Skip to Main Content
Have library access? Log in through your library
Social Forecasting Methodology

Social Forecasting Methodology: Suggestions for Research

Daniel P. Harrison
Copyright Date: 1976
Published by: Russell Sage Foundation
Pages: 104
  • Cite this Item
  • Book Info
    Social Forecasting Methodology
    Book Description:

    A volume in the Social Science Frontiers series, which are occasional publications reviewing new fields for social science development.

    These occasional publications seek to summarize recent work being done in particular areas of social research, to review new developments in the field, and to indicate issues needing further investigation. The publications are intended to help orient those concerned with developing current research programs and broadening the use of social science in the policy-making process.

    A Volume in the Russell Sage Foundation's Social Science Frontiers Series

    eISBN: 978-1-61044-642-6
    Subjects: Sociology

Table of Contents

Export Selected Citations Export to NoodleTools Export to RefWorks Export to EasyBib Export a RIS file (For EndNote, ProCite, Reference Manager, Zotero, Mendeley...) Export a Text file (For BibTex)
  1. Front Matter
    (pp. [i]-[ii])
  2. Table of Contents
    (pp. [iii]-[iv])
  3. Introduction
    (pp. 1-2)

    There is no need to justify research on social forecasting methods; Duncan, de Jouvenel, Helmer, Zarnowitz, Hoos, and others have already done this in some detail. Nor is there a need to argue about the feasibility of doing social forecasting; the empirical record clearly indicates that it is feasible. Nevertheless, the consensus of critics is that social forecasting methodology is in an underdeveloped state. This paper suggests a number of areas in which methodological research could be usefully done. Readers more expert than I in any or all of the forecasting areas mentioned will undoubtedly be able to revise or...

  4. Types of Methods
    (pp. 3-12)

    The usual starting point in discussing forecasting methods is to present an inventory of methods currently in use. Since there are frequently new developments in methodology, there is always a need to update the inventory from time to time. While there are several categorized inventories to choose from, I have found the following designations to be reasonably comprehensive. Further details on these methods can be found in the section entitled “Critical Assessments of the Methods.”

    Consider first the most frequently used forecasting methods.

    The procedure consists of identifying an underlying historical trend or cycle in social processes that can be...

  5. Critical Assessments of the Methods
    (pp. 13-50)

    When one talks about research on social forecasting methods, perhaps one of the first things that comes to mind is the viability of the procedures and algorithms employed in the methods. It is possible to evaluate the properties and limitations of forecasting methods in the abstract; the analysis is given more concrete relevance by considering each method in terms of some aspect of social process it would be likely to be applied to. We will consider needed research in these areas for each of the thirteen types of forecasting methods previously mentioned plus the combined forecasting methods approach.

    There are,...

  6. Evaluating Assumptions
    (pp. 51-54)

    This is an area in which there is a need for a great deal of research for most types of forecasting methods. If we look at the assumptions of the more frequently used forecasting methods, for example, the significance of using untested or unstated assumptions is apparent. In extrapolative forecasting there is little empirical guidance for the underlying assumption of stability or periodicity in social processes. When pushed to the wall the social forecaster will not be able to define the time frame for which his assumption of continuity in process obtains, because he has no precise empirical sense of...

  7. Forecast Accuracy
    (pp. 55-60)

    The literature on forecasting accuracy is quite scattered and much of it unpublished, but, with enough searching, examples can be found for most of the types of forecasting methods described. It may be indicative of the highly underdeveloped state of the art of social forecasting methods (or other kinds of forecasting methods as well) that there is little empirical concern with forecasting accuracy.

    There are few rewards for those who do social forecasting evaluation research and for those who provide the funding for undertaking such evaluative research, and there may, in fact, be penalties for the conscientious execution of such...

  8. Forecasting Method Robustness
    (pp. 61-62)

    Another empirical social forecasting method assessment area lies in the deliberate manipulation of forecasting procedures with artificially patterned or random data to test the discriminability and robustness of such methods, sometimes called “Monte Carlo studies.” This kind of analysis is not reported in the social forecasting literature except for occasional studies of econometric and simulation models. Where poor data plague social forecasting research, robustness is of considerable importance. The matter is of considerable concern, too, in assessing complex algorithmic forecasting machinery that could have inherent biases, say given data with highly disparate magnitudes, as noted for regression forecasts. Where large...

  9. Data Problems
    (pp. 63-66)

    Hoos has argued that a precondition to any successful forecasting effort is the availability of data of relevance and good quality — this, unfortunately, is not the case with government statistics, she claims. Morgenstern, Mincer, Zarnowitz, and others have demonstrated that poor dataalonecan create margins of error in economic forecasts greater than that tolerable to economic policymakers. It appears, too, that the margin of error that is tolerable is highly related to the aspect of social process that is being studied and the kind of expectations that have developed with respect to the forecast accuracy of particular kinds...

  10. Measurement Problems
    (pp. 67-70)

    There are two kinds of measurement problems the social forecaster has to contend with. One is the conventional concern with precision, a matter not unrelated to user needs for forecast accuracy, as noted. Data errors can and do derive from missing data or imprecision in measurement (assuming valid measures are used to begin with) and, as noted, can and do have important effects on forecast precision. Unfortunately, it is difficult to get high quality social forecasting data: data on a drug treatment program are defective because the participants secretly discard medications (see Hoos), a poll fails to call a primary...

  11. Social Processes and Social Forecasting Method Selection
    (pp. 71-74)

    It should be possible to identify social forecasting methods that are best suited to deal with social forecasting problems exhibiting particular kinds of social dynamics. Here, the first task would be to identify the various principal types of social dynamics one is likely to find in social processes. For example, consider the following:

    1. continuous, sustained action

    2. action characterized by a very few major discontinuities

    3. action characterized by numerous discontinuities

    4. accelerating or decelerating action

    5. cyclical action

    6. action characterized by a fundamental transformation in key parameters

    7. action characterized by delayed, feedback, or feedforward effects

    8. continuous rate, constant speed action

    9. catalytic action

    10. parallel...

  12. R & D for New Social Forecasting Methods
    (pp. 75-76)

    A reading of the forecasting method literature suggests a need to develop new social forecasting methods; many of the procedures now available seem to be plagued by arbitrary and unrealistic assumptions that are not consonant with the way social processes are known or thought to work, are too simplistic or clumsy to deal with anything but a minimum amount of social complexity, or are not suited to the analysis of particular aspects of social process. The social indicators movement has arisen from a perceived need to develop better measures of key social processes and to collect data on them. It...

  13. User Forecasting Requirements
    (pp. 77-78)

    Many of the points of this section have already been made. Social forecasting should develop on as sound a scientific footing as possible; however, it should be recognized that in some cases, such as clinical prediction, current scientific best is not good enough. As noted, Ikle suggests that there are many policy uses of forecasting results ranging from public relations gimmickry and bureaucratic politics to a great concern that there be as rigorous and accurate forecasting as possible for serious policymaking analysis.

    In this paper it is assumed that the objective is to do as accurate forecasting as possible and...

  14. Solving Problems with Forecasting
    (pp. 79-82)

    Continuing the argument of the previous section, social forecasting methods specialists should learn how to link up effectively with social theorists and policymakers who are trying to solve problems in their respective domains. This calls for more than the providing of social forecasts having sufficient accuracy to meet user needs for accuracy although this is by no means a small task. It also calls for an understanding of the problem the policymaker(s) is trying to solve or an understanding of the theoretical issue the social scientist is trying to resolve. In addition, there is a need to understand the time...

  15. Costs
    (pp. 83-84)

    It may be rather dull to talk about such things as data gathering costs, computer data processing costs, manpower costs, and time requirements, but in an environment of scarce social science research resources there is a need to assess social forecasting methods in terms of such factors. There may have to be a tradeoff between social forecast accuracy or time horizon on the one hand and data collection and analysis costs on the other. There is a need to present some tabulations of costs per the various types of social forecasting methods applied to various aspects of social process or...

  16. Conclusion
    (pp. 85-85)

    The state of the art of social forecasting clearly provides vast opportunity and need for research on social forecasting methods although fairness suggests that recognition be given to the considerable amount of research that has been done on social forecasting methodology to date. The best stance, I believe, is to treat the matter as one ofprimarilyscientific inquiry and subject to scientific standards. I have taken pains to indicate that this stance is not and should not be incompatible with contributions to public policy problem solving. Whether or not the reader agrees with the point of view or research...

  17. Selected References
    (pp. 86-94)