1Contemporary sociologists rely increasingly on numerical trends, rather than anecdotal and qualitative information, in the study of social change. These trends are expressed as time series, with equally-spaced intervals of time on the x-axis and the numerical values of a social indicator on the y axis Every such series can support some degree of retrospective explanation ; some allow prediction.
2The two functions are analytically distinct. The retrospective explanation of a numerical trend typically requires the analyst to take account of related trends that provide a context and also of isolated events that impinged on the variable. It does not differ from the normal task of the historian, except for a greater dependence on numbers.
3Predicting the future trend of a social indicator is more difficult, of course, than explaining the past trend of the same indicator. The difficulties are not uniform, as I propose to show. The trends that concern sociologists can be usefully placed into a small number of categories with differing predictive potentials
4Several recent developments have been driving sociological research towards more extensive reliance on numerical trends to track social change and on the comparison of trends to describe national and regional differences :
5— small-scale hand-tailored sample surveys designed and administered by individual researchers have been largely replaced by giant omnibus surveys conducted by government agencies and/or research institutes and available for public use ;
6— the replication of these giant surveys at regular intervals has begun to provide longitudinal data on topics – voting behavior, criminal victimization, drug use, food preferences, premarital pregnancies, reading habits, religious observances, and scores of others for which there used to be no reliable measures at all, as well as opportunities to disaggregate such series by gender, education, ethnicity, region and other significant attributes ;
7— the growth of data-generating organizations like OECD and the World Bank has encouraged the development of standard formats for national statistics, greatly simplifying cross-national comparisons. As recently as 1950, hardly any two countries measured fertility, unemployment, homicide, factory wages or anything else in ways that permitted direct comparison. Today, most of them generate numerous trends in standard formats. For example, under the European System of Social Indicators, comparable trend data for more than 200 indicators are now available for 22 countries  ;
8— the measurement of softer indicators – attitudes, opinions, beliefs, values – has benefited from the same tendencies, in particular from the strong demand for such information generated by international social science associations and other non-governmental organizations and the inclination of globalized nations to compare themselves on family values, religious beliefs, reading habits, sexual attitudes, and many other unofficial variables ;
9— computer technology has made it feasible for individual researchers and small teams to access and handle enormous quantities of numerical data ;
10— the present paper is largely based on fifteen years of work with the International Research Group for the Comparative Charting of Social Change, an organization of social research teams from several European and North American nations.
11For at least five centuries, the progress of science-technology has been accompanied by a steady increase of agricultural and industrial productivity together with the steady growth of the world’s population.
12This wave of change began in northwestern Europe and then extended by stages to the rest of the world. Its origins are not entirely clear, but the sailing ship seems to have played a central role, either because (1) as William Cottrell proposed in Energy and Society, it was the first high-energy converter which returned large surpluses of energy for a relatively  small input or (2) as Jean Fourastié argued in Machinisme et bien-être, the improvement of transportation put an end to famines in western Europe or (3) as Wallerstein and other world system theorists maintain, the massive infusion of capital from overseas conquests and trade stimulated European development or (4) as Marx and Engels wrote in The Communist Manifesto, the European bourgeoisie were insatiable in their exploitation of new territories or (5) perhaps all of the above.
13Although it is clear in retrospect that productivity and population began their unprecedented upward movements in northwestern Europe as early as the fifteenth century, hardly anyone noticed. The extensive numerical measurements of social and economic phenomena that we take for granted today did not exist then or for long afterwards. The first serious attempt to guess the population of Great Britain was made in 1688 by Gregory King ; it was an estimate not an enumeration. No census would be taken there until 1801 and it was not until the middle of the nineteenth century that any national census produced much more than a simple headcount. The decennial US censuses from 1790 to 1840, for example, tell us almost nothing about the age distribution of the population.
14If demographic data were scanty before 1850, social and economic data were even rarer. The recording of customs duties and excise taxes incidentally produced a few figures on the production of alcoholic beverages and the importation of machinery, but there was no way for a contemporary to know if the output of grain per acre were rising or if illiteracy were declining.
15When that exceptionally acute observer, Alexis de Tocqueville, arrived in the United States in 1831, steamboats had been plying the Hudson River for a quarter-century and industrial assembly lines had been operating in New England for even longer, but he gave no particular attention to mechanization in his great work on America. The people who lived through the most active phase of the Industrial Revolution were largely unaware of it.
16It was not until the middle of the nineteenth century that increases of population and production and the secondary effects of those increases became too obvious to be ignored .
17It is fair to say that sociology as a distinct field of study grew out of the efforts of nineteenth century scholars to explain the transition from relatively static agrarian societies to more dynamic industrial societies. In the agrarian condition, population and production had fluctuated within a fairly narrow range. The industrial condition was marked by continuous rises in population and production and continual changes in social structure.
18The quest for explanation eventually produced several influential models of social change, including – in the nineteenth century – the linear progress model, the transformation model and the apocalyptic model.
19The linear progress model visualized social development as mankind’s steady ascent up a ladder of predictable stages, each stage marked by parallel improvements in technology, social organization and moral excellence. That was the central theme of Herbert Spencer’s The Study of Sociology, the textbook used in the earliest American and British sociology courses.
20The transformation model identified the transition from traditional to modern forms of social organization as the central event of modern history. The various versions of this model are easily recognized by their paired terms : Tonnies’ Gemeinschaft-Gesellschaft, Durkheim’s mechanical solidarity-organic solidarity, the folk-urban dichotomy of Redfield and other ethnographers.
21A number of apocalyptic models described modern societies as moving towards a culminant struggle that would change them out of recognition. The Marxist version is the most familiar but the Gobineau version eventually did comparable damage.
22All of these nineteenth century models were remarkably ineffective in predicting the major trends of the twentieth century. As every student of the history of our discipline knows, Durkheim regarded nationalism as a spent force while Weber predicted that the European colonial system would endure for centuries and Marxists like Kautsky announced the imminent collapse of capitalism.
23In the twentieth century, two additional models of social change achieved some currency – the linear decline model first popularized by Spengler and the cyclical models of Pareto and Sorokin. These too proved virtually useless for prophecy.
24To understand why these grand models of social change routinely produced defective forecasts, we must go back to the central question which, in one way or another, they had to address : What is the connection between scientific-technical progress and other trends in industrial societies ?
25In recent centuries, scientific-technical progress has been continuous, cumulative, consistent and irreversible. The effort to discover the principles governing physical, chemical and biological phenomena has never encountered a serious interruption or setback. Although natural science theories are challenged and overturned as readily as social science theories, the technologies based on the natural sciences advance inexorably from year to year.
26The linear progress model attempted to equate scientific-technical progress and other forms of social change. But its assumption that social change too was continuous, cumulative, consistent and irreversible could not withstand the weight of contrary evidence.
27The transformation model allowed social change to be discontinuous but required it to be cumulative, consistent and irreversible. That expectation to as well ran counter to the empirical evidence.
28The apocalyptic models did not require social change to be continuous or cumulative, but did insist that it be consistent and irreversible in the pre-apocalyptic phase and nonexistent in the post-apocalyptic phase.
29The cyclical theorists were compelled to ignore the unique elements of modernization in order to insist that social change is consistent and continuous but fully reversible.
30The linear regress model described social change as downwardly continuous, cumulative, consistent and irreversible – an even less plausible scenario than linear progress.
31Each of the grand models assigned some of the features of scientific-technical progress to all types of social change. Their common element was the insistence that social change is consistent, i.e. that all the trends in a social system run roughly parallel. The most cursory examination of empirical data shows this not to be so.
32Because the grand models of social change assume social change to be consistent, they sustain belief in many imaginary trends. Thus, to the average citizen of an advanced industrial nation, mobility seems always to be increasing, patriotism declining, family attachments weakening, craftsmanship disappearing, mental illness spreading, politics becoming more partisan and children less mannerly. All of these supposed phenomena are sometimes linked together under the general rubric of moral decline.
33The most cursory examination of the empirical data on moral attitudes shows just how inconsistent social trends can be. People in the advanced western nations are far more permissive than their grandparents about premarital sex, illegitimacy and bankruptcy ; much less permissive about drug addiction, wife abuse, drunk driving, professional errors and defective goods ; neither more nor less permissive about adolescent crime, political corruption and adultery. Each of these separate trends makes sense when we descend to the examination of details but nothing useful can be said about the trend of morality in general.
34The search for master trends is not necessarily futile, but one needs to know where to look for them.
Types of Trends
35What we find when we examine any considerable number of trends for a given national (or regional or local) society is that they fall into several distinct types which can be arrayed according to their relative dependence on scientific-technical progress :
36Type A (or developmental) trends are directly ascribable to scientific-technical progress and for half a millennium they have been, with the exceptions noted below, continuous, cumulative, consistent and irreversible. They include the uninterrupted increase of people and things, reflected in the growth of population and in long-term upward trends in such indicators as : per capita utilization of fossil fuels, metals, paper and water ; the number of metropolitan cities ; the maximum size of metropolitan cities ; the physical volume of local, regional and international trade ; the velocity and capacity of transport vehicles and communication media ; literacy and numeracy ; occupational specialization ; available leisure ; the output of grain and other vegetable products per hectare and per labor hour ; industrial output per labor hour and per unit of capital ; land area under irrigation ; the size of road networks ; physicians per capita ; telephones per capita ; enrollments in primary, secondary and higher education ; the range and lethality of weapons ; the rates of extinction of plant and animal species ; the erosion of pre-industrial cultures ; atmospheric pollution ; average dwelling size ; average living standards ; per capita ownership of automobiles, television, refrigerators, air-conditioners, washing machines and other consumer appliances ; books published ; pharmaceuticals prescribed ; surgical operations performed ; life-expectancy at birth ; maternal survival ; and dozens of other indicators whose connection with scientific-technical progress is direct and unequivocal. Most Type A trends most of the time are indeed continuous, cumulative, consistent and irreversible.
37A few Type A trends eventually encounter a natural ceiling. Literacy and primary school enrollment cease to be useful measures when their values approach 100 percent. A few other Type A trends have been changed by collective action, as when the increase of per capita energy utilization was halted by conservation policies in some advanced industrial nations.
38Type A trends can be temporarily interrupted by economic or political dislocation. Thus agricultural and industrial output declined in many countries during the economic crisis of the 1930s and in several eastern European countries during their transition from communism. The upward movement generally resumes within a few years.
39The increase of population is a special case. While the total world population continues to expand, a number of the advanced industrial nations now show zero or negative natural growth. But since those conditions generate irresistible immigration pressures, none of the advanced nations has yet suffered an actual decline in population.
40Type A trends have relatively high predictive potential. In the short term (up to 10 years or so) forecasts can be quite precise, although never certain. Forecasts of passenger traffic or petroleum consumption for 2010 are sufficiently reliable to provide a basis for action by transit authorities and oil refiners. A few Type A trends permit much longer forecasts to be made because their components are already in place. The number of elderly pensioners in any advanced nation in 2030 can be estimated within a narrow margin, assuming that neither immigration nor mortality rates nor retirement practices change drastically in the interim.
41Type B (or mixed) trends refer to aggregate phenomena that are clearly affected by scientific-technical progress but also heavily influenced by public policies and accidental variations. Many of them are predictable in the short and middle term but rarely in the long term.
42Typical Type B trends include : workplace injuries ; the incidence of obesity ; the incidence of sexually-transmitted diseases ; the relative size of urban and suburban populations ; fluctuations in prestige systems ; the proportion of immigrants in national populations ; traffic congestion ; architectural styles : attendance at spectacles ; home ownership.
43Type C (or motivational) trends are trends in the attitudes and practices of the members of a population in areas of high-intensity motivation, especially money, sex, health and intoxication. They represent individual choices that are strongly influenced by collective standards and shape those standards in turn.
44Type C trends have relatively close connections to scientific-technical progress and many of them have high predictive potential. They are sometimes continuous, cumulative and consistent, but never securely irreversible.
45Here are some Type C trends currently running in the US ; most of them can be matched in other advanced industrial nations : decreasing proportional expenditures for food and clothing ; increasing proportional expenditures for transportation, health care, entertainment, gambling ; increasing personal debt ; increasing media exposure ; increasing dwelling size ; decreasing average age of first intercourse for both males and females ; increasing non-marital cohabitation ; decreasing extra-marital intercourse ; increasing overt homosexuality ; decreasing tobacco consumption ; decreasing consumption of hard liquor ; increasing aggregate consumption of illicit drugs ; increasing aggregate consumption of prescription drugs ; increasing use of alternative medicine and cosmetic surgery.
46Type D (or institutional) trends occur in the major action systems we call social institutions : family, work, education, religion, leisure, law, government, social services. These huge slow-moving clusters of motives, activities and relationships exhibit a great variety of trends – long-term and short-term, linear and cyclical, rising and declining, local and general, cumulative and episodic, along with surprising continuities and abrupt reversals.
47Examples of Type D trends are marriage rates, divorce rates, age at marriage, birth rates, household composition, occupational distributions, occupational mobility, unionization, income and wealth distributions, tax rates, inflation and deflation, poverty, curricular trends, fashion cycles, attendance at spectacles, crime rates, clearance rates, prisoners and probationers, litigation, electoral participation, legislation, regulation, political corruption, tax rates, pensions and subsidies, philanthropy, military forces, military casualties.
48Some Type D trends are exceptionally long-term, although never as smooth as Type A trends. Tocqueville identified the gradual progress of equality and the increase of administrative centralization as running more or less continuously from the late medieval era to the nineteenth century and they have obviously continued into the twenty-first century, although with many local and temporary reversals.
49A great many Type D trends fluctuate without determinate causation. The homicide rate in any given nation at any given time is likely to be either rising or falling. So long as it moves in one direction, short-term prediction is quite accurate. Thus, since 1994, the US homicide rate has declined from year to year and the decline is expected to continue until the next point of inflection. But the timing of that point is totally unpredictable.
50Advances in technology influence Type D trends in both expected and unexpected ways. The development of oral contraceptives in the 1960s induced sharp declines in fertility in all of the advanced nations. That was expected. But the sharp increase of out of wedlock births that followed the same innovation came as a surprise. Only in retrospect was it clear that in making women solely responsible for their pregnancies, effective contraception, in conjunction with other factors, had undermined paternal responsibility. The other factors included the entry of women into the labor force, the feminist movement and extensive changes in family law and public policy, all intricately influenced by each other and by the innovation of oral contraception.
51Type D trends are very frequently counter-intuitive. Who would have anticipated, for example, that most of the world’s active wars at the outset of the twenty-first century would be wars of religion ?
52The cross-national or cross-regional comparison of Type D trends is complicated by differences in institutional context. For example, both France and Bulgaria experienced sharp increases in street crime in the 1990s – to a much higher level in France. But the French trend was not regarded as a critical problem, while the lesser Bulgarian trend was experienced as a breakdown of social order .
53There can be no general formula for predicting the future course of Type D trends. Their predictive potentials must be assessed case by case – with respectful caution.
54Type E (or adversarial) trends reflect changes in the way a population divides itself into potentially hostile groups by location, ethnicity, religion, party, class, ideology and other sources of contention, together with the internal conflicts produced by those divisions and external conflicts with other populations.
55Type E trends include the incidence and the outcomes of military and naval battles ; riots and insurrections ; electoral campaigns ; legislative struggles ; lobbying efforts ; preference systems ; selective recruitment ; denominational competition ; schisms and feuds ; market corners ; hostile takeovers ; strikes and lockouts ; demonstrations, assassinations, lawsuits ; and terrorist acts, to the extent that these can be counted and distributed over equal intervals of time.
56Since the essence of social conflict is the uncertainty of outcome, we would expect Type E trends to have low predictive potential and, in general, they do. But a distinctions must be drawn between (1) the trend of outcomes in a series of contests between the same parties ; and (2) the trend that represents the occurrence or non-occurrence of a particular type of conflict over time.
57The outcome of any zero-sum contest between evenly matched adversaries is unpredictable in principle, since each of the parties, having information as good or better than a third-party observer, expects to win . When the parties are very unevenly matched, both they and their observers may have a better than even chance of predicting the outcome, although rarely with total confidence. When zero-sum contests are replicated in a series, it must be presumed that the parties are more or less evenly matched or the series would not persist.
58Thus, for example, since the emergence of the two party system in US politics in the nineteenth century, control of the presidency has oscillated between them. Twelve of the 25 presidential elections of the twentieth century were won by Democrats, 13 by Republicans. In seven of these elections, the winning mar  gin was five percent or less. In elections for Congress, the largest Democratic margin in the House of Representatives was 244 in 1937-1938 ; the largest Republican margin was 168 in 1921-1922. At each of these extremes, the demise of the minority party was freely predicted. As of this writing in 2001, both houses of Congress are almost evenly divided between the two major parties. No method is available for predicting their relative strength in the future.
59The trends that reflect the occurrence or non-occurrence of wars, riots, insurrections, terrorist acts or assassinations are similarly unpredictable. No one knows, or can know, when Britain will be involved in another international war or where the next British riot will occur. It is probable, but not certain, that since the organizational machinery for wars and riots is still in place, such events will occur again, but where and when will remain unknown until the moment of occurrence.
60Ideological struggles have the same contingent character. The culture wars between advocates and opponents of abortion and of gun control have been raging in the US for three decades . One of these days, they will be replaced by other issues but when, where and how cannot be foretold.
61In general, a future trend is predictable if all the events by which it can be significantly influenced have already occurred. This is rarely the case in conflict situations, where the most significant events are the actions taken by the contestants in response to each other’s actions as the conflict proceeds. The retrospective analysis of Type E trends can be highly informative, but attempts at extrapolation are usually futile.
62In sum, social trends are not created methodologically equal. Type A trends are exceedingly useful for social forecasting. Type E trends are not. Type D trends are less useful for forecasting than Type C trends, Type C less useful than Type B. These categories are rough but they can help us to distinguish between those social trends that tell us something about the future and those that can only inform us about the past.
63An unexpected implication of this typology is that almost all the predictive potential of contemporary social trends is derived from scientific-technical progress. That seems unlikely at first, but as one works with trends that are not much affected by scientific-technical progress, it becomes more plausible. These include still another type of trend.
64It was Adolphe Quetelet who first called attention to the astonishing stability of certain social indicators from year to year – indicators which turn out to be virtually independent of scientific-technical progress. Suicide, in particular, invites attention because it has attracted so much sociological interest and because it is measured with fair reliability at national, regional and local levels in advanced industrial societies, greatly facilitating cross-national and cross-regional comparison.
65Suicide trends are nearly level for long periods ; the occasional small deviations are attributable to specific events like wars and economic crises.
66As the following table shows, there are spectacular differences in suicide rates among the advanced industrial nations. These differences, which hardly vary at all from year to year, are extremely difficult – perhaps impossible – to interpret in any convincing way .
67Level trends of the Quetelet variety (Type Q ?) are easy to extrapolate but very difficult to explain retrospectively, exactly the reverse of our Type D and Type E trends, which are difficult to extrapolate but very useful for historical analysis.
68If scientific-technical progress accounts for most of the predictive potential in contemporary social trends, how do we explain Tocqueville’s two master trends – administrative centralization and the gradual progress of equality – that seem considerably to antedate the major scientific advances that launched the modern era ?
69The simplest answer is that the roots of scientific-technical progress were clearly discernible in Europe from the fifteenth century onward, with improvements in shipbuilding and firearms ; the exploration of Africa, India and the Americas, by Europeans ; the gradual transformation of alchemy into experimental chemistry and substantial (although not precisely measured) increases in population and production.
70When a quantum leap in science-technology occurred around 1650  the influence of scientific-technical progress on other social trends began to intensify. It continues to intensify today with advances in computerization and biogenetics and will probably do so in the foreseeable future. But for the time being as in the long past, the psychic equipment of homo sapiens shows no trend at all. It is within those two parameters – the ever-increasing influence of science-technology on social change and the astonishing stability of human nature – that the analysis of social trends must advance.
71It will advance further and faster if the several types of social trend are not confounded and the distinctions among them are kept clear, since it becomes much easier to interpret any given trend when its intrinsic character is taken into account. The typology here proposed may be of some assistance in this way.
Www.gesis.org/en/social monitoring/social indicators.
Curiously, the inventions that made the European sailing ship an instrument for world conquest – the compass, the cannon and the square-rigged sail – all came from China, which was more advanced technologically in the preindustrial era than any European nation and had larger and more effective ships, until its rulers chose to turn away from the sea.
Malthus recognized the increase of population somewhat earlier but underestimated the corresponding growth of production.
Mendras, 2001 ; Genov, 2001.
For a more detailed explanation of this principle, see Rapaport, 1960.
Caplow, Hicks and Wattenberg, 2001.
Durkheim’s classic sociological text on suicide is not much help here. Its essential contribution was the partition of suicide into motivational categories – egoistic, anomic, altruistic. But there is no way to distribute the total number of suicides reported for a given jurisdiction in a given year into these categories. They remain purely speculative.
Statistical Abstract of the United States 1998, Table 134.
Merton (1970) provided a detailed account of the acceleration of scientific-technical progress in Britain during the course of the seventeenth century.