Chapter 1: The Scientific Study of Dream Content
The purpose of this book is to search for meaning in dreams through the quantitative study of dream content. It will do so in three senses of that elusive term. First, it will demonstrate an internal coherence or regularity in the dreams of specific groups, such as men, children, or members of hunting and gathering societies. Second, it will show there is consistency in what individuals dream about from year to year and even over decades. Third, it will reveal correspondences between dream content and waking life; more specifically, it will show a direct continuity between dream concerns and waking concerns.
The book will attempt to realize its purpose through the version of content analysis created by Calvin S. Hall and Robert L. Van de Castle (1966) on the basis of earlier work by Hall (1947, 1951, 1953a) and his students (e.g., Polster, 1951; Reis, 1951; Meer, 1955; Cook, 1956; Paolino, 1964). The Hall/Van de Castle coding system was constructed gradually through the empirical study of thousands of dream reports collected from college students in the 1940s and 1950s. It is the most comprehensive and detailed system for the study of dream content developed to date. Given the explicit coding rules developed by Hall and Van de Castle, all of which will be presented in this book, their categories can be used by any investigator willing to take the time to learn the system.
The Hall/Van de Castle system may be unique among methods of dream analysis in that it relies entirely on the dream reports themselves in order to determine whether or not there is meaning in dreams. It does not use free associations, amplifications, biographical information, or any other information provided by the dreamer. Nor does it draw upon metaphoric, linguistic, or literary methods of interpretation. The Hall/Van de Castle system makes comparisons of dream reports in three different ways to search for dream meaning. First, it compares new dream reports with normative information on American college students and other population groups. Second, it compares one dream or type of dream within an individual dream series with other dreams in the series. Third, it compares reports of specific types of dreams collected from many people, such as dreams of flying or of appearing partially clad in public, with each other and with the norms for dreams in general.
The quantitative study of dream content begins with the careful formulation of categories to encompass the many different elements appearing in dreams. In the case of the Hall/Van de Castle system, these categories include characters, social interactions, settings, the activities engaged in by the dreamer and other dream characters, and a wide range of objects. There are also coding categories for emotions, temporal references, successes and failures, good fortunes and misfortunes, and many other aspects of dream content. Categories can be expanded or combined to fit the needs of specific research questions, and new categories can be created.
As will be shown, the Hall/Van de Castle system has a number of advantages over other coding systems, and in fact encompasses most other systems. For one thing, it is possible to achieve high intercoder reliability. For another, its categories have been shown to be psychologi-cally relevant in terms of the waking concerns of those who have contributed dream reports. Furthermore, its normative findings on the dream reports of American college students, replicated several times, provide a comparison point for studies of dream content all over the world. It therefore has been used by investigators in many different countries, including India and Japan, for a wide range of projects. This widespread use makes it possible for findings to be cumulative and therefore to serve as a reference point for new studies.
Studies of over 10,000 dream reports using the Hall/Van de Castle system have yielded consistent developmental, gender, and cross-cultural differences as well as a core of findings stable across gender lines and cultural boundaries. Studies of lengthy dream diaries from a diverse array of individuals reveal that there are large individual differences in dream content as well as a high degree of consistency in what a person dreams about over the space of several months or years, or even 40 and 50 years in the cases of the two longest dream series analyzed to date. There are also striking continuities between dream content and waking life, making possible accurate predictions about the concerns and interests of the dreamers.
In order to link dream content with the waking thoughts and behavior of the dreamer, the Hall/Van de Castle system makes one basic assumption: the frequency with which a dream element appears reveals the concerns and interests of the dreamer. That is, frequency is assumed to be an indicator of intensity (Hall and Van de Castle, 1966:13--14). The idea that frequency reveals concerns and interests means any statistically significant deviation from the norms in either a high or low direction should relate to psychologically unique aspects of the dreamer's waking thoughts or behavior. The formula for the significance of differences between two independent proportions is used to determine statistically significant deviations from the norms; more importantly, the magnitude of any statistically significant differences, that is, the "effect size," is determined by Jacob Cohen's (1977: chap. 6) "h" statistic. For readers unfamiliar with statistics, these methods are explained in an appendix. To make it unnecessary for readers to calculate either of these two statistics, the appendix includes tables from which they can be determined with considerable ease. This statistical appendix is best read in conjunction with the introduction to chapter 4.
Dreams and Dream Reports
Just what is being studied when we say we are analyzing "dream" content? The word "dream" has three possible meanings. It can refer to (1) an experience during sleep; (2) what is remembered upon awakening; or (3) what is reported to others, usually prefaced by "I had this dream" or "last night I dreamt that..." Put another way, there is an experienced dream, a remembered dream, and a reported dream. Unfortunately, there is no way of knowing what people are dreaming or what they remember until they report their recall of the dreaming experience in words. Thus, only the reported dream has an objective, or public, existence. Of necessity, then, this book will deal only with the reported dream, usually referred to as a "dream report" or "dream narrative." The phrase "dream content" will be used to designate what is found in the dream report (Hall and Van de Castle, 1966: 17-18; Hall and Nordby, 1972: 12).
Dream reports are a unique type of document. Most of what is spoken or written is meant to influence other people or communicate with them, but dream reports are descriptive accounts based on memories of an experience that happened during sleep. On occasion they might be used to influence a psycho-therapist or communicate something to a friend, but for the most part they are "representational," not "instrumental," communications. Moreover, dream reports of the kind used by dream researchers are not generally self-initiated. That is, very few of the people who provide dream reports for our studies would have written down their dreams if they had not been asked to do so, although those who keep dream diaries for their own personal reasons are an important exception. Then too, because dreams are usually experienced as something that just happens to the dreamer, and not as something intended, people do not tend to accept as much responsibility for their dreams as they do for what they say or write during waking life. In all these ways, dream reports differ from other types of oral or written reports (Hall and Van de Castle, 1966: 21).
The quantitative study of dream content is not without its difficulties. There can be problems of bias in the collection of dream reports; for example, aspects of dream content may be altered or deemphasized on the basis of instructions given to the dreamer. Nor is there any guarantee that some subjects do not make up dream reports or alter details of what they actually remember. However, none of these problems is a serious one. Most of the dream reports utilized in the studies to be discussed in this book were collected from anonymous volunteer subjects with basically the same instructions. As for the fabrication of some dreams in a sample, or the deliberate alteration of certain details, this presents no problem in terms of our general findings because they are based on very large numbers of dream reports.
Dream Series and Dream Sets
Dream reports can be grouped in two different ways, as "dream series" or as "dream sets." A "dream series" consists of two or more dream reports from the same person; a "short" dream series contains anywhere from two to 99 dream reports, and a "long" dream series contains 100 or more dream reports. The phrase "dream set" is used to describe both (1) dream reports from persons of a certain "type" (e.g., men, women, children, schizophrenics, Americans) or (2) dream reports of a certain "type" collected from a wide range of people (e.g., dreams of being chased, dreams of falling). The same dream reports are sometimes part of both a series and a set. For example, if we took all dream reports of being chased from hundreds of different dream series, we would have a set of chase dreams. For our purposes, the most important example of a dream set drawn from many dream series concerns the normative studies of the dream reports of European-American college men and women reported in detail in chapter 4. This normative study is based upon five dream reports drawn at random from longer series contributed by 100 men and 100 women at Case Western Reserve University and Baldwin-Wallace College in Cleveland, Ohio, between 1947 and 1950 (Hall and Van de Castle, 1966:158).
The Importance of "Blind Analysis"
It is our strong belief that content analyses should be done with no knowledge of the dreamer if such studies are to be convincing evidence for the usefulness of content analysis. Such "blind analyses" are essential because there is always the possibility, even with this objective and quantitative method, that the analyst is reading into the dream reports what she or he already knows from free associations or biographical information, rather than gaining new insights and information from the dream reports themselves.
It is of course necessary and inevitable that a dream analyst will have other information if he or she is treating the dreamer in a clinical setting. But such clinical analyses, however beneficial therapeutically or useful in generating testable hypotheses about dream meaning, always will be suspect by rigorous scientific standards. This is not only due to the fact the analyst has other information available that might be influencing interpretations, but because such dream analyses are a form of "post hoc" interpretation based on reasoning from the present to the past. The criticism of "retrospective" analyses as unable to demonstrate the existence of causality is a cross borne by all clinical theories.
Unfortunately, it is not possible to answer this type of criticism with experimental studies in the case of dreams. Experimental studies of the cognitive process of dreaming have shown that some external (e.g., water drops, hearing significant names) and internal (e.g., thirst) stimuli can sometimes have a modifying influence on dream content, and heightened emotional or motivational states can change the general vividness or emotionality of dreams (e.g., Berger, 1963; Witkin and Lewis, 1967; Witkin, 1969; Bokert, 1967; Hoelscher, et al., 1981). However, beyond a general demonstration of the psychological lawfulness of the dream process, experimental studies have not been able to tell us very much about the meaning of most dream content or about the relationship between dream content and either waking thought or behavior (cf. Antrobus, 1990: 4, and Cartwright, 1990:179).
Given the limited usefulness of retrospective analyses on the one hand and experimental studies on the other in the study of dream content, blind analyses using quantitative methods become the best approach to the scientific study of dream meaning. As we will see, this approach is especially compelling when dream series are used to make many specific predictions about a person's conceptions, concerns, and interests. In those cases where the dreamer is known to the researcher, predictions should be made in advance about the nature of the dream content and the dream reports should be coded by someone who does not know the predictions.
Not all the studies reported in this book are based on blind analyses, but in many studies the person doing the quantitative content analysis knew only the gender and/or nationality of the dreamer, or was unaware of the purpose of the study. Whatever the limitations of previous studies, however, the future clearly lies with blind analyses of dream content if the study of dream meaning is to be taken seriously by social scientists.
Dreaming and Daydreaming
Many different questions can be asked about dreaming and dreams, and there is a large literature on each of them. Since the discovery in 1953 of two different types of sleep--active, fast-wave Rapid-Eye-Movement (REM) sleep and slow-wave, quiescent Non-REM (NREM) sleep--there have been thousands of psychophysiological studies of sleep and dreams by researchers all over the world (e.g., Aserinsky and Kleitman, 1953; Dement, 1955; Dement and Kleitman, 1957; Hartmann, 1967; see Ellman and Antrobus, 1991, for one major summary). In psychological terms, these researchers are searching for physiological correlates of dreaming. Some of them believe dreaming occurs almost exclusively in REM sleep, and that we therefore know something about the neurophysiology of dreaming (e.g., Hobson, 1988; Hobson and McCarley, 1977; McCarley, 1989), but others dispute this conclusion, claiming that the presence of fully developed dreams shortly after sleep onset and during NREM sleep suggests we still know little about the psychophysiology of the dream process (e.g., Foulkes, 1962, 1985, 1993a; Vogel, 1978; Cavallero, et al., 1992; Cicogna, 1994).
There is also an important literature on dreaming as a cognitive process. It builds on the burgeoning research in the area of waking cognition to show that much of what was previously inexplicable about dreaming can be understood in terms of recent findings and concepts developed in the study of waking consciousness and memory. David Foulkes has been the major contributor to this effort; his Dreaming: A Cognitive-Psychological Analysis (1985) synthesizes what is known about dreaming from laboratory studies with findings on waking cognition and presents many new hypotheses for future investigation. The work of John Antrobus (1977, 1990), Harry Hunt (1986, 1989), and Donald Kuiken (1986) also is important in this area. Still others have done ingenious experimental studies in the sleep laboratory suggesting possible psychological functions for dreams. The major contributor to this area of study has been Harry Fiss (1983, 1986, 1991).
As fascinating and informative as this work is, it will not be discussed in this book because it does not provide detailed answers to questions about dream content and its correspondence to waking thought and behavior. The biology of dreaming does not tell us the psychological meaning of dreams (cf. Fiss, 1979, 1991). The dreaming brain and the dreaming mind are two different issues, one neurophysiological, one psychological. Similarly, the study of dreaming as a cognitive process has made a great contribution by showing that dreams are psychologically meaningful in a general sense, but few of these studies help us to understand the meaning of specific dream reports or the relationship between dream content and waking thought or behavior. Then too, dreams could have no "functions" at all in terms of either evolutionary survival or individual adaptation, but still be psychologically meaningful in terms of internal coherence and correlations with waking thought and behavior (e.g. Antrobus, 1993; Foulkes, 1993a).
There is also an interesting literature on daydreaming, reveries, and extraneous thoughts, based in a variety of techniques, including thought-sampling by means of pagers carried by people going about their everyday lives. Just as our studies show that dream content has continuities with waking life, these studies suggest the dreamlike nature of some waking thought (e.g., Singer, 1966, 1975, 1988; Klinger, 1971, 1990, Starker, 1978; Foulkes, 1994). This literature will figure importantly in our theoretical comments in the final chapter. Relevant findings also will be referred to at a few places in the intervening chapters. To underscore the relationship between dream content and relaxed waking thought, we have adopted Klinger's (1971) phrase "current concerns" because it characterizes the major content of both forms of cognition. Our former phrase, "emotional preoccupations," will be used interchangeably. "Unfinished business" and "unfinished intentions" also express part of this concept, but "current concerns" and "emotional preoccupations" allow for the positive interests sometimes appearing in dreams.
For the most part, though, this book will be focused on the quantitative study of dream content and its relationship to such factors as gender, age, nationality, and individual differences. Its findings stand on their own, whatever theory turns out to be right concerning the neurophysiological correlates of dreaming, or whatever cognitive theory eventually explains the production of the dreams reported to investigators. Similarly, there are enough differences between dreams and daydreams to repay a primary concern on dreams as a unique experience in our lives.
The Meaning of Content Analysis
In the most general sense, "content analysis" is the search for meaningful regularities and patterns in written documents. In principle it can be done "qualitatively," as when we use our intuition or our general understanding of language, or it can be done "quantitatively." Historically, however, content analysis as the phrase is used by social scientists has meant the attempt to convert verbal, written, or other symbolic materials into numbers so statistical analysis can be performed. This purpose is accomplished by formulating categories, tabulating frequencies for those categories, and determining percentages, proportions, or ratios. Comparisons are then made with "norms" or control groups.
In practice, there is not a hard and fast line between the qualitative and the quantitative. One can shade into the other. We often begin with an implicit set of categories and develop a rough idea of the frequency of elements fitting into those categories. Next we create more explicit, carefully defined categories, and then we make a more detailed search of the document for exact frequencies. This is in fact the process used by Hall (1951) in his early work.
Thus, quantitative content analysis often develops out of impressionistic qualitative analyses as an attempt to minimize personal bias and make possible greater agreement among investigators who are studying the same type of documents, whether newspapers, plays, folktales, political party platforms, or dream reports. One early content analyst said content analysis consists of "methods in which the bias of the analyst is at least minimized, in which the essential operations can be made explicit and the conclusions thereby more easily replicated, and in which the findings can be communicated in meaningful numbers" (Osgood, 1959:34). Cartwright (1953:466) concluded that the "fundamental objective" of content analysis is to convert the "symbolic behavior" of people into "scientific data," by which he meant (1) objectivity and reproducibility, (2) susceptibility to measurement and quantification, (3) significance for either pure or applied theory, and (4) generalizability. To the degree that the study of dream content can meet the standards set forth by Osgood and Cartwright for content analysis in general, it can be called a "scientific" study of meaning in dream reports.
There is nothing arcane, abstract, or theoretically difficult about the idea of content analysis. Nor are its methodological issues hard to grasp. However, doing content analysis is not an easy task. It is difficult to formulate categories having both reliability and validity. It takes time to learn to use a coding system. Great care must be taken to insure that the material under investigation is categorized, tabulated, and analyzed accurately. Indeed, some of the disagreements in the literature on dream content are due to the fact that some of the studies were not very well done.
No one content coding system works for all kinds of verbal or written materials. Consequently, there has been little or no use of content categories developed for studies of other written material in the study of dream reports. Instead, several different systems have been developed for analyzing one or another aspect of dream content. In the next chapter we will explain the development of one of these systems, that of Hall and Van de Castle, and present an overview of its coding rules. This system then will be compared with other coding systems to suggest that it is more comprehensive and more objective than most other systems. In short, we believe it is a sound basis for a scientific study of dream meaning.
Continue on to Chapter 2
Go back to the Dream Library index.