Chapter 8

Towards Agency–Structure Integration: A

Person-in-Environment (PIE) Framework

for Modelling Individual-Level Information

Behaviours and Outcomes

Sei-Ching Joanna Sin

Abstract

This chapter introduces the Person-in-Environment (PIE) framework, aresearch design and a nationwide empirical study, developed by theauthor, to measure the relative impacts of socio-structural and personalfactors on individual-level information behaviours (IB) and outcomes.The IB field needs to tackle two questions: (1) In a particular situation,how much of an individual’s IB is influenced by personal characteristics?and (2) How much of this behaviour is shaped by one’s environment,such as socio-structural barriers? PIE is a beginning effort to addressthis agency–structure debate, which is a topic that confronts manysocial scientists. This chapter first outlines IB research relevant toagency–structure integration. It then presents six principles of the PIEframework. Personal characteristics (e.g. cognitive and affective factors)and socio-structural factors (e.g. information resources distribution) areconceptualised as interrelated. Thus, these need to be tested simulta-neously. Previously, it was difficult to link individual- and societal-leveldatasets because their units of observation often vary. To overcomethese methodological challenges, this author purposed a research designthat employs secondary analysis, geographic information systems tech-niques and structural equation modelling. An empirical study of the

New Directions in Information Behaviour

Library and Information Science, 181–209

Copyright r 2011 by Emerald Group Publishing Limited

All rights of reproduction in any form reserved

ISSN: 1876-0562/doi:10.1108/S1876-0562(2011)002011a011

182 Sei-Ching Joanna Sin

library usage by 13,000 American 12th graders is presented todemonstrate PIE’s applicability. Discussions on the future directionsof PIE studies conclude the chapter. The PIE framework can contributeto conceptual and methodological development in IB research. It alsooffers scholars and policymakers a way to empirically assess the contri-butions of information services on an individual’s life, while takingpersonal differences into account.

8.1. Introduction

The new millennium has seen much advancement in information commu-nication technologies. In developing and developed countries alike, however,many less-advantaged groups still suffer barriers in accessing informationresources (FreshMinds & Department for Communities and Local Govern-ment, 2008; International Telecommunication Union, 2009; Livingston, 2010).Such an information gap is not limited to the use of digital resources. Varyingaccess to print resources is still prevalent (Duke, 2000; Jue, Koontz,Magpantay, Lance, & Seidl, 1999; Neuman & Celano, 2001; Sin, 2011). Bystudying the causes, processes and outcomes of information practices, theinformation behaviour (IB) field has long contributed to facilitating infor-mation access and use. The extent to which individual-level IB is affected bystructural barriers, however, has seldom been tested. At this juncture, theauthor sees a fruitful but underexplored frontier: agency–structure integrationin IB research.

Social structure is the most basic, enduring and orderly pattern in sociallife. Individual agency refers to the capabilities of an individual to actindependently of the constraints exerted by the social structure (Abercrom-bie, Hill, & Turner, 1994; Calhoun, 2002). In a particular situation, howmuch of an individual’s behaviours and outcomes is influenced by personalcharacteristics? And, how much of these behaviours and outcomes is shapedby one’s environment (e.g. socio-structural factors)? Socio-structural factorsare defined here as factors that are (1) related to the basic, recurring patternof the society where an individual lives; and (2) beyond the individual’simmediate control. This is the classic agency–structure debate that confrontsmany social scientists (Ritzer & Gindoff, 1994). Similarly, in the IB field, wehave yet to fully measure the relative impacts of socio-structural contextsand personal characteristics on individual behaviour. Agency–structureintegration (i.e. the inclusion of both individual- and societal-level factors inone research design) is one way towards addressing these research questions.

Towards Agency–Structure Integration 183

Spink and Cole (2006) illustrate that efforts towards a more integrativeapproach are beneficial to a field’s progress. Similarly, the need for integratingindividual and structural factors is a result of the advancements in the IB field.In the 1980s, there was a shift from a system-centred approach to a user-centred paradigm that emphasises individual agency (Dervin & Nilan, 1986).Since the late 1990s, Information Seeking in Context (ISIC) research hasbrought contextual factors to the forefront (Vakkari, Savolainen, & Dervin,1997). As the field has embraced a holistic view that recognises both indivi-duality and contextual influences, a pressing issue is to differentiate the impactof individual factors from socio-structural factors. There are theoretical andmethodological challenges in achieving agency–structure integration. How dowe study structural factors while keeping the individual as our research focus?How do we recognise the interplay between individual agency and societalinfluences while avoiding the pitfalls of determinism? It is a challenge todevelop and apply frameworks that integrate individual and socio-structuralfactors. This is due to difficulties in operationalising abstract concepts, and inidentifying a suitable research design (Courtright, 2007; Vakkari, 1997). Moreefforts towards developing this research area will be beneficial to expandingthe IB field.

Distinguishing the impact of individual- and societal-level factors alsohas strong policy implications. There is a need to pinpoint the exact natureand extent of information barriers (e.g. individual-level perceptual andaffective barriers vs. societal-level disparities in information provision). Suchknowledge is crucial to formulating suitable policies. For example, tacklingindividual perceptual barriers towards information sources may call fortarget marketing or information literacy training. Deficiency in informationservice provision, on the other hand, requires improvement in service type,level and quality. In addition, there is increasing demand for informationinstitutions such as public libraries to demonstrate their contribution toindividual lives (Durrance & Fisher, 2003). This is especially important intimes of funding shortages. Managers of information services often wonder,‘If we build it, will they come?’ By testing both individual and structuralfactors, the impacts of funding and service levels on individual behaviourcan be assessed while taking personal differences into account.

The time is ripe for our field to untangle the relationships amongindividual agency, structural influences, individual IB and outcomes of suchbehaviour. The author developed and tested a Person-in-Environment (PIE)framework as a first step towards agency–structure integration (Sin, 2009).This chapter aims to provide an overview of the PIE framework. Thechapter first outlines the framework’s background and principles. It thendescribes an accompanying research design that employs secondary analysis

184 Sei-Ching Joanna Sin

of survey data, geographic information systems (GIS) for data linking andstructural equation modelling (SEM) for analysis. An empirical study of13,000 American 12th graders’ library usage is presented to demonstratePIE’s applicability. Discussions on the future directions of PIE studiesconclude the chapter.

8.2. Literature Review

8.2.1. Agency–Structure Debate

Many social science disciplines are interested in the agency–structure issue.Some scholars focus extensively on individual agency (e.g. symbolicinteractionism, ethnomethodology and phenomenology). Others focus onstudying social structure (e.g. structural functionalism). Sociologists refer tothis as the agency–structure or micro–macro debate (or, occasionally, asthe subjectivism–objectivism or individualism–holism debate) (Ritzer &Gindoff, 1994). Psychologists refer to it as the person–situation or person–environment debate (Hogan & Roberts, 2000). There have been attempts todraw attention to both individual and structural factors. These includeecological perspectives such as Roger Barker’s (1968) ecological psychologyand Urie Bronfenbrenner’s (1979) ecology of human development, andalso grand social theories such as Anthony Gidden’s structuration theoryand Pierre Bourdieu’s habitus. In IB research, the user-centred paradigmrecognises the power of individual agency. It emphasises users’ individuality,and views human IB as situational. Dervin and Nilan (1986) characterisedthis new paradigm: ‘It sees users as beings who are constantly constructing,as beings who are free (within system constraints) to create from systemsand situations whatever they choose’ (p. 16). This characterisation toucheson both individual agency and system constraints.

Around the mid-1990s, there was growing agreement that more attentionneeded to be cast on factors beyond the individual user. This crystallisedin interest in the context of IB. The biennial ISIC conference, first started in1996, is particularly prominent in aiding this development. The editors ofthe first ISIC conference proceedings declared, ‘Our strong presuppositionis that a wide variety of contextual considerations (e.g., communitiesand organisations with their structures and cultures) constitute frames ofreference for the IB in individuals’ (Vakkari et al., 1997, p. 8). From thestandpoint of achieving agency–structure integration, this emphasis oncontext is encouraging. With these developments in the IB field, we nowhave gained a better understanding of individual demographics, cognitiveand affective factors, work and everyday life IB, purposive and accidental

Towards Agency–Structure Integration 185

information acquisition and contextual factors such as task and situation(e.g. Allen, 1997; Ellis, Cox, & Hall, 1993; Erdelez, 1997; Fisher & Julien,2009; Kim & Allen, 2002; Kuhlthau, 1991; Leckie, Pettigrew, & Sylvain,1996; Savolainen, 1995).

Perhaps as a reaction to the previous paradigm, which focused more onformal information systems, it has been observed that most studies since theparadigm shift have focused on the individual level (Courtright, 2007;Solomon, 2002; Talja, Keso, & Pietilainen, 1999). The nature and extent ofsocio-structural influences on individual IB have not drawn as muchexamination (Audunson, 1999; Vakkari, 1997). At the moment, the field iswrestling with what context actually constitutes, how to integrate contextualfactors in a coherent conceptual framework and the research methodssuitable for such research.

8.2.2. Socio-Structural Factors in Information Behaviour Research

The following sections focus on select IB research that identifies socio-structural factors and their relationship with individual behaviour. Thechallenges in integrating individual and socio-structural factors will beexamined. For discussion on a broader range of IB paradigms, theories,models, methods, and studies, readers are referred to Case (2007), Court-right (2007), Fisher, Erdelez, and McKechnie (2005), Fisher and Julien(2009), Nahl and Bilal (2007), Spink and Cole (2006), to name a few.

Socio-structural factors are recognised in several IB models. As early asthe 1960s, William Paisley proposed a conceptual framework aimed atthe study of scientists and their work-related IB. Paisley’s frameworkis noteworthy because it identifies many layers of contextual factors. Themodel includes eight systems arranged in concentric circles. The outermostlayer is the cultural system, followed by political system, membership group(e.g. a national association of the scientist’s discipline), reference group (e.g.colleagues from the same specialty), invisible college, formal organisationand work team. At the centre of the circles is the user (including factorsconcerning the user’s perception, cognition, affects and motivation). Twosystems – the legal/economic system and the formal information system –are seen as cutting across the aforementioned eight layers. Paisley (1967)stressed the need for understanding how these various systems influenceusers’ IB.

Wilson’s (1981) model included personal, interpersonal and environ-mental factors (including work, socio-cultural, political-economic andphysical environments) as influential to information needs and informationseeking. In discussing the impact of the environment on work, for example,

186 Sei-Ching Joanna Sin

Wilson identified issues such as the differential stratification of resources.The model also more explicitly recognised the interconnected relationshipsamong all factors (Wilson, 1981). Paisley and Wilson’s nested models arepromising in terms of integrating individual- and structural-level factors.Perhaps due to the models’ comprehensiveness (i.e. many factors need to bemeasured and analysed), empirical IB studies that fully apply theseframeworks have yet to be conducted.

Conceptualisation issues are partly behind this dearth of empiricalagency–structure research. The interest in context has broadened the scopeof IB studies. Further development is hindered by the concept’s ambiguity,however. It is unclear what the term context actually constitutes. In herreview, Dervin (1997) called the term ‘the unruly beast’, and stated that‘There is no term that is more often used, less often defined, and whendefined, defined so variously as context’ (p. 14). Furthermore, she found that‘Virtually every possible attribute of person, culture, situation, behaviour,organisation, or structure has been defined as context’ (p. 14). The blurringof information contexts is highlighted in recent discussion among IBscholars (Burnett & Erdelez, 2010). Currently, context is still a nebulousconcept. Among the types of contextual factors investigated in individual-level IB studies, socio-structural factors are seldom tested empirically ascentral factors. In terms of achieving agency–structure integration, it will bebeneficial to explicitly declare structural factors to be a core category ofcontextual influences.

The field has more success in conceptualising one category of contextualfactors –– the information environment. Zweizig (1973) viewed the individualas being surrounded by an array of information resources. He advocatedfocusing on the library in the life of the user. Scholars have since developedconcepts such as information horizon (Sonnenwald, 1999; Sonnenwald,Wildemuth, & Harmon, 2001), information field and information pathway(Johnson, 1996; Johnson, Case, Andrews, Allard, & Johnson, 2006). Theseconcepts have been applied to study what sources are considered, preferred orused for a particular task, or to identify the sequence of source selection byindividuals. There is room for building upon these constructs and treatingthem as factors that shape individual behaviour. That is, one can begin to usethese concepts of the information environment as explanatory variables. Theycan be used to elucidate variations in individual IB.

Robert Taylor’s (1991) Information Use Environment (IUE) is worthhighlighting; it identifies the information environment as a factor thatinfluences people’s behaviour. The framework includes four elements: sets ofpeople, problems, settings and resolution of problems. Of particular interest isthe inclusion of settings, which cover the nature, structure and attributes ofthe setting, as well as the availability and accessibility of information resources.The model, however, focuses on groups of people –– not individuals. As such,

Towards Agency–Structure Integration 187

individual-level demographics such as age and gender are less consequential inIUE. It also focuses mostly on professionals and entrepreneurs in the workenvironment. Taylor has expressed a hope that the framework would be usedin studies of the general public. Several scholars have indeed used IUE in thestudy of everyday-life IB. Agada (1999) applied IUE to study the IB of 20gatekeepers in Harambee, an inner-city neighbourhood in Milwaukee.Hersberger, Murray, & Sokoloff (2006) used IUE in their study of the IB of15 abused and neglected children. In these studies, the research focus is alsomore on the behaviour of the group rather than individual-level differences.

Fisher’s information ground also pays special attention to the environ-ment in which information is exchanged. The concept is used to examinesocial settings (e.g. community clinics) that foster spontaneous informationsharing (Fisher, Durrance, & Hinton, 2004; Pettigrew, 1999). The charac-teristics of information ground are identified and grouped as being people-related, place-related and information-related. Place-related characteristicsinclude location and permanence, privacy and ambient noises, to name a few(Fisher, Landry, & Naumer, 2007). As the concept is designed to explore atemporal setting, socio-structural factors are not a central component ofinformation ground. Nevertheless, the inclusion of place-related character-istics is encouraging. It helps underline that, in addition to personalcharacteristics, the information environment can also influence behaviour ineveryday life.

Chatman’s concept of small worlds is particularly relevant to the study ofeveryday-life IB. The concept is based on sociological theories and wasdeveloped through exemplary ethnographical research of aging women,janitors and women in maximum-security prison (Chatman, 1991, 1992, 1996,1999). In this framework, individuals in a small world are seen as sharing acultural space, within which members share similar worldviews and socialnorms. These shared norms and expectations shape the boundary ofappropriate behaviour, and thus have the potential to shape the members’IB as well. Because of the focus on marginalised groups, this concept isof strong interest to researchers of structural inequities. Chatman alsointroduced the idea of ‘social types’ to help study sub-groups. On the whole,the interest remains in understanding the worldviews of the group and theimpact of these worldviews on IB. Individual-level variation among group orsub-group members was not the focus of the framework.

For most socio-structural factors other than the information environ-ment, operationalisation –– that is, the process of defining an abstractconcept into concrete, measurable indicators –– often poses a problem.Savolainen’s (1995) study revealed this challenge. In the seminal paper thatpopularised everyday life information seeking (ELIS), Savolainen built onPierre Bourdieu’s concept of habitus. This study is particularly relevantto agency–structure integration. This is because it explicitly recognises

188 Sei-Ching Joanna Sin

structural issues such as class, which are often less directly addressed inother frameworks. The study included middle-class and working-classparticipants (more specifically, teachers and industrial workers). AsSavolainen discussed, the concepts of habitus, way of life and mastery oflife are very abstract. They are difficult to define. Savolainen pinpointedthese difficulties in the conclusion section:

The concepts with large extensions and heterogeneous inten-tions are problematic in that their exact operationalization isdifficult; thus it may not be easy to specify which parts of ELISare really determined by way of life and which would beexplained better by other factors, such as current situation oflife or the degree of difficulty of the problem being encountered.(p. 289)

Not only are structural factors difficult to operationalise but many factorsare also interrelated. Early models in many disciplines tend to conceptualisesocio-structural factors as shaping individual behaviours, and not the otherway round. Since the recognition of individuality, it has been highlighted thatthe individual interacts with, and shapes, the environment as well. Suchbidirectional relations are enshrined in the person-in-situation model proposedand tested by Allen (1997) and Allen and Kim (2001). The model is based onthe theoretical foundation of interactionism. It highlights that individualdifferences and contextual factors act concurrently to influence individualbehaviour. The interaction could be interpreted as a result of a person–situation fit, in which an individual with certain characteristics might fit betterin certain types of situations than others, leading to differences in behaviour.Another interpretation is that individuals could be flexible and change theirbehaviour in order to adapt to different situations.

The person-in-situation model focuses on immediate situations, such asdifferent academic tasks, rather than on the socio-structural focus of thischapter. However, the model is still very relevant to the current discussion.Allen and Kim’s papers expound the methodological implications ofinteractionism –– that is, in order to test for interactions, individual andcontextual factors should be analysed together and not separately. Thismeans that a more complex research design is required.

Another barrier to agency–structure integration is the lack of large-scalestudies. As mentioned above, many individual and contextual factors affectindividual behaviour. These factors are often interconnected. Qualitativemethods excel in revealing the complexity of human IB. However, in aqualitative study, it is more difficult to disentangle and compare theinfluence of different factors (Holstein & Gubrium, 2004). In addition, the

Towards Agency–Structure Integration 189

sample size tends to be small in qualitative research. Large-scale studies ofthe public’s IB, such as those conducted in the 1980s (e.g. Chen & Burger,1985; Chen & Hernon, 1982; Dervin, 1984), are not as common nowadays.Most studies focus on a particular group or a particular geographicallocation. Such a focus is a suitable choice for many IB studies. For researchersinterested in uncovering structural inequality and measuring its impact,however, it is beneficial to include a broader scope or use a larger sample size.

If a study surveyed individuals residing in a homogenous environment,the range of variability in the environment would be limited. It could behard to discern the effects of environmental factors on individual beha-viours. The influences of the structural environment may be underestimated(Duncan & Raudenbush, 1999). A possible remedy is to conduct comp-arative studies. For example, a researcher may compare similar groups livingin different environments. It is observed that, while comparative studiesin organisational settings are often found, they are not as common in theELIS setting (Courtright, 2007). There is much room for conceptual andmethodological improvements towards agency–structure integration in IBresearch. As Hjørland (1997) stated, ‘It has been difficult to reach a synthesisthat would put individual information needs, query formulations, searchbehaviour, and so on, in a sociological perspective. The connecting linkbetween the psychological and the sociological levels has been missing’ (p. 120).

8.3. Person-in-Environment: Conceptual Framework

To address the aforementioned gaps, the author has developed and tested aPIE framework (Sin, 2009). The following sections present the six centralprinciples of the PIE framework.

Principle 1. The Individual is the Centre of PIE Research. The Recom-mended Unit of Analysis is at the Individual Level

This principle is based on the tenet of the user-centred paradigm, whichrecognises individuality. Individual demographic, cognitive and affectivefactors are central to the PIE framework. When the research question callsfor it, PIE can be used for analysing aggregated data about groups. In mostcases, the goal is to measure and analyse personal characteristics, behavioursand outcomes at the individual level.

Principle 2. Environmental Factors as Explanatory Variables

Principle 2 reflects the attention to contextual factors, as advocatedin research on IB in context. Environment is defined as ‘the totality of

190 Sei-Ching Joanna Sin

circumstances surrounding an organism or group of organisms’ (Environ-ment, 2000). Thus, for the PIE framework, this term includes more than thephysical environment; it covers the cultural, social, economic, political andinformation environments as well. Individuals may have more influences oncertain types of environmental factors (e.g. information environment athome) than on other environmental factors (e.g. unemployment rate in aresidential neighbourhood). To tackle the agency–structure debate and toexplore the effects of unequal resource distributions, special attention isgiven to socio-structural factors that are beyond an individual’s immediatecontrol. This focus should not lead to the exclusion of other relevantenvironmental factors. It is important to note that, in the PIE framework,environmental factors are not used simply for background discussionpurposes. Environmental factors should be measured and tested asindependent or mediating variables.

Principle 3. The Information Environment is a Crucial Category of Exp-lanatory Variables

This principle draws on the field’s strong interest in the informationenvironment discussed above. The information environment is seen as acrucial category. This is because of the consistent observations of uneveninformation resource distributions among and within countries. Informationenvironment factors are not treated only as outcome variables but also asexplanatory factors of individual IB. Again, the information environmentshould be empirically measured and tested. Aspects of interest include thetype, quality, quantity, accessibility and relevance of information resourcesand services. By giving this component centre stage, the impacts of unequalresource distribution on individuals can be analysed.

Principle 4. Interactionism Among Individual and Environmental Fac-tors, Not Determinism

Individual actions may be facilitated or constrained by environmentalfactors. Human actions may in turn alter the environment. The PIEframework emphasises that individual and environmental factors interact.These factors can jointly influence individual behaviour and outcomes. Thisis similar to the view of Allen and Kim’s person-in-situation approach,discussed above. This principle has two strong implications for researchdesign: (1) factors are often not independent (thus, they are better treated asexplanatory variables); and (2) factors need to be analysed simultaneously.Multivariate analysis methods are recommended. The person-in-situationapproach focuses on the interaction effects among variables. The PIE

Towards Agency–Structure Integration 191

framework, on the other hand, focuses on the direct and indirect pathwaysamong variables.

Principle 5. Flexible Unit of Observation for Environmental Factors.Neighbourhood Level Is Recommended

Depending on the research focus, environmental factors can be measuredat different scales (e.g. city and state). The author recommends using asrefined a unit as possible. The goal is to measure an individual’s immediateenvironment, within which the individual goes about his or her daily life.This helps detect variations that would otherwise not be observable if datawere aggregated at a distal level (Smith, 1979). The desire to use a veryrefined geographical unit is sometimes hindered by data availability. Fornational research, most studies aim for the neighbourhood level (e.g.neighbourhood effects research). In the United States, for example, zipcode or census tract areas are often used (Sampson, Morenoff & Gannon-Rowley, 2002).

Principle 6. Inclusion of Both Emic and Etic Measures

The PIE framework includes an individual’s view of the environment (i.e.the emic perspective) and measures provided by external parties such asresearchers (i.e. the etic perspective). The IB literature has discussed thesubjective nature of an individual’s knowledge of the world. The value of anemic perspective is recognised. Comparatively, etic measures have receivedless attention in recent years. This may be because the etic perspective issometimes viewed as being too close to positivism or the system-centredapproach. Precisely because IB embraces a subjective epistemology (e.g. socialconstructivism), this author believes that we should include etic measures. Ifeveryone had the same view, emic and etic measures would be the same, andmeasuring both would be unnecessary. But views do differ. We need tocapture a range of perspectives: those from the users and those from thirdparties. Incorporating etic measures is especially useful in studies ofinformation equity. This is because individuals may not be aware of theexistence, or the extent of, the inequities they are experiencing. Including eticmeasures helps better identify disparities in resource distribution. This alsogives researchers the opportunity to map the differences between the emic andetic perspectives. Further research can then be designed to examine why suchperceptual differences exist. Including etic measures would not privilege theoutsider’s view over an individual’s view. The multivariate aspect of PIEprevents such an assumption. How each view affects individual behaviour in aparticular context is empirically tested and not assumed.

192 Sei-Ching Joanna Sin

8.4. Person-in-Environment: Research Methods

The influences of individual and structural factors can be examined usingdifferent methods. This chapter focuses on quantitative methods and large-scale analysis, but the PIE framework need not be limited to the methodsdiscussed below. The current focus is chosen because these methods ––secondary analysis, GIS techniques and SEM –– are rarer in IB research.This chapter aims to illustrate their utility.

8.4.1. Data Collection: Secondary Analysis

The high monetary, personnel and time costs involved in data collection areamajor barrier in conducting large-scale representative studies. This authorrecommends taking advantage of the many datasets offered by governmentand other reputable sources, which often provide representative samples.A limitation of such secondary analysis, however, is that an existing datasetmay not contain all the variables of interest to a researcher. Scholarsoften have to link different datasets, or combine primary data withsecondary datasets. This data-linking problem can be exacerbated whenattempting agency–structure integration. Datasets with an individual focusand those with a structural focus often have a different unit of observation(i.e. the entity that is being observed and about which data is beingcollected). For example, a dataset about student behaviour may havethe individual student as the unit of observation. Data on public libraryfunding levels, on the other hand, may have a library system or a state as theunit of observation. This makes it difficult to merge different datasets.

Researchers may also aggregate the data of each dataset until there is acommon unit by which to merge two datasets (e.g. merging at the county orstate level). However, findings from aggregated data such as state-levelanalysis cannot be generalised to individuals. Doing so would be committingecological fallacy, which is an erroneous interpretation in which character-istics or associations found based on aggregated data for a group aremistakenly thought to be found for each individual belonging to that group.As emphasised in Principle 1, it is most beneficial to retain individual-leveldata whenever possible. When measuring an individual’s environment,Principle 5 recommends measuring the individual’s immediate environmentusing as refined a geographical unit as possible. Finding existing data abouta small geographical unit can be difficult. Statistics in many datasets areoften already aggregated into a larger unit. Different datasets seldom have ashared data element readily available for linking the datasets in a sufficientlyrefined manner. In an empirical study of U.S. students, the author used GISto tackle these data-linking issues.

Towards Agency–Structure Integration 193

8.4.2. Data Linking: Geographic Information Systems

This author proposes linking diverse datasets with GIS. Spatial location canbe used as the key for merging. GIS are systems used for ‘capturing, storing,checking, integrating, manipulating, analysing and displaying data that arespatially referenced to the Earth’ (University of Edinburgh & Associationfor Geographic Information, 1996). GIS techniques can be used on datasetsthat include geospatial information (e.g. when a dataset includes the addressof a facility or the zip code, city or county where an individual resides).Researchers can use commercial GIS software such as ESRI’s ArcGIS, orfree/open-source programs such as GRASS GIS and Quantum GIS, toperform data linking. To link different datasets, the data first need to bemapped digitally. This can be done through processes such as geocoding, aprocedure that identifies the latitude and longitude of a place based on itsstreet address. Digital maps of administrative boundaries or census areas areoften available online without charges. Once all the datasets are mapped, themaps can be overlaid on top of each other and merged spatially using GISoperations. For example, one can identify which census block a publiclibrary is located in, or estimate the characteristics of a neighbourhood. GIScan also compute spatial measures such as density and distance. An exampleof GIS data linking can be found in an analysis of U.S. public libraryfunding and service levels. Data from the Public Libraries Survey (PLS) wasspatially merged with U.S. census tract data using ArcGIS (Sin, 2011).

8.4.3. Data Analysis: Structural Equation Modelling

Data from a PIE study can be analysed using various multivariate methods.This chapter explores the use of SEM. SEM consists of two parts: (1) themeasurement model and (2) the structural model. An important characteristicof SEM is its focus on latent variables. Latent variables are the variables thatare not directly measured. They are measured indirectly through indicators(i.e. observed variables). For example, a researcher may represent the abstractconcept of socio-economic status (SES) using the observed variables ofeducation level, occupation and household income level. The measurementmodel involves specifying what indicators are used to represent each latentvariable. The analysis of the measurement model is a variant of confirmatoryfactor analysis (CFA). It helps evaluate measurement issues such as validityand reliability. The structural model involves specifying the relationships (e.g.associations and feedback loop) among the latent variables (Kline, 2005). Du(2009) offers an overview of SEM and its application in LIS research. SEMsoftware includes SSI’sLISREL, IBMSPSS’sAMOS,andMuthen&Muthen’s

194 Sei-Ching Joanna Sin

Mplus. SEM can also be conducted with the free statistical softwareR using thesem package (Fox, 2006).

SEM is especially suitable for the PIE framework. In SEM, explanatoryvariables are not assumed to be independent. This fits Principle 4 of PIE,which recognises the interaction among individual and environmentalfactors. SEM can test the relationships among different variables of interest.The pathways among variables can be represented visually through a pathdiagram (for an example, see Figure 8.2). In addition, measurement errorsare explicitly modelled. This fits the view of a subjective epistemology, asdiscussed in Principle 6. SEM acknowledges that observed variables are onlyapproximations of the abstract constructs one wants to measure. Adrawback of SEM is that it requires a large sample size. As a bareminimum, at least 100 subjects are needed (Kline, 2005). This author electsto discuss secondary analysis in the data collection section above, in partbecause secondary datasets often have a large number of respondents whichcomplements SEM well.

8.5. Empirical Study

To test the applicability of the PIE framework and the research methodsoutlined above, the author conducted an empirical study of American12th graders’ library usage. The PIE framework is applied to model thoseindividual and structural factors influencing 12th graders’ frequency of usingthe library for schoolwork, non-schoolwork and Internet access. Thefollowing sections provide an overview of this PIE application study.Selected findings are highlighted. Readers are referred to Sin (2009) for detailsof the research.

8.5.1. Research Design

The empirical study followed the six PIE principles. It centred on individualstudents and their IB (in this case, the frequency of students’ public libraryusage). The unit of analysis was individual (Principle 1). This research useddata from the Education Longitudinal Study (ELS) collected by theNational Center for Education Statistics (NCES). The analysis included anationally representative sample of more than 13,000 American high schoolstudents.

Figure 8.1 presents the study’s initial conceptual model. The variables canbe broadly grouped into three categories: (1) personal characteristics, (2)neighbourhood socio-economic environments and (3) information environ-ments. These variables were derived through reviews of public library

Figure 8.1: Initial conceptual model.

TowardsAgency–Stru

cture

Integ

ratio

n195

196 Sei-Ching Joanna Sin

literature (Agosto, Paone, & Ipock, 2007; D’Elia, 1980; D’Elia, Abbas,Bishop, Jacobs, & Rodger, 2007; Kronus, 1973; Rees & Paisley, 1968; Sin &Kim, 2008; Whitmire, 2002; Zweizig & Dervin, 1977). The variables’reliability and validity were tested through the SEM’s measurement model.

In the final measurement model, there were 21 variables. The personalcharacteristics category included the student’s SES, ethnicity, sex, achieve-ment motivation, learning activities, reading, level of social participationand perception of the school library. The student’s usage of school librariesfor schoolwork, non-schoolwork, and Internet access were also included asmediating factors. For environmental variables (Principle 2), the studyincluded the characteristics of each student’s residential neighbourhood ––namely its income and urbanisation levels. The unit of observation of theseenvironmental factors was at the neighbourhood level (Principle 5).Residential neighbourhood was defined as the zip code area where thestudent resided at the time of the survey. In terms of the informationenvironment (Principle 3), the study examined the student’s home printresources, home computer resources, school information environment andresource levels at the student’s neighbourhood public library. This studyrecognised the interactions among individual and environmental factors(Principle 4). Among the 18 explanatory variables, 7 of them served asmediating factors.

The study employed secondary analysis. When measuring an individual’senvironment, etic measures were included (Principle 6). Neighbourhoodcharacteristics were based on data collected by the U.S. Census Bureau.Statistics about neighbourhood libraries were drawn from the PLS. PLS is anationwide survey previously collected by NCES and now collected by theInstitute of Museum and Library Services (IMLS). In total, this researchused three datasets: ELS, PLS, and the 2000 U.S. Census. These datasetshave different units of observation (i.e. individual student, public librarysystem and census tract). This hindered data linking with a regular databaseprogram. With GIS techniques such as those discussed earlier, this authorused ArcGIS to link the data. Public libraries were mapped using the libraryaddresses listed in PLS. The public library layer was overlaid with the zipcode and the census tract layers. Measures of distance and density were thenconducted. With these spatial joins, the author can identify which libraryoutlets were in a student’s residential neighbourhood. The characteristics ofa neighbourhood were interpolated based on the area weighting method(Goodchild & Lam, 1980).

SEM was conducted with SSI’s LISREL. The study went throughiterations of model specification and testing. This analysis included non-continuous variables such as ethnicity. Thus, the program PRELIS was usedto pre-process the data. SEM models were estimated using the weighted-least squares (WLS) method.

Figure 8.2: Final structural model (direct effects).

Towards Agency–Structure Integration 197

8.5.2. Study Findings

The final structural model is presented in Figure 8.2.1 The path width in thediagram is proportional to the standardised structural coefficient (b). bIndicates the amount of change in the outcome variable with a unit change

1. To avoid overcrowding, the covariances/correlations among exogenous variables (i.e.

variables that are not predicted by any other variables in the model) and the disturbances/

measurement errors are not shown in the diagram.

198 Sei-Ching Joanna Sin

in the explanatory variable. It can be used to compare the relative effects ofvariables. The model fits and variance explained (R2) can be found in Tables8.2 and 8.3, presented later.

The analysis reveals a prevalent disparity in students’ informationenvironments. At the neighbourhood scale, public library resource andservice levels varied substantially with neighbourhood income levels(b ¼ 0.34) and urbanisation levels (b ¼ 0.37). Public libraries in higher-income or urbanised neighbourhoods had higher resource levels than theircounterparts. This coincides with the findings of Sin (2011). There were alsonotable individual-level disparities in the availability of print and digitalresources. Home, school and neighbourhood public library resource levelsvaried with SES, race/ethnicity and gender.

Other factors being constant, students with lower SES tended to havefewer print and digital resources at home. They were also slightly more likelyto be attending a school that had fewer information resources. Inaddition, these students were more likely to reside in lower-income or ruralneighbourhoods, which tended to have public libraries with lower levels ofresources. The study finds that ceteris paribus, ethnic minorities and femalestudents had fewer computer resources at home than their counterparts.Such disparate access to information resources shows that there is a strongneed for funding and initiatives to overcome the information barriers stillexperienced by different groups.

Equally important, the study shows that committing more resources tolibraries did make a difference in individual behaviour. The analysisdemonstrates that library usage was not solely a matter of individualdisposition. The school information environment, frequency of schoollibrary use and race/ethnicity were the top three factors affecting students’public library use (Table 8.1). That is, environmental factors werestatistically and substantially significant. Even after individual differences(such as SES and achievement motivation) were taken into account, higherresource levels at public libraries encouraged more frequent use of publiclibraries for schoolwork (b ¼ 0.12), non-schoolwork (b ¼ 0.14), andInternet access (b ¼ 0.12). It is worth discussing that the direct effect ofSES on public library usage was not significant. Lower-SES individuals wereoften found to be library non-users or infrequent users. The PIE frameworkand SEM analysis allow exploration of the factors behind such infrequentusage. They distinguish the various pathways between SES and library use(Figure 8.2).

The model finds that personal characteristics did play a role, but theywere not the only factors. For example, high SES was correlated with highparticipation in social activities (r ¼ 0.3), which was in turn associated withhigher library usage for schoolwork (b ¼ 0.3). At the same time, socio-structural factors such as neighbourhood library resource levels also made a

Table 8.1: Top factors affecting the frequency of students’ public libraryusage.

Outcome variables: Frequency of public library usage

For schoolwork For non-schoolwork For Internet access

Rank Explanatory

variables

Coef. Explanatory

variables

Coef. Explanatory

variables

Coef.

1 School

information

environment

�0.44 School library

usage for non-

schoolwork

�0.38 School library

usage for

Internet access

�0.45

2 School library

usage for

schoolwork

�0.37 Ethnicity

(Caucasian)

�0.30 Home computer �0.26

3 Ethnicity

(Caucasian)

�0.30 School

information

environment

�0.26 Ethnicity

(Caucasian)

�0.21

4 Social

participation

0.30 Reading outside

of school

0.26 Perception of

school library

0.18

5 Perception of

school library

0.29 Perception of

school library

0.23 Reading outside

of school

0.16

Towards Agency–Structure Integration 199

difference. In this study, the infrequent use of public libraries by lower-SESindividuals was partly explained by the following:

Lower SES - higher probability of living in low-income neighbourhood -lower public library resource levels- lower individual-level library usage.

What’s more, the PIE framework is capable of showing the counterforces at play. For example, some lower-SES students attended schools withfewer information resources, which could contribute to more frequent publiclibrary use for schoolwork (b ¼ �0.44), non-schoolwork (b ¼ �0.38) andInternet access (b ¼ �0.45). With the PIE framework and method, we canbegin to tease out the complex picture behind individual IB.

8.6. Discussion

This empirical study was conducted to test whether the PIE framework andthe research design are applicable. It also evaluates whether PIE is worthusing –– that is, can PIE help further our understanding of IB? The result is

200 Sei-Ching Joanna Sin

favourable. The findings show that PIE can be successfully applied to studyindividual-level behaviour. It also highlights the benefits of using thisframework. Table 8.2 shows the overall fit of the final structural model.

Table 8.3 shows the variance explained (R2) of two structural models.Model A is the final structural model selected for the empirical study. Itincluded socio-structural factors. This final model accounted for 35% ofvariations in students’ public library usage for schoolwork, 25% of publiclibrary usage for non-schoolwork and 21% of public library usage forInternet access. An alternative model, Model B, included only individual-level factors. Without the socio-structural factors, the R2 of Model Bdropped notably. The drop in explained variance is as large as 20% for thevariable of public library usage for schoolwork. The findings establish thateven after controlling for individual differences, structural conditions had asignificant impact on individual behaviours. If structural factors were notincluded (e.g. Model B), the researcher would miss several significant factors(e.g. school or public library environments). It is thus paramount for IBresearch to move towards agency–structure integration, such that poten-tially influential socio-structural factors are not left out of the analysis.

Table 8.2: Overall fit of the final structural model.

Index Recommended level Final structural model

AGFI W0.9 0.98NNFI W0.9 0.96RMESA r0.05 0.037ECVI Model with smaller ECVI is preferred 1.26AIC Model with smaller AIC is preferred 16,595.32

Table 8.3: Variance explained (R2) of the two structural models.

Model Frequency of public library usage

For

schoolwork

For non-school

work

For Internet

access

Model A(Includes individual andsocio-structural factors)

0.35 0.25 0.21

Model B(Includes individual-levelfactors only)

0.15 0.15 0.16

Towards Agency–Structure Integration 201

In summary, the benefits of the PIE framework are as follows: (1) socio-structural factors such as disparities in information resources distribution,important but hitherto underexamined aspects, are fully incorporated inPIE; (2) a multivariate method is used and thus the relative impact ofindividual and structural factors can be compared; (3) pathways amongexplanatory variables are modelled. The sometimes contradicting forcesbehind individual behaviour can be hypothesised and tested and (4) byintegrating both individual and socio-structural factors, PIE can offer higherexplanatory power than testing individual-level factors alone. This PIEstudy has information policy implications. It shows that one cannot dismisslower levels of information use as simply a matter of personal disposition.Structural inequity in information resources distribution can depress anindividual’s library usage. This provides evidence to policymakers thatdevoting more resources to schools and library systems in disadvantagedneighbourhoods can be beneficial.

The findings have theoretical implications which underscore the need forholistic approaches. Measuring only individual-level factors or onlyinformation environment factors, for example, would paint an incompletepicture of IB. The IB field is enriched by frameworks and theories from awide range of disciplines (Fisher et al., 2005), and continual investigation ofvarious approaches is essential to the field’s development. At the same time,further steps are needed to capitalise on this rich and diverse body ofliterature. These include a review of the landscape and evolution of the IBfield (Bates, 2005) and development of holistic approaches that integratethese wide-ranging frameworks (Spink & Cole, 2006). While these areno simple undertakings, they are essential to understanding the complextapestry that is IB.

8.7. Further Research

The study presented above explores factors influencing library usage. ThePIE framework, however, is not limited to studying a specific informationsource or channel. PIE is equally applicable to study, for example, the use ofthe Internet, Web 2.0, mobile communication or journal databases. The PIEframework is versatile, as it is not tied to a specific theory. Researchers arefree to test their own theories, models or hypotheses. PIE underscores theneed to include a wider range of variable categories that are hitherto lessexamined (e.g. socio-structural factors, information environments and eticmeasures as explanatory variables). The exact choice of constructs andindicators rests with each researcher. The inclusion of socio-structuralfactors alongside individual factors is a distinctive feature of PIE. The

202 Sei-Ching Joanna Sin

framework is especially useful when studying individuals who are residing inheterogeneous environments. Given the variations in social and informationenvironments across nations, using PIE for cross-cultural research is worthexploring.

In the empirical study of 12th graders’ library usage, structural factorsfocused on socio-economic aspects. This is because these factors have beenfound to exert a direct impact on the functioning of public libraries(Williams, 1980; Wilson, 1938). Future PIE studies could include more typesof explanatory factors. First, researchers may investigate subtler aspects ofsociety, such as societal norms and values. Governmental and institutionalpolicies are another category worth considering. Compared to socio-economic variables, norms and policy factors are more difficult toconceptualise and operationalise. More efforts in developing measurementinstruments will be helpful. Researchers may draw from other disciplines,such as anthropology, management, political science, social psychology orsociology. Historical and critical research in library and information science(Garrison, 2003; Pawley, 1998, 2006; Robbins, 2000; Wiegand, 1989) canalso shed light on how the norms and operations of LIS institutionsperpetuate or challenge social structures. Literature in these areas maysuggest salient societal and institutional factors for inclusion in future IBresearch.

Future PIE studies can include an individual’s social networks asexplanatory variables. People often turn to family, friends and colleagues forinformation. Individuals in different social positions may have varying typesof social networks and levels of social capital. Social network analysis(SNA), which focuses on relational data such as the connections and tiesamong individuals and groups, offers a promising research avenue. SNA isnot new to library and information science researchers. It is often used inbibliometric research and citation analysis (Otte & Rousseau, 2002). There isrising interest in SNA, particularly in the study of computer-mediatedcommunication. The utility of SNA in IB research has been identified(Haythornthwaite, 1996; Sonnenwald, et al., 2001). Still, empirical IBstudies using SNA are fewer than expected. This may be partly due to thefact that gathering social network data from participants can mean a moreintensive data collection process. This author is interested in SNA because,in addition to visually mapping one’s social network, SNA offers manyquantitative measures of network structure such as its density and centrality(Scott, 1994). These measures can be calculated using free software such asPajek. Such quantitative measures can readily be included as observedvariables in PIE studies.

In terms of data analysis, this chapter focused on a quantitative method,SEM. It must be highlighted that this does not preclude the use of

Towards Agency–Structure Integration 203

qualitative data collection methods. In fact, a mixed-method approach isfavourable — and encouraged. Researchers can conduct field studies toobserve both the environment and an individual’s interaction with theenvironment. Neuman and Celano (2001) offer an example of fieldwork andobservation of neighbourhood information environment. The value ofethnographic methods is clearly demonstrated in many IB studies (e.g.Chatman’s ethnographic work). Qualitative methods provide rich portraitsof individuals’ lives and environments. These primary field data can becoded systematically using content analysis (Krippendorff, 2004). The codeddata can then be included in a PIE study and analysed using multivariatemethods. Another fruitful avenue to pursue is to use PIE to model theoutcomes of IB. To illustrate, in the example study, the outcome variablesmeasure the frequency of public library usage. The author is conductingfurther research. The goal is to model the individual and structural factorson library usage, and to test how these in turn affect a student’s academicperformance. Because SEM does not assume variables to be independent,one can simultaneously model the pathways among individual character-istics, environmental factors, IB and the outcomes of such IB.

8.8. Conclusion

Structural factors are rarely included in individual-level IB research. Thisauthor developed the PIE framework as a first step towards agency–structure integration. To tackle the methodological problems in conductingsuch research, this chapter discussed the use of secondary analysis, GIStechniques and SEM. The empirical test on 12th graders’ library usagedemonstrates that such integration is both feasible and beneficial. PIE is abeginning effort to evaluate the intricate relationships among individualcharacteristics, socio-structural factors, individual behaviours and out-comes. With the PIE framework, researchers are not bound to individual-level factors or emic measures. We are also not limited to testing whetherfactors are related. One can theorise the direct and indirect paths betweenvariables and test the various forces at work. This paves the way towardsuncovering how individual agency and social structure shape individual IB.Untangling individual and structural influences is a complex venture; furtherresearch on various models and methods would benefit such an endeavour.It is hoped that the PIE framework will encourage more inclusion ofstructural factors in individual-level IB studies. The IB field has seen manyaccomplishments in the past decades. The author believes that our field hasa strong foundation to offer a variety of frameworks and research methodsfor bridging the agency–structure gap.

204 Sei-Ching Joanna Sin

Acknowledgements

This chapter is an extension of the author’s dissertation from the School ofLibrary and Information Studies, University of Wisconsin-Madison. Theauthor would like to thank her Dissertation Chair, Dr. Kyung-Sun Kim,and Dissertation Committee members, Drs. Louise Robbins, ChristinePawley, Ethelene Whitmire and Se-Kang Kim, for their invaluable guidance,insights and mentorship.

References

Abercrombie, N., Hill, S., & Turner, B. S. (1994). The Penguin dictionary of sociology

(3rd ed.). London: Penguin Books.

Agada, J. (1999). Inner-citygatekeepers:Anexploratorysurveyof their informationuse

environment. Journal of the American Society for Information Science, 50(1), 74–85.

Agosto, D. E., Paone, K. L., & Ipock, G. S. (2007). The female-friendly public

library: Gender differences in adolescents’ uses and perceptions of U.S. public

libraries. Library Trends, 56(2), 387–401.

Allen, B. (1997). Information needs: A person-in-situation approach. In: P. Vakkari,

R. Savolainen & B. Dervin (Eds), Information seeking in context (pp. 111–122).

London: Taylor Graham.

Allen, B., & Kim, K.-S. (2001). Person and context in information seeking:

Interactions between cognitive and task variables. The New Review of Information

Behaviour Research, 2, 1–16.

Audunson, R. (1999). Can institutional theory contribute to our understanding of

information seeking behaviour. In: T. D. Wilson & D. K. Allen (Eds), Exploring

the contexts of information behaviour (pp. 67–81). London: Taylor Graham.

Barker, R. G. (1968). Ecological psychology: Concepts and methods for studying the

environment of human behaviour. Stanford: Stanford University Press.

Bates, M. J. (2005). Information and knowledge: An evolutionary framework for

information science. Information Research, 10(4). Retrieved from http://Information

R.net/ir/10-4/paper239.html

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by

nature and design. Cambridge, MA: Harvard University Press.

Burnett, G., & Erdelez, S. (2010). Forecasting the next 10 years in information

behaviour research: A fish bowl dialogue. Bulletin of the American Society for

Information Science and Technology, 36(3), 44–48.

Calhoun, C. (Ed.) (2002). Dictionary of the social sciences. Oxford: Oxford University

Press.

Case, D. O. (2007). Looking for information: A survey of research on information

seeking, needs, and behaviour (2nd ed.). Amsterdam: Elsevier.

Chatman, E. A. (1991). Life in a small world: Applicability of gratification theory to

information-seeking behaviour. Journal of the American Society for Information

Science, 42(6), 438–449.

Towards Agency–Structure Integration 205

Chatman, E. A. (1992). The information world of retired women. New York, NY:

Greenwood Press.

Chatman, E. A. (1996). The impoverished life-world of outsiders. Journal of the

American Society for Information Science, 47(3), 193–206.

Chatman, E. A. (1999). A theory of life in the round. Journal of the American Society

for Information Science, 50(3), 207–217.

Chen, C.-C., & Burger, L. B. (1985). Assessment of Connecticut citizens’ information

needs and library use study: Final report. Hartford, CT: Connecticut State

Library.

Chen, C.-C., & Hernon, P. (1982). Information seeking: Assessing and anticipating

user needs. New York, NY: Neal-Schuman Publishers.

Courtright, C. (2007). Context in information behaviour research. Annual Review of

Information Science and Technology, 41, 273–306.

D’Elia, G. (1980). The development and testing of a conceptual model of public

library use behaviour. Library Quarterly, 50(4), 410–430.

D’Elia, G., Abbas, J., Bishop, K., Jacobs, D., & Rodger, E. J. (2007). The impact of

youth’s use of the internet on their use of the public library. Journal of the

American Society for Information Science and Technology, 58(14), 2180–2196.

Dervin, B. (1984). The information needs of Californians–1984. Davis, CA: Institute

of Governmental Affairs, University of California.

Dervin, B. (1997). Given a context by any other name: Methodological tools for

taming the unruly beast. In: P. Vakkari, R. Savolainen & B. Dervin (Eds),

Information seeking in context (pp. 13–38). London: Taylor Graham.

Dervin, B., & Nilan, M. (1986). Information needs and uses. Annual Review of

Information Science and Technology, 21, 3–33.

Du, Y. (2009). A review of structural equation modeling and its use in library and

information studies. Library & Information Science Research, 31(4), 257–263.

Duke, N. K. (2000). For the rich it’s richer: Print experiences and environments

offered to children in very low- and very high-socioeconomic status first-grade

classrooms. American Educational Research Journal, 37(2), 441–478.

Duncan, G. J., & Raudenbush, S. W. (1999). Assessing the effects of context

in studies of child and youth development. Educational Psychologist, 34(1), 29–41.

Durrance, J. C., & Fisher, K. E. (2003). Determining how libraries and librarians

help. Library Trends, 51(4), 305–334.

Ellis, D., Cox, D., & Hall, K. (1993). A comparison of the information seeking

patterns of researchers in the physical and social sciences. Journal of Documenta-

tion, 49(4), 356–359.

Environment. (2000). In: J. P. Pickert (Ed.), The American heritage dictionary of the

English language (4th ed.). Boston, MA: Houghton Mifflin.

Erdelez, S. (1997). Information encountering: A conceptual framework for accidental

information discovery. In: P. Vakkari, R. Savolainen & B. Dervin (Eds),

Information seeking in context (pp. 412–421). London: Taylor Graham.

Fisher, K. E., Durrance, J. C., & Hinton, M. B. (2004). Information grounds and the

use of need-based services by immigrants in Queens, New York: A context-based,

outcome evaluation approach. Journal of the American Society for Information

Science and Technology, 55(8), 754–766.

206 Sei-Ching Joanna Sin

Fisher, K. E., Erdelez, S., & McKechnie, L. (2005). Theories of information

behaviour. Medford, NJ: Information Today.

Fisher, K. E., & Julien, H. (2009). Information behaviour. Annual Review of Infor-

mation Science and technology, 43, 317–358.

Fisher, K. E., Landry, C. F., & Naumer, C. (2007). Social spaces, casual interactions,

meaningful exchanges: ‘Information ground’ characteristics based on the college

student experience. InformationResearch, 12(2). Retrieved from http://informationr.

net/ir/12-2/paper291.html

Fox, J. (2006). Structural-equation modeling with the SEM package. Structural

Equation Modeling, 13(3), 465–486.

FreshMinds, & Department for Communities and Local Government. (2008). Under-

standing digital exclusion-research report.Retrieved from http://www.communities.

gov.uk/documents/communities/pdf/1000404.pdf

Garrison, D. (2003). Apostles of culture: The public librarian and American society,

1876–1920. Madison: University of Wisconsin Press.

Goodchild, M. F., & Lam, N. S.-N. (1980). Areal interpolation: A variant of the

traditional spatial problem. Geo-Processing, 1, 297–312.

Haythornthwaite, C. (1996). Social network analysis: An approach and set of

techniques for the study of information exchange. Library & Information Science

Research, 18(4), 323–342.

Hersberger, J. A., Murray, A. L., & Sokoloff, S. M. (2006). The information use

environment of abused and neglected children. Information Research, 12(1).

Retrieved from http://InformationR.net/ir/12-1/paper277.html

Hjørland, B. (1997). Information seeking and subject representation: An activity-

theoretical approach to information science. Westport, CT: Greenwood Press.

Hogan, R., & Roberts, B. W. (2000). A socioanalytic perspective on person-

environment interaction. In: W. B. Walsh, K. H. Craik & R. H. Price (Eds),

Person-environment psychology: New directions and perspectives (2nd ed.).

Mahwah, NJ: Lawrence Erlbaum.

Holstein, J. A., & Gubrium, J. F. (2004). Context: Working it up, down and across.

In: C. Seale, G. Gobo, J. F. Gubrium & D. Silverman (Eds), Qualitative research

practice (pp. 297–311). London: Sage.

International Telecommunication Union. (2009).Measuring the information society-the

ICT development index. Retrieved from http://www.itu.int/ITU-D/ict/publications/

idi/2009/material/IDI2009_w5.pdf

Johnson, J. D. E. (1996). Information seeking: An organizational dilemma. Westport,

CT: Greenwood Publishing.

Johnson, J. D. E., Case, D. O., Andrews, J., Allard, S. L., & Johnson, N. E. (2006).

Fields and pathways: Contrasting or complementary views of information seeking.

Information Processing & Management, 42(2), 569–582.

Jue, D. K., Koontz, C. M., Magpantay, J. A., Lance, K. C., & Seidl, A. M. (1999).

Using public libraries to provide technology access for individuals in poverty: A

nationwide analysis of library market areas using a geographic information

system. Library & Information Science Research, 21(3), 299–325.

Towards Agency–Structure Integration 207

Kim, K.-S., & Allen, B. (2002). Cognitive and task influences on web searching

behaviour. Journal of the American Society for Information Science and Technology,

53(2), 109–119.

Kline, R. B. (2005). Principles and practice of structural equation modeling. New

York, NY: Guilford.

Krippendorff, K. H. (2004). Content analysis: An introduction to its methodology (2nd

ed.). Thousand Oaks, CA: Sage.

Kronus, C. I. (1973). Patterns of adult library use: A regression and path analysis.

Adult Education, 23, 115–131.

Kuhlthau, C. C. (1991). Inside the search process: Information seeking from the

user’s perspective. Journal of the American Society for Information Science, 42(5),

361–371.

Leckie, G. J., Pettigrew, K. E., & Sylvain, C. (1996). Modeling the information

seeking of professionals: A general model derived from research on engineers,

health care professionals, and lawyers. Library Quarterly, 66(2), 161–193.

Livingston, G. (2010). The Latino digital divide: The native born versus the foreign

born. Retrieved from http://pewhispanic.org/reports/report.php?ReportID¼123

Nahl, D., & Bilal, D. (2007). Information and emotion: The emergent affective paradigm

in information behaviour research and theory. Medford: Information Today.

Neuman, S. B., & Celano, D. (2001). Access to print in low-income and middle-

income communities: An ecological study of four neighborhoods. Reading

Research Quarterly, 36(1), 8–26.

Otte, E., & Rousseau, R. (2002). Social network analysis: A powerful strategy, also

for the information sciences. Journal of Information Science, 28(6), 441–453.

Paisley, W. J. (1967). Information need and use studies. Annual Review of

Information Science and Technology, 3, 1–30.

Pawley, C. (1998). Hegemony’s handmaid? The library and information studies

curriculum from a class perspective. Library Quarterly, 68(2), 123–144.

Pawley, C. (2006). Unequal legacies: Race and multiculturalism in the LIS

curriculum. The Library Quarterly, 76(2), 149–168.

Pettigrew, K. E. (1999). Waiting for chiropody: Contextual results from an

ethnographic study of the information behaviour among attendees at community

clinics. Information Processing & Management, 35(6), 801–817.

Rees, M. B., & Paisley, W. J. (1968). Social and psychological predictors of adult

information seeking and media use. Adult Education Journal, 19, 11–29.

Ritzer, G., & Gindoff, P. (1994). Agency-structure, micro-macro, individualism-

holism-relationism: A metatheoretical explanation of theoretical convergence

between the United States and Europe. In: P. Sztompka (Ed.), Agency and

structure: Reorienting social theory (pp. 2–23). Yverton: Gordon and Breach.

Robbins, L. S. (2000). The dismissal of Miss Ruth Brown: Civil rights, censorship, and

the American library. Norman, OK: University of Oklahoma Press.

Sampson, R. J., Morenoff, J. D., & Gannon-Rowley, T. (2002). Assessing

‘‘neighborhood effects’’: Social processes and new directions in research. Annual

Review of Sociology, 28, 443–478.

208 Sei-Ching Joanna Sin

Savolainen, R. (1995). Everyday life information seeking: Approaching information

seeking in the context of ‘‘way of life’’. Library & Information Science Research,

17(3), 259–294.

Scott, J. (1994). Social network analysis: A handbook. London: Sage.

Sin, S.-C. J. (2009). Structural and individual influences on Information behaviour: A

national study of adolescents’ use of public libraries. Doctoral dissertation,

University of Wisconsin-Madison, Madison, WI. Retrieved from ProQuest

Dissertations & Theses database (UMI No. 3384534).

Sin, S.-C. J. (2011). Neighborhood disparities in access to information resources:

Measuring and mapping U.S. public libraries’ funding and service landscapes.

Library & Information Science Research, 33(1), 41–53.

Sin, S.-C. J., & Kim, K.-S. (2008). Use and non-use of public libraries in the

information age: A logistic regression analysis of household characteristics and

library services variables. Library & Information Science Research, 30(3), 207–215.

Smith, D. M. (1979). Where the grass is greener: Living in an unequal world.

Baltimore, M.D: John Hopkins University Press.

Solomon, P. (2002). Discovering information in context. Annual Review of Infor-

mation Science and Technology, 36, 229–264.

Sonnenwald, D. H. (1999). Evolving perspectives of human information behaviour:

Contexts, situations, social networks and information horizons. In: T. D. Wilson &

D.K.Allen (Eds),Proceedings of the Second International Conference in Information

Needs (pp. 176–190). London: Taylor Graham.

Sonnenwald, D. H., Wildemuth, B. M., & Harmon, G. L. (2001). A research method

to investigate information seeking using the concept of information horizons: An

example from a study of lower socio-economic student’s information seeking

behaviour. The New Review of Information Behaviour Research, 2, 65–86.

Spink, A., & Cole, C. (2006). Human information behaviour: Integrating diverse

approaches and information use. Journal of the American Society for Information

Science and Technology, 57(1), 25–35.

Talja, S., Keso, H., & Pietilainen, T. (1999). The production of ‘context’ in

information seeking research: A metatheoretical view. Information Processing &

Management, 35(6), 751–763.

Taylor, R. S. (1991). Information use environments. In: B. Dervin & M. J. Voigt

(Eds), Progress in communication sciences (vol. 10, pp. 217–255). Norwood, NJ:

Ablex Publishing Corporation.

University of Edinburgh, & Association for Geographic Information. (1996).

Geographic information system. AGI GIS Dictionary. Retrieved from http://

www.geo.ed.ac.uk/agidict/welcome.html

Vakkari, P. (1997). Information seeking in context. A challenging metatheory. In:

P. Vakkari, R. Savolainen & B. Dervin (Eds), Information seeking in context

(pp. 451–464). London: Taylor Graham.

Vakkari, P., Savolainen, R., & Dervin, B. (1997). Foreword. In: P. Vakkari,

R. Savolainen & B. Dervin (Eds), Information seeking in context (pp. 7–9).

London: Taylor Graham.

Towards Agency–Structure Integration 209

Whitmire, E. (2002). Academic library performance measures and undergraduates’

library use and educational outcomes. Library & Information Science Research,

24(2), 107–128.

Wiegand, W. A. (1989). An active instrument for propaganda: The American public

library during World War I. New York, NY: Greenwood Press.

Williams, R. W. (1980). Sources of the variability in level of public library

development in the United States: A comparative analysis. Library Research, 2(2),

157–176.

Wilson, L. R. (1938). The geography of reading: A study of the distribution and status

of libraries in the United States. Chicago, IL: University of Chicago Press.

Wilson, T. D. (1981). On user studies and information needs. Journal of

Documentation, 37, 3–15.

Zweizig, D. (1973). Predicting amount of library use: An empirical study of the role of

the public library in the life of the adult public. Ph.D. dissertation, Syracuse

University.

Zweizig, D., & Dervin, B. (1977). Public library use, users, uses: Advances in

knowledge of characteristics and needs of the adult clientele of American public

libraries. Advances in Librarianship, 7, 232–255.