Observation Oriented Modeling

Analysis of Cause in the Behavioral Sciences

James W. Grice

• Publisher: Academic Press
• Year: 2011
• ISBN:
  9780123851956
• Topic: Psychologie
• Subtopic: Experimentelle Psychologie

This book introduces a new data analysis technique that addresses long standing criticisms of the current standard statistics. Observation Oriented Modelling presents the mathematics and techniques underlying the new method, discussing causality, modelling, and logical hypothesis testing. Examples of how to approach and interpret data using OOM are presented throughout the book, including analysis of several classic studies in psychology. These analyses are conducted using comprehensive software for the Windows operating system.

• Describes the problems that statistics are meant to answer, why popularly used statistics often fail to fully answer the question, and how OOM overcomes these obstacles
• Chapters include examples of statistical analysis using OOM

Chapter 1. Introduction

Contents

Metaphysical and Methodological Errors1

Observation Oriented Modeling and Philosophical Realism3

Metaphysical and Methodological Errors

Mortimer wrote candidly of his experiences as a young psychologist on his way to earning a doctoral degree from Columbia University in the 1920s. He described how psychology had been divided into various “schools of thought” and how he learned to recite the dogmas of these diverse schools. Through careful study, he learned to tell the difference between behaviorists, structuralists, Freudians, Gestalters, and Jungians—not because they had made significantly different discoveries regarding the human psyche but, rather, because they had developed their own “point of view.” He also studied human physiology, endocrinology, and neuroanatomy, and he was schooled in the various techniques of psychological experimentation. In one study pertaining to emotions, he would carefully measure pupillary responses as his colleague lowered a boa constrictor onto the head of a wary participant. Yet, upon completing his dissertation, Mortimer was confronted with a frightening reality:

I could not tell my students, my colleagues, or myself, what psychology was about, what its fundamental principles were, or what was the theoretical significance of all the data and findings that thousands of young men like myself had been collecting and assorting ever since the Ph.D. industry and the research foundations had encouraged such labors.

The astute reader will recognize this student of psychology as Mortimer J. Adler, one of the most visible and controversial philosophers of the 20th century (Adler, 1941, p. ix). He wrote of his experience with psychology in 1941, and nearly 70 years later his words accurately describe an academic discipline adrift in diversity with no apparent hope for real unity. From Adler’s perspective, psychology lost its way through a series of metaphysical errors made early in its history. Other distinguished scholars, such as Rom Harré, Daniel Robinson, and Joseph Rychlak, have made similar arguments, noting that in the rush to divorce itself from philosophy, psychology fell into a number of metaphysical traps from which it has yet to fully recover (Harré & Secord, 1973; Robinson, 1986; Rychlak, 1988).

Many of the metaphysical difficulties faced by psychology pertain specifically to its prevailing research tradition and whether or not the premises underlying this tradition have been examined sufficiently. Psychology is in a unique position among the sciences because the object of study (a human person) is also the subject of study. A psychologist, for instance, who concludes from her research that individuals cannot view factual evidence in an unbiased manner must wonder if this conclusion is itself the result of her own biases. Such quandaries are not new to psychology, yet its prevailing methodology has been derived almost exclusively from a philosophical position (i.e., positivism) that denies the significance of the subject–object dialectic. This fact raises the specter of an important question: Are the prevailing research methods capable of yielding an authentic science of psychology? Asked in more specific forms, is significance testing the appropriate tool for evaluating data? Is the randomized controlled experiment sufficient for determining causality? Are parametric statistics appropriate for the attributes studied by psychologists? Such questions have been a nagging thorn in the side of psychology because they are routinely answered in the negative. In fact, more than a few prominent researchers in the field have publicly denounced the prevailing research methodology. For instance, in a book chapter titled “What’s Wrong with Psychology, Anyway?” David Lykken (1991) stated plainly yet forcefully that psychology’s research tradition is fundamentally flawed, describing most grant applications, submitted manuscripts, and published research as simply “bad.” He also indicted the majority of published psychological studies as nonreplicable and as yielding very little in the way of cumulative knowledge. Interestingly, Lykken was simply sharpening the same criticisms made by Paul Meehl nearly 15 years earlier (Meehl, 1978). Jacob Cohen offered a more precise critique of psychology’s prevailing research tradition, arguing that the primary tool for judging the outcome of any given study (i.e., the significance test) actually impedes the advance of psychology as a science. He stated, “I argue herein that NHST [null hypothesis significance testing] has not only failed to support the advance of psychology as a science but also has seriously impeded it” (Cohen, 1994, p. 997). Like Lykken, Cohen was simply restating a valid criticism made nearly 30 years earlier, this time by David Bakan (1967):

I will attempt to show that the test of significance does not provide the information concerning psychological phenomena characteristically attributed to it; and that, furthermore, a great deal of mischief has been associated with its use. If the test of significance does not yield the expected information concerning the psychological phenomena under investigation, we may well speak of a crisis; for then a good deal of the research of the last several decades must be questioned.

(p. 2)

In a similar vein, Benjamin Wright lamented the failure of social scientists to heed previous arguments regarding the importance of establishing the quantitative structure of measured attributes. He concluded, “That is why so much social science has turned out to be no more than transient descriptions of never-to-be reencountered situations, easy to contradict with almost any replication” (Wright, 1997, p. 35). Finally, in two organizational efforts, the American Psychological Association (APA) commissioned a task force to offer recommendations for reforming the statistical reporting practices of psychologists, and the National Institutes of Mental Health (NIMH) gathered the editors of 24 scientific journals to endorse a statement calling for radical change in the prevailing research tradition of psychology (NIMH, 2000):

We believe that traditional, variable-oriented, sample-based research strategies and data analytic techniques alone cannot reveal the complex causal processes that likely give rise to normal and abnormal behavior among different children and adolescents. To a large extent, the predominant methods of our social and psychological sciences have valued quantitative approaches over all others, to the exclusion of methods which might clarify the ecological context of behavioral and social phenomena.

(p. 66) ¹

¹The conclusions from the APA’s task force were published in Wilkinson et al. (1999).

Despite the efforts of these two organizations and those of the prominent and respected academics mentioned previously, no substantial, lasting change in psychology’s research methodology has occurred.

Observation Oriented Modeling and Philosophical Realism

Given that the issues plaguing psychology’s research tradition are at their root metaphysical, any valid way forward must be accompanied by a fundamental shift in philosophy; and it is on the basis of this premise that observation oriented modeling is herein presented as an entirely new way of conceptualizing and analyzing psychological data. As a novel set of methods, observation oriented modeling seeks to explain patterns of observations in terms of their causal structure. This approach stands in stark contrast to the prevailing research strategy that is centered around variable modeling and the estimation of parameters (e.g., means, variances, and proportions) for populations of events that often exist only in theory. By focusing on causal structure, observation oriented modeling aligns itself with philosophical realism and stands in opposition to the positivism that produced psychology’s current research tradition.

Philosophical realism has taken on a variety of forms over the course of history, but it nonetheless represents a continuous line of thought that can be traced all the way back to Plato and Aristotle. It has recently gained a strong foothold in the relatively young philosophy of science discipline and has found a number of prominent voices among physical and social scientists (e.g., see Bhaskar, 1975; Harré, 1987; Manicas, 2006). In general terms, philosophical realism holds (1) there is a world of real existence that is not made or constructed by humanity, (2) a person can know this existing reality in which he or she is an active participant, and (3) such knowledge is the most reliable guide to individual and social conduct (Wild, 1948). ² These tenets may seem to represent common sense, and indeed realist philosophy is sometimes described as “reasoned common sense.” It is also sometimes said that all scientists are realists in the laboratory, but most adopt idealism when participating in armchair philosophy and subjectivism when confronting questions of morality. Regardless, philosophical realism provides a number of distinct advantages over the positivism that has given rise to the prevailing research tradition in psychology. These advantages will be made manifest in subsequent chapters as observation oriented modeling is introduced, but a number of prefatory thoughts will greatly aid its introduction.

²Daniel Sullivan also provides a brief and perhaps more enjoyable introduction to philosophical realism in An Introduction to Philosophy: Perennial Principles of the Classical Realist Tradition (1957/2009).

For the realist, the subject–object dialectic is valid and has merit because it means that a person can come to know the things of nature with increasing clarity and depth. For the moderate realist, following Aristotle and Thomas Aquinas, an act of knowing occurs when a person (subject) intentionally possesses the immaterial form of a thing (the object), and this possession involves both sensation and intellection. ³ When presented with a red book, for instance, a person is, through his senses, immediately aware of the book as an individual existing thing; and at the same time his intellect abstracts from the red book the universal concepts “red” and “book.” These concepts exist within the mind but also in some way within the book as well. Such real concepts are therefore not independent inventions of the mind, although they can be combined, separated, and manipulated in various ways within the mind; and they can be evaluated with other abstract concepts that exist solely within the mind, such as logical concepts and certain mathematical concepts (e.g., zero and null set) (Wallace, 1996, p. 255). New formulations of concepts can moreover be judged for their conformity with the things of nature, such as when a psychologist posits memory as separable into long-term and short-term components and then devises experiments to test these conceptualizations.

³The term “moderate realism” is taken from Wallace (1983). It is meant to convey realism in the tradition of Aristotle and Thomas Aquinas. Wallace exemplifies this viewpoint in his book. See also Wallace (1996).

A consequence of this position is that a scientist can know something of the “things themselves,” and it is perfectly legitimate to speak of the powers and properties of the things of nature. Although modern academics find it fashionable to quote Molière’s clever criticism of Aristotle’s notions of powers and properties, they run the risk of undercutting the possibility of scientific knowledge (an example can be found in Gigerenzer, 2009). Truly, it is circular to argue that opium causes drowsiness because of its dormative properties, but it is a graver mistake still to assume that caffeine or some other compound can all equally be used to explain drowsiness. This is simply not the case because there is something in the nature of opium that plays a causal role in drowsiness and that makes it different from caffeine. Scientists indeed depend on these different natures, at least implicitly, when making general and predictive statements about these very compounds. A realist will furthermore hold that powers and properties can be discovered and made known through the discernment of patterns in normal experience or in the special experience that constitutes a scientific experiment:

What are best termed the “abstract sciences” aim at an understanding of the fundamental processes of nature. Such inquiry may be motivated by discerning a pattern, but not all patterns will be of concern. Indeed, patterns which emerge from experimentally generated data, e.g., the results of Lavoisier’s painstaking use of the chemical balance, are of high importance.

(Manicas, 2006, p. 25) ⁴

⁴Rom Harré (1986, p. 35) also states, “Theories are seen as solutions to a peculiar style of problem: namely, ‘Why is it that the patterns of phenomena are the way they are?’ A theory answers this question by supplying an account of the constitution and behavior of those things whose interactions with each other are responsible for the manifested patterns of behavior.”

The concept of pattern also brings to the forefront other important concepts, such as whole–part relations, unity, and integration, all of which figure prominently in discussions of observation oriented modeling in subsequent chapters.

Presuming things to have particular powers because of their natures brings causality to the center of scientific reasoning as well. Consistent with the notion of pattern described previously, particular causes normally entail particular effects, and it is the primary goal of the scientist to create a model that explains these ordered (i.e., patterned) structures and processes. The models of the atom or human cell are excellent examples of such explanatory, causal models. These models show whole structures with different parts that interact and change over time and that are impacted by various forces or other structures, both internal and external. Psychologists and other social scientists should take as their goal the creation of such models, but they must recognize that reaching this end will require a richer definition of causality than the one supplied by positivism. In observation oriented modeling, Aristotle’s four species of cause (material, efficient, formal, and final) in fact provide this more complete picture of causality. Final cause, in particular, proves important in subsequent chapters because it provides the means for modeling purpose, which is one of humanity’s more enigmatic and important powers.

Finally, the realist claims that things have properties, some of which are essential for making a given thing to be what it is, whereas others are nonessential and referred to as accidental. By stating “this red book is 100 pages in length,” for instance, a person is expressing a species concept, “book,” that denotes the object’s essence. Two accidental qualities of this particular book are that it is “red” and composed of “100 pages.” Other books will naturally have different colors and numbers of pages, but their unity will still be recognized through the species concept of “book.” For the Thomistic psychologists of the early 20th century, the starting point of inquiry was the essential unity of each person, and from this vantage point psychology was defined generically as a “study of the acts, powers, habits, and nature of man” (Brennan, 1941, p. 53). In other words, a person is a unified whole and the object of study; but as an organized composite of distinguishable parts, any given person must be understood using a variety of concepts, such as acts, powers, and habits. The aspects of personhood to which these concepts apply might also be referred to as qualities (or attributes), and from the realist perspective, any scientific investigation of these qualities must attend to their different forms of existence. Aristotle spoke of a variety of different types of qualities, only some of which are quantitative; however, contemporary psychology is largely built on the premise that qualities such as intelligence, emotional states, and personality traits exist as continuous quantities. ⁵ This assumption will be challenged in the pages that fol...

Table of contents

• Cover Image
• Table of Contents
• Front Matter
• Copyright
• Foreword
• Dedication
• Acknowledgments
• Chapter 1. Introduction
• Chapter 2. Data at Its Core
• Chapter 3. Rotating Deep Structures
• Chapter 4. Modeling with Deep Structures
• Chapter 5. Statistics and Null Hypothesis Significance Testing
• Chapter 6. Modeling and Inferential Statistics
• Chapter 7. Modeling and Effect Sizes
• Chapter 8. Measurement and Additive Structures
• Chapter 9. Cause and Effect
• Chapter 10. Coda
• References
• Index