Search

We discuss how sampling design, units, the observation mechanism and other basic statistical notions figure into modern network data analysis. These considerations pose several new challenges that cannot be adequately addressed by merely extending or generalizing classical methods. Such challenges stem from fundamental differences between the domains in which network data emerge and those for which classical tools were developed. By revisiting these basic statistical considerations, we suggest a framework in which to develop theory and methods for network analysis in a way that accounts for both conceptual and practical challenges of network science. We then discuss how some well-known model classes fit within this framework.

The new European Organic Regulation 2018/848 has announced the phasing out of derogations for the use of untreated non-organic seed by 2036. However, the use of organic seed by organic farmers is currently limited. This paper aims to identify the factors affecting the use of organic seed. It is based on data collected from 749 organic farmers in 20 European countries, by conducting an online survey and using a network sampling. Results of the descriptive statistics and linear mixed models indicate that: (1) the situation of organic seed use is not consistent across geographical regions and crop sectors, (2) the use of organic seed is higher on farms selling directly to consumers than on those selling to supermarkets, (3) larger and more recently converted farms use less organic seed than established organic farms. In the second part of the paper, we analyse farmers&rsquo, attitudes towards organic seed use. The structural equation model (SEM) suggests that the highest contribution to explaining intention to use organic seed comes from social norms, i.e., farmers&rsquo, perception of societal expectations, particularly from the consumer and the organic certifier. Such expectations, if communicated in the public and political discourse, could stimulate the use of organic seed.

CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior Redes temporais (ou dinâmicas) são frequentemente usadas para modelar conexões que ocorrem ao longo do tempo entre partes de um sistema por meio de nós e arestas. Nessas redes, todos os nós, arestas e instantes de tempo são conhecidos e estão disponíveis para serem utilizados na análise. Entretanto, em várias situações reais, dados são produzidos de forma massiva e contínua, o que é conhecido como fluxo contínuo de dados (FCD). Nesse tipo de aplicação, o volume de dados pode ser tão grande que o armazenamento deles pode ser impossível e as tarefas de mineração se tornam ainda mais desafiadoras. Em redes provenientes de FCD, arestas são continuamente adicionadas em distribuição não-estacionária. Tanto em redes temporais quanto em redes em FCD, padrões relacionados à atividade de nós e arestas são tipicamente irregulares ao longo do tempo, o que torna a visualização dessas redes útil para obter insights sobre a estrutura e dinâmica delas. Por outro lado, a distribuição não-estacionária aumenta a complexidade e torna a visualização de redes em FCD ainda mais desafiadora. Vários layouts visuais foram propostos até hoje, mas todos possuem limitações. O principal desafio é a quantidade de informação visual, que aumenta dependendo do tamanho e densidade da rede e causa poluição visual devido à sobreposição de arestas, resolução temporal e proximidade dos nós. Nesta tese, nós propomos métodos para melhorar a visualização de redes em FCD por meio da manipulação das três dimensões da rede: nó, aresta e tempo. Mais especificamente, nós propomos: (i) CNO, um método de ordenação de nós visualmente escalável; (ii) SEVis, um método de amostragem de arestas em FCD; (iii) um método para FCD que adapta a resolução temporal de acordo com níveis locais de atividade de nós. Também apresentamos um estudo comparativo considerando a combinação destes métodos. Por meio de estudos de caso com redes reais, mostramos que cada um dos métodos melhora bastante a legibilidade do layout, levando a uma tomada de decisão rápida e confiável. Temporal networks (also known as dynamic networks) are often used to model connections that occur over time between parts of a system by using nodes and edges. In temporal networks, all nodes, edges, and times, are known and available to be used in the analysis. However, in several real-world applications, data are produced in a massive and continuous way, which is known as data stream. In this case, the volume of data may be so large that the storage may be impossible and mining tasks become more challenging. In streaming temporal networks, edges are continuously arriving in non-stationary distribution. In both temporal and streaming temporal networks, patterns related to node and edge activity are typically irregular in time, which makes the visualization of such networks helpful to gain insights about network structure and dynamics. Nevertheless, the non-stationary distribution of incoming data increases complexity and turns the streaming temporal network visualization even more challenging. Several visualization layouts have been proposed, but they all have limitations. The main challenge in this context is the amount of visual information, that increases depending on the network size and density, and causes visual clutter due to edge overlap, fine temporal resolution, and node proximity. In this thesis, we propose methods to enhance the visualization of streaming temporal networks through the manipulation of the three network dimensions, namely node, edge, and time. Specifically, we propose: (i) CNO, a visual scalable node ordering method; (ii) SEVis, a streaming edge sampling method; and (iii) a streaming method that adapts the temporal resolution according to local levels of node activity. We also present a comparative study considering the combination of these methods. We show through case studies with real-world networks that each of these methods greatly improves layout readability, thus leading to a fast and reliable decision making. Tese (Doutorado)

We consider the use of sample survey data to estimate dyadic characteristics of family networks, with an application to noncoresident parent–child dyads. We suppose that survey respondents report either from a parent or child perspective about a dyad, depending on their membership of the dyad. We construct separate estimators of common dyadic characteristics using data from both a parent and a child perspective and show how comparisons of these estimators can shed light on data quality issues including differential missingness and reporting error. In our application we find that a simple missingness model explains some striking patterns of discrepancies between the estimators and consider the use of poststratification and a related redefinition of count variables to adjust for these discrepancies. We also develop approaches to combining the separate estimators efficiently to estimate means and frequency distributions within subpopulations.

Internet users are increasingly becoming self-publishing consumers. Applying the Uses and Gratifications (U&G) perspective, this study investigated how people use TikTok in terms of consuming, participating and producing behaviors, and examined the role of personality traits and users’ motivation as predictors to this integrated usage behavior. An online survey was conducted to recruit 385 TikTok users using online network sampling technique. Our findings suggest that it was users’ motivations, not personality traits, that have significant influence on TikTok use. Results show that users’ motivations – namely archiving, self-expression, social interaction and peeking – are significant predictors to TikTok usage behaviors but differ in levels and influence. This study contributes to both the theoretical and the empirical understanding of media use in a user-generated media (UGM) context.

There is a tendency to consider covert networks as separate from overt networks. Drawing on data from the Democratic Republic of the Congo, we demonstrate that this is not the case and identify how covert and overt networks are mutually constitutive. While most studies of African brokers have relied on network metaphors like Big Men and social membranes, we consider the embeddedness of covert networks in overt networks explicitly. We perform two analyses on a large original dataset encompassing 396 partially overlapping ego-nets obtained from a hybrid link-tracing design. An ego-net analysis reveals a large degree of homophily and a deep embeddedness of the different networks. A multilevel exponential random graph model fitted to the reconstructed network of a 110-node subset shows that demobilised combatants are the actors likely to broker between armed groups, state forces, and civilian blocs, suggesting their capacity to broker peace or foment war. Funding Agencies|UK Aid from the UK Department for International Development (DFID); Centre for Public Authority and International Development at LSE; ESRC GCRFUK Research & Innovation (UKRI)Economic & Social Research Council (ESRC) [ES/P008038/1]; Collecting and analyzing secondary covert social network data project - Leverhulme TrustLeverhulme Trust [RPG-2013-140]; DFID-ESRC Poverty Alleviation ProgrammeUK Research & Innovation (UKRI)Economic & Social Research Council (ESRC)

Agent-based modeling holds great potential as an analytical tool. Agent-based models (ABMs) are, however, also vulnerable to critique, as they often employ stylized social worlds, with little connection to the actual environment in question. Given these concerns, there has been a recent call to more fully incorporate empirical data into ABMs. This article falls in this tradition, exploring the benefits of using sampled ego network data in ABMs of cultural diffusion. Thus, instead of relying on full network data, which can be difficult and costly to collect, or no empirical network data, which is convenient but not empirically grounded, we offer a middle-ground, one combining ABMs with recent work on network sampling. The main question is whether this approach is effective. We provide a test of the approach using six complete networks; the test also includes a range of diffusion models (where actors follow different rules of adoption). For each network, we take a random ego network sample and use that sample to infer the full network structure. We then run a diffusion model through the known, complete networks, as well as the inferred networks, and compare the results. The results, on the whole, are quite strong: Across all analyses, the diffusion curves based on the sampled data are very similar to the curves based on the true, complete network. This suggests that ego network sampling can, in fact, offer a practical means of incorporating empirical data into an agent-based model.

Respondent-driven sampling is a commonly used method for sampling from hard-to-reach human populations connected by an underlying social network of relations. Beginning with a convenience sample, participants pass coupons to invite their contacts to join the sample. Although the method is often effective at attaining large and varied samples, its reliance on convenience samples, social network contacts, and participant decisions makes it subject to a large number of statistical concerns. This article reviews inferential methods available for data collected by respondent-driven sampling.

Social network analysis is now used widely to study social behaviour in humans and non-human animals, and missing individuals can represent a problem for network studies. This problem is becoming especially frequent in studies using bio-logging to collect interaction data, which is an approach used particularly frequently in the construction of animal networks. This therefore represents an important audience for Smith et al. (2017) who investigate how sub-sampling from networks impacts the outcome of subsequent analysis. Here I take advantage of the progress made by this paper to outline key issues that still require addressing to understand the effect of missing individuals on social network analysis.

Around the year 2000, the problem of reconciling the estimate of loans advanced by the banks and the estimate of loans incurred by the rural farmers was studied in the Indian Statistical Institute. Some approximately unbiased estimates were examined along with approximately unbiased estimates of their approximate variances. Utilizing “Constrained Network” sampling technique exactly unbiased counterparts are presented as alternatives.

In this paper, I discuss some of the wider uses of adaptive and network sampling designs. Three uses of sampling designs are to select units from a population to make inferences about population values, to select units to use in an experiment, and to distribute interventions to benefit a population. The most useful approaches for inference from adaptively selected samples are design-based methods and Bayesian methods. Adaptive link-tracing network sampling methods are important for sampling populations that are otherwise hard to reach. Sampling in changing populations involves temporal network or spatial sampling design processes with units selected both into and out of the sample over time. Averaging or smoothing fast-moving versions of these designs provides simple estimates of network-related characteristics. The effectiveness of intervention programs to benefit populations depends a great deal on the sampling and assignment designs used in spreading the intervention.

Both random walk and respondent-driven sampling (RDS) exploit social networks and may reduce biases introduced by earlier methods for sampling from hidden populations. Although RDS has become much more widely used by social researchers than random walk (RW), there has been little discussion of the tradeoffs in choosing RDS over RW. This paper compares experiences of implementing RW and RDS to recruit drug users to a network-based study in Houston, Texas. Both recruitment methods were implemented over comparable periods of time, with the same population, by the same research staff. RDS methods recruited more participants with less strain on staff. However, participants recruited through RW were more forthcoming than RDS participants in helping to recruit members of their social networks. Findings indicate that, dependent upon study goals, researchers' choice of design may influence participant recruitment, participant commitment, and impact on staff, factors that may in turn affect overall study success.

Sampling graphs is an important task in data mining. In this paper, we describe Little Ball of Fur a Python library that includes more than twenty graph sampling algorithms. Our goal is to make node, edge, and exploration-based network sampling techniques accessible to a large number of professionals, researchers, and students in a single streamlined framework. We created this framework with a focus on a coherent application public interface which has a convenient design, generic input data requirements, and reasonable baseline settings of algorithms. Here we overview these design foundations of the framework in detail with illustrative code snippets. We show the practical usability of the library by estimating various global statistics of social networks and web graphs. Experiments demonstrate that Little Ball of Fur can speed up node and whole graph embedding techniques considerably with mildly deteriorating the predictive value of distilled features., Comment: Code is available here: https://github.com/benedekrozemberczki/littleballoffur

In a social network individuals or nodes connect to other nodes by choosing one of the channels of communication at a time to re-establish the existing social links. Since available data sets are usually restricted to a limited number of channels or layers, these autonomous decision making processes by the nodes constitute the sampling of a multiplex network leading to just one (though very important) example of sampling bias caused by the behavior of the nodes. We develop a general setting to get insight and understand the class of network sampling models, where the probability of sampling a link in the original network depends on the attributes $h$ of its adjacent nodes. Assuming that the nodal attributes are independently drawn from an arbitrary distribution $\rho(h)$ and that the sampling probability $r(h_i , h_j)$ for a link $ij$ of nodal attributes $h_i$ and $h_j$ is also arbitrary, we derive exact analytic expressions of the sampled network for such network characteristics as the degree distribution, degree correlation, and clustering spectrum. The properties of the sampled network turn out to be sums of quantities for the original network topology weighted by the factors stemming from the sampling. Based on our analysis, we find that the sampled network may have sampling-induced network properties that are absent in the original network, which implies the potential risk of a naive generalization of the results of the sample to the entire original network. We also consider the case, when neighboring nodes have correlated attributes to show how to generalize our formalism for such sampling bias and we get good agreement between the analytic results and the numerical simulations., Comment: 11 pages, 5 figures

Traversing, searching and sampling approaches have been investigated very much in graph theory and, then, applied in a large variety of research fields. However, in bioinformatics and biomedicine, sampling complex networks is a new and little investigated task. For instance, it is used to classify knowledge for creating semantic maps, summarizations and multi-label classifications, or to search motifs. In this article, first we introduce a formalism to represent complex networks. Then, we provide both three taxonomies of network sampling approaches and a brief overview to each of them. After this, we present a comparison of these approaches. Finally, we draw some conclusions and take a look at the future.

Due to the rapid development of various social networks, the spatial autoregressive (SAR) model is becoming an important tool in social network analysis. However, major bottlenecks remain in analyzing large-scale networks (e.g., Facebook has over 700 million active users), including computational scalability, estimation consistency, and proper network sampling. To address these challenges, we propose a novel least squares estimator (LSE) for analyzing large sparse networks based on the SAR model. Computationally, the LSE is linear in the network size, making it scalable to analysis of huge networks. In theory, the LSE is $\sqrt{n}$-consistent and asymptotically normal under certain regularity conditions. A new LSE-based network sampling technique is further developed, which can automatically adjust autocorrelation between sampled and unsampled units and hence guarantee valid statistical inferences. Moreover, we generalize the LSE approach for the classical SAR model to more complex networks associated with multiple sources of social interaction effect. Numerical results for simulated and real data are presented to illustrate performance of the LSE.

Online social media contain valuable quantitative and qualitative data, necessary to advance complex social systems studies. However, these data vaults are often behind a wall: the owners of the media sites dictate what, when, and how much data can be collected via a mandatory interface (called Application Program Interface: API). To work with such restrictions, network scientists have designed sampling methods, which do not require a full crawl of the data to obtain a representative picture of the underlying social network. However, such sampling methods are usually evaluated only on one dimension: what strategy allowsfor the extraction of a sample whose statistical properties are closest to the original network? In this paper we go beyond this view, by creating a benchmark that tests the performance of a method in a multifaceted way. When evaluating a network sampling algorithm, we take into account the API policies and thebudget a researcher has to explore the network. By doing so, we show that some methods which are considered to perform poorly actually can perform well with tighter budgets, or with different API policies. Our results show that the decision of which sampling algorithm to use is not monodimensional. It is not enough to ask which method returns the most accurate sample, one has also to consider through which API constraints it has to go, and how much it can spend on the crawl.

Respondent-driven sampling (RDS) is a method for sampling from a target population by leveraging social connections. RDS is invaluable to the study of hard-to-reach populations. However, RDS is costly and can be infeasible. RDS is infeasible when RDS point estimators have small effective sample sizes (large design effects) or when RDS interval estimators have large confidence intervals relative to estimates obtained in previous studies or poor coverage. As a result, researchers need tools to assess whether or not estimation of certain characteristics of interest for specific populations is feasible in advance. In this paper, we develop a simulation-based framework for using pilot data-in the form of a convenience sample of aggregated, egocentric data and estimates of subpopulation sizes within the target population-to assess whether or not RDS is feasible for estimating characteristics of a target population. in doing so, we assume that more is known about egos than alters in the pilot data, which is often the case with aggregated, egocentric data in practice. We build on existing methods for estimating the structure of social networks from aggregated, egocentric sample data and estimates of subpopulation sizes within the target population. We apply this framework to assess the feasibility of estimating the proportion male, proportion bisexual, proportion depressed and proportion infected with HIV/AIDS within three spatially distinct target populations of older lesbian, gay and bisexual adults using pilot data from the caring and Aging with Pride Study and the Gallup Daily Tracking Survey. We conclude that using an RDS sample of 300 subjects is infeasible for estimating the proportion male, but feasible for estimating the proportion bisexual, proportion depressed and proportion infected with HIV/AIDS in all three target populations.

Objetivo. La presente investigación tuvo como objetivo identificar los factores de riesgo que influyen en la violencia contra la mujer en el municipio de Ventaquemada (Boyacá). Asimismo, determinar los tipos de violencia ejercidos hacia este género. Método. La metodología utilizada es el enfoque cualitativo descriptivo. Se tomó una muestra de quince mujeres en situación de violencia, pertenecientes a las veredas Boquerón, El Carmen y Frutillo, seleccionadas mediante un muestreo por redes. Las técnicas aplicadas fueron la encuesta social y entrevista semiestructurada. Resultados. A partir de una lectura desde trabajo social, se estudió la teoría ecológica de los sistemas de Uriel Bronfenbrenner, en la que se determinó que una de las causas relacionadas es la de los factores individuales, relacionales y sociales, en los que se ubica el déficit económico, la alta influencia del medio y la baja autonomía, liderazgo y participación, lo que da paso a que esta población se enfrente a dinámicas de violencia física, económica, verbal, psicológica o sexual, que ocasionan daños y cambios en el ciclo vital de la mujer. Conclusiones. La normalización de la violencia contra la mujer en las veredas Boquerón, El Carmen y Frutillo del municipio de Ventaquemada se determina como uno de los factores de riesgo más relevantes. Objetive. The objective of this research was to identify the risk factors that influence violence against women in the municipality of Ventaquemada (Boyacá). Also, determine the types of violence exercised towards this gender. Method. The methodology used is the descriptive qualitative approach. A sample was taken of fifteen women in a situation of violence, belonging to the villages of Boquerón, El Carmen and Frutillo, selected through network sampling. Result. The techniques applied were the social survey and semi-structured interview. From a reading from social work, the ecological theory of the systems of Uriel Bronfenbrenner was studied, in which it was determined that one of the related causes is that of the individual, relational and social factors, in which the deficit is located economic, the high influence of the environment and low autonomy, leadership and participation, which gives way to this population is faced with dynamics of physical, economic, verbal, psychological or sexual violence, which cause damage and changes in the life cycle of the woman. Conclusions. The normalization of violence against women in the villages of Boquerón, El Carmen and Frutillo in the municipality of Ventaquemada is determined to be one of the most relevant risk factors.

We study the relaxation time in the random walk with jumps. The random walk with jumps combines random walk based sampling with uniform node sampling and improves the performance of network analysis and learning tasks. We derive various conditions under which the relaxation time decreases with the introduction of jumps., 13 pages