The lack of a reliable and feasible method for collecting good samples of dream reports is one of the major obstacles to the systematic study of dream content with both adults and children. For example, only a small percentage of teenagers asked to keep dream journals for one or two weeks on a voluntary, nonpaid basis are likely to record five or more dreams, with boys providing even fewer dreams than girls (e.g., Buckley, 1970; Howard, 1978), and there are inevitable questions about whether those who record their dreams differ from those who do not. Then, too, participants may report recurrent dreams or memorable childhood dreams that are not likely to be typical of dream life. Laboratory studies can provide representative samples of dream content (Foulkes, 1982, 1993), but they are time-consuming and expensive.
The Most Recent Dream Method was developed for use with adult populations to overcome these difficulties (Hartmann, Elkin, & Garg, 1991). It simply asks participants to write down the last dream they can remember having, "whether it was last night, last week, or last month" (Domhoff, 1996, p. 310). The Most Recent Dream instructions also prime participants to focus on the last dream they recall by asking them to write down the date and time when they recalled the dream. Asking for the date of the dream also makes it possible to exclude dreams from months or years in the past if the researcher so desires.
The overwhelming majority of adults participants are able to provide a report that takes them 15 to 20 minutes to record. Over 90% of the reports state that the dream happened within the past six months. Thus, this method makes it possible to collect dozens or hundreds of dream reports in a short time when large groups of people are congregated in one place (e.g., classrooms, convention halls, and waiting rooms).
The potential utility of this method was first demonstrated by drawing many subsamples of 25, 50, 75, 100, and 250 dreams from Hall and Van de Castle's (1966) normative sample of 500 dreams provided by 100 college men between the ages of 18 and 22. Samples of 100 to 125 single dreams from each subject came close to duplicating the norms, thereby establishing the sample sizes that are minimally necessary for replicable results. A study of 100 Most Recent Dreams written down by college women between the ages of 18 and 25 at the University of California, Santa Cruz, in the early 1990s showed that the findings did not differ from the Hall and Van de Castle (1966) female norms based on 500 dream reports provided by 100 college women between the ages of 18 and 22 (Domhoff, 1996, p. 67).
The purpose of this study was to see whether the Most Recent Dream Method can be useful with teenagers, it focuses on 12-13 year-olds for two main reasons. First, limited time and resources precluded collecting and coding 100-125 Most Recent Dreams from boys and girls at several age levels. Second, it was decided that the first study should concern children just entering their teenage years on the assumption that the method could be used with older teenagers if it proved useful with those aged 12-13.
Since the focus of the study is methodological, no attempt will be made to review findings from the small published literature on the dreams of adolescents. Aside from the two longitudinal studies by Foulkes (1982) and Strauch and Lederbogen (1999) that are employed later in this paper as benchmarks for evaluating the present findings, this literature is not relevant to the present study. It consists of two studies of dreams from disturbed adolescents (Amanat, 1974; Langs, 1967); a study correlating Hartmann's Boundary Questionnaire with answers to questions about dream recall frequency, nightmare frequency, and the waking impact of past nightmares (Cowen & Levin, 1995); a questionnaire study on themes in adolescent dreams (Potheraju & Soper, 1995); a study of most recent "bad" (anxiety) dreams from German children ages 10-16 (Schredl, Pallmer, & Montasser, 1996); a study of age and gender differences in "object representations" with a theoretical rating scale (Winegar & Levin, 1997); and a study of the "linguistic features" of adolescents' dream reports (Azzone, Freni, Maggiolini, & Provantini, 1998).
Personal contacts developed by the first author were used to enlist the cooperation of several teachers and the principal at a suburban middle school near Santa Cruz, California. The children were overwhelmingly Caucasian and from middle-class backgrounds. Detailed demographic information was not collected because of the emphasis on whether or not children of this age would be able to respond with dream content similar to that found in other studies. The nature of the project was fully explained to participants. The objective nature of the analysis and the anonymity of the participants were emphasized. The teachers were promised a visit to each classroom by the third author after the data were collected so that he could answer students' general questions about dreams as well as explain the objectives of the project. The importance of such a follow-up visit in an attempt to give something back to the teachers and their students in return for their help cannot be overemphasized to those who might want to do similar studies.
The first author was introduced in each classroom by the teacher. She first said that she would like the participants to write down the last dream they could remember having, and that she would explain why after they were finished. She next passed out the Most Recent Dream form, read the instructions on the form to the students, and answered any questions that they asked. There were relatively few questions because of the promise to explain study objectives afterwards. Most answers to questions involved reassurances that it was indeed the "most recent dream" ("like this morning," "like if it was when I slept in last Saturday?") that was being requested. The instructions that were read, which also appeared on the form, are as follows:
We would like you to write down the last dream you remember having, whether it was last night, last week, or last month. But first please tell us the date this dream occurred. Then tell us what time of day you think you recalled it.
Usable Most Recent Dream forms were collected from 110 boys and 162 girls in 16 classrooms. They were coded for several main categories within the Hall and Van de Castle system by the first and second authors. By the "method of perfect agreement" (Domhoff, 1996, p. 28), in which the number of agreedupon codings made by two coders is divided by the sum of A their codings, the reliabilities were above 0.80. The few differences in coding were resolved through discussion so that there would be one set of codings. Codings were then entered into DreamSAT, a spreadsheet available to researchers through a Web site on quantitative dream research (Schneider & Domhoff, 1999). The spreadsheet makes instantaneous computations of several percentages and ratios explained in Domhoff (1999, Table 1).
Percentages and ratios were analyzed to overcome two major problems that impair many published studies of dream content. First, percentages and ratios provide a good and understandable way to correct for differences in dream length. Second, percentages and ratios are useful with nominal coding categories, which are employed in the Hall and Van de Castle system to avoid the untenable psychological assumptions built into some ordinal scales for analyzing dream content (Domhoff, 1996, 1999; Hall, 1969a, 1969b; Van de Castle, 1969).
Once the analyses are completed, DreamSAT compares the results with the appropriate Hall and Van de Castle (1966) norms, and displays the results in both a table and a bar graph, along with significance levels and effect sizes for each of the analyses. The tests of significance and the effect sizes utilize Cohen's (1977, p. 180) h statistic, which corrects for the fact that standard deviations cannot be determined in a distribution of percentages.
Eighty-three percent of the girls and 60% of the boys reported a Most Recent Dream. The rest of the children turned in a blank sheet or wrote that they could not recall a dream. Judging by when they stopped writing and began to read or look around, writing the dream reports took 20-25 minutes on average. The range was from 5 minutes or less for very brief
reports to 35-40 minutes. Those who took longer often thought for several minutes before beginning to write or else wrote very lengthy reports (over 300 words). The median length was 125 words for the girls, with a range of 5 to 463, and 89 words for the boys, with a range from 11 to 360. For the girls, 92.5% of the reports were 50 words or more, whereas 74.3% were of this length or greater for the boys.
The results of the content analysis were consistent with those of a smaller, unpublished pilot study carried out by the first author and a co-worker using data collected at two middle schools different from the one used in this study (McNicholas & Avila-White, 1995). The main results of that study, based on 64 male and 80 female 12-13 year-olds, are reported in Domhoff (1996, p. 95). Because the sample size for girls approached the minimum necessary (100), the findings for girls in the pilot study and the present study are presented in Table 1. The table shows that the major differences concern the amount of friendliness and aggression. This replication is important evidence for the reliability of the Most Recent Dream method.
The gender similarities and differences found in the present study are consistent with those in the Hall and Van de Castle (1966) norms for young adults. This age comparison is presented in Table 2. The most general conclusion is that boys differ from girls in the same way that men differ from women on almost every indicator. For example, women have a male/female percent of 48/52, and men have a male/female percent of 67/33, a difference that holds for all traditional societies where anthropologists have collected dreams and for most current nations (Hall, 1984).
The largest differences between 12-13 year-olds and young adults were on the aggressions per character ratio (A/C index) and physical aggression percent. For both variables, the 12-13 year-olds were much higher than the young adults. Nonetheless, the same gender differences were present, with the boys deviating even further from the men than the girls did from the women. Girls and boys also differed from their young adult counterparts in the settings of their dreams. The 12-13 year-olds were more likely to be outdoors and in unfamiliar settings. On these indicators, the girls deviated further from the women than the boys did from the men.
There are two categories of friendliness in dreams where the deviations of the boys and girls from the adult norms did not go in the same direction. On the friendliness per character ratio (F/C index), the girls were higher than the women and the boys were lower than the men. This greater amount of friendliness in girls' dreams also appeared in the aggression/friendliness percent: the girls' percentage was lower than it was for women because their dreams had more friendliness; the boys' aggression/friendliness percent was higher than that for men because there was less friendliness and more aggression in their dreams. A somewhat similar difference showed up on the befriender percent. The normative figure for the befriender percent is 47 for women and 50 for men. The girls were very close to the women's norms, but the boys were far below the men's norms.
These present results indicate that it is feasible to collect Most Recent Dreams from young middle-class teenagers within the time frame of a standard classroom period. The fact that more girls than boys provided reports is consistent with findings on greater recall by girls in the laboratory (Strauch, 1996) and with the larger number of dreams reported by girls in dream diaries (Howard, 1978; Strauch & Lederbogen, 1999; Winegar & Levin, 1997).
The median word lengths of 125 for the girls and 89 for the boys compare well with the mean of 100 for girls and 94.9 for boys at ages 11-13 in Strauch and Lederbogen's (1999) study using dream diaries. These findings are also roughly comparable to the results reported by Foulkes (1982, p. 334) when the participants in his study were between ages I I and 13; in his study, the median word length was 83 for the girls and 103 for the boys. The lengths of the Most Recent Dreams in the present study are also in keeping with those reported by Winegar and Levin (1997) for teenagers aged 15-18; in their study, the mean word length was 149 for girls, just slightly above the median of 125 for girls in the present study, and 119 for boys, which is 30 words above the median of 89 for boys in the present study. Moreover, it is likely that results from these two studies would be even more similar if median word lengths could be compared.
It is also noteworthy that the Most Recent Dreams of the girls in our study were very similar in word length to those in a Most Recent Dream study of women at the University of California, Santa Cruz (Domhoff, 1996). In that study, 7% of the women provided reports of 50 words or less, which is almost the same as the 7.5% for the girls in the present study. The university women actually provided fewer reports over 200 words (15%) than did the girls in the present study (28%). Although no Most Recent Dreams are available for men, it is likely that many of the boys' reports are shorter than those of men. This judgment is based on our study of the first 8 dreams in each of 41 male dream series containing from 8 to 37 dreams (Hall, 1963). In this comparison, only 10.4% of the men's reports were under 50 words, compared to 25.7% for the boys in this study.
Regarding dream content, the overall results of this study are similar to those for the same age group in a six-year longitudinal study of 24 Swiss children (12 girls, 12 boys) that used the Hall and Van De Castle coding system to analyze two-week dream journals kept at home (Strauch & Lederbogen, 1999). For example, the animal percent, male/female percent, and aggressor percent are virtually identical for both boys and girls in the two studies. The biggest difference is that there is less aggression, especially physical aggression, in the dreams of the Swiss boys, but that is also true of Swiss men compared to American men (Domhoff, 1996, pp. 101-102).
The overall findings in this study also have much in common with those reported by Foulkes (1982, chap. 7) when his participants were ages 11-13. In this phase of his study, 8 girls provided 88 dreams and 12 boys provided 106. However, caution is required when comparing the results of these two studies. First, the girls in Foulkes' study were about a year younger than the girls in the present study and typically in the sixth grade, whereas the girls in the present study were all in the seventh grade (Foulkes, 1982, p. 179). (The typical boy in his study, on the other hand, was in the seventh grade, as were the boys in the present study.) Second, and even more importantly, the coding system used in his study is not directly comparable to the Hall and Van de Castle system. It is therefore necessary to speak in more general terms, except in the case of the male/female percent, which was the subject of a separate study conducted by Hall (1984).
As part of his study of the male/female percent in a wide range of dream samples from around the world, Hall (1984) included an analysis of the dreams collected in Foulkes' study when the children were between the ages of 9 and 15. Finding no age differences, he reported the overall findings. For the boys, the male/female percent was 76/24, which is very similar to the 74/26 figure for the 12-13 year-olds in the present study, and for girls it was 43/57, very similar to the 47/53 finding for our study.
The two studies are not directly comparable on the frequency of animals as characters, but they agree that the role of animals declined from a high level in young children's dreams to very close to the adult level in young teenagers. Foulkes (1982, p. 335) reports that 17% of the girls' dreams and 9% of the boys' dreams have at least one animal character. In our study, girls have an animal percent of 9; boys have an animal percent of 12.
Foulkes (1982, pp. 192-193) reports that his participants showed more gender differences at ages 11-13 than when they were younger. Most of these differences concern character categories, such as peers and strangers, or sensory activities, such as seeing or hearing, that are not comparable with any results in the present study. However, there also were more "antisocial acts or other misfortunes" in the boys' dreams, which may be comparable to their higher A/C index in this study.
In general, then, it seems that the Most Recent Dreams collected from suburban Caucasian 12-13 yearolds within one class period are rather similar in content to dreams collected from young teenagers of about the same social background in two independently conducted longitudinal studies. The implication of this similarity is that the Most Recent Dream method may provide a reasonably representative sample of such teenagers' dreams in an efficient and economical manner. However, further studies are needed to determine whether socioeconomic, ethnic, or cognitive variations may limit the utility of the method.
If the findings of the present study prove replicable, then the Most Recent Dream Method opens up four new types of research possibilities. First, it is clearly feasible to do cross-sectional developmental studies of dream content with teenagers if the cooperation of a school system can be enlisted. An undergraduate or graduate student research team could collect hundreds or thousands of teenage dream reports in a single day. Such large samples could then be analyzed very quickly for the six indicators that simply require coding for presence or absence in each dream report (aggression, friendliness, good fortune, misfortune, success, and failure). However, the "at least one" method of analysis has one drawback that must be kept in mind. Unlike the other Hall and Van de Castle content indicators, which provide a correction for differing report lengths by using percentages and
ratios, this method of analysis does not contain such a correction. It should be used only if median dream length is similar from age group to age group.
Second, it becomes feasible to do same-day studies of the Most Recent Dreams of teenagers in different parts of the United States and in other countries. Thus, it would be possible to study regional and national similarities and differences in dream content, and to see if any major events of the previous week are ever incorporated into the dreams of teenagers.
Third, it would be possible to turn the group findings into norms for each age level. Such norms would make it feasible to think of the individual dream journals kept by some teenagers as useful "nonreactive" personal documents (Allport, 1942; Baldwin, 1942; Webb, Campbell, Schwartz, Sechrest, & Grove, 1981). Any deviations from the norms in the individual journals could be studied to see if they correspond to atypical concerns, interests, or waking behavior (Domhoff, 1996, chap. 8). However, it should be noted that studies of subsamples drawn from lengthy individual dream journals suggest that at least 75 to 100 dream reports are necessary to have a reasonably representative sample of a person's dream life (Domhoff, 1996, chap. 7; 1999).
Fourth, the Most Recent Dream method makes it easy to determine whether dream content relates to other data collected in developmental studies of teenagers. Do high achievers, for example, tend to experience more success or initiate more aggressive interactions in their dreams? Do popular students have a higher F/C index and a higher befriender percent? Data to answer such questions could be gathered with one extra sheet of paper in a battery of tests and 15 to 30 minutes more testing time, depending on the age of the teenager.
This study shows that many different analyses of dream content during the teenage years are feasible using the Most Recent Dream Method. 'Me similarities of the findings to those for young teenagers in two longitudinal studies suggest that this method generates samples that are at least reasonably representative of dream life. The present study also shows striking similarities and intriguing differences between 12-13 year-olds and their young adult counterparts. Cross-sectional studies of teenagers from the ages of 14 to 18 are needed to fill in the gaps between the young teenagers and the young adults, and to see if and how Most Recent Dreams might add to the understanding of cognitive or psychosocial development.
Go back to the Dream Library index.