Stark, Statistics, and the Early Church

In Cities of God (New York: HarperCollins, 2006), popular sociologist Rodney Stark attempts to bring quantitative, statistical analysis to bear on the study of the history of early Christian expansion. The description of the 'triumph' of Christianity, Stark argues, is incomplete, indeed unscientific, without the utilization of quantitative methods. "A major purpose of this book," writes Stark, "is to demonstrate that quantitative methods can help to resolve many debates about early church history" (22). He explains the book attempts "to identify adequate, quantifiable indicators of key concepts and then to properly test important hypotheses" (22, emphasis original). In so doing, the chapters of the book "will attempt to place the rise of Christianity within the appropriate social and cultural contexts, but they will do so in a more fully social scientific way than as ever been attempted" (22). Although the book is a provocative protreptic exercise, the book ultimately fails to convince because the author has not understood the pitfalls of the field he attempts to address.

Stark's tone is condescending and abrasive toward historians, most of whom he seems to think are out to disaffirm traditional Christianity. He characterizes all historical work by the ideology of what he calls 'higher criticism,' by which he means the admittedly subjective stance of many historians toward their subject matter. Rather than subjective, ideologically-motivated (whether explicit or implicit) theses and theories of Christian origins, Stark makes a plea for a more empirical model of investigating early Christian expansion. This is a fine and noble goal. But the subjective emphases of some should not be taken as the working methodologies of the majority of historical scholars. Not all reticence on the part of historians to engage quantitative methods stems from ideological commitments in antagonism toward traditional Christianity, which I will discuss below. At any rate, given the occasional blatant piece of misinformation in the book (e.g. that Acts is the only writing of the NT not written by a Jew, apparently unaware that Luke was written by the same author and that the authors of Hebrews and the deutero-Paulines need not have been Jewish [136]), one might hope Stark would have embraced a more humble tone. If, as the last chapter of the book says, historians ought to count, then perhaps sociologists should also read. It is clear from his limited use of the secondary literature on Christian origins that he is not really familiar with the field beyond the status of an amateur. It is refreshing when he admits specialists in the field are better positioned to provide important contributions than he is, if they would engage in quantitative analysis. After bringing out a laundry list of scholars he has employed in the book, Stark admits that none thought their work would be quantified. And why not? "Because nobody does this sort of thing" (213). I will discuss later the very good reasons (really one important reason) historians have for not doing this sort of thing. In the meantime, let us look further at Stark's arguments.

Stark is at his most provocative when he uses quantitative analysis to argue that early Christianity expanded not by mass conversions, but by gradual network expansion. Stark reasonably argues that if the Christian population of the Roman empire grew at a steady rate of 3.4% from 1,000 in the year 40 CE, by 100 CE the Christian population would be a modest 7,434, but by 350 CE the population could have grown to an impressive 31.7 million. This model of growth is very closely consistent with similar projections drawn from Christian epigraphs in Rome. This seems to me an entirely plausible growth curve for early Christianity, and it helps explain why Christianity was adopted by the emperor Constantine. Stark writes,

In 312, the year of Constantine's conversion, these projections show nearly 9 million Christians, making up about 15 percent of the population. ...Constantine was not responsible for the triumph of Christianity. By the time he gained the throne, Christian growth already had become a tidal wave of exponential increase. If anything, Christianity played a leading role in the triumph of Constantine, providing him with substantial and well-organized urban support (68, 189).

This is a provocative thesis, and one worth considering. Even so, it tells us very little we did not know before. In fact, Stark himself points out that these numbers corroborate numbers already projected by Robert Wilken and Robin Lane Fox. Indeed, Stark goes on to corroborate the commonplaces of scholars of early Christianity, namely, that Christianity grew more rapidly in larger Hellenic (rather than Roman) port cities in relatively close proximity to Jerusalem. This work is necessary, Stark argues, to show the validity of the quantitative methods he employs; their value, on the other hand, must be shown by their ability to illuminate disputed questions.

But it is precisely here that Stark fails to impress. Stark offers three studies that he thinks sheds new light on the early expansion of Christianity. The first case study Stark embarks upon is that of the Cybele and Isis cults. He argues that the cults of Cybele and Isis, as 'oriental' predecessors of Christianity, paved the way for Christian expansion. Stark follows Franz Cumont in asserting that the appeal of 'oriental' religions for Romans lay in three features: the emotionality of 'oriental' religions, their focus on the individual as opposed to the community, and their more sophisticated and moral picture of the gods. All three are questionably ascribed to so-called oriental religions, much less to Christianity (one might think of Paul's emphasis on the body of Christ, or of the communal notions of soteriology from Qumran; conversely, of the theurgical elements of Plato or the sophistication of Aristotle's theology). Despite the correlations Stark manages to draw between interest in Cybele and Christian expansion, the explanation of the connections, so dependent on the obsolete theories of Cumont, cannot be maintained, and Stark's quantitative analysis does not show its explanatory power, i.e. its value. The other two case studies are even less interesting than the first, as Stark proceeds to 'demonstrate' Paul's real missionary success was not among gentiles, but among Hellenistic Jews, and to further demonstrate that 'Gnosticism' was not an outgrowth of conventional Christianity but a motley assortment of pagan creations (following, but apparently ignorant of his significant modification of, Michael Allen Williams).

The main problem with Stark's analysis, though, is precisely that of numbers. Why is quantitative analysis important? "First of all," writes Stark, "the statistics inspire confidence because they are so stable and consistent" (220). Inspiring confidence is great...if that in which they inspire confidence is correct. Stark presents his quantitative method as more disciplined than historical research. That may be true depending on what we mean by disciplined: "In the end, what quantification mainly contributes to historical discussions and disputes is discipline. To work with quantitative data, one must make systematic arguments and draw clear conclusions--no dancing around and having it both ways" (222). Speaking of having it both ways, what do you call someone who admits that numbers for antiquity are incredibly hard to substantiate (34, 63), but then complains about scholars who shy away from quantitative analysis because such numbers are difficult to substantiate (63, 19)? In fact, it is a mark of the historian's discipline to admit when the evidence does not allow for certain kinds of analysis--including quantitative analysis. If the numbers cannot be substantiated the disciplined approach is the approach that understands its hypotheses to be less certainly verifiable based on the available evidence.

Let us take one example of how this plays out in Stark's analyses. In his discussion of the Christianization of the Roman Empire, Stark's data is not raw numbers. His indicator for Christianization is the presence of a Christian congregation. Those cities with a Christian congregation are scored with a one, those without are scored zero. (Let us forget for the moment that Stark has already narrowed his data to cities with over 30,000 inhabitants, for reasons that are not strictly quantitative.) This method of coding relies on Harnack's examination of early Christian expansion, but it raises serious questions. What constitutes a Christian congregation? Should congregations with dedicated buildings be included and/or coded differently? Should cities with multiple congregations be coded differently? How many Christians worshiped in each community, and what can this tell us about the population growth of early Christianity? Moreover, the working out of this position vis-a-vis Gnosticism is biased from the beginning. Why should Gnostic communities not be scored as Christian communities? The failure to do so already skews the data by the subjective decision to treat them as separate entities. This is a qualitative decision that is prior to the quantitative analysis, with profound effect on the outcome of the analysis. The coding of Hellenistic Jewish and Christian communities contains the same problem, since they were in many ways mutually-existing, indeed interwoven, communities in many places. In other words, Stark commits two fallacies here. First, he does not rely on hard numbers, because there are so few hard numbers on which to rely. Instead, he focuses on whether or not a Christian congregation was at a particular place (drawing on secondary literature), with complete disregard for the size or character of the congregation (this hardly inspires the confidence of hard statistics). Second, Stark makes a priori qualitative decisions about how to arrange the data he uses that potentially skews the results.

The most important problem here is that Stark fails to take account of missing data. The problem of quantitative analysis of early Christianity is precisely that we have such little extant data. We do not have firm numbers for the first centuries of Christian expansion. Stark's extrapolation of a consistent growth of roughly 3.4% is reasonable--but it is still an extrapolation. We do not know whether some communities grew faster or slower, whether some died (besides Jerusalem) and some boomed (in mass conversions or the like). The primary supposition Stark employs here is one of consistency and stability. Although Stark admits there was probably a decline in the rate of expansion by 350, he fails to admit that the rate of growth need not have been consistent either in any one place or from place to place. This mechanistic view of population growth can hardly be more detrimental to the results of scholarship than the materialist emphases Stark charges historians of highlighting. History is not statistics. Although I find Stark's extrapolation of early Christian growth plausible, I likewise recognize that there is nothing in Stark's model that guarantees its verdicts. While I think the historical record can probably back up his conclusion, I do not think the supposition of stability is itself a stable supposition.

Stark tries to quantify with qualitatively derived indicators, but without hard evidence. That is risky business--no less risky than the historical hypotheses advanced by the historians he wishes to refute. Even so, the book stands as an exhortation to historians to remember to develop models that responsibly include as much of the data as possible. Likewise, the book reminds us that the most fitting models we develop end up looking very much like the ones we received, and that far from indicating our failure, such outcomes may only confirm the reliability of the historical reconstruction at hand.

Corrections made 9 January 2010