A complete parties’ scholar

Janda as teacher/researcher

Kenneth Janda is the epitome of the teacher/researcher; in his research he is always teaching, and in his teaching he instructs how to do research. Literally from the beginning of his professional career, Janda has incorporated research into his teaching, and some of his most impactful research has benefited from that association.

In the early 1960s, he introduced judgmental data collection into his undergraduate parties’ class at Northwestern University, and thus were developed the first building blocks of the International Comparative Political Parties (ICPP) project (which was not officially founded until 1967). As Janda tells the story himself in his 1980 data tome:

This study had its genesis in a course on ‘political parties and elections’ taught in the spring of 1962 during my first year on the faculty at Northwestern University. I thought then (and am convinced now) not only that the cross-national study of political parties was a worthy end in itself but that a proper understanding of American parties required cross-national comparisons. The class project was to gather data on some of Maurice Duverger’s major concepts for political parties in countries across the world with the objective of testing his major propositions. Although the project brought forth no usable data that year, I made some improvements in our library research procedures and tried it again the following year. Working mostly with published literature, my 1963 parties class coded in full or in part some 205 parties in 55 countries. The data they produced validated several of Duverger’s propositions, for example, that parties organized on a ‘caucus’ basis are more likely than ‘branch-based’ parties to restrict their activities to contesting elections, while disproving others. (1980b: xv)

While for most of his peers the ICPP project was first and foremost the producer of a unique data set, for Janda it was also – and at least equally so – a teaching/learning tool. The very production of the data taught the project’s undergraduates and graduate students not only about the nature of the parties they were studying – which was important in its own right – but also how to think creatively in producing data on difficult concepts.

Would it really be possible to produce quantitative data for such theretofore ‘immeasurables’ as party centralization of power, degree of organization, or interest aggregation and articulation, and to do so cross-nationally (indeed, even holonationally)? Janda thought so. It may not have been easy – and certainly wasn’t – but he proceeded to carefully develop coding schemes that students and others could use as guides for applying numerical codes to textual information gleaned mostly from secondary literature on parties. The process was painfully time-consuming and relatively capital-intensive, but the payoffs in learning – both along the way and at the end – more than justified the cost. Though Janda himself said in the preface to the tome ‘If transported back in time and given the chance to decide about doing the project, would I do it again? The answer is no’ (Janda, 1980b: xviii), others are certainly pleased that he did it the first time. The data set itself – which has fueled many interesting projects by Janda, his students, and others – is arguably less important than what was demonstrated by the process: that with creativity and hard work, it is possible to extend the reaches of quantitative testing of empirical theory well beyond the ‘naturally measurables’.

In his most recently published book (Janda and Kwak, 2011), Janda and his co-author again combine reporting of original research results with instruction and motivation for students to seek answers to their own empirical questions. The first sentence of their preface reads: ‘This book targets three audiences: students, teachers, and researchers’ (Janda and Kwak, 2011: xiii). Later, Janda says:

It invites researchers to consider more innovative approaches to cross-national analysis of party systems, studying a greater range of countries, considering new measures of party system competition, and suggesting that new measures be not merely proposed and mathematically adjusted through scholastic exchanges in professional journals but actually applied in testing party theory with empirical research. (Janda and Kwak, 2011: xiii)

Indeed, that latter sentence does a nice job of summarizing the multiple foci of Janda’s career of combining teaching with research. While advocating for a broader comparative perspective, for more creative and precise measures for difficult-to-measure concepts, and for extending empirical theorizing and testing into more areas of party research, Janda has consistently taught by example. And by example he has helped to teach both his students and his peers in the study of parties not only that we should extend our horizons in the study of parties, and why, but also that it can be done, and how.

Rather than just admonishing his fellow American students of parties on the self-centric focus of their research, Janda set out to demonstrate the utility of learning about one’s own parties and party system through comparison with others. Indeed, it could be argued that the lesson extended to showing Eurocentric peers (many of whom were already more ‘comparative’ than American counterparts) what could be gained from adopting a more holonational perspective. In Party Systems and Country Governance, after discussing a number of countries ranging from dictatorships to failed governments to established democracies, Janda asks ‘Is it possible to meaningfully compare such diverse nations concerning how well they deliver the benefits of government?’ and then characteristically answers ‘We think so’ and sets out in the rest of the book to demonstrate how it can be done (Janda and Kwak, 2011: xxiv).

Not that any of that would be surprising to those familiar with the much earlier ICPP project! There Janda was not content to limit his study to the American parties, or just the parties of English-speaking countries, or just the parties of established democracies. No, he was convinced that in order to produce and test general theories of parties, we needed to be dealing in concepts that were broadly (if not universally) applicable and with data covering the whole range of parties and party situations (e.g. see Janda, 1980b: xii; see also Janda, 1993: 163–166.) As testimony to the relative novelty of that idea at the time, Janda included the following justification in a 1984 paper on the project’s conceptual framework:

One school of thought would argue that such a collection of disparate entities called ‘parties’ is nothing more than a stew of apples and oranges and that little can be expected from any effort at ‘comparing’ the German Democratic Party, for example, with the Kabaka Yekka of Uganda or the Paraguayan Liberals. To the contrary, the intellectual impetus behind the ICPP Project is that the enormous diversities among political parties throughout the world can be accommodated within a relatively few major concepts or dimensions of variation. Moreover, diversities within these dimensions conform to patterned relationships, specified in advance, which hold among political parties of all types and across cultural settings. (Janda, 1984: 16)

While limitations of time and labor would preclude production of data for all parties in every existing country, it might be possible – and indeed proved possible (with some well-documented exceptions) – to produce data for a large sample of significant parties across the full range of party situations. 158 parties operating from 1950 to 1962 in 53 countries, representing all 10 major geo-cultural regions of the world, to be precise!

Once again, Janda had taught not only his students, but his peers and future generations of party researchers as well. Not only is it important to work outside the box – or in this case, outside one’s own familiar parties and party systems – but it is also possible to do so.

Janda’s proselytizing for extending the range of comparison has applied not only cross-sectionally, but longitudinally as well. In a 1980 article, he argued for – and demonstrated the utility of – extending measurement of ‘party system change’ to a longer timeframe than the inter-election comparisons that were previously dominant. ‘Granted that we must begin with the dynamics of change from time 1 to time 2’, he noted, ‘we seldom limit attention to changes between two time points and instead seek measures of system instability over a series of time points’ (Janda, 1980a: 418). He then went on to illustrate one relatively simple alternative for doing so. In his later data collection effort with Harmel (the Party Change Project), where the ultimate objective was testing of empirical theory on party change, the variables and procedures of the ICPP project were adapted for producing annualized data for the period 1950–1990 for 19 parties in four countries.

When Janda undertook these efforts, he may not have been the only advocate for broader generalization in the study of parties and party systems, but he was certainly a significant and early contributor to the advances we have now seen in these areas – not just or most importantly by doing it all himself, but rather by ‘teaching’ the rest of us through the examples of his own research.

More generally, Ken Janda has taught by example the importance of taking great care in:

Defining concepts;

For Janda, teaching and research are not separable responsibilities; they are inseparable components of the scholar’s purpose.

Janda as conceptualizer and operationalizer

In addition to encouraging parties research that was more truly ‘comparative,’ across both space and time, Janda’s work has also motivated growth in the empirical study of party organization. As Janda himself posed the problem in a 1983 article:

Party organizations are the leprechauns of the political forest, legendary creatures with special powers who avoid being seen. Because no one has ever photographed a party organization, descriptions vary widely and many scholars do not take them seriously enough to investigate their being. Very few scholars study party organization in comparison to the study of concrete things like voters and nations, and you could put into a briefcase all the studies that systematically and empirically compare party organizations across nations. Why don’t more people study party organization? It is not for lack of interest, for scholars often discuss party organization. It is that party organizations are so intractable for research. Unlike voters who politely answer questions and unlike nations which dutifully generate statistics, party organizations rarely tell about themselves. Because one must imagine what these invisible creatures look like, it is very uncertain business to measure and compare party organizations. (Janda, 1983: 319)

Through his ICPP project (self-described (1983: 326) as ‘the first systematic, empirically based, comparative analysis of political parties across the world’), he was already demonstrating one way of tackling the problem, and since then many others have followed his lead – either using his measures and data or producing their own. (As Janda noted in that 1983 piece, others were already beginning to address pieces of the problem, but their general tendency to focus on just competitive parties in democratic countries significantly limited their conceptual scope. See Janda, 1983: 320–321.)

But the data production project was not conducted in a theoretical/conceptual vacuum! To the contrary, each of the ICPP’s more than 100 variables for which data were collected was tied to one of 12 ‘main concepts’ gleaned from the theoretical literatures on political parties and, especially, organizational theory. These included eight on parties’ ‘external relations’ (institutionalization, governmental status, issue orientation, goal orientation, autonomy, and social attraction, concentration, and representation) and four on their ‘internal organization’ (degree of organization, centralization of power, coherence, and involvement). Starting from theory-derived concepts and working down to the appropriate measurement was key to Janda’s approach. In the 1983 article, Janda noted two main advantages to building from the abstract concepts: ‘one is that the more general concepts facilitated comparisons among very different political parties in various regions of the world, by giving additional meaning to the parties’ particular organizational features,’ and the other ‘that organizational theory can help one theorize about party processes’ (1983: 329).

At the operational level, each of the 12 main concepts would be measured by multiple basic variables. Organizational complexity, for instance, would be operationalized through measurement of seven indicators: structural articulation, frequency of local meetings and of national meetings, maintaining records, and intensiveness, extensiveness, and pervasiveness of organization. In all, data would be collected on 111 such ‘basic variables’.

While the ICPP project provides an excellent illustration of the meticulous effort Janda has put into identifying, defining, and operationalizing key theoretical concepts, it is but one of many examples. In a paper where he devotes more than 10 manuscript pages to discussion of the concept ‘governance’ (Janda, 2010: 2), he states the principle that should guide all conceptual/definitional work in social science: ‘So the issue in answering the question – “What is governance?” – is whether its definition advances understanding. In other words, is the concept linked to the term useful to inquiry?’ Throughout his career, Janda has been known for his adherence to that principle, always being very precise in his definitional work, and always driven by the goal of defining concepts so as to maximize their value for building and testing empirical theory.

Whether distinguishing ‘degree of organization’ from ‘centralization of power’ (Janda, 1980b: Chapters 9 and 10), or ‘Party Law’ from mere ‘party law’ (Janda, 2005: 4), or laws that ‘proscribe’ from laws that ‘promote’ parties (Janda, 2005: 8), or ‘competitive party’ from ‘subversive’ and ‘restrictive’ varieties (Janda, 1980b: 5), Janda has aided the research of others through his careful definitional work. Indeed, in distinguishing among the party types, Janda effectively broadened the concept of ‘party’ well beyond its traditional usage. By defining parties as ‘organizations that pursue a goal of placing their avowed representatives in government positions’ – thereby leaving out any reference to elections as the vehicle for doing so, as is commonly included in other definitions – Janda intentionally made way for inclusion of parties that use other vehicles (subversion or restriction) for attaining or retaining public office.

Always the believer in and practitioner of the multiple-indicator approach, Janda’s work also exemplifies ‘best practice’ in the realm of precise operationalization. Guided by theory and especially by clear conceptualization, the next step in each of his projects was to be just as meticulous in identifying or creating operational indicators. As evidence of the care taken in matching ICPP project indicators with concepts, and then the subsequent care taken to make certain that the indictors were performing as anticipated, Janda formed 18 scales and subscales related to the original conceptual framework. He reported:

The reliabilities of these scales ranged from .69 to .96 and averaged .82. Scholars who wish to utilize the existing data base … should find it helpful to know that the basic variables in that data base do tend to interrelate as originally conceptualized. (Janda, 1984: 22)

Beyond the ICPP project, Janda has since then tackled – with some success – operationalization of party performance (using ICPP project indicators; Janda and Coleman, 1998) and two dimensions of governance (using World Bank indicators; Janda, 2012; Janda and Kwak, 2011), grappled with operational issues regarding various dimensions of party systems (Janda, 2010; Janda and Kwak, 2011), and provided guidance for operationalizing multiple types of party law (Janda, 2005).

Janda as data builder

One can’t fully appreciate Janda’s impact on the study of parties without acknowledging his contributions to the art and science of data collection.

To quote Janda himself: ‘Cross-national research on the effects of party organization has been lacking due to the absence of suitable cross-national data’ (1983: 330). Characteristically, Janda accepted the challenge embedded in those words.

How could one possibly test hypotheses requiring ‘quantitative’ data on political party organization? For his ICPP project, Ken developed an answer. After carefully constructing ‘judgmental coding schemes’ linking numbers to ordinal descriptors, Janda set about gathering and organizing relevant information on nearly 150 parties from the extant literatures, both primary and mainly secondary sources. Since computers were not yet advanced enough to be very helpful with such a text-intensive enterprise in the early 1960s, Ken employed the best available option: Eastman Kodak’s ‘MIRACODE’ technology (see Janda, 1967, 1982). After three years of collecting and indexing information from more than 60,000 pages of material (Janda, 1984), Janda’s coders – mainly graduate and undergraduate students who had been trained in application of the coding schemes – began more than seven years of coding. Finally the data set was completed and deposited at the ICPSR, and Janda’s first book from the project – Political Parties: A Cross-National Survey – was published (Janda, 1980b).

While Ken would later question whether it had all been worth it (1980b: xviii), there was another important lesson for the rest of parties’ scholars: don’t assume that it is impossible to produce quantitative data just because a concept is not automatically quantifiable! ‘Obtaining the proper data to operationalize and test a complex social theory is often more a matter of creating the data than finding them’ (Janda, 1984: 14). Without that lesson, much of the recent development and testing of empirical theory on party organization would have been impossible.

Though Janda would not again delve into such a vast project, he did join with Harmel in producing a more modest ‘party change’ data set, based substantially on procedures developed for the ICPP, but now with the objective of producing annualized data for parties in just four countries. And later he again faced challenges of too little data for addressing important questions, and again he stepped in and produced holonational data sets on party laws (Janda, 2005) and with Jin-Young Kwak on party systems (Janda and Kwak, 2011).

Janda’s impact as data-builder has reached beyond the sharing of his data with others. Always creative (in some cases, ‘ingenious’ may not be too strong a word) in developing approaches for producing empirical data with which to address important research questions on parties, Janda’s work has encouraged others to think outside the box in similar pursuits. ²

Janda as theorizer and tester

As with any complete social scientist, ‘theory’ has always been prominent in Janda’s work – at both the beginning and the end of each of his projects. Though perhaps known best by some of his peers for the data he has produced, theory has always driven the conceptualizing, the operationalization, and ultimately the data production.

Whether to test Duverger’s theoretical insights, to guide research on party change, or to drive his work on party systems and governance, Janda’s primary role as theorizer has been that of ‘formalizer’. Janda began formalizing theory at a time when formalizing meant clearly identifying and distinguishing between critical assumptions and propositions, demonstrating how each proposition was derivable from some combination of assumptions and/or other propositions, and then deriving testable hypotheses from the more abstract propositions. The yield was always a set of testable statements which ultimately served the social science goal of supporting or rejecting logically plausible explanations.

Always drawing from and building upon extant theories in the literature, Janda formalized Duverger’s theories on party structure (Janda and King, 1985), drew upon the works of Downs and Strom and others and formalized an ‘integrated theory of party goals and party change’ (Harmel and Janda, 1994), and formalized a theory of party systems and party governance (Janda and Kwak, 2011). Each time, the refinement of theory has supported Janda’s own empirical work, bringing what have often been data he has produced himself to bear in testing key hypotheses.

Janda and King (1985), for instance, not only formalized Duverger’s theories of party structure and organization (Duverger, 1954), but then operationalized and rigorously tested several of the hypotheses using ICPP data. In formalizing key propositions and demonstrating their testability, Janda’s and King’s efforts brought Duverger’s seminal work to life and moved it back into the mainstream of empirical political science research.

Since the 1990s, Janda’s theoretical work on party change has stimulated others to explore the question of why parties change, when, and how, in spite of being large and inherently conservative organizationally. Janda’s contributions in this area began in 1990 with a paper presented at the Sociology World Congress and titled ‘A performance theory of party change’. Later, Janda and Harmel built upon that theory and drew also upon the works of Downs (1957), Strom (1989), Panebianco (1988), and others and formalized their ‘integrated theory’ (Harmel and Janda, 1994). Published in the Journal of Theoretical Politics in 1994, it is currently one of that journal’s most highly cited pieces. In subsequent works, Janda joined with others in testing elements of the integrated theory (Harmel et al., 1995; Janda et al., 1995), using data collected by the Party Change Project.

Beyond his own research, Janda’s theoretical work has always fueled the work of others as well. For instance, the integrated theory of party change has been addressed in work by Bille (1997), Muller (1997), Tan (2000, 2002, Burchell (2001), Hines (2003), Rihoux (2006), Duncan (2007), Fell (2009), Pedersen (2012), Skinner et al. (2012), and more.

Conclusion: Janda’s lasting impact on the study of parties

In his 50+ years (so far) of studying political parties, Kenneth Janda may not have collected all of the necessary data nor developed all of the relevant hypotheses nor answered all of the important research questions, but he has shown the rest of us the way (or at least a way), and in that is the greatest potential impact of Ken Janda on the study of political parties.

Ultimately, though, ‘impact’ requires not just a leader/teacher, but followers/students as well. If the study of parties were no different today from when Janda began his work in the 1960s, it would indeed be difficult to argue that there has been much actual impact on the rest of the field.

But in spite of the fact that Janda himself found it necessary to again address his earlier concerns in his 1993 ‘state of the discipline’ review, a survey of the more recent work does indicate some measurable – albeit modest – change. While the study of party organization still takes a distant back seat to study of voting behavior, there is no doubt that there is more empirical study of party organization today than was the case when Janda started his ICPP project.

While the arduous task of data building still stands as a roadblock to developing and testing theory on a broad range of important research questions, it is nonetheless clear that some (e.g. those involved in the Katz and Mair project, 1992; those involved in the ongoing Political Party Data Base project directed by Susan Scarrow, Paul Webb, and Thomas Poguntke) have taken up the challenge implicit in Janda’s insight that ‘obtaining the proper data to operationalize and test a complex social theory is often more a matter of creating data than finding them’ (Janda, 1984: 14).

Likewise, while single-nation remains the dominant form of studies of party organization, truly comparative (i.e. cross-national) studies have certainly increased since Janda critiqued the field in the 1960s. At the time, he was one of just a handful of American students of parties to challenge their peers to put the American parties in comparative context, and since then more have answered that call.

To attribute these changes to any single advocate/exemplar would certainly be foolhardy, but it would be just as foolhardy to suggest that Ken Janda has had anything less than an important role in these significant changes in the agenda and approaches for the study of political parties. ³