PETER LOR_Methodology in Comparative Politics.pdf

Chapter 4 in the broad sense outlined there. Ragin (1987:165) put it this way:...social science methodology does not concern mere technique; it Methodology in comparative studies...

Chapter 4 in the broad sense outlined there. Ragin (1987:165) put it this way:...social science methodology does not concern mere technique; it Methodology in comparative studies concerns the relationship between thinking and researching. The key concern here is the impact of the organization of the investigation and the structure of the data analysis on how the investigator thinks about Outline the subject. Comparative research Terminology While there has been little recent reflection on the methodological aspects of Quantitative and qualitative methodologies comparative librarianship, there has been ongoing rethinking and discussion in Quantitative vs. qualitative other comparative social science fields, such as in comparative education (Kelly et Mixed methods al. 1982:509-511; Raivola 1986; Crossley 2002, Cowen 2007), comparative Comparative strategy ethics (Lewis 2000); comparative history (Skocpol & Somers (1980), comparative How many countries? politics (Ragin 1987; Pennings et aI. 1999; Landman 2008), comparative social Variable-oriented vs. case-oriented strategies policy (Jones 1985; Mabbett & Bolderson 1999; Kennett 2001), and comparative Comparative research designs social research generally (Hantrais 2009) as well as in cross-cultural studies such as Single-country studies (case studies) cross-cultural social work (Tran 2009), among many others. Thus this chapter Many-country comparisons draws heavily on methodological writings in these comparative fields. In this I Few-country comparisons Trade-offs between cases and variables follow in the footsteps of Foskett (1977) who provided the LIS profession with an The time dimension overview of comparative studies in other disciplines at a time when there was Further design decisions much new interest in comparative librarianship. In some of these disciplines which Classification and typologies are richer in theory than LIS there is much emphasis on the development and Conclusions testing of more wide-ranging, higher-level theory than is usual in LIS. Not all the References considerations set out in these texts can be readily transposed to our field, but we have much to learn from them. Where relevant, reference is also made to work on comparative librarianship, such as Simsova and MacKee (1975) and Krzys and Litton (1983). In Chapter 3 the relationship between metatheory, methodology and method was This chapter focuses on comparative librarianship. Much of what is said here will examined. Following Dervin (2003:136-137) and Pickard (2007:xv-xvii) I also be relevant to non-comparative research in international librarianship or proposed to consider methodology as the bridge between metatheory, the general research into library and information phenomena more generally in other higher-level assumptions (also referred to as paradigms or world views) that countries. underlie researchers‟ work, and method, the specific practical procedures they use in collecting, analyzing and interpreting data. This chapter deals with methodology Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 1 Comparative research properties of these units. They can take their existence for granted. This is different for comparativists, because they compare macrosocial units as such: Comparison is inherent in all science, including the social sciences, where comparative research has historically played a significant role in their development At a very general level, comparativists are interested in identifying the as scientific disciplines. However, there is little agreement in the social sciences on similarities and differences among macrosocial units. This knowledge the question whether the comparative method should be considered a distinct provides the key to understanding, explaining and interpreting diverse subfield (as suggested by terms such as comparative education or comparative historical outcomes and processes and their significance for current institutional arrangements. Cross-societal similarities and differences... politics) or as a methodology. Many comparative methodology texts present at constitute the most significant feature of the social landscape, and, least a brief discussion of this issue (e.g. Hantrais 2009:5-9; Pennings et al.1999: consequently, these researchers have an unmistakable preference for 21-26). In an influential article on comparative politics, Lijphart (1971:682) explanations that cite macrosocial phenomena... Most comparativ- situated the comparative method as a basic method in its own right, alongside the ists... are interested in the cases themselves, their different historical experimental, statistical and case study methods. Sartori (1991:243) stated experiences in particular, not simply in relations between variables categorically that comparative politics is a “field characterized by a method”. characterizing broad categories of cases. (p.6) However, this did not end the disagreement as to the status of the comparative method. Kelly et al. (1982:511-515) discussed in some detail the question Similarly, Pennings et al. (1999:50) argue that comparisons are made across whether comparative education is a method or an area of content. More recently, political and social systems that are defined in relation to territorial space. Arnove Mabbett and Bolderson (1999:34) stated that “many of the issues surrounding the et al. (1982:5) discuss disagreement in comparative education on whether sub- theories and methods in comparative work are not exclusive to cross-national units of national systems can be utilized as units of comparison in addition to the studies... There is no distinct social science „cross-national method‟ although such national systems themselves, and whether these can be compared at different research highlights some of the issues in making scientific as opposed to points in time. There are advantages and disadvantages to selecting countries as impressionistic comparisons”. The idea that comparative social science is no „comparators‟ (the units being compared). One disadvantage is that sometimes different from any other form of social science and that it does not have any within-country differences are obscured, since in some national units, e.g. post- unique methodological issues is attractive from a positivist perspective because it unification Germany, internal diversity may be greater than the diversity observed suggests that all social sciences use basically the same methods and because it when comparing countries with one another, e.g. Germany with other EU underlines the „scientific‟ nature of comparative social science (cf. Ragin 1987:2). countries (Hantrais 2009:54). Lijphart (1975:166-167) has critically discussed the issue of “whole-nation bias” and the arguments for and against the focus on However, Ragin (1987:1-6) points to significant differences between the countries. A wide-ranging and conceptually rich discussion of the concept „nation‟ orientations of most comparativists and most „noncomparativists‟. These is found in Galtung (1982). differences have methodological implications. The distinctive orientation of comparative social science is that it is concerned with what he calls “large In this chapter I follow the approach that emphasizes comparisons between macrosocial units”, a term he uses to refer to countries, nations and other larger territorially distinct macrosocial units, i.e. international (or cross-national) political entities. Although all social scientists claim to study societies or things that comparisons. However, much of the discussion is also relevant to comparisons of happen in society, most do not feel the need to define the macrosocial units within cultural, societal or linguistic groups that are distributed within or across which their research is conducted and they are not much concerned with the countries. I further adopt the point of perspective that comparative studies are Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 2 sufficiently distinct to justify considering the comparative method at the level of necessarily numeric. Variables may be at different levels (or scales) of methodology as defined above. measurement, as routinely described in most LIS methodology texts, e.g. Conna- way & Powell (2010:65-66). Terminology Any phenomenon can be studied at various levels of analysis. For example, if information literacy education is studied, we could investigate aspects of such Before we look more closely at comparative methodology, it is helpful to clarify education at the level of countries, provinces, school districts, or individual some key terms. A useful way to start is to consider a data matrix as depicted in schools, classes, teachers or students. At each level of analysis, different units of Table 4.1: analysis might be appropriate. The unit of analysis refers to the type of entity or object that is studied. For example, in Table 4.1, the unit of analysis is libraries. Table 4.1: Data matrix for six public libraries (hypothetical data) Table 4.2 lists some levels of analysis with a selection of appropriate units of Variables analysis for a hypothetical study of education for information literacy. A: B: C: D: Population Number of Number of Number of Table 4.2: Levels and units of analysis in a study of information literacy education served registered books in loans per users stock year Levels of analysis Units of analysis Library 1 8,100 5,887 17,600 33,245 Library 2 18,700 12,465 31,900 68,432 Observations Country Countrywide policies, curricula, syllabi, materials, standards; total Library 3 4,100 2,345 8,600 13,911 resources; aggregate student performance measures; Library 4 23,100 9,855 38,100 71,313 international rankings Library 5 14,700 8,288 25,500 68,113 Library 6 7,500 3,853 12,250 23,911 Province, School District Provincial or district-wide policies, curricula, syllabi, materials, standards; total resources; aggregate student performance Table 4.1 presents a data matrix containing a set of data on six public libraries. The measures; national rankings libraries are the observations, also referred to as the units of observation. (Units of observation may sometimes be referred to as cases. To avoid confusion this is not School Classes in which instruction is given; number of hours of instruction; number of teachers certified to give instruction; types advisable here, as will become clear later.) Each observation occupies a row. The of materials used; aggregate student performance measures; number of observations is often abbreviated using the capital letter N. Here N=6. provincial or district rankings For each library four categories of information have been collected. These categories are the variables, sometimes referred to as units of variation. Each variable Student Class in which enrolled, number of hours instruction received; occupies a column. The number of variables is sometimes abbreviated using the performance on tests capital letter K. Here K=4. At the intersection of an observation and a variable we find a data value (or data point).Thus for Library 3 the value for the number of Note that a picture at a higher level of analysis may be built up by aggregating data registered users (Variable B) is found at cell B3, namely 2.345. This data matrix from a lower level. For example, the test scores of all the individual students in has 6x4=24 data values. All this will be familiar to users of Excel or other Grade 5 may be summarized using measures of central tendency (mean, median spreadsheets. Note that there are sometimes missing values, and that values are not etc.) and dispersion (range, standard deviation etc.) by class, school, school Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 3 district, province or country. At each higher level we get further from the nitty- Table 4.3 illustrates that data collected at more than one level can be used in a gritty detail and some information is sacrificed for the bigger picture. comparison of countries. Confusion about levels of analysis can lead to aggregate fallacies or „wrong level‟ Table 4.3: Data matrix for six countries (hypothetical data) fallacies, which result from making inferences about units of analysis at one level Variables based on observations of units of analysis at another level (Hantrais 2009:55). This A: B: C: D: can happen in particular when data are collected about individual persons and Population Number of Number of Number of about territorial units such as countries. There are two kinds of aggregate fallacy. public books in loans p.a. The ecological fallacy occurs when we make inferences about individuals on the libraries public by public basis of data about larger units, for example, inferring that Jessica, a student at libraries libraries x 1,000,000 x 1,000,000 x 1,000,000 Central High, has a low level of information literacy skills because the curriculum Country 1 12.4 687 27.2 89.8 used in her school‟s school district has been rated as sub-standard. Variables Country 2 53.5 865 32.6 108.4 Observations measured at different levels of analysis may look the same, but are often not. In Country 3 2.8 17 0.1 0.1 this example, information literacy skill has been measured at the individual level Country 4 14.1 858 32.5 103.7 while the quality of the curriculum has been measured at a higher level. The Country 5 87.1 2282 150.5 312.8 opposite of the ecological fallacy is called the individualistic fallacy. For example, it Country 6 8.6 113 2.8 3.9 would be inappropriate to arrive at conclusions about the school district‟s curriculum for information literacy simply on the basis of test scores from Jessica‟s Here the observations are countries, and for each country only one data value is class at Central High. Further examples are found in comparative social sciences given for each variable. In columns B, C and D data on each country‟s public texts, such as Landman (2008:43) and Hantrais (2009:55). libraries are presented. The level of analysis in Table 4.3, where countries are compared, is different from that in Table 4.1, where libraries are compared. Table In comparative studies it is particularly important to be clear about the levels and 4.3 presents data about the libraries in each country. The data in column B units of analysis. Thus in a comparative study of public libraries in different represent a simple count of the number of libraries in each country. These are countries, we could use data collected at various levels: country-level data. The data in columns C and D are aggregated data based on statistics that would have been kept in each individual library and would have been individual library users (e.g. their attitudes to libraries, frequency of use, reported in surveys or statistical returns to yield the information depicted in Table number of books borrowed per year) 4.1. Note that the detail provided there is lost in Table 4.3. individual librarians (qualifications, salaries, length of service, etc.) In Table 4.4 a comparison between two countries is depicted. For each country individual libraries (population served, number of registered users, etc.) there is a data matrix as in Table 4.1. library consortia or districts (number of libraries, holdings in union catalog, volume of resource sharing, management software used, etc.) countries (total population and area, number of libraries, total registered users in all the libraries, national library legislation and policies, etc.) Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 4 Table 4.4: comparison of two countries reported at more than one point in time, then the time-units included indicate the number of cases, e.g. two countries at four points in time: Country P Country Q eight cases. Variables Variables A B C D A B C D The upshot is that, in reading comparative methodology texts or evaluating 1 1 comparative studies, one should be aware that comparativists do not always agree on terminology. Observations (e.g. libraries) 2 2 Observations (e.g. libraries) 3 3 4 4 Quantitative and qualitative methodologies 5 5 Pickard (2007:xvi) echoes a fairly common standpoint that there are only two 6 6 basic methodologies: quantitative and qualitative. The choice between these two is 7 7 the highest level methodological decision. The metatheoretical assumptions discussed in the previous chapter have a strong influence on this methodological choice. Quantitative methodology is usually associated with a positivist and post- The two countries are referred to as cases or „comparators‟ (Hantrais 2009). Each positivist metatheoretical stance, and qualitative methodology with an case has seven observations. Is N=2 or N=14? Opinions differ on this point. interpretivist or allied metatheoretical stance (cf. Hantrais 2009:57-59). Figure 4- Some authorities use N to refer to the number of cases (so that N=2), thus they A depicts the relationship between the three main metatheoretical positions that use the term „small-N‟ studies to refer to studies of a small number of countries were discussed in the previous chapter and methodological choices in comparative and „large-N‟ studies to refer to studies covering many countries. These studies. To remain within the iceberg metaphor, the diagram should be read from expressions are frequently seen in the literature. Others, e.g. Landman (2008) and the bottom (the metatheoretical level) upwards. Roughly following the levels Gerring (2007) use N to refer to the number of observations (so that N=14). distinguished by Pickard (2007), the methodological level has been divided into Differences in the terminology used by various writers can lead to confusion. three sublevels, those of general methodology, comparative strategy, and compar- Pennings et al.(1999:10-11) discuss the distinction between cases and observations ative research design. in some detail. Ultimately, in their view, it depends on the research question. If the research question concerns: an international comparison (comparison of countries), the number of cases is identical to the number of countries included (In Figure 4.3: N=6) a cross-national comparison (where in each country a number of units such as libraries or library systems might be studied), the number of cases is defined by the units of observation (In Figure 4.4: N=14) a comparison of change over time, where data for each country are Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 5 Figure 4-A: Relationship of comparative methodological choices to metatheory comparative education positivism inspired a drive for the use of educational data from a large number of countries, on the assumption that large-scale studies would be more „scientifically‟ reliable. However, as a counter-trend this period saw a move towards more interpretivist methodology, including phenomenology and Comparative Many-country Few-country Single-country ethnography, in comparative education (Hayhoe & Mundy 2008:10-13). The trend research design comparison comparison study towards more use of qualitative methodology is also visible in other social science disciplines Comparative Variable-oriented Case-oriented Strategy In their introduction to the Handbook of qualitative research, Denzin and Lincoln (1994:3-4) offer the following definition of qualitative research: General Quantitative Mixed Qualitative Qualitative research is an interdisciplinary, transdisciplinary, and some- Methodology methods times counterdisciplinary field. It crosscuts the humanities and the social and physical sciences. Qualitative research is many things at the same time. It is multiparadigmatic in focus. Its practitioners are sensi- Metatheory Positivism Postpositivism Interpretivism tive to the value of the multimethod approach. They are committed to the naturalistic perspective, and to the interpretive understanding of human experience. At the same time the field is inherently political and shaped by multiple ethical and political positions. Quantitative vs. qualitative Qualitative research embraces two tensions at the same time. On the one hand it is drawn to a broad, interpretive, postmodern, feminist, and critical sensibility. On the other hand, it is drawn to more In most general social science research methods texts, including texts in LIS, the narrowly defined positivist, postpositivist, humanistic, and naturalistic majority of chapters are devoted to quantitative methods, with the emphasis on conceptions of human experience and its analysis (pp.3-4). the formulation of hypotheses, operationalization of concepts, measurement (a metaphor derived from the physical sciences), the development of instruments (a From the overview given by Denzin and Lincoln in their introduction, a picture similar metaphor), the design of experiments or surveys, sampling, and the emerges of an extremely diverse methodology with a confusing array of statistical testing of hypotheses. In such texts quantitative methods are regarded as competing paradigms. the standard or default approach. Often a single chapter is devoted to qualitative (or naturalistic or ethnographic) methodology. We may find here that qualitative Many texts (e.g. Mouton & Marais 1990:160-162; Cresswell 2009:12-17; Hantrais approaches are subsumed under a positivist methodology. This seems to be the 2009:98) present tables contrasting quantitative and qualitative methodology. It is case in the political science texts of Pennings et al. (1999) and Landman (2008). hardly necessary to add to this discussion, but for the convenience of the reader, Landman considers what he calls “conceptual description” to be the first Table 4.5 summarizes some important characteristics. Terms used in the table and “objective” (where „objective‟ refers to a step, activity or procedure) in a process not encountered in Chapter 3, are explained in the course of this chapter. leading to quantitatively conceived hypothesis testing and prediction (p.7). In Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 6 Table 4.5: Characteristics of quantitative and qualitative methodology pluralism‟ in the social sciences generally and in comparative studies specifically. Characteristic Quantitative Qualitative The use of multiple methods or mixed methods is now covered in many research methods texts (e.g. Mason 1996; Teddlie and Tashakkori 2003) and the topic has Metatheory Positivist, Postpositivist Interpretivist also been addressed in LIS (Cresswell 2009; Pashaeizad 2009). Hantrais (2009: Nature of reality Singular, stable, independent of Multifarious, culturally 109-113)) deals specifically with multiple methods in comparative social research, observer; external reality determined, socially constructed; distinguishing between three approaches: holistic reality, Relation of External, observing from outside; In the study setting, observing Triangulation: two or more different research strategies are used to investigator to what is in artificial setting from within; in real-life setting studied investigate the same phenomenon so that findings or insights from one Relation to social Neutral Engaged strategy can be corroborated by the other(s); specifically quantitative and phenomenon Empirical Normative qualitative approaches are used in parallel. Research aim Nomothetic; hypothesis testing; Idiographic; hypothesis Facilitation: more than one approach is used, but one of them is generalizing generating; contextualizing dominant and different techniques may be used sequentially (for example Strategies Structured, theory-derived Unstructured, open-ended, a qualitative study to generate hypotheses before a quantitative study is variables identified beforehand; theory developed during research; undertaken). controls; operationalization & concepts that are rich in meaning measurement Complementarity: different approaches are integrated rather than used in Typical methods Experiments, surveys Participant observation, case parallel or sequentially, as when researchers shift repeatedly from the one studies to the other. Criteria for judging Validity & reliability; objectivity Credibility, transferability, research dependability; authenticity While the quantitative and qualitative methodologies are complementary and while there are advantages to combining them, there is a risk that the results will There is a huge literature discussing the pros and cons of quantitative versus be irreconcilable. In general it seems that one of the two dominates and the other qualitative approaches. It is routinely dealt with in current social science research is secondary and supplements it (Ragin 1987:69-78). Problems can arise when methodology texts. It is also reflected in debates among comparativists about the mixed methods are used by researchers who are insufficiently aware of the respective merits of large surveys in many countries, in-depth comparisons of few metatheoretical implications of the methods they are using. Mason (1996:79) countries, and single-country case studies – an issue to be discussed in the next advises that “a researcher must think strategically about the integration of multiple section. methods, rather than piecing them together in an ad hoc and eclectic way.” This implies that the researcher must be aware of the ontological, epistemological and other assumptions underlying their methodology. Mixed methods Hantrais (2009:59, 103-108) points out that the quantitative/qualitative divide may have been exaggerated and that for many researchers it is no longer so important. In recent years there has been a greater acceptance of „methodological Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 7 Comparative strategy Figure 4-B: Number of countries studied and degree of detail One of the most prominent issues discussed in comparative methodology texts in the social sciences is the question of how many cases (where cases refer mostly to countries) should be studied. In fact, the distinction between studies with many High countries (often referred to as large-N studies) and those with few countries (often referred to as small-N studies) has given rise to a major typological division of comparative social science research. For example, Lijphart (1971:683-684) Level of detail distinguished between the statistical, comparative and case study methods. By the latter Lijphart meant single case studies. By the “statistical” method he meant quantitative comparative research using large amounts of data. For Lijphart the crucial difference between the statistical method and the comparative method was that the latter uses fewer cases – too few for the statistical control that can be Low exercised in the analysis of survey data. His point of departure is essentially One Few (50) positivistic. It accepts the experimental method as the norm which other methods try to approximate. Similarly, Landman (2008:26) adopts a three-part division of Number of countries covered comparative studies into “comparing many countries, comparing few countries, and single-country studies”. This depiction suggest that comparative studies lie on a continuum and that the major differences between studies at the two ends lie in the number of countries covered and the degree of detail provided about each. Landman (2008:26) indi- How many countries? cates that the continuum can also be looked at from another angle: the level of ab- straction. The more countries are included in the study, the higher the level of Generally, a study of a single country can be very intensive and conducted in abstraction (Landman 2008:26). Abstraction here refers to the concepts used. This considerable detail, but the more countries there are, the less intensively each one is an issue that will be dealt with later in this chapter. Nevertheless, Landman will be studied. This is illustrated in Figure 4-B, in which I have used Landman‟s (2008) insists that all comparative studies (regardless of the number of cases) are (2008:26) categories for the number of countries dealt with. “grounded in one logic of inference” (p.45), which is part of a program of hypothesis testing (p.7, 9), theory building and prediction. If this implies that there is a single ontological and epistemological basis for all comparative research, such a view appears to be an over-simplification. This is suggested by the two exhibits that follow, which illustrate how a putative relationship between public libraries and democracy might be investigated using quantitative and qualitative methodologies. Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 8  e.g. degree of industrialization, strength of labor unions, literacy level, population EXHIBIT 4-I: Quantitative comparative study of the relationship between democracy density, per capita GDP, etc., which might also influence the relationship. If and free public libraries relevant variables are omitted, the relationship shown by the data may be spurious. The hypothesis is that there is a direct relationship between a democratic form EXHIBIT 4-II: Qualitative comparative study of the relationship between democracy of government in a country and the prevalence of free public libraries. Once we and free public libraries have clarity on the two variables (in italics), they need to be operationalized. For the first variable, we might use a typology borrowed from political We have no formal variables or hypothesis in mind when we start. We select science which could (for example) enable us to divide the countries of the two or three countries, e.g. Egypt, France and Colombia, study the available world into a number of basic categories ranging from fully democratic to literature on the public libraries, their history, legislation, funding, current dictatorship, using an appropriate index or classification that has already statistics on their status, also public library rules, curricula of courses on been devised by political scientists. We can operationalize the second public library work and textbooks prescribed for this topic in used in library variable by using a set of indicators such as the following. schools... We visit websites, including websites of pro-democracy Number of public libraries related to country‟s population, e.g. organizations and www.humanrightsdata.com. Then we travel to these mean population per public library countries, and in each of them visit a number of public libraries, observe their Percentage of population who are registered members of public collections and use during opening hours, conduct interviews with library staff libraries and users, as well as with some senior government officials, labor union Mean number of books borrowed from public libraries per capita officials, democracy activists, office bearers of the national library associations, The data on the library conditions in the countries can be collected from professors of LIS, professors of political science and other experts who are sources such as UNESCO statistical yearbook. knowledgeable about the constitutions and politics of the countries, etc. As We then perform statistical tests on the data to see whether there is a we do this, we write up notes describing what we observe and recording the significant relationship between the two variables. If the null hypothesis (that insights we gain, including variables and possible hypotheses that we discover there is no significant relationship) is rejected, the substantive hypothesis as we reflect on our observations. We discover and obtain more literature and (that there is a direct relationship between the two variables) is supported. documentation to study back home. Back home, while doing the additional We report our findings in a rather dry scientific report in which the text is reading and writing up our notes on the visit, we correspond further by e-mail supplemented by tables and graphs. with contacts in the countries to fill in gaps or clarify things we may have misunderstood. Finally, we write a report describing the interactions and relationships that we The research suggested in Exhibit 4-I can be done without leaving one‟s desk. It is uncovered. The text of our report will be embellished using interesting not necessary to get close to the context. This research would enable us to make quotations from informants and descriptions of conditions and events we fairly confident general statements about the relationships between democracy and observed. free public libraries. However, because we have had to specify the variables of interest beforehand, there is a danger that we may have overlooked other variables, Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 9 The study outlined in Exhibit 4-II may prove to be quite a messy and time- from the phenomenon that is studied. All this reflects an underlying positivist consuming exercise, and it is possible that it may not turn out quite the way it was ontology and epistemology. anticipated. The researcher will discover a huge range of factors and influences that she had not thought of before visiting the countries. She would gain an excellent In case-oriented studies a single country or a small number of countries is studied. understanding of how these various factors may have influenced the establishment, The focus is on the individual country in its historical specificity and its full development and current status of libraries. She would find that these are context. Each case is considered as a whole, taking into account the total configured rather differently in each of the three countries, and that it is quite configuration or constellation of factors and conditions. Ragin (1987:20) difficult to write all this up in a lucid, comprehensible and insightful report. emphasizes the need to unravel the “multiple conjunctural causation” that However, if she succeeded it could make fascinating reading and convey an in- characterizes social phenomena: depth understanding of public librarianship in the three countries. Although we would not have much confidence that the pattern or patterns detected are also...social phenomena are complex and difficult to unravel not because applicable in other countries, the study would suggest relationships that might there are too many variables affecting them... but because different fruitfully be studied in other countries, or that might be tested in a study such as causally relevant conditions can combine in a variety of ways to that suggested in Exhibit 4-I. An early example of an in-depth study, albeit based produce a given outcome. In short, it is the combinatorial, and often complexly combinatorial, nature of social causation that makes the largely on published literature and statistical data, was a comparison by problem of identifying order-in-complexity demanding (p.26). Hassenforder (1967) of public library development in France, the United States and Great Britain, which yielded striking insights into the social and other This embrace of complexity and the use of „thick description‟ rather than statistics conditions that favor public library development. in case-oriented studies reflect a greater affinity for interpretivist metatheory. However, the two approaches have complementary strengths and they may meet in mixed methods studies in the grey area between the two extremes. Variable-oriented vs. case-oriented strategies The two exhibits illustrate the differences between what Ragin (1987) has called Comparative research designs variable-oriented and case-oriented studies, in which quantitative and qualitative methods respectively are applied in comparative studies. These orientations refer In this section we consider how the mainly qualitative case-oriented approach and to more dimensions than merely the number of cases. the mainly quantitative variable-oriented approach are manifested in the three main comparative research designs: single-country studies, many-country Typically, in variable-oriented studies many countries are studied. The focus is on a comparisons, and few-country comparisons. limited number of variables, which are abstracted and removed from the concrete reality and context of the countries that are studied by means of simplifying assumptions. As Ragin says, the approach tends to “eliminate complexity instead of Single-country studies (case studies) deciphering it” (p.xiv). Formal hypotheses stating universal relationships, the use of operational definitions and emphasis on quantitative data obtained by means of There has long been controversy about whether single-country studies (case „measurement‟ and the use of „instruments‟ combine to distance the researcher studies proper) should be considered to be comparative studies. This is also Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 10 reflected in the literature of comparative librarianship, e.g. in the difference of comparative politics the nation-state is the dominant type of case, but other social opinion between Danton (1973:46-52) and Krzys and Litton (1983:27-29) on the and political units or institutions can also be chosen (pp.94-5).A case study is “the one hand, who reject single-country studies as being comparative, and Collings intensive study of a single case for the purpose of understanding a larger class of (1971) and Simsova and Mackee (1975:30-32) on the other, who accept them. In cases (a population)”, while case study research may include several cases. The political science Sartori (1991:252) insisted that the single case investigation number of cases is limited by the extent to which they can be investigated “cannot be subsumed under the comparative method (though it may have intensively. At a given point such intensive study is no longer possible, and the comparative merit)” (Sartori‟s emphasis).On the other hand, Landman (2008:28) emphasis of a study will shift from the individual case to a sample of cases. Gerring states that refers to such a study as a “cross-case study” and he sees case studies and cross-case studies as lying on a continuum (Gerring 2007:95). It should be noted that when...a single-country study is considered comparative if it uses concepts Gerring discusses case studies, his discussion is not limited to single cases. that are applicable to other countries, and/or seeks to make larger inferences that stretch beyond the original country used in the study A key decision in single-country studies concerns the selection of countries. (p.28) Countries may be selected simply because the researcher is familiar with them or has access to them, because they have not yet been studied, or, because they are Even if a case study does not itself constitute comparative research, good seen as being important in relation to other cases or studies. Thus countries may descriptions of individual cases are useful as raw material for comparisons, or as be chosen because they are considered to be representative of a category or group the first step in a comparative study (cf. Landman 2008:5). Lijphart (1971: 691- of countries, exceptional, or counterfactual. Much depends on whether the 693) described the “scientific status of the case study method [as] somewhat country is chosen for purposes of generating or testing hypotheses. ambiguous” (p.691), but distinguished six types of case studies on the basis of their potential contributions to theory development in political science. From having If case studies are used as a substitute for experimentation with the intention of been treated with some suspicion, the case study is making a comeback. A very testing hypotheses, comparativists may seek counterfactuals, situations in which the thorough and lucid overview of the case study is found in Gerring‟s (2007) chapter conditions that supposedly gave rise to the phenomenon or situation being studied in the Oxford handbook of comparative politics. He suggests that there is a growth in are absent. Counterfactuals can be theoretical and imaginary, or real cases, where interest in case study research design, possibly a movement away from the the required counterfactual situation exists naturally (Landman 2008:14-15). To variable-centered approach due to a number of factors, including growing take an example from LIS: In a text widely used in „foundations‟ courses intro- discontent with “cross-case observational research” (many-country comparisons) ducing American students to library and information science, Richard Rubin and an epistemological shift away from the positivist model of explanation. How- (2004:260) identified three “prerequisite conditions... for libraries to prosper”: ever, the case study is still viewed “with extreme circumspection”. Paradoxically, centralization, economic growth, and political stability. A country with thriving Gerring (2007:93) points out that while case studies have taught us a great deal, libraries but lacking one or more of these prerequisite conditions, does not not much is understood about the case study method. conform to the expectations generated by Rubin‟s theory and would constitute a counterfactual to challenge it. Gerring (2007:94-95) defines a case as “a spatially delimited phenomenon (a unit) observed at a single point in time or over some period in time. It comprises the In surveys of many countries, deviant or „outlier‟ countries, which do not fit the sort of phenomena that an inference attempts to explain” (p.94). This implies that general pattern, may be identified. Such countries may be chosen for more the case is selected or delimited on account of its potential explanatory value. In Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 11 intensive study to determine why they do not conform to the theory. Thus single- can be seen as units, that the features being compared can be measured, that these country studies can be used to confirm or infirm accepted theory and to provide features are sufficiently similar, and that variations in features in one country are insights for refining it (cf. Landman 2008:87-89). largely independent of variations of the same features in other countries. The latter assumption is referred to as „unit independence‟. Vast differences between Countries may also be chosen because particular characteristics are present in countries call into question the assumption that their features are comparable. For them to an extreme degree, because the case appears to it lend itself to the study example, in 2010 the smallest member of the United Nations, Nauru, had a of causal mechanisms, or because a policy of interest has been implemented there. population of under 10.000, while that of the most populous UN member, China, On the other hand, a country may be chosen because it is thought to be repre- was estimated at 1,3 billion. The assumption of unit independence can also be sentative of a group or category of countries. This raises the question of classi- questioned (Landman 2008:52-54). It is possible that some of the cases are not fications or typologies of countries, which is briefly discussed under few-country independent of one another. This is referred to as „Galton‟s problem‟: a comparisons below. relationship empirically determined between presumed independent variables P, Q and R and a dependent variable Y within three countries A, B and C may result Case studies are particularly useful for generating hypotheses, exploring from the fact that country A influenced countries B and C, rather than from causal phenomena, determining causal relationships, tracing causal mechanisms or relationship between the independent variables P, Q and R and the dependent pathways, offering in-depth insights, and dealing with heterogeneous entities. variable Y. Thus the causal relationship was not within-country but across countries Ontologically speaking, “case study researchers tend to have a „lumpy‟ vision of the (cf. Lijphart 1975:171). Globalization further calls into question the assumption of world: they see countries, communities and persons as highly individualized unit independence, particularly in the case of smaller countries which are highly phenomena” (Gerring 2007:98-109). While this suggests an affinity for susceptible to outside influences, such as those exercised by western education and interpretivist metatheory, Gerring (2007:115-116) points out that case studies media. may take on many forms and can be used within any paradigm. Nevertheless, many-country comparisons lend themselves to the formal testing of hypotheses. When hypotheses are to be tested, a relationship holds between the Many-country comparisons number of variables and the number of cases. The more variables that may exert a potential influence on the phenomenon under investigation, the more cases are Many-country studies are also referred to in the literature as survey studies, cross- needed to test all the possible combinations of several variables. As an example, let sectional studies, cross-case research or large-N studies. The methodology is us assume that we wished to test the hypothesis that the integration of school usually quantitative and typically involves multivariate analysis, i.e. simultaneous media centers in the school curriculum is more advanced in English-speaking statistical analysis of data collected on more than one variable. Use of qualitative countries where school media specialists are formally certified and are required to methods in analysis of many-country comparisons is unusual because “a richer have dual qualifications in library science and education, than in other countries level of information” is needed, including “deep history”, which would be difficult where there is no formal certification and dual qualifications are not required. to collect and analyze if large numbers of countries are involved (Landman Here we have one dependent variable (degree of integration of the media center in 2008:52). the curriculum), and three independent variables: language of country, presence or absence of formal certification, and qualification requirement (single or dual). Ontological assumptions underlying many-country comparisons are that countries Simplistically, to test the relationship formally we would need a three-dimensional Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 12 contingency table as in Table 4.6: concept of democracy by using an operational criterion that would not exclude too many countries. Thus we might decide that a country is considered to be Table 4.6: Contingency table for three independent variables democratic if its current leader was elected in a general election, regardless of the Independent variables Dependent variable: fact that such elections are often rigged. Degree of integration of School Media Center in Curriculum Since the total number of countries is relatively small and this number tends to be Language Certification Single/dual qualification None Low Med High further reduced by the factors just mentioned, comparativists do not commonly English Yes Single select countries by means of sampling. Instead, all the countries that satisfy given criteria or belong to defined types and for which data are available, tend to be Dual included. Sampling may, however, be used in comparative studies in which more No Single numerous sub-national units (e.g. provinces, counties) constitute the cases. Of Dual course, sampling may be used within cases if data are collected at a lower level of Other Yes Single analysis. For example, the libraries in Table 4.4 could have been selected using Dual random sampling even if the countries were selected purposively. No Single In statistically-oriented many-country comparisons there may also be problems Dual relating to the validity and reliability of measures used in comparisons. An example would be per capita GDP, which says nothing about the distribution of Using dichotomous variables as here, we need a minimum of 2x2x2=8 cases to income, or literacy rate, which is measured differently in different countries. The control for all possible conditions. Failing that, there would not be enough degrees dichotomous variables that were used in the example depicted in Table 4.6 of freedom for a statistically valid test. If we allowed more values per independent illustrate a measure taken to prevent the occurrence of empty cells. However, variable (e.g. for Language: English, French, Spanish, Other) we would need more dichotomizing this variable holds a threat to the validity of the study: by cases (4x2x2=16). If we added another dichotomous variable (e.g. Governance of characterizing certification as either „certification‟ or „no certification‟, no education system: centralized or decentralized) we would need 4x2x2x2=32 cases cognizance is taken of different forms or procedures of certification, which may be to avoid having lots of empty cells and not enough degrees of freedom. Thus many- germane to the relationships being investigated. To use the distinction discussed in country (large-N) comparisons are needed for performing valid statistical tests. the previous section, many-country comparisons are essentially variable-oriented. Cases are disaggregated into variables. Variables are measured, but a major However, there are limitations. The number of variables that can be included in a weakness of quantitative many-country comparisons is that variables tend to be statistical model is quite limited. There are around 220 countries and inhabited conceptualized and measured at a shallow level. We can determine with some territories, of which some 30 have fewer than 100.000 inhabitants. Statistical data degree of confidence that relationships exist between the variables, but this may may not be available from all of them. Data may be out-of-date or unreliable. not tell us very much about the nature of the relationships (cf. Lijphart 1975; Some countries may be disqualified from the study for other reasons. The problem Ragin 1987). is exacerbated when the study population is limited by other criteria, for example, if we decided to limit the study to democratic countries. To ensure that we obtain enough cases for our statistical analyses we would be tempted to „stretch‟ the Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 13 Few-country comparisons The goals of case-oriented investigation often are both historically interpretive and causally analytic. Interpretive work... attempts to In terms of the number of cases being compared, few-country comparisons are account for significant historical outcomes or sets of comparable found on the continuum between single-country studies and many-country outcomes or processes by piecing evidence together in a manner sensitive to historical chronology and offering limited historical comparisons. The countries can be as few as two. Two or three appear to be the generalizations which are sensitive to context. Thus, comparativists most prevalent number in recent comparative studies in LIS. The deciding factor, who use case-oriented strategies often want to understand or interpret however, is not so much the number of countries, but the methodological specific cases because of their intrinsic value. Most, but not all, case- approach. oriented work is also causal-analytic. This companion goal is to produce limited generalizations concerning the causes of theoretically Various terms are used for studies comprising a small number of cases. For some defined categories of empirical phenomena... common to a set of authors (e.g. Lijphart 1971, 1975) this is “the comparative method”; he also refers cases. to the “comparative-cases strategy” (Lijphart 1975:163). Ragin (1987: 34-52) places it under the rubric of “case-oriented comparative methods”. Smelser (1976, What this implies is that the case is of interest in itself and not merely as a bearer quoted in Ragin 1987:31) refers to it as the “method of systematic comparative of a set of variables, and that relationships within a case are of at least as much illustration”, „illustration‟ suggesting that it is an adjunct method, not suited for the interest as the generalized relationships among variables across cases. Because in serious task of testing hypotheses. Indeed, the terminology often reflects the few-country comparisons the comparativist studies the selected countries in depth methodological orientation (quantitative/qualitative) of the writer. Quantitatively- and is closer to the data, the problems of comparability and concept stretching oriented authorities tend to see a few-country comparison as a less desirable or (referred to above in connection with many-country comparisons) are alleviated: watered-down version of studies using larger numbers of cases, and they appropriate countries can be chosen, and richer, multidimensional, less abstract emphasize methods of compensating for its perceived weakness by approximating concepts can be employed. Furthermore, considerable attention can be paid to the inferential value of many-country comparisons as far as possible (e.g. Landman unraveling complex relationships, including relationships of multiple and 2008). Lijphart (1975), who has identified a number of advantages of few-country conjunctural causation, within each country, and over time. As Ragin (1987:23- comparisons in relation to many-country comparisons, nevertheless sees them as a 26) has pointed out, the complexity of social phenomena is not only a function of “method of testing hypothesized relationships among variables” (p.164), using the the many causes that may be responsible for a given effect. It also derives from the same logic as many-country comparisons with the difference that countries are effects of conjunctures, where a particular combination of factors has to be in carefully selected to compensate for the inability to sample from a large popu- place or in sequence before a given effect can occur. The depth of analysis makes lation. for a high level of internal validity. On the other hand, despite the considerable investment in time and resources needed for such in-depth studies, their findings On the other hand, qualitatively-oriented authorities tend to consider few-country cannot readily be applied to develop broad generalizations explaining phenomena comparisons on their own terms as insight-generating, in-depth studies of cases as in countries not studied – hence their external validity is low compared to that of wholes and as opportunities to study multiple and conjunctural causation. This is many-country comparisons. more aligned with interpretivist metatheory. Thus they adopt the case-oriented approach as described by Ragin (1987:35): A critical question in few-country comparisons, as it is in single-country studies, is which countries to select. In few-country studies the countries are not selected by Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 14 sampling. Instead they are carefully selected for the purpose of the study (Ragin Figure 4-III: Public libraries and literature: most similar systems design 1987:15). Selection of countries for single-country studies was touched on earlier. Principles applied there are relevant here as well, but additional factors come into play when we consider few-country comparisons. It is intuitively obvious that MSSD: Most Similar Systems Design there is little point in comparing entities that are so different that hardly any Country A commonality can be found (e.g. Nauru and China). Neither would it be useful to Former British colony Local public libraries Medium literacy compare entities that are so similar that little difference of interest can be found. Medium GDP When countries are selected for comparison, they should be comparable in Multiple languages respect of the phenomenon or theory that is primary interest in the study. Sartori Country B (1991:246) has stated that entities to be compared should have both shared and Former British colony Local public libraries Medium literacy non-shared attributes. They should be at the same time “similar” and Medium GDP “incomparable”. Multiple languages Country C A first step in selecting relevant countries may be to narrow the field to countries Former British colony Low literacy No Local public libraries in particular regions or in particular categories, such as democratically governed Medium GDP countries, francophone countries, Islamic countries or developing countries. This Multiple languages raises the issue of classification and typologies, which is dealt with in a separate 7 section below. If it is intended to uncover causal relationships or conditions associated with Here we are interested in the role of literacy in relation to the presence of local particular developmental pathways, there are two basic design strategies for public libraries. By selecting countries that are very similar in respect of other selecting countries for comparison. These strategies are related to the methods for characteristics (in this case, their colonial history, GDP, and number of languages determining causation that were formulated by the British philosopher J.S. Mill. spoken) we in effect control for the influence of those variables, which otherwise The basic choice is between the „Most Similar Systems Design‟ (MSSD), which might have been thought to influence the presence of local public libraries. We can corresponds to Mill‟s „Method of Difference‟ and the „Most Similar Systems therefore say that ceteris paribus (all things being equal) there is a relationship Design‟ (MDSD) which corresponds to Mill‟s „Method of Agreement‟ (Pennings et between literacy level and the prevalence of public libraries. The ceteris paribus al. 1999: 43-49; Landman 2008: 70-76; Hantrais 2009:59-64). principle is important. What we are doing here is in effect to simulate the operation of experimental controls (which we cannot exercise in real life In a most similar systems design (MSSD) we select countries that are very similar in all situations) by matching the countries on the variables we need to control for. Note respects except in respect of the particular factor or variable of which we want to that we did not say that literacy levels determine or cause the prevalence of public study the effect. This is illustrated in Figure 4-III. libraries. The relationship of cause and effect may well operate in both directions. The most different systems design (MDSD) is depicted in Figure 4-IV. Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 15 Figure 4-IV: Public libraries and literature: most different systems design Trade-offs between cases and variables MDSD: Most Different Systems Design Country D Earlier I referred to the trade-off between cases and variables. Essentially, in many- Nordic Local public libraries country comparisons (which tend to have greater inferential power in terms of the High literacy High GDP ability to generalize with confidence) it is not possible to deal with as many Two languages. variables, or to deal with them in as much depth, as in few-country comparisons Country E where we can have greater confidence that we fully understand the complex Latin America Local public libraries relationships, interactions and causal mechanism among variables. This raises the High literacy Medium GDP question as to what can be done to combine the strengths of the two designs, and One language specifically to increase the inferential power of few-country comparisons. Country F Landman (2008: 27-30; 68-70; 79-82) puts much emphasis on combining Former British colony High literacy Local public libraries quantitative and qualitative methods and on methods of statistical inference when Medium-low GDP few cases are studied. Ragin (1987) developed a method of “qualitative Multiple languages 8 comparative analysis” using Boolean truth tables, which has been lucidly summar- ized by Landman (2008:79-81). Another approach is to multiply the number of cases by repeated measurement over time. In the most different systems design we take the opposite approach. We select a number of very different countries that do, however, share the phenomenon we are interested in, in this case again, the presence of local public libraries. The The time dimension countries depicted in Figure 4-IV differ in respect of their cultural-linguistic groups, their GDP, and the number of languages spoken. Because local public In the literature there has been some discussion as to whether the historical or libraries are present in spite of the differences in these factors, this suggests that diachronic perspective is appropriate in comparative studies. there is a relationship the presence of public libraries and the one factor they do have in common, a high literacy rate. Again it is worth pointing out that we In qualitative studies, the historical perspective is seen as an essential component cannot say that the high literacy rate is the cause of the presence of public libraries. of the detailed description that is expected. Sweeting (2005) discusses the relat- We can only say that there is probably a relationship between these variables. ionship between comparative education and history of education and asserts that they complement each other. The historical dimension adds another level of There has been a great deal of philosophical discussion and criticism of Mill‟s complexity, as it requires the special expertise needed in finding, evaluating and methods and various weaknesses have been pointed out (Hantrais 2009:62-64). In utilizing primary sources, but Sweeting adds that the historical perspective can add the decision on a design, other factors such as the number of cases also play a role much value in cases where and further variants and refinements of the methods are possible (Landman 2008:70-78; Pennings et al. 1999:43-49)....the characteristics of historical analysis – its concern for evidence, its tentativeness, its utilization of historical consciousness, its interest Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 16  in provenance, agency, seminality, and significance, its interest in other at the sub-national level, which can include local communities or individual connections, and its distrust of teleological 1 explanations – are human beings. Landman (2008:19-20) makes an analogous distinction between adopted in the process of comparing (p.40). macro and the micro levels, whereas three levels, macro, meso and micro levels, are discussed by Kennett (2001:6-7) and Hantrais (2009:54-55). The decision on In quantitative studies a historical dimension may be added when the number of the level of analysis in a study is closely related to the choice between a variable- cases is small and researchers want to increase the number of cases for purposes of oriented and a case-oriented approach, as discussed earlier. Macro level studies statistical validity (cf. Lijphart 1971:689).Thus in a study of eight countries, tend to be variable oriented and micro level studies tend to be case-oriented. measurements taken four times at say, 5-year intervals, could be used to increase the number of cases to 32. This is called the “pooled cross-sectional time-series However, there are also underlying ontological beliefs affecting the decision. In analysis” (Landman 2008:32). It should be noted that this is a fundamentally political science this is referred to as the „structure-agency‟ problem: ahistorical approach, as it is the situations as at those points in time that are of interest (so that they are conceived as equivalent to separate countries), not the Micro-analysts believe that the world of politics is shaped by the development of the situation over time. In this connection Lijphart (1975:171- actions of „structureless agents‟, while macro-analysts believe that the 172) warns against the danger of Galton‟s problem and “case-stretching” when world is shaped by the unstoppable processes of „agentless-structures‟ using this method to increase the number of cases. Clearly, if data are repeatedly [sic] (Landman 2008:19). collected in respect of the same country, these „cases‟ cannot be considered to be independent. Thus, in international comparisons of public library development, theorists who emphasize the role of large-scale structures and processes, such as the shifting More information on the time dimension in comparative studies can be found in relations of social classes, democratizing and popular educational movements and Pennings et al. (1999:49-54), who discuss the role of space and time, government policies, would focus on these phenomena at the macro level. On the distinguishing between designs that are located in the time dimension only (time other hand, theorists who emphasize the role of individual activity would focus at series and cross-sectional designs in single countries) and designs that are located the micro level on the roles of cultural, educational and library leaders and in both time and space (including pooled time series designs in multiple activists, and may want to examine how they interact at the local level. In studying countries). information service development in developing countries, macro-theorists may want to emphasize matters such as the telecommunications infrastructure, the cost of internet connectivity and government policies on the importation of IT Further design decisions equipment, while micro-theorists may want to examine quite closely the social and cultural interactions that take place within rural communities when The term level of analysis was introduced earlier. Selecting an appropriate level of telecenters are put in place there, or the ways in which more and less receptive analysis is an important decision in designing a comparative study. In cross- school principals and teachers integrate the use of information sources in their national studies Novak (1977:12) distinguished two levels, one essentially at the curricula and classrooms. national level (“the human aggregate corresponding or equivalent to a nation”), the Increasingly international comparisons are conducted at more than one level at the 1 same time. The examples cited above illustrate that combining analysis at the two Here “teleological explanations” refers to explanations of historical events in terms of larger frameworks such as „God‟s plan for our people‟, or „manifest destiny‟. levels would provide a more balanced assessment. Multilevel studies make possible Peter Lor: International and Comparative Librarianship, Chapter 4 draft 2011-04-20 page 17 combinations of methodological approaches, thus providing richer sources of data individual example or case (Mouton & Marais 1990:137). The distinction made by and applying the principle of triangulation that was referred to earlier. In doing so, Knuth (1995) between “American” and “British” models of school library however, care must be taken not to fall into the trap of confusing the levels of development approach such ideal types. In typologies phenomena are often analysis and committing aggregate (ecological or individualist) fallacies (Hantrais classified in terms of more than one variable or dimension. An example from 2009:55). In this connection it is worth mentioning the effect of distance from the political science is Lijphart‟s well-known typology of democratic political systems phenomenon being observed, as discussed by Hantrais (2009:56-57). For in terms of (a) the nature of the political elite (which can be competitive and example, a „long-distance‟ study of library development in sub-Saharan Africa adversarial or coalescent and cooperative ) and (b) the political elite (which can be undertaken from Europe or North America would reveal a much more uniform homogeneous or fragmented). Applying these two variables produces a typology of situation than a „close-up‟ study looking at community libraries on the Cape Flats four cells, as in Table 4.7, adapted from Lijphart (1968). around Cape Town, South Africa, and rural village reading rooms in Botswana. Depending on the focus, Hantrais suggests that the research design needs to be Table 4.7: Lijphart’s typology of democratic political systems adjusted to ensure an appropriate level of analysis and the right focus (Hantrais Political culture 2009:57). Homogeneous Fragmented Coalescent Depoliticized Democracy Consociational Democracy (e.g. (e.g. Nordic countries) Netherlands) Classification and typologies Elite Competitive Centripetal Democracy (e.g. Centrifugal Democracy behavior UK, United States) (e.g. Italy) As mentioned earlier, in comparative studies we usually need to select or group countries, or both. Classifications are developed by comparativists to group In addition to serving purposes of exploration and explanation, such a typology phenomena such as countries into “distinct categories with identifiable and shared can be used as a frame of reference for the collection and analysis of data (Mouton characteristics” (Landman 2008:5-6). In addition to facilitating the selection of & Marais 1990:138). Ragin (1987:20) points out that such typologies are countries for study, classifications help to simplify the complexity that emerges important because they set boundaries on comparability. As an example he from contextual description by grouping entities into simpler categories that can mentions the comparability of “dependent” countries. Among developing form the basis for cross-national comparisons alo