Gender differences in the analysis of 240 dream reports from Brazilian participants in Seminars

Abstract This study asked the questions, “Are there significant content differences between male and female dream reports obtained in dream seminars conducted in Brazil?” Each of the 240 (137 female, 103 male) research participants volunteered recent dream reports (one per person) during dream seminars that he or she attended between 1990 and 1998. Dreams were scored according to Hall-Van de Castle criteria. Comparative Cohen h - statistics revealed several gender differences. Further study is recommended because the dream reports did not represent Brazil’s social-economic diversity, and may not have been characteristic of the totality of participants’ dream lives.

Introduction Most investigators realize that the dreams with which they work are simply reports; the actual dream as directly experienced can not be studied. The shaman, the psychotherapist, and the dream researcher all deal with verbal or, in some cases, pictorial dream - reports. These reports may be incomplete, poorly remembered, or completely fabricated; they may change or undergo revision depending on the social or temporal context. Not only does a dream report represent a dialogue between one’s waking and sleeping mentation, it reflects a discourse between the dreamer and the listener. Dreamers may provide one version of the recalled dream to family members, a second version to their friends, and a third version to their psychotherapist. What is disclosed about the dream may vary considerably, depending on how the dreamer forgets, embroiders, or reconstructs different portions of the report. In postmodern terms, dream reports are fluid texts rather than fixed texts; rather than remaining static, their meaning and even their words vary as they are told and retold at different times in different settings. Hence, dreamwork of any type needs to be done with care, with attentiveness, and with modesty. The association between dream reports and the dreamer’s everyday activities and concerns has been demonstrated both for individuals (e.g., Winget, Kramer, & Whitman, 1972) and cultures (e.g., D’Andrade, 1961; Prasad, 1982). Dream reports have been used to study cross-cultural differences which have often yielded striking results (e.g., Heynick, 1993). The manifest content of reported dreams also has been used to study groups within cultures. Gender differences, for example, constantly have emerged in the literature (e.g., Soper, Rosenthal, & Milford,1994). Therefore, the question asked in this study was, “Are there significant content differences between male and female dream reports obtained in dream seminars conducted in Brazil?”. Only one previous study on this topic has been reported. Luciano Ribeiro Pinto, Jr. (in Ludwig & Cristiane, 1999, pp. 66-67) of the Institute of Sleep, Federal University of Sao Paulo, queried 70 Brazilian men and women about their dreams. Overall, the most frequent content items reported were friends, family, travelling, and sex, in that order. Women recalled their dreams more frequently than men; men reported less auditory dream content as well as fewer vivid and repetitive dreams; women reported more dream content concerning family members, friends, work, and emotions.

Methodology Content analysis is a research procedure, not a research method. However, it does employ an explicit, organized plan for assembling data, quantifying them to measure the concepts under study, examining their patterns and interrelationships, and interpreting quantitative comparison of verbal reports elicited by research participants (Riley & Stoll, 1968). In this case, Brazilian men and Brazilian women were compared. The research method utilized to obtain data on gender differences was J. Cohen’s (1977) h-statistic.

Research Participants The 137 female and 103 male research participants for this study were members of dreams seminars that one of the authors (SK) conducted in Brazil between 1990 and 1998. These events were held in various Brazilian cities, specifically Belo Horizonte, Brasilia, Curitiba, Fortaleza, Porto Alegre, Recife, Rio de Janeiro, Salvador, and Sao Paulo. The age range spanned people from their 20s to their 70s, with a few outliers on each end of the spectrum. Using Stephen’s (1997) classification terms, the “upper” and “middle” classes were over-represented as there were entrance fees for most of the seminars; however, a few scholarships were available for other individuals. Many seminars were held at colleges and universities; as a result, the educational level of the participants was higher than would have been found in the general population. A variety of ethnic groups were represented in the sample.

Research Instrument Content analysis is a research procedure developed to systematically and objectively quantify textual characteristics and themes. It also identifies the relative extent to which these schema pervade a given communication, document, or other text. In 1966, The Content Analysis of Dreams (Hall & Van de Castle, 1966) was published; this book outlines a thorough coding system with which the authors investigated 1,000 dreams collected from 200 undergraduates from Case Western Reserve University between 1947 and 1952. Since that time, the Hall-Van de Castle system has been used to compare groups of varying gender, age, cultural background, and diagnostic category (e.g., Kane, Mellen, Patten, & Samano, 1993; Krippner, Lenz, Barksdale, & Davidson, 1994; Krippner, Posner, Pomerance, Barksdale, & Fischer, 1974; Lortie-Lussier, Schwab, & De Konick, 1985). The Hall-Van de Castle (1966) system is one of several which has been developed over the years in an attempt to reflect the relation between a dream report and the dreamer’s environment, or, in the case as such scales as “ego strength” (pp. 208-210), a theoretical conjecture about the dreamer’s inner world. Reliability and validity concerns have been addressed by Van de Castle (1969) and the system’s utility has been evaluated by Winget and Kramer (1979). Hobson (1988), however, makes a distinction between dream-content and dream-form, holding that the latter (e.g., discontinuities, incongruities, uncertainties) are the most interesting features of a dream because this dream “architecture” more clearly differentiates it from waking consciousness (pp. 233-234).

Research Procedure At the beginning of each seminar, participants were invited to write down “a recent dream,” providing no information on this report other than their gender, and to place it on a centrally located desk. Several of these dreams were used to illustrate concepts explored in the seminar, and a few dreams were explored in detail, with the donor’s participation. Following a procedure approved by the Saybrook Institutional Review Board (for the protection of human subjects), participants were told that the entire collection of dream reports would be used for research purposes unless anyone objected. Those who did, were requested to withdraw their dream report at the end of the seminar. Following a procedure often used in the content analysis of dreams (Domhoff, 1996), all dream reports below 50 words and above 250 words were eliminated from consideration. Four judges, working blind and independently, coded the dream reports following the rules outlined by Hall and Van de Castle (1966) as well as those more recently contributed by Domhoff (1996), whose book contains the complete HallBvan de Castle system. The Hall-Van de Castle categories coded were total characters, total aggressive interactions, total friendly interactions, total sexual interactions, total settings, total emotions, total activities, total successful outcomes, total failure outcomes, total outcomes ending in misfortune, total outcomes ending in good fortune, total objects, total modifiers, total negative descriptions, total temporal descriptions and dramatic intensity score (the sum of the categories aggressive interactions, friendly interactions, sexual interactions, successful outcomes, failure outcomes, outcomes ending in misfortune and outcomes ending in good fortune). Many of these categories are subdivided in the Hall-Van de Castle system; those subcategories coded were male characters, female characters, strangers, family members and relatives, children, animals, dreamer as aggressor, dreamer as victim, dreamer witnessing but uninvolved in the aggression, aggression by a male, aggression by a familiar person, aggression by a stranger, dreamer as both aggressor and victim, indoor settings, outdoor settings, architecture, household objects, food and eating, tools and implements, travel objects and conveyances, streets and roadways, regions and land areas, nature, body parts, clothing, communication objects and money. An Aggression per Character (i.e., A/C Index) score was determined; this index indicates the number of characters involved in aggressive interactions. In much the same way, Friendliness per Character (i.e., F/C Index) and Sexuality per Character (i.e., S/C Index) scores were derived. Finally the number of dreams with aggressive, friendliness, sexuality, success, failure, misfortune, and good fortune content was tallied. (See Figure 1 for a listing of these categories, subcategories, and indices). Each of the 240 dream reports was coded for these 53 categories, subcategories, and indices. However in reviewing the data, it must be recalled that percentages of objects in each subcategory, characters in each subcategory, settings in each subcategory and aggression subcategories were calculated from the total number of objects, characters, settings and aggressive interactions. Hence when percentages are given, these are the percentages of the content item in their categories, not the percentages of the total dream reports. The exception, of course, is the final tally in which all dreams were assessed to determine aggression, friendly, sexual, success, failure, misfortune, and good fortune content.

Data Analysis For 30 of the 53 content categories, subcategories, and indices coded, the mean frequencies were computed between genders and analyzed statistically. No correlations could be computed for the other categories (e.g., emotions) because there are no proportions involved in the total findings; the data simply represent numerical differences between one series and another. For the data for which statistical tests could be performed, J. Cohen’s (1977) h- statistic was used for all percentage differences finding p by using a weighted N and doing a z -score transformation. Cohen’s h-statistic shows effect size and prevents one from regarding some statistically significant differences as containing important meaning. A reliability check was made between the four coders, yielding intercoder reliability scores by the method of common agreement with a range between 90% and 100%. Table 1 presents the intercoder reliability report for two of these judges.

Results For 30 comparisons, one or two would be significant at.05 by chance alone (i.e., “false positives”). Instead, 7 comparisons obtained significance (see Table 2). In comparison with Brazilian females, dream reports from Brazilian males contained more references to household objects (p<.000), and fewer references to food (p=.008) and body parts (p=.001). Strangers appeared more frequently in male (37%) than in female (26%) dream reports (p<.008), while children appeared more frequently in female (6%) than in male (2%) dream reports (p<.028). The A/C Index was about the same for males (12%) and for females (14%); this index measures how many dream characters engaged in aggression. The F/C Index was higher for females; more characters (23%) in female dream reports engaged in friendly interactions than characters (14%) in male reports. However, is it improper to apply Cohen’s h-statistic to these ratios, so statistical comparisons could not be made. More female dream reports contained friendliness (p=.035) and had successful outcomes (p=.004). Female dream reports contained the same amount of sexual content as did male reports according to both the S/C Index which reflects the number of characters who engage in sexual interactions (6% vs. 6%) and a tabulation of the dreams themselves (12% vs. 11%). The areas in which no differences were reported are also of interest. Both genders displayed about the same proportions of male characters (56% for female dream reports, 60% for male reports) and female characters (44% for female reports, 40% for male reports). There were no differences regarding aggressive interactions, sexual interactions, or the mention of tools and implements, or of travel objects and conveyances. The average number of words per dream was 88 for females and 84 for males; therefore, gender differences could not be attributed to disparate word frequencies. Table 2 presents the content categories, subcategories, and indices, noting for which ones statistical analysis was possible. An earlier analysis of Brazilian dream reports used a subset of these data (Krippner, Winkler, Rochlen, & Yashar, 1998). After comparing 60 female and 66 male reports, it was observed that witnessed aggression was more frequent in female dream reports (p=.024) and that there were more communication objects in female reports (p=.004), findings not repeated when the number of dream reports was increased. However, the other major gender differences were repeated in this study. Perhaps the two other differences were “false positives,” or perhaps the larger sample presented a more accurate picture of Brazilian dream content.

Discussion Bateson (1972) has written a thoughtful perspective on the issue of “national character,” and several studies of dream reports in non - U.S. societies have been conducted in an attempt to explore this topic. Monroe, Monroe, Brasher, Severin, Schweickart, and Moore (1985) studied gender differences in dream content of 325 secondary school children who were members of the East African Gusli, Kipsigic, and Logoli tribes, observing that women’s dream reports included as much physical aggression and more verbal aggression than male dream reports. In other aspects of aggression, however, the data were similar to those reported by Hall and Van de Castle. Gregor (1981) studied Mehinaku dream reports, noticing a much higher amount of aggression, especially initiated by animals, for both genders than had Hall and Van de Castle. However, the gender differences emerging from Gregor’s study were virtually identical to those found in the U.S. sample. These latter data, and data from studies in some 30 different social groups, support Hall’s (1984) suggestion that there was an “ubiquitous sex difference in dreams”, i.e., the percentage of male characters is higher in male dreams than in female dreams. This finding was not confirmed in this study, nor was Hall’s earlier report that women dreamed about indoor settings and family members more frequently than men, and less frequently about outdoor settings, tools and implements, and successful outcomes. However, some of our other Brazilian data resemble Hall’s earlier work: Women’s dream reports contained more references to children. Hall’s “ubiquitous sex difference” was not found in six other groups outside the United States (Hall, 1984; Van de Castle, 1994, p. 320); therefore, our results are not singular. Of the Brazilian female dream reports, 38% contained friendly interactions as compared with 25% of the male dream reports, resembling the U.S. data. Sexual content was about the same for both Brazilian genders, as opposed to U.S. data where men report sexual interactions far more often than do women (Domhoff, 1996, p. 327). Some groups may be reluctant to share sexual dreams, even if their personal identities are withheld. However Hall’s (1953) and D. Cohen’s (1973) continuity view of dreams finds support in these data. Stephen (1997) has documented the growth of the women’s movement in Brazil, and the growing availability of birth control devices and family planning procedures, including sterilization. Espinoza (1996), in reporting on sexual behavior in Rio de Janeiro, asserts that the average interviewee claimed to have sex 2 or 3 times a week, with 17% having sex every day. The amount of time spent per sexual encounter was 45 minutes for Brazilians in contrast to 8 for the Italians, 6 for the French and U.S. residents, and 3 for the British. In contrast to an estimated 27% of women worldwide who have orgasm during virtually every sex act, 55% of Brazilian women make such a claim (Espinoza, 1996). The earlier study by Luciano Ribeiro Pinto, Jr., did not utilize a standard content analysis technique, but reported that women’s dream reports contained more emotional content. In our sample, the dramatic intensity of female dreams was stronger than in male dreams, reflecting Ribeiro Pinto’s notation that female dreams are more vivid. A number of Ribeiro Pinto’s dreamers noted that the dream they volunteered was repetitive; almost all of these dreamers were women. Travel was a common theme in these dream reports, as well as in ours; 6% of female dream reports and 7% of male reports contained references to travel objects and conveyances. When Ribeiro Pinto checked the educational level of his dreamers, he found that those who had completed more years of education contributed more dream reports containing work settings and travel. Those with less education contributed more dream reports containing references to death and aggression. In other words, there seem to be class differences in Brazilian dream reports, and Ribeiro Pinto’s data helps inform the questions raised by our data. Stephen (1997) has described four classes of Brazilian women: upper class, middle class, peasants and rural workers, working class; future research would be advised to examine dream report content differences among these classes. These interpretive comments reflect the perspectives of Mary Calkins (1893), Alfred Adler (1938) and Calvin Hall (1953), all of whom pointed out the congruence of dreaming life and waking life. Some dreams may rehash past traumas or fantasized desires (e.g., Freud, 1933/1965), others may rehearse future activities (e.g., Jung, 1956), and others may be a “knitting together” of images that occur during brain activation during sleep (McCarley & Hobson, 1979, pp. 124-125). However, there is a body of research that supports the idea that there is a basic continuity between dream content and the waking emotional concerns and cognitive style of the dreamer (e.g., Cartwright, 1986). Hendricks and Cartwright (1978) reported high correlations between subjects’ cognitive style during waking activity and cognitive style in dream reports. Foulkes (1981) found that the test-assessed cognitive development of children is mirrored in their dreams. Domino (1976) and Urbina (1972) attained significant correlations between dream analyses and data from both projective techniques (e.g., Rorschach) and standardized personality measures (e.g., MMPI). Winget and Kramer (1979), in summarizing research in this area, concluded that “the content of dreams has most often, but not always, been found to be continuous with, rather than compensatory to, waking life” (p. 23). This concept of similarities and continuities between dream life and waking life is supported by cross-cultural research. Van de Castle (1971) studied the Cuna Indians in Panama, noting that their dreams included very few acts of aggression against other people - a trait observable in their daily lives. However, Punamaki and Joustie (1998) found that Palestinian children living in violent and dangerous environments recorded dream reports (in a 7-day diary) the contents of which “incorporated persecution and aggressive themes” in comparison with Palestinian and Finnish children living in peaceful areas. In all three groups, girls’ unpleasant dreams typically incorporated negative feelings, whereas boys’ dreams involved horror scenes, ventures, and actions. Gender differences were greater in both national groups from peaceful areas. Monroe, Nerlove, and Daniels (1969) reported that East Africans living in areas of high-density population where food is often scarce, have an unusually low frequency of food consumption in their dreams. Kane, Mellen, Patten, and Samano (1993) used the Hall-Van de Castle scales to compare dream content of Mexican, Mexican American, and Anglo American college women, finding that Mexican American college women were the median group insofar as similarities were concerned. For example, Mexicans and Mexican Americans did not differ regarding dream content but Anglos reported significantly fewer emotions. Levine (1966) investigated three groups of male Nigerian students, finding that dream content differed in relation to their tribal values. For example, the Ibo culture has a value system and social structure favoring upward mobility of its members. The Hausa culture does not support social mobility and individual achievement, while the Yoruba culture takes an intermediate position. Dream reports from Yoruba students contained more achievement themes than those of Hausa students, but fewer than those of Ibo students, which is what one would predict if dream life reflects waking life. Globus (1987) proposes that dream life and waking life share more similarities than differences, and that both are “thought” into existence in a manner not unlike the way in which the Upanishads described how Vishnu “dreamed” human beings and their world into existence. In the case of waking life, environmental information passes freely across a person’s sensory receptors. As these receptors match the “tunings” of neural filters, they constitute that person’s life world. In dreaming life, information from the preceding days, and from earlier life experiences, become reoperative. But the dreamer creates a specific life world out of many possibilities; “dream life is our own formative creation” (p. 173). Again, Globus echoes Hindu scripture’s description of dreaming sleep as an opportunity for human beings to create as the gods create, by emitting images. However, Hindu philosophy used a divine artisan as its model, while Globus’ mechanism is “a possible world machine” that creates by selection from a plenum of enfolded possibilities that includes genetic predispositions, life experiences, and the random stimulation of brain centers during sleep (p. 174). In this manner, dreamers and their culture operate in tandem, “making each other up” (Shweder, 1993). To understand a person’s dreams, one must also understand his or her cultural milieu (Hall, 1991). To thoroughly describe and/or understand a culture, the investigation of dreams is a necessity. Dreams have social roots, and society not only reflects its members’ dreams but might be influenced and often changed by them. Unfortunately, psychology has too often ignored culture as a source of influence on human behavior (Segall, Lonner, & Berry, 1998, p. 1101).

Limitations A number of limitations caution against engaging in overinterpretation of these data. A large number of comparisons were made, and some of the statistically significant results may have been artifacts, especially those hovering near the.05 significance level. Our sample was not representative of the general population of Brazilian males and females, under-representing the less affluent social-economic classes. A plethora of literature demonstrates the importance of class differences in Brazil (e.g., Stephen, 1997; Surratt & Inciardi, 1998; Wood & Magno de Carvalho, 1988). Furthermore, the dreams submitted may have been selected because they were particularly memorable or provocative; mundane dreams, often more reflective of one’s daily attitudes and activities, may have been excluded. Hall and Van de Castle collected five dreams per person, tapping into each dreamer’s nighttime mentation more broadly than our procedure of asking for a single dream. The time of year that dreams are collected could influence content, reflecting weather conditions (e.g., the indoor settings subcategory), holidays (e.g., the good fortune subcategory), or publicized behavior of celebrities (e.g., the sexual interactions subcategory). However, the dreams in this study were obtained at many different times of the year and never during holidays as no dream seminars were held at that time. Nor was our research instrument without its limitations. In addition to the criticism already cited that more can be learned from the form of a dream report than from its content (Hobson, 1988), it is apparent that content analysis permits only a partial assessment of dream reports. Content analysis does not deal with the dream report as a whole, with the life context of the dreamer, or with factors in waking life that might underlie particular items. One dreamer may be the victim of an aggressive act in a dream because he lives in a dangerous neighborhood, another because she is being verbally attacked in the workplace, another because he engages in paranoid fantasies, and another because she saw a violent film before retiring for the night. Other research strategies could use the same collection of dream reports and discover important levels of meaning and application only hinted at in this study. When such scales as that developed by Hall and Van de Castle are used in other cultural settings, additional problems emerge. The selection mechanisms that to into the act of volunteering a dream may be quite different from country to country. Translations, even when carried out by native speakers, may lead to distortions when subjected to an analysis originally designed for English language dream reports. There are several Hall-Van de Castle measures (e.g., the psychodynamic scales) we did not utilize; there are other content analysis systems that have been developed by other researchers. Any and all of these may have produced results that would have been of interest to dream researchers. Finally, this project suggests the need to investigate such topics as how different cultures actually interpret their dreams, as well as the accompanying differences in attitudes about dreams.

Conclusion Phenomenologically, dreams are a series of images that are experienced during sleep, and reported in narrative form during wakefulness. The dream report can be conceptualized as a text, hence its content is influenced by the linguistic style of the subject. Differences in dream content among individuals or groups may reflect their differences in verbal behavior more than any other measure (Winget & Kramer, 1979, p. 14). In one of the few studies relevant to this issue, people who were asked to make up a dream while awake produced accounts that judges could not discriminate from written reports of their nighttime dreams (Cavallero & Natale, 1988-1989). There are dangers in accepting language as an accurate representation of experience. Instead, it exists in relation to its world; the ensuing back-and-forth communication makes it difficult to compare dream reports even from a single culture or group, much less between groups. Culture can be conceptualized as a shared reality or way of life around which people have developed values, norms, life-styles, and social roles (Kane, Mellen, Patten, & Samano, 1993); imperfect though it may be, the role of language is paramount in understanding the interaction between individuals and their cultural setting. The use of dream reports may become an important research tool in the emerging field of cultural psychology, the discipline that studies interactions between individuals and their cultural environment. It is the premise of cultural psychology that there is no population, least of all urban, Euro-American males, whose activities, practices, and ideals can be presumed to be a universal normative base line for human development and mental health (Shweder, 1991). Dream reports can provide a glimpse into the variety of human worldviews and experiences. Dream reports could be discounted as providing dependable data for the study of culture/person interaction as long as they were considered meaningless, on the one hand, or contradictory to daily experience on the other. In contradistinction to Freud, it was Adler (1938) who stressed the congruent relationship of dreams to the lifestyles of their dreamers. For Adler, the dream is not significantly different from waking thoughts; like all cognitive and emotional activity, the dream becomes part of the process of rehearsal for future behavior and achievements. Bonime (1960) also insisted that dreams are less disguised than uncensored and, therefore, authentic self-presentations that express the individual’s shifting motivations, attitudes, and actions. However, we agree with Tedlock (1987) that waking consciousness itself is not unitary but is constantly shifting between foreground and background, between the internal world and the external world, between arousal and dissociation. This paradox has not kept psychology from attempting to study waking experience. Why should the paradox of dream life present obstacles to disciplined inquiry into dream reports?