Common questions about statistics
This is not a formal Reality Check, but rather a brief recap of exactly what an estimate is, and how we compile them--particularly in the light of the many statistics we use in the World Christian Encyclopedia and our associated publications.
First of all, some times people ask where we get our statistics from. They are compiled from literally thousands of sources: newswires, denominational reports, statistical surveys, e-mails, field interviews, books, studies, lists, directories, correspondence, confidential communiques, computer searches, bibliograhies--and that's just a few of the sources! We have literally thousands of pages on the single subject of martyrdom alone.
Second, some people take issue with the statistics. Please, do so! The whole process of "critiquing" work is very much part of the research journey. Sometimes we will agree with you and change the number. Sometimes we will not. Nonetheless, the process of contribution and critique is very important. And if you have information you'd like to contribute that you don't think we've seen--such as where your mission is operating, details about the peoples you are reaching, etc.--then now is the time to submit it. You can send it to me via this e-mail address, or to: GEM, PO Box 6628, Richmond, VA 23230.
Third, people ask us about the precision of the numbers. This is perhaps the most oft-asked question: "How in the world do you get it down to the last digit?" The main feature of the statistical presentation in the WCE and in all our work is that its primary object is to establish broad areas of magnitude--to give the general order of magnitude of the situation, whether denominational, local, tribal, national, regional, racial, continental or global. From the point of view of the planner, development officer, Bible society executive, broadcaster, journalist or researcher, the important thing is to know (for example) whether Protestants in a particular country number 1,000, or 100,000, or 1 million, or 10 million; the exact size to the last digit may be of interest but of little use. In the same way, many other totals enumerating approximately the entire Christian enterprise have been computed and presented, such as radio audiences, unevangelized populations and so forth. The word "approximately" is the operative word in the survey: absolute precision and acuracy are not to be expected, nor in fact are they necessary for practical working purposes. This means that, although the tables and statistics may help readers who want specific individual figures, they are mainly designed to give this general-order picture set in the total national and global context. To this end, where detailed local statistics compiled from grass-roots sources were not available or incomplete, our tables supply general-order estimates provided by people intimately familiar with and authoriative on the local statistical situation.
Churches and religious bodies do not have at their disposal the vast networks of enumerators and analysts that government censuses and public-opinion polls employ, hence churches' statistics are not able to claim accuracy to the last digit. Similarly, though a government census in, say, 1968 may report the number of Christians to the last digit (e.g. 2,450,793), projections into the future based upon this figure cannot claim complete accuracy but can only be approximations indicating the general order of size of the statistics required. Statistics therefore fall in two categories: general-order estimates and multi-digit aggregates. General-order estimates can be recognized throughout by their rounded nature (e.g. 5,000, 20,000, 100,000). Where they occur, in Table2, they usually represent estimates of their own size by the churches themselves; where they occur in Table1 they represent estimates by persons familiar with the nation, or by the editors. Multidigit aggregates can be recognized as figures appearing to claim accuracy to several digits (e.g. 5,291, 21,684, or 102,735). In a number of cases, a church or diocese returned this kind of statistic rather than rounding it to the nearest hundred or thousand as one would do with a general-order estimate. In such cases Table2 repeats the rounded number, for 2 reasons: (1) to indicate the church's claim that the statistic is _not_ a rough guess but is based on some kind of aggregate head count, and (2) to enable readers familiar with the church's statistics to recognize the particular figure and hence to know its source and the exact date it applies to, in case more up-to-date or reliable figures later become available. It is remarkable how often such a multi-digit aggregate may be quoted for years after its original computation, and how easily recognizable each one is when it turns up again. Preserving the digits in this way is therefore an aid to the proper use of these statistics; but our practice must not be taken to imply bogus precision or any claim to accuracy to the last digit.
Further, because our survey must of needs be comprehensive, in certain countries where no hard statistical data or reliable surveys were available, we have had to rely on the informed estimates of experts in the area and subject. In the WCE, we have made no detailed attempt at a critique of each nation's censuses and polls or each church's statistical operations. After examining what is available, we have then selected the best data available until such time as better data comes into existence.
There are a number of areas of church or religious life where it is impossible to obtain accurate statistics, usually because of state opposition to Christianity or religions. Thus it will probably never be possible (nor perhaps desirable) to get exact head counts of crypto-Christians or of isolated radio believers, or of annual conversions to underground churches and the like. Where such information is necessary to our surveys, we have made reasonable and conservative estimates.
In each table, a number of totals are given at the end. When figures which are multidigit aggregates (e.g. 102783) are added together iwth others which are general order estimates (e.g. 110,000), the resulting total (212,783) is printed to the last digit in the interests of consistency and exactness of analysis. However, when such a total is quoted or used outside this survey, it should be rounded (to 210,000) to indicate only the general order of accuracy that it is possible to claim. This avoids the quoting and spreading of totals which appear to have greater accuracy than is justified.
A major feature of our surveys is that care has been taken to make the various statistical categories, the national tables, and the international totals all fully consistent with each other and without internal statistical discrepancies or contraditions. This means that (1) each statistical category we use has been given a single clear operational definition applicable worldwide and in all countries, churches and religions; and (2) in all complete enumerations, all subtotals of absolute numbers add up to exactly the relevant total. A number of other consistency checks come readily to mind: e.g. in large infant-baptizing state churches with few adult conversions each year, the annual number of infant baptisms should be smaller than the total number of annual births in the country. Vast numbers of new single facts or items of data can be checked in this way before being entered into the tables.
Finally, in a few places when data was being collected, different authorities gave, for the same situation, radically divergent data. In most cases the discrepancy was solved and the situation resolved by examining the exact context of the data. It was then found that in almost all cases the differing data referred to either different points in time, or to slightly different geographical areas, or to different definitions of the item in question. As a result our surveys are able to give a single figure for every clearly defined entry at a given point in time. Similarly, where various authorities differ concerning quantifiable matters, we have attempted to present a single statement rather than a series of range of contradictory statements.
