Gordon Welty and Alfred Beradino

[//953] This discussion of the problems of rigorous research in the dynamic organizational context is specifically methodological, addressing itself to the demonstration of possibility by indirect proofs. For illustrative purposes, we concern ourselves with organizations seeking to effect resolution of social problems. Although research has more frequently addressed itself to these particular organizations in the guise of "action research," it is not to be assumed that this restricts the general applicability of our findings.

While experimental designs are frequently used in research within an organization, for example, in industrial research applications, designs are rarely applied to the study of an organization and its management. Price has pointed out that rigorous research designs are virtually unknown in the study of organizations./2/ Seashore, in his discussion of problems and prospects of the utilization of rigorous designs, points out that the five or ten attempts to use designs reported in the literature are "primitive, pioneering ventures."/3/

It is not difficult to explain why the organizational researcher has tended to case studies and anecdotal evidence. The universe of organizations is ill-defined; the number of units accessible to an investigator is small. The organizational patterns are myriad; although the development of graph theory has shown promise, in general the analytical tools of mathematics are of limited application. All this has combined to favor description rather than rigorous inference.

In addition, some organizational members themselves resist such research. For example, in many cities today, the Department of Justice, through the Omnibus Crime Control Act of 1968, is supporting extensive and expensive police training programs. The value of higher education for rank and file policemen has been seriously questioned./4/ It is a moot point [953/954] whether such organizational change in police forces will benefit the wider community. The organizations involved, however, give little indication that they appreciate the necessity, and particularly in the case of an organization such as the police, the possibility of rigorous study to answer these questions.

Of course, causal inference in research can be permitted only when rigorous designs are employed. Without designs it is impossible to construct a genuine theory of organizations and their management. Work undertaken in the absence of experimental designs has given rise to a plethora of untested taxonomies. This lack of design is particularly striking when organizations in a dynamic context are considered. Here management is the active element driving the organization closer and closer to an approximation of an ever-modified set of goals, the movement toward which is always subject to environmental constraints. In this explicitly dynamic context, the necessity of rigorous designs in theory construction is patent, but the application of these designs is quite problematic.

If the problems encountered in the use of designs were simply the practical ones enumerated by Seashore,/5/ they could in principle be resolved by diligence and adequate resources. It has been proposed, however, that there is a logical incompatibility of the application of design and the dynamic study of organizations. The first premise of this logical argument is that if preliminary research findings are disseminated during the research cycle, in simplest terms, between pre- and post-measures, then the management may use these findings to modify operations or organization during the cycle. This sort of modification practice, to achieve organizational improvement, so the argument goes, will corrupt the design and render the research meaningless. The other premise of the logical argument is that the researcher either is ethically prescribed to publish the preliminary findings, or else is under a monetary sanction to do so. From these premises, the conclusion of incompatibility is drawn.

We have discussed elsewhere this logical argument as it applies to the management of educational programs/6/ and family planning programs./7/ The claim that early dissemination of research findings will render the study invalid has also been made with regard to juvenile delinquency prevention organizations/8/ and programs in the War on Poverty/9/ and certainly had a [954/955] bearing on the Ford Foundation's refusal to fund organizations employing research designs, preferring to fund "action research" projects./10/

A question arises at this point concerning the theoretical assumptions the argument makes about the rationality of the organizational decision maker. Will the decision maker necessarily employ the information of the research findings for organizational improvements? There appear to be two factors relevant to the questions: (1) The decision maker is an optimizer, a satisficer or irrational; these possibilities appear to be exhaustive, though perhaps not exclusive. ( 2 ) The information of the research findings is of various degrees of credibility./11/

Only the irrational decision maker is patently unresponsive to feedback. Only if the information has zero credibility will the research findings not be a factor in the decision making of the optimizer or satisficer. It is an empirical matter whether decision makers are irrational or incredulous. The rise of operations research, and other institutional changes fostering accountable management, would suggest they are neither.

To return then to the incompatibility argument, it appears plausible that a decision maker would use research findings for organizational improvement. Patrick Suppes has pointed out that problems of an impossibility theorem are acute. He suggests that more care is needed to prove a negative argument than a positive argument./12/ Yet a large number of alleged impossibilities and incompatibilities are continually asserted in the social sciences. Most of these arguments are vague, proceeding from nebulous premises. The argument before us is a case in point. As formulated, it is vague. On the other hand, when precisely formulated, the argument is seen to be false.

Suppose that an organization seeks to remedy a social problem. Implicit in the formulation of most social problems is the notion of an intolerable distance from a goal state, and a variable on which that distance is measured. Thus we can refer to the variable Y as the measure of a social state, the existence and magnitude of a social problem being determined by a difference between the actual and the desired values of the variable (A.1)./13/ So we hear that the crime rates, the accident rates, and the suicide rates are too high, while the employment rates, the Gross National Product, and the grade point averages are too low.

Given a set of resources of level X, some technology f transforms inputs into an output described by the dependent variable Y. Thus organizational [955/956] behavior is a function f of a complex state of affairs X, intending to meliorate the social problem (A.2).

A rigorous design can be symbolized for the one independent variable case as follows (the extension to the n-dimensional case is straightforward). Consider

y = f ₀ + x₁ [f ₁ + f ₂x₂] + e

where e is a stochastic variable of zero expectancy and unit variance. Let X₁ and X₂ be dummy variables such that x₁ = 1 if the i-th case is under f ; x₁ = 0 otherwise; and, x₂ = 1 if research findings are disseminated; x₂ = 0 otherwise. Hence f ₁ and the independent variable X are sufficient for the prediction of the organizational outcome in the absence of publication of f _i or X. This can be represented by the two-dimensional diagram shown in Figure 1. We note that for a given interval of program states [x₀, x_i], Y will have a range of [y₀, y_i]. This Y might be a measure of the lifetime earning power of blacks in contemporary America; X might be a measure of prejudice,

[956/957] or of individual aptitude and achievement, or of socioeconomic status, etc. (D.3).

If f is a social change program, we can imagine that if the decision maker discovered a credible research report disclosing that some aspects of the organization was tending to decrease the lifetime income of blacks, then given the provisos about his behavior noted above, he would react to the report and change the organization, thus varying f (A.3). It is possible to describe the induced difference between f _t and f _{t + 1} by a suitable distance function d (D.2). Thus the space of treatments is conceived as a metric space. The value of the output of the organization would presumably change as well. Thus the prediction of program effects is made false by the act of publishing the prediction./14/

Let there be introduced a reaction function R, which is a behavioral parameter descriptive of the dependence of the actual organizational outcome on the organizational decision maker's knowledge of the published judgment (or prediction, i.e. x₂ = 1). Given the relevant range [y₀, y_i], we can represent the reaction function for the two-dimensional case by the schematic diagram in Figure 2.

[957/958] With the variation of Y through the range y₀ to y_i will be associated a variation of the reaction between y₀* and y_i*.

By the contractive mapping theorem, for a metric space <f , d > for any reactions Rf ₁, Rf ₂ if

d [Rf ₁, Rf ₂] £ Kd [f ₁, f ₂]

for the coefficient of adjustment 0 < K < l, then R has a fixed point./15/ At the fixed point, the social system described by the diagrams is in equilibrium, which is to say the organizational decision maker will cease to react to Y. Thus, that value of Y is the correct public prediction of the organizational outcomes,/16/ as well as the optimal state of the organization.

Thus we have shown that the argument against the use of research designs in a dynamic research context is based on an untenable first premise. The dissemination of early research findings and the subsequent modification in the operation or organization, instead of corrupting the design and rendering the research useless, can be viewed as making the design and the research more realistic by permitting the consideration of explicitly dynamic elements in the theory.

We are now prepared to note that if the decision maker is considered a significant factor in organizational outcomes, then his behavior must be explicitly considered in the research plan. Of course the unequivocal partitioning of degrees of freedom will not allow the decision maker's behavior and the behavior of the subject of his decisions, the organization, to simultaneously introduced as variables. In testing a model of the organization we are confronted by precisely the problem of evaluating ex ante forecasts in economics. But it is well known that the identification problem is not unique to econometrics. Thus we are reminded of the necessity of a Lewinian field theory for a comprehensive theory of organization./17/

It can thus be concluded that rigorous research designs not only are applicable to the study of organizations but also permit the study of the decision maker's response to feedback in the real world situation. The use of such research designs will permit the establishment of a causal theory of management and organizations. [958//]

Footnotes
1. Revision of "Structural Models and Dynamic Organizational Research: A Possibility Theorem" presented at the 1970 Annual Meeting of the Eastern Psychological Association, Atlantic City, New Jersey (April 2, 1970). We would like to thank Professor William Beck and John Curry for their kind criticism and comments. Of course, responsibility for any errors remaining rests with the authors.

2. James L. Price, "Design of Proof in Organizational Research," Administrative Science Quarterly, Vol. 13 (June, 1968), pp. 121-122.

3. Stanley E. Seashore, "Field Experiments with Formal Organizations," Human Organization, Vol. 23 (Summer, 1964), p. 166.

4. Paul Chevigny, Police Power (New York: Pantheon, 1969), pp. 272-274.

5. Seashore, "Field Experiments," p. 167.

6. Gordon A. Welty, The Logic of Evaluation (Washington, D.C.: Educational Resources Institute, 1969).

7. Gordon A. Welty, "A Methodological Note on Research Design," Social Biology, Vol. 20 (June, 1969), p. 128.

8. James F. Short, "Action Research Collaboration and Sociological Evaluation," Pacific Sociological Review, Vol. 10 ( Fall, 1967 ), p. 52.

9. Michael P. Brooks, "The Community Action Program as a Setting for Applied Research," Journal of Social Issues, Vol. 21 (Jan., 1965), p. 38; and Peter Marris and Martin Rein, Dilemmas of Social Reform (New York: Atherton Press, 1969), pp. 204-207.

10. Gilbert Sax, Empirical Foundations of Educational Research (Englewood Cliffs: Prentice-Hall, 1967), p. 337.

11. We follow K. W. Rothschild, "Cobweb Cycles and Partially Correct Forecasting," Journal of Political Economy, Vol. 72 ( June, 1964), p. 304.

12. Patrick Suppes, "The Desirability of Formalism in Science," Journal of Philosophy, Vol. 65 (Oct., 1968), pp. 661-664.

l3. The references in brackets refer to the appropriate axioms (A) and definitions (D) given in the appendix of our paper "Structural Models and Dynamic Organizational Research" cited above. The reader is invited to examine that discussion for a complete and rigorous proof.

14. Historical observations on this phenomenon are available in Gordon Welty, "The History of the Prediction Paradox: 1928-1968," presented at the 1970 Annual Meeting of the International Society for the History of the Behavioral and Social Sciences, Akron, Ohio (May 10, 1970).

15. The formal argument here is based on the Cacciopoli-Banach Principle. Cf. Leonid V. Kantorovich and G . P. Akilov, Functional Analysis in Normed Spaces (New York: Pergamon, 1964), pp. 627 ff. For an elementary discussion of fixed point theorems, cf. Richard Courant and H. Robbins, What is Mathematics? (New York: Oxford University Press, 1941), pp. 251-255.

16. For a further discussion of the complex issues of the effects of public predictions in general social contexts, cf. Emile Grunberg and F. Modigliani, "The Predictability of Social Events," Journal of Political Economy, Vol. 62 (Dec., 1964), pp. 465-478 and Herbert A. Simon "Bandwagon and Underdog Effects of Election Predictions," Public Opinion Quarterly, Vol. 18 ( Fall, 1954), pp. 245-253.

17. Cf. Kurt Lewin, "Frontiers in Group Dynamics: II," Human Relations, Vol. 1 (Nov., 1947), pp. 143-153.