JOURNAL

OF EXPERIMENTAL

CHILD

PSYCHOLOGY

52,

319-335

(1991)

Sex Differences in Elaborative Strategies: A Developmental Analysis HARRIET

SALATAS WATERS AND LINDA L. SCHREIBER State Univeristy

of New

York

at Stony

Brook

The present study examined sex differences in the use of elaboration in paired associate learning in adolescence and young adulthood. In Experiment 1,48 eighth grade and 48 tenth grade students were asked to recall 24 word pairs, half of which were high frequency pairs, and half low frequency pairs. After recall, students reported the type of strategy used for each word pair (reading the pairs carefully, rehearsing the words, using imagery or constructing a verbal connection). Females used elaborative strategies more often, were more likely to recall elaborated pairs, and recalled more word pairs than males at both ages. These effects were observed in both high and low frequency word pairs. There was also a main effect of frequency, with elaboration more common with high frequency word pairs. In Experiment 2, college students performed the same paired-associates learning task, but with the added instruction to describe their elaborations in a sentence. At this age, there was a sex by materials interaction, with sex differences in strategy use only present with low frequency word pairs. These findings indicate that sex differences diminish under more favorable task conditions that encourage strategy use (high frequency word pairs) as males and females become more proficient strategy users, but remain under less favorable circumstances. An examination of the types of elaborations generated by college students indicated although males and females produced similar types of elaborations to the word pairs, sex differences in the recallability of low frequency words appeared with less interactive and more idiosyncratic elaborations. o 1991 Academic press, IX.

One of the most consistent findings in the literature on learning and memory strategies is that task variables play an important role in initiating strategy use in childhood (Brown, Bransford, Ferrara, & Campione, 1983; Kail & Hagen, 1977; Ornstein, 1978). With age, strategy use generalizes to a wider range of task conditions, and initiation of strategy use depends less on task variables. In response to these findings, Waters and Andreassen (1983) presented a developmental framework that organizes the changing patterns of strategy use across age. They proposed that the Requests for reprints should be sent to Dr. Harriet Waters, State University of New York, Stony Brook, NY 11794.

Department of Psychology,

319 0022~0965191 All

$3.00

Copyright 0 1991 by Academic Press, Inc. rights of reproduction in any form reserved.

320

WATERS

AND

SCHREIBER

developmental pattern of increasing strategy use across varied contexts can be summarized by three principles. First, a particular memory strategy or skill will first appear under more favorable processing conditions (e.g., organization during recall with semantic processing task instructions, Waters & Waters, 1976; elaboration of word pairs when more time is given to study each pair, Pressley & Levin, 1977). Second, a particular memory strategy will first appear with materials that encourage the use of that strategy (e.g., organization during recall with blocked presentation of categories, Furth & Milgram, 1973; elaboration with picture stimuli vs verbal materials, Hughes & Walsh, 1971, Pressley & Levin, 1978). And the third and final principle, individuals become more active in initiating strategy use as they become older and more experienced. Thus, strategy use begins under both favorable processing conditions and materials and then generalizes to less favorable conditions with age. A similar analysis of strategy use has been presented by Rohwer (1973) in his discussion of effective prompt conditions for elaborative strategies and is, in part, reflected in concepts such as “production deficiency” (Flavell, 1971). Cox and Waters (1986) have proposed that this general developmental framework should be used to investigate and interpret differences in strategy use across subgroups of the population, in particular, differences between males and females. Their own results, and the results of an earlier study on prose recall (Waters, 1981), have supported the hypothesis that there is a developmental lag between males and females in strategy use, with females initiating strategy use under more favorable task conditions before males, and generalizing to less favorable conditions sooner. Thus, the earliest evidence of sex differences in the Cox and Waters data occurs with favorable processing conditions, as females begin to consistently use the appropriate strategy under these conditions. Males, on the other hand, do not show any consistent strategy use, under any of the task conditions, producing a sex by condition interaction. In the next phase, males begin to use the appropriate strategy, particularly under more favorable task conditions, as females continue to use the strategy and begin to generalize to the less favorable task condition. This pattern of sex differences produces two main effects, one of sex, the other of task condition, and no interaction. In the next phase, males continue to use the strategy, catching up to females under favorable task conditions, but continuing to show a lag under less favorable task conditions. Once again, there is a sex by task interaction. The final phase would be one in which both males and females are now equally proficient at strategy use, eliminating any sex differences. The present study was designed to determine whether sex differences in the rate at which males and females progress through phases of strategy use in prose recall (Waters, 1981) and free recall (Cox & Waters, 1986)

SEX DIFFERENCES

321

are also evident in the development of elaborative strategies in pairedassociate learning. Elaboration in paired-associate learning is particularly interesting because unlike organization of prose passages, or category clustering or subjective organization in free recall, it is a late developing strategy (junior high school and high school grades). If successful in finding sex differences, the present study would represent the third of a series, and the three together would represent a wide range of memory paradigms and ages in which sex differences in strategy use can be found. Possible explanations for the results would have to account for the pervasiveness of the sex differences. In particular, continued sex differences at this late stage of development would undermine explanations that focused on age-specific characteristics (e.g., sex differences are only present in early elementary school in the transition from perceptually based processing to more semantic, languagebased representation-a la Bruner). In addition, the presence of sex differences with elaboration would indicate that sex differences in strategy use go beyond organization (with word lists, with prose passages). Whereas organization usually involves identifying structural features already present in the material (subjective organization is the exception), elaboration relies more heavily on constructive processes. Sex differences in elaboration would shift the focus of explanation from the salience of structural features in the materials to the role of constructive processes and the knowledge base in strategy use. Bjorklund (1987, 1988) has recently made a compelling case for how changes in the knowledge base facilitate the use of deliberate memory strategies (organizational and elaborative strategies alike). Sex differences in elaboration would suggest that investigators turn to differences in semantic elaboration and encoding processes for an explanation of sex differences in strategy use across age. The strategy use framework for understanding sex differences also contrasts with more traditional conceptualizations of differences in terms of verbal ability (Maccoby & Jacklin, 1974). According to the Cox and Waters (1986) analysis, sex differences in memory for verbal materials will be both age and context specific. In no sense does this perspective imply a trait-like conceptualization of verbal ability in which females can be expected to outperform males each and every time. Instead, whether sex differences are found under a particular set of circumstances will depend on the developmental level of the individuals and how readily materials and procedures elicit appropriate strategy use. These constraints, along with researchers’ tendencies not to break down data by sex, are likely to account for the memory literature’s failure to identify sex differences in elaboration to date. For the present study, the task variable that was manipulated in order to encourage elaboration was type of materials. Two different kinds of word pairs were used, high frequency word pairs and low frequency word

322

WATERS

AND SCHREIBER

pairs. Previous research has shown that individuals spontaneously produce more elaborations to high frequency word pairs, which are then recalled better than low frequency word pairs (Waters, 1982). Thus, the high/low frequency difference is consistent with principle 2 of the Waters and Andreassen (1983) framework, that some materials are more likely to encourage strategy use than others. In Experiment 1, equal numbers of males and females are tested at two grades, eighth and tenth. Individuals receive a mixed paired associate list composed of high frequency and low frequency word pairs. They are asked not only to recall the word pairs, but to indicate which of several strategies they had used to study each pair, a verbal report technique which has been used successfully in previous research (Waters, 1982). It is predicted that there will be sex differences in the use of elaborative strategies, and that these differences will be consistent with the developmental patterns established in previous research. EXPERIMENT

1

Method Subjects and design. A total of 48 eighth graders and 48 tenth graders participated in the experiment. Equal numbers of males and females were used at each age. All eighth and tenth grade students were recruited from two suburban school districts. Students from different classrooms and schools were equally distributed across conditions. There was no disproportionate selection of males or females from one classroom or another, and academic level of subjects was comparable across conditions and age, i.e., students were recruited from average academic level classrooms. Each student was asked to study and recall a list of 24 paired associates. The pairs were presented one at a time every 8 s. Twelve of the 24 paired associates were high frequency concrete noun pairs. The other 12 were low frequency concrete noun pairs. After students completed the recall test, they were interviewed about strategies they had used during study. Each student was asked to indicate which of four possible strategies they had used to study each of the 24 word pairs [read carefully, rehearse pairs, combine the words in an image (visual elaboration), combine the words in a verbal phrase (verbal elaboration)]. Materiuls. Two lists of 24 paired associates were constructed for the experiment. Each list consisted of 12 high frequency (AA) concrete noun pairs (e.g., mouth-sky, doctor-machine, kitchen-mountain, circle-grass, cat-soldier, paint-arm, etc.) and 12 low frequency (less than 16 per million) concrete noun pairs (e.g., thimble-strap, acorn-bathtub, sardineumbrella, gravel-walnut, pajamas-dictionary, clover-shelf, etc.) obtained from the Thorndike-Lorge (1944) frequency count. Slides were made of the individual word pairs and high and low frequency pairs were randomly

SEX DIFFERENCES

323

arranged for presentation during the study phase of the paired-associate learning task, with the restriction that no more than two pairs of high frequency word pairs or low frequency word pairs occurred in sequence. List assignment to students was counterbalanced across age and sex. Procedure. Students were tested in groups of no more than six individuals. They were told that a list of 24 word pairs would be presented and that they would be asked to remember the pairs after presentation. They were also told that an interview about how they learned the pairs would follow recall. Each word pair was presented for 8 s for study. When the presentation of word pairs was over, students were reminded of the recall procedure, and then asked to recall the second member of each pair on a sheet of paper as the experimenter read the stimulus terms one at a time (approximately 1 min after presentation). Stimulus terms were presented in the order in which they had been studied, from first to last, for each word list. After the recall phase, the experimenter went on to the interview phase. In the interview phase of the experimental session, four possible study strategies were described. Students were told that past discussions with high school students had revealed that (1) some students simply read the pairs carefully, (2) some say the words over and over to themselves (rehearsal), (3) some generate an image involving the objects referred to in each pair of words (visual elaboration), and (4) some connect the pairs of words with a verbal phrase or sentence (verbal elaboration). The students then reported on their own strategy use. They received a sheet of paper listing each of the 24 pairs they had studied, along with spaces in which they placed checks to indicate which of the four strategies they had used to study the individual pairs. They were free to check off as many strategies per pair as they wished, although checking off more than one was rare. If one of the strategies checked off was an elaborative strategy in the few cases where multiple strategies were selected, students were given credit for using an elaborative strategy. Checking off more than one nonelaborative strategy, or both types of elaboration strategies, had no impact on the analysis of strategy use which broke down student approaches into elaboration or not. Results and Discussion

Individual student data was tabulated in terms of number of high and low frequency word pairs elaborated (visual or verbal), recall of high and low frequency word pairs, and probability of recall of elaborated pairs (high and low). Students were only given credit for recall if the actual word was remembered, synonyms were rare and were not accepted as accurate recall. Table 1 presents mean scores on the three dependent variables across sex and age. There were no list effects on any of the dependent variables, and list is not included in the subsequent analyses.

TABLE

1

MEAN NUMBER OF WORD PAIRS ELABORATED, MEAN NUMBER RECALLED, AND MEAN PROBABILITY OF RECALL OF ELABORATED PAIRS-~TH AND 10~~ GRADE DATA

Mean number of word pairs elaborated High Frequency

Low Frequency

8th grade

M SD Males M

SD Females M

SD Males M

SD

6.75 3.94 4.58 4.15 10th grade

5.54 4.10 3.25 4.15

8.16 3.49

5.58 3.42

5.50 3.69

4.20 3.65

Mean number of word pairs recalled High frequency

Low frequency

8th grade Females

M SD

5.08 3.18

4.04 3.47

Males

M SD

4.20 2.82 10th grade

3.33 3.42

Females

M SD Males

M SD

7.16 2.72

5.29 3.09

4.54 2.66

3.50 2.65

Mean probability of recall of elaborated pairs High frequency

Low frequency

8th grade Females (N = 22)

M SD

65% 31.8

60% 37.8

Males (N = 13)

M SD Females (N = 22) M SD Males (N = 19)

M SD

55% 32.9 10th grade

53% 39.3

78% 22.2

73% 28.8

62% 31.7

52% 34.6

SEX DIFFERENCES

325

A 2(age) by 2(sex) by 2(word frequency) mixed design analysis of variance on number of word pairs elaborated produced main effects of sex and word frequency, F(1, 92) = 8.12, p < .Ol, and F(1,92) = 44.04, p < .OOl, respectively. Females elaborated a mean number of 13.02 word pairs, compared to 8.77 for males. A mean number of 6.25 high frequency word pairs were elaborated, compared to a mean number of 4.64 low frequency word pairs. A similar analysis of variance on the recall data produced main effects of sex and word frequency as well, F(1, 92) = 6.75, p < .Ol, and F(1, 92) = 30.62, p < .OOl, respectively. Females recalled a mean number of 10.79 word pairs, compared to 7.79 for males. A mean number of 5.25 high frequency word pairs were recalled, compared to a mean number of 4.04 low frequency pairs. There was no main effect of age for either analysis, although means suggested superior performance for 10th graders. There were no significant interactions. In order to determine probability of recall for elaborated high and low frequency word pairs the proportion of elaborated pairs which were also recalled was determined for each student. Students who did not elaborate any high frequency word pairs, low frequency word pairs, or both, did not contribute any scores to the analysis. Table 1 indicates the total N for each sex at each age, after dropping students who did not elaborate either of the two types of word pairs. A 2(age) by 2(sex) by 2(word frequency) mixed design analysis of variance with unequal N produced a main effect of sex, but no main effect of frequency, F(1, 92) = 4.15, p < .05. Females at both ages were consistently more likely to recall word pairs they had elaborated compared to males. However, elaborated high frequency word pairs were no more likely to be recalled than low frequency elaborated pairs. The frequency result parallels that of Waters (1982) who also found that frequency of word pairs had no effect on strategy effectiveness (probability of recall), but did have a significant effect on the likelihood that elaboration would be used (see our earlier analysis on number of pairs elaborated). In evaluating her results, Waters (1982) suggested that task variables impact on strategy use by influencing the likelihood that a particular strategy is used, but that strategy effectiveness is influenced by how different individuals implement the strategy. The probability of recall of elaborated pairs versus nonelaborated pairs was also examined, to determine whether students’ verbal reports of different strategies paralleled recall probabilities. One indication of the reliability of the verbal reports would be better recall of pairs reported to have been elaborated compared to those not elaborated. Since this was a within-subjects analysis, students who elaborated all 24 pairs, or who used nonelaborative strategies on all 24 pairs, were dropped from the analysis. This resulted in nine male and five female 8th graders, and four male and three female 10th graders being dropped. A 2(age by 2(sex) by 2(elaborated vs nonelaborated pairs) mixed design analysis of variance

326

WATERS

AND SCHREIBER

with unequal N produced a main effect of type of word pairs (elaborated versus not), and a significant interaction of type of word pairs by sex, F(1, 71) = 140.56, p > .OOl, and F(1, 71) = 4.55, p < .05, respectively. The mean probability of recall for elaborated pairs for females and males was 67 and 51% in 8th grade, and 72 and 57% in 10th grade. Females’ probability of recall for elaborated pairs was higher than males’ at each age. In contrast, there were no sex differences for nonelaborated pairs at either age, 15 and 18% in 8th grade, and 21 and 18% in 10th grade, for females and males, respectively. These results are consistent with Waters (1982) who used the same verbal report measure and with the work done by Pressley and Levin (1977) who used a more global verbal report measure of elaboration. Since the results from all three studies, present study included, are consistent with evidence that recall increases when subjects are explicitly instructed in elaborative strategies, it is reasonable to conclude that self-reported strategy use accurately reflects actual strategy use. EXPERIMENT

2

The results from Experiment 1 indicate that there are sex differences in use of elaboration, in the probability that elaborated pairs will be recalled, and in recall performance. The data fit the description of an intermediate phase in the developmental pattern of sex differences in strategy use characterized in the introduction. In this phase, both females and males are already using the appropriate strategy (elaboration), but females use the strategy more frequently, and more effectively. Since there were no interactions between sex and task variables in the two age groups tested, an addtional experiment was conducted with an older age group (college students). Based on the Waters and Andreassen (1983) framework, we would expect to see an attenuation of sex differences in strategy use with older students, as males begin to catch up. The attenuation should be more pronounced with more favorable materials, i.e., high frequency word pairs. The only significant difference in procedure between Experiments 1 and 2 is that in Experiment 2 subjects are also asked to describe their elaborations in a sentence or two. It was hoped that this information could shed some light on differences in strategy effectiveness between females and males. The elaborations were scored on two dimensions, an active/ passive dimension and a typical/atypical dimension. Interactive elaborations have been repeatedly shown to be more effective than more passive elaborations (Bower, 1970; Wollen, Weber, & Lowry, 1972), and typicality has been shown to influence the recall of prose content, with more typical content more readily recalled (Smith & Graesser, 1981). In light of the sex differences in prose memory reported by Waters (1981), and stable individual differences in typicality of prose content in self-generated

SEX DIFFERENCES

32-7

passages reported by Waters (1985), typicality of elaborations seemed a fruitful dimension on which to make comparisons and perhaps bridge the data on sex differences in paired-associate learning with that of prose production and recall. Differences in the typicality of elaborations had appeared in preliminary work, suggesting that this would indeed be an interesting dimension on which to evaluate elaborations. Method Subjects and design. A total of 24 female and 24 male college students participated in the experiment. They were recruited from introductory psychology courses and received course credit for their participation. The university students came primarily from nearby suburban and metropolitan areas, with similar backgrounds to the junior high school and high school students recruited in Experiment 1. Each student was asked to study and recall 24 paired associates, half of which were high frequency word pairs, and half low frequency word pairs. Students were also asked to indicate which of four study strategies they had used to study each word pair (read carefully, rehearse word pair, visual elaboration, verbal elaboration). In addition, if students indicated they had used some form of elaboration, they were asked to describe that elaboration in a sentence or two. Materials and procedure. The same word lists used in Experiment 1 were used in the present experiment. Instructions and procedures were also similar. The only procedural differences in the present experiment were that the college students were tested individually, not in small groups, and they were asked to describe their elaborations as well as indicate whether they had used an elaborative strategy. We have used both small groups and individual sessions to collect paired-associate learning data from high school students and college students in the past, and this minor procedural change has not effected our results. In both situations, stimulus presentation and recall procedures are exactly the same. Results and Discussion As in Experiment 1, individual student data was tabulated in terms of number of high and low frequency word pairs elaborated, number of high and low frequency word pairs recalled, and probability of recall of elaborated high and low frequency word pairs. Table 2 presents the means for all three dependent variables for the college data. A 2(sex) by 2(word frequency) mixed design analysis of variance on number of word pairs elaborated produced a main effect of word frequency, F(1, 46) = 36.12, p < .OOl, and a significant sex by word frequency interaction, F(1, 46) = 4.39, p < .05. As indicated by the means presented in Table 2, the difference in number of word pairs elaborated by females and by males for high frequency word pairs is only .70, whereas the difference between the two sexes for low frequency word pairs is 2.00.

328

WATERS

AND SCHREIBER TABLE

MEAN

2

NUMBER OF WORD PAIRS ELABORATED, MEAN NUMBER RECALLED, AND MEAN PROBABILITY OF RECALL OF ELABORATED PAIRS-COLLEGE DATA

Word pair frequency High Number

Low

of elaborated pairs

Females

M SD Males M

SD

7.45 3.36

6.25 3.17

6.15 2.99

4.25 3.62

Number of recalled pairs Females

M SD

6.45 2.71

5.04 3.48

6.75 2.95

4.45 3.38

Males

M SD

Probability of recall Females

M SD

78% 25.0

73% 28.0

81% 19.6

59% 33.5

Males

M SD

As predicted, there is an attenuation of sex differences with high frequency word pairs at the college level. A similar analysis of variance for recall performance only produced a significant main effect of word frequency, F(1, 46) = 35.34, p < .OOl. There were no significant sex effects in the recall analysis. Males did show a greater recall difference between high and low frequency pairs than did females (a difference of 2.30 word pairs vs 1.41 for females), although this interaction was not significant @ < .17). A 2(sex) by 2(word frequency) mixed design analysis of variance was also conducted on probability of recall for elaborated high and low frequency word pairs. As in Experiment 1, students who did not elaborate both types of word pairs were not included in the analysis. Only four males were dropped from the analysis. There was a significant main effect of word frequency, F(1, 42) = 8.92, p < .Ol, and a significant sex by word frequency interaction, F(1, 42) = 3.94, p < .05. Once again, sex differences are attenuated for high frequency word pairs, whereas a large difference in probability of recall remains for low frequency word pairs. In addition, a 2(sex) by 2(elaborated vs nonelaborated pairs) mixed

SEX

DIFFERENCES

329

design analysis of variance with unequal N was conducted to assess the impact of elaboration, dropping any students who elaborated all 24 pairs (one male). The analysis yielded a significant main effect of elaborated pairs vs nonelaborated pairs, F(1, 45) = 164.16, p < .OOl. Porbability of recall for elaborated pairs was 73% for males and 72% for females. Probability of recall for nonelaborated pairs was 24% for males and 19% for females. Interestingly there are no sex differences in the probability of recall of elaborated pairs, as was present in the 8th and 10th grade data of Experiment 1. This is undoubtedly due to the fact that there is no sex difference in the probability of recall of elaborated high frequency word pairs, and that the sex difference in elaborated low frequency word pairs is notably less than in the 8th and 10th grade data. In sum, the analyses conducted on the three dependent variables, number of pairs elaborated, pairs recalled, and probability of recall, support the hypothesis that as males catch up in strategy use and strategy effectiveness, sex differences are first attenuated under more favorable processing conditions, i.e., with high frequency word pairs. It may also be interesting to note at this time that the mean levels of performance of the college females are quite comparable to that of the 10th grade females, suggesting no further progress in the use of elaborative strategies. Males however do show mean changes in level of performance from the 10th grade data to the college data, a pattern which is consistent with the view that males eventually catch up to females in strategy use. Scoring of individual elaborations. A categorization scheme was developed in which elaborations were judged to be interactive elaborations, passive elaborations, or atypical elaborations. The categorization scheme could be applied to both verbal and visual elaborations, although most elaborations with these highly concrete materials were visual elaborations (70%).

An interactive elaboration was one in which the stimulus and response term interacted in some way (e.g., for friend-ball, “imagined playing ball with my friend,” or for snail-windmill, “a snail crawling on a windmill”), one in which the stimulus and response term occurred in a well-integrated and often familiar image (e.g., for door-flower, “imagined flower on door,” or for teacher-cup, “cup with the word teacher on it”), or one in which a verbal elaboration related the stimulus and response term in a meaningful way (e.g., for tree-book, “paper comes from trees and books are made of paper,” or for sardine-umbrella, “sardines live in water and water falls on an umbrella”). Interactive elaborations were neither passive, nor did they represent an unusual (atypical) way of relating the stimulus and response terms. The definition of passive elaboration was more restrictive than in previous studies of different types of elaborations. Passive elaborations could

330

WATERS

AND SCHREIBER

be ones that identified some characteristic that the two words had in common, but were not particularly significant characteristics, at least not for both words (e.g., for mouth-sky, “both are openings,” or for propeller-banana, “I remembered that they are both long”). Passive elaborations could also be those that passively placed the two terms in some type of spatial relationship, one that was not a typical occurrence (e.g., for kitchen-mountain, “kitchen in a mountain”, or for pill-hive, “imagined a pill on top of a hive”). Elaborations in which the two words were placed in a situation which was highly meaningful were not included in this’ passive group, but fell into the interactive category because they represented familiar images (e.g., for chair-salt, “salt spread on a chair”). Included in the passive elaboration group were also those elaborations in which the individual commented on one term, and then the other, without integrating the two (e.g., for broom-parachute, “broom used on the floor; parachute used in the air,” or for dandelion-sausage, “one’s a plant; one’s a meat”). Atypical elaborations were those elaborations in which the two words are related to one another in an unusual manner, that is, one which is unlikely to be repeated by another individual. Sometimes this is the result of focusing on a secondary meaning of the stimulus or response term, and not its most common meaning (e.g., for door-flower, “Jim Morrison (of the Doors) singing a song about life” or for friend-ball, “bringing a friend to a dance (ball)“). In other cases, the two words are interacting in an unusual and unlikely manner (e.g., for pajamas-dictionary, “a pajama party where everyone reads dictionaries”, or for water-party, “saw a birthday party in a large amount of water”). The elaborations produced by the college students (both males and females) were collected and randomly arranged within word pairs. Two raters scored each elaboration produced for the various word pairs as interactive, passive, or atypical. There was rater agreement on 91% of the elaborations, and all disagreements were resolved by discussion. The proportion of total elaborations produced by females and males that were interactive, passive, or atypical are presented in Table 3. Relatively fewer interactive elaborations were produced for low frequency word pairs, z = 5.55, p < .OOl, and relatively more passive elaborations, z = 15.39, p < .OOl, with atypical elaborations constant across type of word pair. In addition, females and males show the same sensitivity to type of word pair, responding with more passive elaborations, z = 4.96, p < .OOl and z = 4.43, p < .OOl respectively, and fewer interactive elaborations for low frequency word pairs, z = 3.70, p < .OOl and z = 3.74, p < .OOl, respectively for females and males. There are no sex differences in the kinds of elaborations produced. The next analysis conducted on the different types of elaborations involved calculating probability of recall for the different elaborations. For

331

SEX DIFFERENCES TABLE DIFFERENT

3

TYPES OF ELABORATIONS PRODUCED BY GDLLEGE PROBABILITY OF RECALL Type

Type

of word

High frequency Females (N Males (N = Low frequency Females (N Males (N =

of word

High frequency Females Males Low frequency Females Males

of elaboration

Interactive

Passive

Atypical

= 180) 162)

70% 78%

12% 6%

18% 16%

= 150) 102)

50% 56%

35% 25%

15%

Interactive

Passive

Atypical

75% 77%

70% 71%

71% 91%

pair

Probability Type

STUDENTS AND THEIR

pair

19%

of recall

67%

67%

77%

66%

60%

66%

’ The number in parentheses is the total number and males to high and low frequency word pairs.

of elaborations

produced

by females

each set of interactive, passive, and atypical elaborations across type of word pair and sex, the probability with which the elaborations were followed by recall was determined. These probabilities are presented in Table 3. t-test analyses comparing mean probability of recall scores between males and females for each type of elaboration across high and low frequency word pairs indicated that the probability of recall of different elaborations did not change for females with type of word pair. For males, there was no change with interactive elaborations from high to low frequency word pairs. However with passive and atypical elaborations combined (necessary because not enough male subjects produced these types for both high and low frequency word pairs), there is a significant decrease in probability of recall from high to low frequency word pairs, t(l1) = 1.81, p < .05, one-tailed test. In sum, the descriptive information provided by the data on types of elaborations produced by males and females suggest it is not in the type of elaboration per se that differences in strategy efficiency occur, but perhaps in the overall depth of processing afforded each elaborative effort. Support for the elaboration category scheme per se was found in the differential rates of passive elaborations for high and low frequency word pairs. Because low frequency words are less meaningful, and are less

332

WATERS

AND

SCHREIBER

easily brought together into some type of interactive elaboration, we would expect more passive elaborations with those pairs, and that is exactly the change in produced elaborations which occurred. GENERAL DISCUSSION The present results, along with those of Waters (1981) and Cox and Waters (1986), demonstrate that sex differences occur for a wide range of strategies, across a wide range of ages. The results also demonstrate the usefulness of the developmental framework of strategy acquisition and generalization proposed by Waters and Andreassen (1983), and highlight the importance of identifying the mechanism or mechanisms which account for this pattern of increasing strategy use with age. Identification of such mechanisms would not only explain the overall pattern of strategy generalization to less favorable task conditions, but should be the key to understanding the sex differences in strategy use which have been uncovered. Although the existence of sex differences on global measures of verbal ability is currently in dispute (Maccoby & Jacklin, 1974; Hyde & Linn, 1988), the context-specific and age-specific differences in strategy use now documented in prose processing, free recall, and paired-associate learning cannot be explained in terms of trait-like sex differences. Whether or not sex differences are found depends on the developmental level of the individuals and the particular task circumstances that they are presented with. These findings are not consistent with a quantitative difference in verbal ability in which females are always expected to do better than males. Instead, the patterns of sex differences in strategy use which have been demonstrated (Waters, 1981; Cox & Waters, 1986; present study) are more easily understood as differences in the ages at which females and males progress through the stages of strategy acquisition outlined by Waters and Andreassen (1983). The question then is not why do females and males differ in strategy use, but rather why do they acquire strategy use on different schedules. Plausible explanations are likely to involve reference to meaningfulness and semantic representation. Chi (1978; 1981) has suggested that content knowledge plays an important role in memory performance, with greater content knowledge tied to more sophisticated memory behavior. One hypothesis for these sex differences is therefore that males lag behind females in semantic development, with the words (and the concepts they represent) less elaborately represented in semantic memory. From this hypothesis would follow the prediction that females would be more likely to initiate the use of elaborative strategies, and be more likely to construct a more elaborate representation of the connection they generate. It would also suggest that they would be more likely to construct organizational connections, particularly when subjective organization is involved. One

SEX DIFFERENCES

333

problem with this hypothesis is that it predicts that as long as meaningfulness differences exist there should be sex differences. Waters and McLean (1991) however gave interactive elaborations to college students while they studied word pairs (the same lists as the present study) and the experimenter-provided elaborations eliminated any sex differences in recall of the word pairs, and even eliminated recall differences between high and low frequency word pairs. Another possible explanation acknowledges that females initiate strategy use at an earlier age (for whatever reason), but focuses on the effect of practice in implementing strategies more efficiently and effectively. This explanation hypothesizes that with practice, individuals can elaborate more easily, and are thus more likely to engage in the type of in-depth semantic processing that leads to a semantically rich representation of the connection generated between word pairs. The difference between more efficient and less efficient elaborators should be manifested in different rates of elaboration, and in different probabilities of recall for elaborated pairs. Assuming that females have been practicing elaboration for a longer period of time, this explanation can account for the data presented in the present study (elaboration data, probability of recall data, individual elaboration analysis). It is also consistent with a recent proposal by Ornstein, Baker-Ward, and Naus (1988) that increasing strategy efficiency and automatization is a key mechanism for strategy generalization to less favorable processing contexts (whatever the strategy). A third possibility is to focus on motivational differences between males and females, proposing that females are more motivated to do well in learning situations, and thus develop appropriate strategies sooner and use them more often in appropriate contexts. There are two variations of this hypothesis. One proposes that if motivational differences are eliminated, sex differences can be eliminated. The other suggests that over long periods of time, real differences in strategy implementation develop through differential practice effects, so that sex differences can only be lessened by the elimination of motivational differences at an early stage of strategy development. Any or all of the factors suggested above may play a role in the developmental differences between males and females in strategy use. The present results help us recast the questions that should be addressed in future research. There is more than one way to conceptualize sex differences in performance. The trait-like model of old, in which sex differences are described in terms of quantitative differences in verbal ability, is not supported by the results on sex differences in strategy use across different memory paradigms. Future research should focus on developmental processes in strategy use rather than on trait interpretations of sex differences in verbal ability and strategy use. Fruitful avenues of approach include investigations of how efficient

334

WATERS

AND SCHREIBER

males and females are in implenting memory strategies, and whether the cognitive demands of strategy use are greater for males than females. Potential differences in the knowledge base for females and males and their impact on strategy use should also be explored. Bjorklund for example has proposed that age changes in the knowledge base lead to increasing accessibility of information, and more effortless activation of relations among concepts, improving the efficiency of strategy implementation (Bjorklund & Bjorklund, 1985; Bjorklund, 1987, 1988). In light of the fact that sex differences in strategy use can be found in a wide range of memory paradigms (Waters, 1981; Cox & Waters, 1986; present study), researchers should focus on processing explanations that can account for the differential implementation of a wide range of strategies. REFERENCES Bjorklund, D. F. (1987). How age changes in the knowledge base contribute to the development of children’s memory: An interpretive review. Developmental Review, 7, 93-130. Bjorklund, D. F. (1988). Acquiring a mnemonic: Age and category knowledge effects. JOU~MI of Experimental Child Psychology, 45, 71-87. Bjorklund, D. F., & Bjorklund, B. R. (1985). Organization versus item effects of an elaborated knowledge base on children’s memory. Developmental Psychology, 21,11201131. Bower, G. H. (1970). Imagery as a relational organizer in associative learning. Journal of Verbal Learning and Verbal Behavior, 9, 529-533. Brown, A. L., Bransford, J. D., Ferrara, R. A., & Campione, J. C. (1983). Learning, remembering, and understanding. In P. H. Mussen (Ed.), Handbook of child psychology: Cognitive development (Vol. III). New York: Wiley. Chi, M. T. H. (1978). Knowledge structures and memory development, In R. S. Siegler (Ed.), Children’s thinking: What develops? Hillsdale, NJ: Erlbaum. Chi, M. T. H. (1981). Knowledge development and memory performance. In J. P. Das & N. O’Conner (Eds.), Intelligence and learning. New York: Plenum. Cox, D., & Waters, H. S. (1986). Sex differences in the use of organizational strategies: A developmental analysis. Journal of Experimental Child Psychology, 41, 18-37. Flaveli, J. H. (1971). First discussant’s comments: What is memory development the development of? Human Development, 14, 272-278. Furth, H. G., & Milgram, N. A. (1973). Labeling and grouping effects in the recall of pictures by children. Child Development, 44, 511-518. Hughes, S. E. D., & Walsh, J. F. (1971). Effects of syntactic mediation, age, and modes of representation on paired-associate learning. Child Development, 42, 1827-1836. Hyde, J. S., & Linn, M. C. (1988). Gender differences in verbal ability: A meta-analysis. Psychological Bulletin, 104,53-69. Kail, R. V., Jr., & Hagen, J. W. (Eds.). (1977). Perspectives on the development ofmemory and cognition. Hillsdale, NJ: Erlbaum. Maccoby, E. E., & Jacklin, C. N. (1974). The psychology of sex differences. Stanford, CA: Stanford Univ. Press. Ornstein, P. A. (Ed.). (1978). Memory development in children. Hillsdale, NJ: Erlbaum. Omstein, P. A., Baker-Ward, L., & Naus, M. J. (1988). The development of mnemonic skill. In F. E. Weinert & M. Perlmutter (Eds.), Memory development: Universal changes and individual differences. Hillsdale, NJ: Erlbaum.

SEX DIFFERENCES

335

Pressley, M., & Levin, J. R. (1977). Task parameters affecting the efficacy of a visual imagery learning strategy in younger and older children. Journal of Expeimental Child Psychology,

24, 53-59.

Pressley, M., & Levin, J. R. (1978). Developmental constraints associated with children’s use of the keyword method of foreign language vocabulary learning. Journal of Experimental

Child

Psychology,

26, 359-372.

Rohwer, W. D., Jr. (1973). Elaboration and learning in childhood and adolescence. In H. W. Reese (Ed.), Advances in child development and behavior (Vol. 8). New York: Academic Press. Smith, D. A., & Graesser, A. C. (1981). Memory for actions in scripted activities as a function of typicality, retention interval, and retrieval task. Memory and Cognition, 89, 550-559. Thorndike, E. L., & Lorge, I. (1944). The teacher’s word book of30,ooO words. New York: Bureau of Publications, Teacher’s College, Columbia University. Waters, H. S. (1981). Organizational strategies in memory for prose: A developmental analysis. Journal of Experimental Child Psychology, 32, 223-246. Waters, H. S. (1982). Memory development during adolescence: Relationships between metamemory, strategy use, and performance. Journal of Experimental Child Psychology, 33, 183-195. Waters, H. S. (1985, April). Typicality of passage content in narrative production. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, Maryland. Waters, H. S., & Andreassen, C. (1983). Children’s use of memory strategies under instruction. In J. R. Levin & M. Pressley (Eds.), Cognitive strategy research: Psychological Foundations. New York: Springer-Verlag. Waters, H. S., & McLean, A. J., Jr. (1991). On the generality of typicality effects in memory, manuscript submitted for publication. Waters, H. S., & Waters, E. (1976). Semantic processing in children’s free recall: Evidence for the importance of attentional factors and encoding variability. Journal of Experimental

Psychology:

Human

Learning

and Memory,

2, 370-380.

Wollen, K. A., Weber, A., 8c Lowry, D. H. (1972). Bizarreness versus interaction of mental images as determinants of learning. Cognitive Psychology, 3, 518-523. RECEIVED:

August 15, 1988;

REVISED:

January 5, 1990 & June 17, 1991