to study the effect of the patient sharing network on the monthly average number of CDI cases per hospital, and we estimated the proportion of CDI cases attributable to the network.Results:We found that 13% (95% confidence interval [CI], 7.6%–18%) of CDI cases were due to diffusion through the patient-sharing network. The network autocorrelation parameter was estimated at 5.0 (95% CI, 3.0–6.9). An increase in the number of patients transferred into and/or an increased CDI rate at the hospitals from which those patients originated led to an increase in the number of CDIs in the receiving hospital.Conclusions:A minority but substantial burden of CDI infections are attributable to hospital transfers. A hospital’s infection control may thus be nontrivially influenced by its neighboring hospitals. This work adds to the growing body of evidence that intervention strategies designed to minimize HAIs should be done at the regional rather than local level.