HERBERT SIMON, PAUL THAGARD, PAT LANGLEY AND OTHERS ON DISCOVERY SYSTEMS
APPENDIX II- Page 13
Rejections and Rejoinders
In science as in life generally, discovery produces novelty, which is a disturbance that produces negative reaction. And the greater the novelty, the greater the disturbance and consequent reaction. Hickey’s computational metamodel produced a disturbingly novel macrosociological theory, and the disturbance produced a predictable reaction from the conformist academic sociology establishment, which was incapable of assimilating it. The reaction is exhibited below.
Hickey submitted his “A Post-Classical Quantitative-Functionalist Theory of Macrosocial Change in the American National Society” to four peer-reviewed sociology journals in succession. All four rejected the paper. In this appendix he describes his correspondence with the editors of the journals, the attempted criticisms written by their chosen referees, and his rejoinders to the attempted criticisms. The chosen referees are an editor-selected sample presumably representing the best and the brightest that American academic sociology has to offer. But the sample is a dismal exposé of sociologists’ technical incompetence and their philosophical naïveté, the fact that sociology is manifestly retarded. A preface for the following criticisms and rejoinders is set forth above in the sections titled “A Pragmatist Critique of Academic Sociology’s Weltanschauung” and “The ‘Last Sociologist’”.
Sociological Methods and Research
The first academic sociological journal to which Hickey had sent his paper was Sociological Methods and Research published by Sage Publications, Inc. This journal did not acknowledge receipt of the paper, but Hickey’s U.S. Postal Service receipt documents that the paper was received on 18 December 1978. The macrosociometric model itself was actually developed in the latter half of 1976, which is the year of the paper’s registered copyright and the year that Hickey uses to document his priority. It is also the year in which the U.S. Department of Commerce published the statistical compendium Historical Statistics of the United States, which is the principal source of input data for Hickey’s computerized discovery system. On 22 May 1979 – five months later – Hickey received a letter from the editor, a Mr. George W. Bohrnstedt of Indiana University, rejecting the paper for publication. In his letter Bohrnstedt says he is in agreement with the criticisms. With the letter were enclosed the following two referee criticisms, which are paraphrased in detail below together with Hickey’s rejoinders.
Bohrnstedt’s first referee attempted criticisms and Hickey’s rejoinders
Bohrnstedt: Bohrnstedt’s first chosen referee stated that the paper is “theoretically a reification of the worst type”, and that nothing is said about how values are constituted within the population or how they change.
Hickey: The term “reification” is highly ambiguous. But given the romantic character of the remainder of this attempted criticism Hickey can guess at Bohrnstedt’s first referee’s jargon. In Georg Lukács’ History and Class-Consciousness “reification” disapprovingly refers to objectifications of human activity that become estranged from the subjects who produced them thereby masking their social genesis. His agenda is Marxist, self-consciously social psychological and antipositivist. Reification understood as objectification is not always viewed pejoratively. In a more neutral sociological context reification is merely the objectivity of the social world’s institutions, because institutions are independent of and beyond the average individual’s ability to change them. Institutions are indeed real, they change slowly, and Hickey’s macrosociometric model describes the interinstitutional progression of those changes through history.
Talcott Parsons opposed the “reification” that he believed he found in the writings of the positivists. Apparently this referee concluded incorrectly that because Hickey’s paper sets forth a quantitative model built with measurement data, the model must be positivist in spite of Hickey’s discussion of pragmatism in his paper. Parsons considered reification to be fallacious and objectionable, because he viewed it as “monistic” realism requiring that all scientific theories be reduced to one, if they are not to be regarded as fictional. The monistic view is the Unity-of-Science agenda of the Vienna Circle positivists. Hickey is a pragmatist, not a positivist.
Parsons proposes his own alternative ontological thesis of “analytical realism”, according to which the concepts of science grasp aspects of reality. Hickey’s pragmatism is also realist, and Parsons’ “analytical realism” suggests the contemporary pragmatists’ thesis of ontological relativity, which also admits multiple ontologies. But Bohrnstedt’s chosen referee is apparently innocent of this pragmatist thesis, which enables empirical scientists to avoid the fallacy of overriding empirical criticism with one or another prior ontological prejudice. However Parsons is inconsistent with his analytical realism, because he is also a romantic that demands a mentalistic ontology.
This referee is a romantic, because he demands description of how values are constituted and how they change. His claim that nothing is said about how values are constituted within the population or how they change is actually two claims. Firstly the demand about how values are constituted is a throwback to social-psychological reductionism, and to the classical “mechanisms” of socialization and social control, which as Hickey made clear in his paper, are incapable of addressing the macrosociological problem addressed by his paper. The critic is attempting to force the author to change the topic of his paper to what the critic understands, because he evidently knows little about the modeling techniques set forth in the paper. As the author of this paper Hickey claims the right to decide what he will write about, and rejects this critic’s attempt to dictate his own favorite topics.
Secondly the demand that Hickey describe how values change is irrelevant to the validity of the model. Hickey’s model is very much about values, institutions and consensus. And some values associated with the various institutional groups undoubtedly have changed over the fifty years of the sample period. But that fact does not invalidate the sociological significance of the per capita voluntary group associational rates as measures of consensus. If this referee’s criticism were valid, then every economist’s market model would also be invalid, because products such as automobiles change over time, but the price, quantity and value relationships described by the supply and demand equations are still valid without having to describe how automobiles have changed. This naïve critic does not understand that his criticisms are irrelevant. To repeat the language in Hickey’s paper: the macrosocial changes that Hickey’s model describes are changes in the population’s degrees of consensus about values that are characteristic of the represented types of institutional groups as measured by their aggregate voluntary institutional group-associational behavior relative to the aggregate population.
Bohrnstedt: Bohrnstedt’s first chosen referee rejects equation (6) complaining that it is a reversal of Weber’s causal ordering, saying it is “offensive” and calling Hickey’s causal claim “trite”.
Hickey: Webster’s Dictionary of the English Language defines “trite” as boringly obvious, but why that applies to Hickey’s reference to Weber is mysterious given the referee’s criticism. There is nothing in Hickey’s paper that any sociologist could soberly call “trite” beyond Hickey’s opening recitation of classical functionalism. Bohrnstedt’s chosen referee rejects Hickey’s reference to Weber, saying that Hickey has the direction of causality reversed. Does Bohrnstedt’s chosen referee think this is trite? Furthermore Hickey is well aware that Weber says in Protestant Ethic and the Spirit of Capitalism that religion influenced the values of the capitalist economy. The relation Hickey mentions in connection with Weber was occasioned by Weber’s thesis of “elective affinity”, which has been interpreted by many Weberian scholars to imply reciprocity or mutual reinforcement. Hickey thinks that Weber was more sophisticated than this referee. Causality in human interactions is not a simple unidirectional influence. Causal factors are always located in a network of relationships wherein there is reciprocal interaction.
In Hickey’s model this interaction is demonstrated in the execution of the model through its iterations for its simulations, where the religion variable affects every other variable and vice versa. Furthermore since the equations are linear and therefore monotonic, each endogenous variable can be transformed to be expressed as a function of any other, thus showing that Bohrnstedt’s first chosen referee’s presumption of unidirectional causality is naïvely simplistic. But in future versions of the paper Hickey deleted any reference to Weber, because his paper is not a gloss on Weber, and because for critics such as this referee issues about interpretations of Weber are needless distractions.
Bohrnstedt: Bohrnstedt’s first chosen referee claims that Hickey’s “value-based modeling” is “inferior” to the demographic accounting framework developed by Kenneth Land, because he believes that Hickey’s model does not generate interpretable structural parameters. He also objected to Hickey’s not referencing Land, and claims that interpretability of parameters is one of the biggest advantages of Land’s demographic accounting approach.
Hickey: Kenneth C. Land is a Duke University sociologist with his own agenda for sociological modeling. Hickey’s model can indeed be described as “value-based”. It uses demographic data but is nothing like Land’s agenda. Hickey since discovered that Bohrnstedt had previously published an article by Land titled “A Mathematical Formalization of Durkheim’s Theory of the Causes of the Division of Labor” in Sociological Methodology (1970). This fact occasions Hickey’s belief that Bohrnstedt is a patronage-driven editor, that he is not objective.
Bohrnstedt’s first chosen referee calls Hickey’s value-based modeling “inferior”. This claim is false. Bohrnstedt’s chosen referee clearly recognizes that Hickey’s value-based modeling is a competitive alternative to Land’s agenda, and the critic therefore seems to find Hickey’s modeling threatening. In fact Bohrnstedt’s chosen referee objected to Hickey’s not referencing Land. But Hickey’s paper did not reference anything by Land, because Hickey’s modeling has no need for it. Hickey is not indebted to Land or to Land’s approach much less to any of Land’s analyses. And Hickey is unwilling to be conscripted by Bohrnstedt and his chosen referees to support Land’s agenda as a condition for publication. Bohrnstedt and his chosen referee are playing intermural academic politics.
Hickey’s construing per capita rates as measures of consensus enables giving sociological significance to the vast watershed of data collected and released by the several cognizant Federal government agencies. The sociological relevance of these time series gives national demographic data more than just demographic significance, because it enables distinctively macrosociological modeling. For example Hickey finds sociological relevance (i.e., interpretability) in the high-school graduation rates in his model, because the completion of high school reveals voluntary group-associational behavior in response to values characteristic of the educational type of institution. Obviously voluntary high-school school dropouts do not value education. Thus the high-school graduation rates measure degrees of consensus about this educational institution’s characteristic values.
As for interpretability of the coefficients: the statistically estimated coefficient measures the causal impact of changes in the phenomenon described by each associated explanatory variable of the equation upon changes in the phenomenon described by dependent variable of the equation. This practice of relativized or contextual semantics is characteristic of contemporary pragmatism as well as linguistics. It justifies a macrolevel representation that identifies macrosociology as a perspective in sociology separate from classical social psychology, just as starting with Keynes macroeconomics became recognized as a perspective in economics separate from classical economic psychology.
Furthermore the coefficient for each independent variable in an equation is interpretable as an “elasticity coefficient” as understood by economists, because the relation of the dependent to the independent variables can be expressed as the ratio of the two change rates to each other. Understanding this interpretation is not “rocket science”, but amazingly Bohrnstedt’s chosen referee is in obdurate denial of this evident interpretation, and with Bohrnstedt’s complicity has had Hickey’s informative empirical findings suppressed.
At the time Hickey was not aware of Bohrnstedt’s blissful innocence of the contemporary pragmatist philosophy of science, until he later found that Bohrnstedt is a co-author of an undergraduate-level textbook titled Statistics for Social Data Analysis. The textbook effectively advocates an ersatz version of Haavelmo’s structural-equation agenda first published in Econometrica in 1944, which implements the neoclassical economists’ romantic philosophy. Like Haavelmo these authors distinguish unobserved “conceptual variables” from observable “indicators” thus revealing ignorance of the contemporary pragmatist thesis of relativized semantics.
Furthermore in his textbook Bohrnstedt demands identifying causality prior to statistical modeling and testing thus revealing ignorance of the contemporary pragmatist thesis of ontological relativity. This amounts to announcing the findings of the modeling before the modeling construction is performed. Bohrnstedt’s textbook as well as his selection of referees reveals that he is an agenda-driven editor, and unfortunately for his journal and its readers his agenda is technically inadequate and philosophically retarded.
Bohrnstedt: Bohrnstedt’s first chosen referee says that Hickey’s system is a step-wise, “self-cooking” program that generates a model without researcher intervention, and that such “atheoretical routinization” is “inappropriate” for structural-equation modeling.
Hickey: If this rant is not by Bohrnstedt himself, it suggests that Bohrnstedt’s chosen referee is or has been an undergraduate student taking a sociology course using Bohrnstedt’s textbook. The phrase “structural-equation” modeling is code language from the Haavelmo agenda, which this critic is intent upon imposing on Hickey’s modeling, although this sociologist has probably never heard of Haavelmo. Hickey readily agrees that his approach is “inappropriate” for structural-equation modeling, because Hickey does not subscribe to the Haavelmo agenda, and therefore his macrosociological theory is not a structural-equation model.
Furthermore Hickey’s paper did not say that he used step-wise regression, and Bohrnstedt’s critic’s claim that Hickey uses it is presumptuously wrong. Had this referee read Hickey’s Introduction to Metascience, he would know better than to make such a distorting misrepresentation. Bohrnstedt’s critic’s “self-cooking” rhetoric is a Luddite’s diatribe. In the “Introduction” to his Models of Discovery Nobel-laureate economist Herbert Simon, a founder of artificial intelligence wrote that “dense mists of romanticism and downright know-nothingness” have always surrounded the subject of scientific discovery and creativity. What can an author think of an editor of a peer-reviewed journal who chooses a Luddite for a referee for a paper like Hickey’s? He must conclude that the world still awaits the disenchantment of academic sociology.
Bohrnstedt: Bohrnstedt’s first chosen referee believes that he perceives autocorrelation of the residuals in a number of the graphic plots included with this version of the paper.
Hickey: There is only small serial correlation in Hickey’s equations, which is innocuous due to the very small residuals. Furthermore Hickey’s model is very acceptable for showing the interinstitutional adjustment patterns in social change. But Hickey included the Durbin-Watson statistic and removed the graphic plots in future versions of the paper submitted to other journals to forestall such an objection, which was not made by any other referee.
Bohrnstedt: Bohrnstedt’s first chosen referee also claims that there is “unnecessary inefficiency” caused by the pooling of data into four-year units, and prefers single-year observations.
Hickey: Bohrnstedt’s chosen referee should invest in a dictionary of the English language; his “inefficiency” rhetoric is as misconceived as his claim of “triteness”. In fact the combination of observations into four-year periods enhances efficiency in the use of computer resources by reducing computation in Hickey’s very computer-resource-intensive discovery system using the generate-and-test design. The four-year periods also remove the need for any distributed lags, which greatly reduce degrees of freedom.
But as it happens, Hickey’s use of four-year-period time increments is entirely incidental due to the fact that one of the inputs to the discovery system was the political party of the U.S. President, who holds office in four-year terms. But this political variable is a social-conflict variable, and was not selected by the discovery system for any outputted model.
Bohrnstedt’s second criticism and Hickey’s rejoinders
Bohrnstedt: Bohrnstedt’s second chosen referee says that he can’t quite figure out whether or not Hickey’s paper is “a put on”.
Hickey: Hickey thought the first referee was bad, but the second is a grotesque parody. The “put on” comment is a defamatory slur. Sociologists who study the behavior of crowds know that anonymity fosters irresponsibility. Hickey believes that Bohrnstedt’s releasing such a comment as criticism exposed Sociological Methods and Research as disreputable, and he consequently refrained from submitting any rejoinders to Bohrnstedt. Hickey believes that Bohrnstedt’s selection of such a referee for his Sociological Methods and Research journal is a misfortune for his journal’s reputation and a disservice to its readers. In fact Hickey believes that selection of Bohrnstedt to be the editor is a misfortune for this journal’s reputation and its readers.
Bohrnstedt: Bohrnstedt’s second chosen referee demands a “theory” to inform the specification of the models (i.e. the choice of explanatory variables) and to explain why equations of the particular form employed were actually used. He demands “justification” for the particular variables introduced, and says that statistical inference is totally ignored, that serial correlation is neglected, and that the results are not addressed to any “depth”. He claims therefore that the findings are “almost without meaning”, and that if there is meaning, Hickey does not communicate it. He concludes that he wishes that there were some improvements that he could suggest, and says, “I really can’t understand what the paper is trying to do.”
Hickey: The Encyclopedia of the Social Sciences defines functionalism as a theory of how major social patterns operate to maintain the integration or adaptation of larger social systems. It is remarkable that Bohrnstedt’s chosen referee should not recognize sociologist’ conventional view that functionalism is a sociological theory. The critic’s belief that there is no “theory” reflects his romantic concept of theory. But Hickey’s paper sets forth the findings from original research, and therefore implements the contemporary Pragmatist philosophy of science, according to which empirical testing is the pragmatics of theory, such that the model is the theory and the theory is the model, because the model is the language that is proposed for testing.
The equation specifications are extracted from the data, and the selection of variables is justified by the empirical adequacy of the equations that have been estimated over fifty years of American history. Does this referee really believe that the model represents fifty years of coincidence? Contrary to Bohrnstedt’s chosen referee’s comments statistical inference is not ignored; it is employed. Hickey addressed the erroneous claim of serial correlation by adding the Durbin-Watson statistic in future versions of the paper sent to other journals.
In this version of the paper Hickey had set forth only the estimated equations to display the interinstitutional structure of the national society. This is sufficient information to preclude this critic’s complaints that the findings are “almost without meaning”. But to prevent future such nonsense Hickey added the static equilibrium analysis to display the macrosociety’s malitegration and the three types of dynamic simulation analyses to display the national society’s interinstitutional adjustment patterns through time. These analyses offered additional insights and occasioned improvements in the paper but occasioned no change to the model itself.
Nonetheless as the paper says (to repeat
for the obdurate sociologists),
the meanings of the
variables are that they measure voluntary group associational
behavior and therefore represent degrees of macrosocial
consensus about institutional values characteristic of the
different types of institutional groups. If this critic’s phrase “depth” is anything but obscurantism, it
is code language for social-psychological reductionism. The critic’s failure to
understand what Hickey’s paper is “trying to do” is the result
of the critic’s asking the old questions, questions that have
social-psychological answers usually obtained by small-group
is not the kind of question addressed by a macrosociological
analysis, so it is not surprising that the critic cannot
understand Hickey’s answers or their significance. Hickey also
admonishes Bohrnstedt that since this referee admits that he
cannot understand the paper, Bohrnstedt should have selected one
that can. No referee
can be persuaded by what he cannot understand.
American Journal of Sociology
The second sociological journal to which Hickey sent his paper is the American Journal of Sociology published by the University of Chicago Press and edited by an Edward O. Laumann, then the Sociology Department Chairman at the University of Chicago. The journal’s stationary lists a Winfred Hunt Benade as editorial manager. The journal acknowledged receipt of Hickey’s paper on 19 October 1979, and on 21 November 1979 Hickey received a rejection letter signed by Laumann together with two criticisms cited as reasons for rejection. Hickey submitted rebuttals to the journal, which yielded another referee criticism together with a second rejection letter dated 30 July 1981 and signed by Laumann stating that “several internal” referees had reviewed and rejected Hickey’s rebutting rejoinders. The Internet web site for the University of Chicago identifies Laumann as a 1964 Ph.D. sociology graduate of Harvard University, and the site lists Parsons as the first among his teachers who influenced him. The criticisms by Laumann’s chosen referees are the most dogmatically romantic that Hickey had received from any of these sociological journals. Laumann’s choice of such referees led Hickey to believe that Laumann is another agenda-driven editor.
The first sociology department in the United States was founded at the new University of Chicago by Albion W. Small, and the American Journal of Sociology, the first sociology journal in the United States, was founded by Small in 1895. In his A Short History of Sociology Heinz Maus reports that under Small the journal always carefully avoided narrowness and one-sidedness in its outlook. The criticisms made by Laumann’s selected referees show that Laumann has not followed in Small’s diverse outlook. The criticisms of Hickey’s paper reveal that Laumann enforces social-psychological doctrinairism. By selecting his anachronistic referees and their narrow outlook Laumann has marginalized this University of Chicago’s flagship sociology journal. The three referee criticisms enclosed with Laumann’s rejection letters are paraphrased below.
Laumann’s first criticism and Hickey’s rejoinders
Laumann: Laumann’s first chosen referee says that Hickey’s paper is “useless” because: (1) the data are not sufficient to the purpose, (2) the theory is not developed, (3) the indicators are not discussed to justify their theoretical use, (4) no other types of data or theorizing are brought to bear to assess the general value of the theory developed, and (5) the “general kind of enterprise” of developing theories automatically “without thought” is “not generally fruitful”.
Hickey: Firstly with respect to “useless” – The referee’s statement is false. Utility is validating for applied research, but it is not necessary for basic research. Much basic research (e.g., astronomical cosmology) is useless. Laumann’s chosen referee does not understand basic research, and perhaps should have chosen social work instead of sociology for his occupation. Nonetheless in the basic-science perspective Hickey’s model is useful, because it is informative. If this referee had not been so careless and prejudiced in his reading of the paper, he would have seen that the simulation analyses demonstrate the strategic rôle both of rising per capita real income growth rates and of rising secondary-education completion rates for increasing macrosocial stability including rising compliance with criminal law proscribing homicide. The current problems of slow economic growth and of large high-school dropout rates have had manifestly disintegrating macrosocial consequences for the American national society. Thus one useful implication of Hickey’s macrosociometric model shown by the simulation analyses is that generous public funding for education is beneficial for the macrosociety. Another is that rising per capita real income growth due to macroeconomic expansion is beneficial for the macrosociety.
Furthermore this sociometric model’s finding later suggested the development of an econometric model for an optimized State public investment fiscal policy, while Hickey was the senior economist for the State of Indiana Department of Commerce. The State fiscal policy model showed the optimum level both of private-sector employment and of State fiscal revenues. It showed that the optimum expenditure allocation of increases in the State budget is for primary and secondary education. At the Lt. Governor’s request Hickey drafted a speech describing his econometric findings for the Speaker of the Indiana House of Representatives in support of the Governor’s successful “A+” legislative initiative to increase expenditures for K-12 education in Indiana. This finding also corroborated the importance of K-12 education indicated by Hickey’s Post-Classical Quantitative Functionalist Macrosociological Theory of the American National Society, which he constructed with his METAMODEL discovery system.
Furthermore with respect to usefulness Hickey later incorporated the model’s equation specifications into a larger macrosocio-econometric model of the national economy, which he had developed for the Division of Economic Analysis of the Indiana Department of Commerce. This model was used for long-term economic analyses for economic development of the State economy that supported increased public financing for the State government’s Indiana Corporation for Science and Technology. Economists call this mixed economic-sociological type of model an Institutionalist model.
All of these five comments by Laumann’s first chosen referee are either irrelevant or wrong. Hickey’s rejoinders to the above five itemized objections are as follows:
(1) Firstly to say that the data are not sufficient to the purpose is false, because the sufficiency of the data is demonstrated by the satisfactory statistical properties and retrodictive out-of-sample performances of the equations that were estimated from a lengthy sample representing more than fifty years of American history. And invoking the contemporary pragmatist principle of ontological relativity, Hickey maintains that the equations describe causal influences. Hickey’s equations do not describe fifty years of incredible coincidence!
Furthermore the per capita rates were transformed into first-differences, i.e., growth ratios. This transformation not only scaled the data to prevent ill conditioning in the computer calculations, but also enhanced the variances in the data, so as effectively to eliminate collinearity from the independent variables in the estimated equations. This transformation enabled the equations to measure more accurately the causal impacts of the explanatory variables on the dependent variables. This transformation to first differences also increases the difficulty in obtaining satisfactory statistical properties for a model, yet notwithstanding this added difficulty the statistical properties of Hickey’s equations indicate that they are empirically acceptable stochastic models.
(2) Secondly to say that the “theory” is not developed reflects the critic’s romantic concept of “theory”. Laumann’s chosen referee is ignorant of the pragmatic concept of theory that is actually operative in successful basic research. Hickey rejects the social-psychological reductionist agenda operative in this critic’s classical concept of sociological theory. Hickey’s paper is a contemporary pragmatist project, in which the theory is the model and the model is the theory, because the model is the language that is proposed for testing and that is tested. The pragmatics of theory language in science is empirical testing.
(3) Thirdly to say that the “indicators” are not discussed to “justify” the “theoretical” use made of them is more social-psychological reductionism. Were this referee’s reductionist demands carried to another level, he would further require reduction of social psychology to sociobiology, which invalidates social psychology as valid theory unless “justified” by biology. And the biology in turn would have to be “justified” by biochemistry. This romantic critic would not accept these successive reductions, but why not if he can disregard the problems that motivational analyses encounter in social psychology due to unintended outcomes? In fact the variables in Hickey’s model are discussed quite adequately in his macrosociological paper and are appropriate for a macro perspective and his model describing determinants of outcomes rather than motives.
(4) Fourthly the demand for other types of data is gratuitous. Laumann’s chosen referee has not referenced any variables in any empirically superior models. His demand is cynical, because it can be made of any paper at any time by the obstructionist referee, and is indicative of the critic’s mental state of denial. For him there could never be sufficient data or evidence, and the demand is a wild goose chase. In other words, he is “sandbagging”. Furthermore no other data are necessary to affirm the model’s empirical adequacy and thus its explanatory efficacy.
(5) Fifthly Laumann’s chosen referee says that the “general kind of enterprise” of developing theories automatically without thought is not “generally fruitful”. This rhetoric is a red herring, because Laumann’s chosen referee is not referencing Hickey’s model. Hickey’s model is not “generally”; it is the one in his paper, and his discovery system has been fruitful. Furthermore the criticism is pretentious, because Laumann’s chosen referee gives no evidence that he commands the competence or experience with mechanized quantitative data analysis to say responsibly what has been “generally fruitful”. He is manifestly ignorant of the requisite computational analysis, statistical modeling and pragmatist philosophy.
Laumann: Laumann’s chosen referee further elaborates on this fifth objection. He claims that the idea of replacing thought by the automatic working of a theory-building computer program requires a computer program with somewhere near the degree of complexity of a scientist’s “intuition”, and he rejects developing theories “automatically without thought”.
Hickey: “Intuition” is a nondescriptive weasel word invoked by ignorant persons in the pretense of explaining what they cannot explain. The word is uninformative. Imagine that you are not an automobile mechanic, but you wish to make repairs to the transmission in your automobile. So you ask a professional automobile mechanic for some free advice. In response he nods, smiles and says, “Just use your intuition!” Be satisfied that you got your money’s worth of free advice. For the romantic such as Laumann’s chosen referee “intuition” is an evasive escape from responsible empiricism.
Laumann’s chosen referee is a latter-day Luddite. As it happens, in his Extending Ourselves (2004) the University of Virginia philosopher of science and cognitive scientist Paul Humphreys, reports that computational science for scientific analysis has already far outstripped natural human capabilities, and that it currently plays a central rôle in the development of many physical and life sciences. The Luddite criticism by Laumann’s chosen referee explains why no such development (excluding Sonquist’s AID system developed as a dissertation at the University of Chicago’s sociology department prior to Laumann’s ascension as department chairman) can be found in academic sociology. Today’s functioning computerized discovery systems are existential proof that mechanized theory development for empirical science is not only possible, but is ongoing. Hickey believes that this Luddite is the kind of critic that Laumann wanted; such is sociology in Laumann’s reactionary and retarded Sociology Department at the University of Chicago.
Laumann: Laumann’s chosen referee makes additional criticisms. He says that inadequacy of the data is illustrated by equation (3), because the shape of the time series for the homicide rate is almost level in the 1920’s and early 1930’s, drops in the thirties and in the 1950’s, and then rises markedly. He thus claims that there are “two pieces of information”, and that any fluctuation in any other curve, which has that general shape, will result in the crime rate being either a predictor or an effect. He refers Hickey to an “ancient paper” by Sergeant in Review of Economics and Statistics, and describes the data as “sticky”.
Hickey: This is the kind of criticism made in the 1930’s by ignorant economists who incorrectly thought that linear equations cannot produce cyclical findings. The critic’s description of the time series does not invalidate the empirical adequacy of the equations containing the crime variable. On the contrary, the existence of inflection points in the time-series data is helpful for making valid statistical inferences for longitudinal equations. Furthermore as the model is iterated through its recursive structure each endogenous variable is both cause and effect. In microeconomics reciprocal causality is recognized even in a static model representing the price-quantity relations in the equations for market demand and supply.
The Review of Economics and Statistics started publication in 1910, and the name Sergeant never appears in any issue in the journal’s publication history. This spurious reference is as bogus as Laumann’s chosen referee’s whole critique. Laumann has selected a referee that is not just incompetent but is actually deceptive. This attempted deception suggests the corruption in Laumann’s American Journal of Sociology peer-review process.
The image of “sticky” data truly staggers the imagination. Hickey can only retort that his data are not nearly as “sticky” as this sociologist’s lame criticism is tacky.
Laumann: Laumann’s chosen referee goes on to claim that the homicide rate time series does not measure any uniform condition of the social body, because violent crime has been concentrated more in the ghetto in recent years than it used to be – i.e. the distribution of homicide has been changing. Thus he concludes that the condition of the country as a whole is not a good indicator. He claims that what the indicators mean is not discussed with an elementary sociological sophistication. He also denies that the rate of formation of new businesses is “group associational behavior in the economic institutional group”, and calls it “mere theoretical hocus-pocus”. He makes similar comments about the heterogeneity of the marriage rate over time, noting that the proportion of all marriages that are first marriages has been going down.
Hickey: Hickey’s model is not a regional model; it is a macro model of the national macrosociety. The variables in Hickey’s model measure what they say they measure, and if this critic wants anything else, he should write his own paper instead of attempting to force Hickey to change the topic of his macrosociological paper.
The heterogeneity of the national data is irrelevant to the subject addressed by the model, because it bears no relation to the empirical adequacy of the equations based on the national aggregate variables. If heterogeneity invalidated these equations, then every macroeconometric model containing such heterogeneous variables as aggregate national consumption or aggregate national investment would be invalidated given the heterogeniety of consumer or investment goods and services and their diverse geographical distribution in the national economy. No economist today is as naïve as Laumann’s chosen referee, and none would commit the logical non sequitur made by this critic and say that that the aggregate consumption functions or investment functions in macroeconometric models are invalid due to the heterogeniety of the aggregate national data. Furthermore experienced modelers know that the more aggregate the data, the more statistically reliable the model and the more accurate its forecasts. This referee does not know what is relevant to the macrosociological perspective.
Also with respect to concentration of violence in the ghettos, it is noteworthy that the discovery system did not select the urbanization rate in the equation for conformity to criminal law prohibiting homicide. But it is also noteworthy that the shock simulation shows that exceptionally rapid urbanization increases crime. Thus urbanization as such does not produce crime. But accelerated massive urbanization is a large human ecological disturbance resulting in social disorganization that does increase crime. The change rate of the demographic ratio of Negroes to Caucasians was a variable inputted to the discovery system, and it was not selected in the outputted model.
Likewise changes in the proportion of all marriages that are first marriages are another irrelevant heterogeneity; the marriage rate variable is a measure for all marriages.
Laumann’s chosen referee’s dismissive “hocus-pocus” rhetoric is a frivolous attempt to trivialize Hickey’s sociological interpretation and findings. Referees use such rhetoric when they cannot criticize. Laumann’s chosen referee is in obdurate denial of the sociological significance of the per capita rates for the institutional variables. To repeat (and repeat and repeat!): The per capita rates having numerators that represent voluntary group-associational behavior show degrees of consensus about the values characteristic of the relevant types of organized institutional groups. It is impossible to speak of voluntary group-associational behavior without reference to values, because voluntary behavior is in response to values.
Hickey adds that his interpretation of per capita rates to reveal cultural values is comparable to Nobel-laureate economist Paul Samuelson’s interpretation of unit prices to reveal economic values in the latter’s thesis of “revealed preference” set forth in “A Note on the Pure Theory of Consumer’s Behavior” (1938) and in “Consumption Theory in Terms of Revealed Preference” (1948). Samuelson rejected the Austrian school’s romantic concept of subjective utility, which is an introspectively perceived and unmeasurable psychological experience of consumer satisfaction that motivates consumer behavior. Instead he describes consumer preferences in terms of observed relative prices. Thus a commodity’s relative per-unit-price measurements reveal economic value in the observable choices of the market transaction, even though the per-unit-price data do not characterize the economic value except in association with an identified type of consumer product or service. Similarly the per capita-rate measurements of institutional group-associational behavior reveal the degree of consensus about the set of cultural values characteristic of the type of institutional group, even though the per capita rate does not characterize the cultural value except in terms of the associated type of institutional group.
Laumann: Laumann’s chosen referee claims that interpretation of the coefficients in the theory is “vague”, as when Hickey compares the sizes of the coefficients with different numeraires; specifically the reciprocal of the murder rate is a very large number, so when the murder rate doubles the reciprocal increases by a very large number, while the rate of business formation per capita can double without varying by such a large number. Laumann’s chosen referee thus concludes that the “dimension” of these coefficients is left very vague.
Hickey: This is a critic who needs a dictionary, because there is no problematic vagueness here. Laumann’s first chosen referee has confused dimension with magnitude. All of the data used in the model are firstly transformed into per capita rates, which are then further transformed into change ratios of those per capita rates and then into index numbers of the change ratios with the out-of-sample period set as the base period for measuring each generated model’s forecast accuracy. Thus coefficients relating these change ratios to one another in the equations are nondimensional like the economist’s nondimensional price elasticity coefficients that relate change rates. Each coefficient in Hickey’s equations measures changes in the dependent variable in response to changes in the associated independent variables in an equation.
Laumann: Laumann’s first chosen referee then returns to his problem of replacing thought by the automatic working of a theory-building computer program, and he claims that this requires a computer program somewhere near the degree of complexity of a scientist’s “intuition”. He again claims that Hickey’s discovery system is a stepwise regression, and that it therefore cannot really be called a “meta-theoretical program”. He adds that there is a lot of “windy garbage” about the language in which these equations are described, i.e., “about semantic and syntactic and whatnot.”
Hickey: This exhibition of shallow know-nothingism is truly astonishing. Romantics love the term “intuition”, because it is useful as a strategically uninformative weasel word that they invoke, so they can appear to explain what they cannot explain. Hickey finds no evidence that Laumann’s first chosen referee has any idea what a metatheoretical computer system looks like, and finds the critic’s pretension ludicrous. Hickey’s METAMODEL system described in his Introduction to Metascience and referenced in his paper’s bibliography produced the theory that is displayed in his paper; the theory was not created by a stepwise regression.
Discovery systems have been in use for many years including the sociologist Sonquist’s AID system. Hickey’s system has been of great value to him professionally for more than thirty years for his Institutionalist econometric modeling in both business and government. This critic’s rejection of mechanization is Luddite rant betraying his dismissive mentality. The “windy garbage” conceit is another manifestation of shallow knownothingism. Generative grammar enables mechanized syntactical construction, and Quine’s contemporary pragmatist philosophy of language enables semantical interpretation of the outputs.
Laumann: Laumann’s chosen referee denies that Hickey’s paper describes measurements of the values by which people live, so as to construct a model of how values make the social system stable. He claims that there is not in the whole paper a single measure of values, nor is there anything about how our values have changed over time. Laumann’s chosen referee recognizes only “a bunch of simulations” with the model, which show that it is “absolutely useless”, because it produces a positive feedback between the birth rate and the marriage rate of a cyclical sort. He thus compares Hickey’s model to the Lotka equations of “hares and foxes”, which produce explosive cycles until the whole world is covered with no hares and starving foxes. He concludes that Hickey’s model is eighteenth-century demography.
Hickey: Laumann’s selected romantic critic has chosen to be dismissive about the semantics of the institutional variables, to be in obdurate denial about the model, and to make statements that are blatantly false. This is due to his fixation for the description of values, to which he wants Hickey to conform. To repeat still again: Hickey’s paper is not a social-psychological description that is characteristic of classical sociologists like Laumann’s chosen referees. Parsonsonian motivational analysis is not capable of explaining the outcomes described by the model’s simulations. Hickey reserves an author’s right to decide the subject of his paper, and he has been clear about his subject, which this critic ignored.
Hickey does not say that the data are measures of values; indeed there can be no such a thing as a measure of the subjective experience of value any more than there can be a measure of the subjective experience of utility used in classical economics. To repeat (and re-repeat for this obdurate referee) the relevant discourse in Hickey’s paper: The numerators of the per capita rates measure degrees of macrosocial consensus about values characteristic of the several institutional types of groups as revealed by aggregate voluntary group-associational behavior. Typically no one voluntarily joins or remains in a group, if social controls in the group enforce values that he rejects. “Voluntary” means behavior in response to values. What are measured are the degrees of consensus in the general population or other relevant denominator in the per capita rate. The Federal government offers a vast watershed of sociologically relevant longitudinal data, but sociologists’ obdurate refusal to recognize its sociological relevance is a huge missed opportunity that has retarded the profession’s development of quantitative empirical macrosociology.
Laumann’s chosen referee equated the positive feedback between the change in birth rates and the change in the marriage rate to the models created by Alfred J. Lotka. The Lotka-Volterra equations are viewed as quite reputable by contemporary biologists to analyze inter-species population dynamics. If this critic’s comparison were correct, it might independently corroborate Hickey’s model instead of criticizing it.
But the comparison is wrong. Lotka applied his model to biology in 1920 to describe predator-prey interspecies equilibrium dynamics in the demographics of wildlife. Hickey’s U.S. macrosociometric model differs mathematically from Lotka’s, because the latter’s is nonlinear, continuous and has only two variables and two equations, while Hickey’s macrosociometric model is a first-degree, higher-order difference system of nine equations, nine endogenous variables and three exogenous variables, and his model is statistically estimated from discrete time-series data.
Hickey references Nobel-laureate Paul Samuelson’s multiplier-accelerator interaction model published in 1939, the type of mathematical model most commonly used by econometricians today and used by Hickey in his paper. The behavior of Samuelson’s model depends on the values of its statistically estimated parameters. Likewise when a Lotka model is empirically estimated, as Pat Langley did with his IPM discovery system in his “Discovering Ecosystem Models from Time-Series Data”, the model’s behavior will depend on the statistically estimated values of the parameters in the equations, just as the behavior of the equations in the simulations with Hickey’s model. But this critic references no empirical sociological applications of the Lotka-Volterra type of model in his comparison.
As it happens the actual U.S. birth rates during first three-quarters of the twentieth century exhibited wildly cyclical fluctuations with a declining trend ending in 1975 during the sample period of the model, and the model’s empirically estimated coefficients captured the fluctuations and trends. It was not until the last quarter of the century and beyond that the fluctuations stabilized, and this latter period was not in the sample data. Specifically during the sample period from 1920 to 1972, the period available at the time the model was made, birth rates dropped by 47%, while since then they have dropped by only 15%. Had the most recent forty years following 1972 been available at the time that the model was made, the simulation results for the birth-rate equation would have been different. Curiously in his How Civilizations Die David Goldman maintains that in the latter half of the twentieth century birth rates have been falling in many countries (excluding Israel and the U.S.), and that like many civilizations that became extinct due to depopulation those countries are entering the “fourth Great Extinction” to occur in the twenty-first century. Hickey’s model is not eighteenth-century demography; it is an accurate empirical capture of mid-twentieth-century demography. And consequently it was necessary to make the birth-rate variable exogenous for some of the simulations, in order to isolate the interinstitutional adjustment patterns that the model is designed to explore.
More relevantly the positive and negative interinstitutional feedbacks described in Hickey’s model crucially depend on the values assigned to the exogenous variables that are not found in Lotka’s equations. In addition to the change rate in birth rates, another relevant exogenous variable in Hickey’s model simulations is the change rate of the per capita rate of aggregate real income, which is strategic for the twentieth-century Great Depression era represented in the sample data that Hickey used. Historically both resident population and per capita real income approximately doubled in the entire fifty years covered by the model. There was no long-term Malthusian starvation due to diminishing returns, because in the U.S. there were abundant natural resources, the “green revolution” in crop yields due to technological innovation, and increasing production to scale. Thus instead of Malthusian destitution there were rising living standards except during the Great Depression years due to declining aggregate demand and stagnation during the liquidity-trap period. But in one of Hickey’s “what-if” simulations and contrary to history per capita real income is exogenously made constant at zero-growth rate, thus making the society what economist Lester Thurow in 1980 called a “zero-sum society”, in which anyone’s gain must be someone else’s loss.
But most fundamentally Hickey’s model does not describe any predatory demographic dynamics like Lotka’s model; the macrosociological model contains no predator variables. Thus the referee’s comparison with Lotka is not even wrong; it is frivolously irrelevant.
Laumann: Laumann’s chosen referee wrote that Hickey’s paper is not publishable in The American Journal of Sociology or anywhere else, “for it is thoroughly incompetent in every respect”.