EFFECTS OF STRELTEGY MEG EN EHRED‘ASSOOIATE LEARNENG‘ A DEVELOPMENTAL STUDY Thesis fer the Degree of M; A. MORGAN STATE SWEET? IHESIS ' h n“. I LIBRARY Michigan State University ann . “E 0" LY ‘JV 3 V ABSTRACT EFFECTS OF STRATEGY AIDS IN PAIRED-ASSOCIATE LEARNING: A DEVELOPMENTAL STUDY by David L. Cox One purpose of this study was to determine whether children in the elementary grades could effectively verbalize about cues which they used to form associations on a paired associate (PA) task. If so, this would enable the experimenter (E) to analyze the types of strategies reported as a function of develOpmental level. It was also of interest to determine the relationships between these re- ported strategies and learning and retention as a function of age and sex. Another purpose was to investigate whether the administration of high level strategy aids to elementary school children would fa- cilitate acquisition and retention of the PA material. It would consequently be of interest to determine the differential effect of giving associational cues on the performance of slow (S) and fast (F) learners. Two hundred forty-two students from three hth, 6th, and 8th grade classes served as subjects (85) in a lowahigh, PA task with 3 identical treatment conditions at each grade level. Subjects given treatment 1 received no aid and served as the control group at each grade level. Subjects given treatment 2 received high level strategy aids on half the pairs, and, in treatment 3, Ss received strategy aids on all pairs. The developmental analysis of the unaided control groups showed that elementary school children could effectively verbalize about cues which they used to form associations. Moreover, these stra- tegies could be rank ordered along a continuim of complexity and quantified. In brief, the data for groups which had treatment 1 revealed that acquisition and retention scores increase with age as do the number of higher level strategies, and that better performers used higher level strategies at every grade level. Females also performed better than males at all grade levels and consistently used higher level strategies. The analysis of groups which received treatments 2 and 3 re- vealed a significant facilitation on acquisition and retention when Ss were provided with strategy aids. Furthermore, giving associa- tional cues on all pairs seemed to be more beneficial than giving aids on only half the pairs. Subjects also performed better on the acquisition and retention of unaided items within a list which had other items aided. This was shown to be related to an increase in the number of high level strategies associated with these pairs. Finally, the data revealed that strategy aids facilitated the per- formance of slow learners significantly more than fast learners. The significance of these findings is discussed in the light of developing more appropriate teaching materials and methods for the classroom. EFFECTS OF STRATEGY AIDS IN PAIRED-ASSOCIATE LEARNING: A DEVELOPMENTAL STUDY By David L. Cox A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS College of Education 1965 ACKNOWLEDGEMENTS To the pe0ple who made this study possible I wish to express my gratitude. I am indebted to Dr. Clessen J. Martin, advisor and project director, who afforded the opportunity for a study of this magnitude, and for his valuable suggestions and guidance throughout my program. To Dr. Frederick.J. Boersma and Cornelius vanderveen I also owe a special word of thanks. Without their kind assistance in experimentation and performing the many analyses, completion would not have been possible. Finally, to the 2H3 grade school children, their teachers and principals, who eagerly gave their time and participation, I am grateful. ii ACKNOWLEDGEMENTS . . . LIST OF TABLES . . . . LIST OF FIGURES . . . . LIST OF APPENDICES . . INTRODUCTION . . . . . DESIGN AND METHODOLOGY Subjects . . Materials . . Procedure . . RESIJLTS O O O O O O O O DevelOpmental Analysis Analysis of Treatment Effects . . . . . Retention Analysis Analysis of Fast and Slow Learners . . BIBLIOGRAPHY . . . . . . TABLE OF CONTENTS . DISCUSSION AND CONCLUSIONS APPENDICES . . . . . . . . . Page 0 O O O O O O O O O O O O O O l l O O O O I O O O O O O O O O C 1v 0 O O O O O O O O O O O O O C Vl O O O 0 O O O O O O O O O O 0 viii . . . . . . . . . . . . . . . 21 . 21 °°'°°°°IIII.I 33 OOOOCOOOOOOOOCO 56 61 iii Table 10 ll 13 1h 15 LIST OF TABLES Classification and Rank Order of Associative S‘brategies O O O O O O O O O O O O O O O O O O O O 0 Sample Characteristics and Treatment Designations for the Nine Classes . . . . . . . . . . . . . . . . . . Paired-Associate List and Strategy Aids for Criterion TaSk I O O O O O O O O O O O O O O O O O O O O O O 0 Percent Frequency of Strategy Level Use for A, 6, and 8th Grade Control Groups . . . . . . . . . . . . . . Analysis of Variance on Total Correct Responses on Criterion Task for h, 6, and 8th Grade Control Gr Cup S O O O O O O O O O O O O O O O O O O O O O O 0 Mean Correct Responses and Variances for Males and Females in h, 6, and 8th Grade Control Groups . . . Analysis of Variance on Total Strategy Scores for h, 6, and 8th Grade Control Groups . . . . . . . . . . Mean Strategy Scores and Variances for Males and Females in h, 6, and 8th Grade Control Groups . . . Analysis of Variance on Practice Task Scores . . . . . Mean Correct Responses and Variances for Practice TaSk O O O I O O O O O O O O O O I O O O O O O O O 0 Analysis of Variance on Criterion Task Scores . . . . . Mean Correct Responses and Variances for Criterion TaSk O O O O O O O O O O 0 O O O O O O O O O O O O 0 Analysis of Variance on Total Correct Responses on B Pairs for C and E-h Treatment Conditions . . . . . . Mean Correct Responses and Variances on B Pairs for C and E-h Treatment Conditions . . . . . . . . . . . . Collapsed Percent Frequency of Strategy Level Use on B Pairs for h, 6, and 8th Grade C and E-h Treatment conditions 0 C C C C . O C O C O . C O O . O C O O . iv 1% 15 2h 27 28 29 30 3h 35 37 39 145 A6 A7 Table l6 17 18 19 2O 21 22 23 2h 25' 26 Analysis of Variance on Total Strategy Scores on B Pairs for C and E-h Treatment Conditions . . . . . . Mean Strategy Scores and Variances on B Pairs for C and. E‘h' Treatment Conditions 0 o o o o o o o o o o 0 Mean Correct Responses and Variances on A and B Pairs for Criterion Task . . . . . . . . . . . . . . . . . Analysis of Variance on Total Retention Scores . . . . Mean Correct Responses and Variances for Retention TaSk O O O O O O 0 O O O O O O O O C O O I O O I O 0 Mean Correct Responses and Variances for A and B Pairs on Retention Task . . . . . . . . . . . . . . . . . Intercorrelations Between Acquisition Scores on Practice and Criterion Tasks . . . . . . . . . . . . Mean Correct Responses for Fast and Slow Learners on PraCtiCe TaSk O O O O O O O O O 0 O O O O O 0 O O 0 Mean Correct Responses for Fast and Slow Learners on criterion TaSk 0 O O O O O O I O O I O O I O O O O O Collapsed Percent Frequency of Strategy Level Use for Fast and Slow Learners on Control Groups . . . . . . Collapsed Percent Frequency of Strategy Level Use for Fast and Slow Learners on Control Groups When Collapsed Over Grades . . . . . . . . . . . . . . . Page A9 50 55 58 59 6O 62 67 72 7h 75 LIST OF FIGURES Figure Page 1 Acquisition Curves for A, 6, and 8th Grade Control Groups on Criterion Task . . . . . . . . . . . . . 22 2 Mean Number Correct Responses by Strategy for Control Groups on Criterion Task . . . . . . . . . . . . . 25 3 Percent Correct Responses on Retention as a Function of Strategy Level Used on Criterion Task for Control Groups . . . . . . . . . . . . . . . . . . 31 A .A Geometric Interpretation of Interaction of Grade Level with Treatment Condition on the Practice Task . . . . . . . . . . . . . . . . . . . . . . . 36 5 A Geometric Interpretation of Interaction of Grade level with Treatment on the Criterion Task . . . . ho 6 Fourth Grade Acquisition Curves on Criterion Task for E‘s-8, E-Ll" and C Treatments 0 o o o o o o o o o o 0 (+1 7 Sixth Grade Acquisition Curves on Criterion Task for 3-8, E-Lt, and C Treatments 0 o o o o o o o o o o o 1-1-2 8 Eighth Grade Acquisition Curves on Criterion Task for E-8, E-A, and C Treatments . . . . . . . . . . . . SB 9 Fourth Grade Acquisition Curves for A.and B Pairs . . 52 10 Sixth Grade Acquisition Curves for A and B Pairs . . . 53 ll Eighth Grade Acquisition Curves fer A and B Pairs . . 5h 12 Fourth Grade Acquisition Curves for Fast and Slow Learners on Practice Task . . . . . - . . . . . . . 6h 13 Sixth Grade Acquisition Curves for Fast and Slow Learners on Practice Task . . . . . . . . . . . . . 65 1% Eighth Grade Acquisition Curves for Fast and Slow Learners on Practice Task - . . . . . . .N. . . . - 66 15 Fourth Grade Acquisition Curves for Fast and Slow Learners on Criterion Task . . - . . - . . . - . . 68 vi vii Figure Page 16 Sixth Grade Acquisition Curves for Fast and Slow Learners on Criterion Task . . . . . . . . . . . . 69 1? Eighth Grade Acquisition Curves for Fast and Slow Learners on Criterion Task . . . . . . . . . . . . 7O LIST OF APPEND ICES Appendix Page A Instructions for the Practice Task . . . . . . . . . . 93 B Instructions for the Criterion Task . . . . . . . . . . 96 C Instructions for Strategy Collection Task . . . . . . . 97 D Instructions for Retention Task . . . . . . . . . . . . 98 viii INTRODUCTION The study of conscious thought processes in scientific psychology has an old and interesting history. Early discussions of consciousness were either philosophical and/or theological in character and were most often couched in terms of a dualism like mind vs. matter, rational vs. irrational, or soul vs. body. From the ancient Greeks through the time of Descartes discussions of this nature dominated much of the conversations of learned.men. Then in the 19th century British Empiricism brought dualism and the concept of consciousness into psychology. Locke, Berkeley, and Hume, the main representatives of this philos0phy, were all inter- ested in how the mind gets to know about the external world (Russell, l9h5). But their investigations of this external world led to a departure from traditional dualism and to a doctrine of association which dealt mainly with the relations among the items of mind, with a corresponding de-emphasis on the other half of the dualism. The founders of the new experimental psychology, Fechner, Wundt, Mach, Helmholtz, etc., were interested in studying con- sciousness and their chief method was introspection. In general they believed conscious events were dependent on brain events, but completely separate and different from them. Therefore, some kind of introspection or inner perception through which one obtains evidence about mental events was necessary. James (1890) was of the opinion that introspective observation is all human beings - 1 - - 2 - have to rely on to know of their existence. He defined intro- spection as looking into our own minds and reporting what we dis- cover there. He further stated, "Every one agrees that we there discover states of consciousness." By this reasoning the intro- spective technique soon came to be used in the experimental labo- ratories. If conducted correctly it was regarded as a process not subject to error. In fact, wundt and his followers placed a great deal of emphasis upon'training observers fer the accurate descrip- tion of their conscious processes. Classical introspectionism soon boasted such members as‘Wundt, Kulpe, Muller, and Titchener (Boring, 1953). They shared the belief that description of consciousness reveals patterns of sen- sory elements which could be further broken down into more basic elements. Thus elementism and sensationiam'became synonymous with introspectionism. Titchener further'maintained that these descrip- tions of consciousness should exclude statements of meaning since these were merely inferences on the part of the observer and lacked the accuracy of reports of sensations. But introspection with inference and meaning left out soon became extremely dull and with- out function. And psychologists began to note that examinations of the mind did not reveal stable images and sensations, but rather intentional activities directed toward an object or goal. Such shortcomings combined with the advent of Gestalt and Behavioristic psychology culminated in the downfall of classical introspec- tionism. - 3 - In the early 1900's Gestalt psychology, founded by Wertheimer, developed as a protest against the constraints and inadequacies of classical introspectionism. It took root particularly in the area of perception where classical introspectionism.could not explain the phenomenon of seen movement. Within this theory, learning involved changes in the phenomenal field through an organization of the incoming stimuli in a way which achieved.maximum.simplicity. As such, the perceptions could be analyzed and studied in and of themselves using an introspective technique, but without reference to conscious mental processes, images, or sensations. The main difficulty in applying Gestalt psychology to problems of learning was that the system, while describing certain aspects of the learning process, did not indicate how the environment should be arranged in order to facilitate learning. And even though Gestalt psychology exerted extensive influence on educational thought and remained respectable, its popularity in America soon began to wane. As Boring (1953) states, “American psychology tended all along to be practical and functional and...was destined to become behavior- istic." - The greatest rejection of classical intrOSpectionism.came with the advent of behaviorism. Its principal proponent, J.B.‘Watson, reacted against the fermalism and inaccuracy of intro- spection and attempted successfully to replace it with a psychology of behavior. {At the same time, studies of conditioning by Pavlov helped spread this behavioristic position in.American psychology. In brief, behaviorism espoused the doctrine that psydhological - h - theory should be based on data recorded from observable actions of subjects (Ss). These data were to be collected under controlled experimental conditions in which the experimenter (E) manipulated various properties of the stimulus materials presented. Subjects in these tasks were to be naive, rather than trained, and intro- spective references to consciousness were to be excluded. Watson, in attacking introspection was not objecting to the use of words by the subject, but rather to trusting the subject to place correct and interpretable meanings into his words. He thus criticized this method as being extremely unreliable. Introspec- tive reports lacked consensual validity, since several persons viewing the same material often gave different interpretations as to its content. Because of these factors psychologists became very leary about obtaining verbal reports from 83 in psychological studies. It was assumed the experimenter could not believe the subject or accurately interpret what he said. Most studies, there- fore, dealt just with the quantitative aspects of learning and recall and with modes of presentation of the material. This opera- tionism further tended to prevent psychologists from.speculating about processes inside the memorizer and emphasized what the ex- perimenter, rather than the memorizer, was doing. Such data were seldom supplemented by reports of how the learner represented the material to himself. Certain psychologists took an extreme position in the condem- nation of introspectionism and denied that a person's conscious mediational processes had any necessary connection at all with the - 5 - learning process. One such investigator was E.L. Thorndike. Throughout the 1930's Thorndike accumulated evidence which showed that a great many 33 who performed well in learning situations could not verbalize how they had learned. He maintained that con- scious awareness of what one is doing is not necessary for learning to take place. ‘Any relationship between introspective reports of awareness and performance merely shows that awareness depends on how much learning there is of which to be aware (Thorndike, 1935; Farber, 1963). In other words, awareness may be a consequence of learning rather than a prerequisite. More recently, B.F. Skinner expressed the view, "The private event is at best no more than a link in a casual chain and it is usually not even that (Farber, 1963) . But even the behaviorists could not afford to deny completely that such processes were going on. Comprehensive theories of learning, postulated by such noted men as Hull, Tolman, Levin, and others, acknowledge the central processes involved in learning. But their theories, according to Travers (1963), made little pro- vision for the measurement of the mediating processes which their systems discussed. various studies of mediation were conducted, but usually within some sort of transfer paradigm. Subjects were typically asked to learn several similar lists of paired-associates, and.mediation was indirectly inferred by'E from.the changes in Ss performance from one list to the next. As Barclay (1961) has sug- gested, however, transfer paradigms are not the most effective designs for studying mediation. He concluded that within st'n - 6 - designs mediation either has no effect or its effect is so closely interwoven with general transfer that it cannot be distinguished. An introspective technique, when considered with respect to other variables, might minimize this problem and provide valuable infor- mation about mediational processes. Recently there has been a resurgence of interest in conscious mediational processes. A.number of psychologists have argued that not only do private experiences exist, but also they may to some extent explain a person's behavior in a learning situation (Miller, Galanter, &.Pribram, 1960). Among these investigators there is a growing acceptance that learning is not accomplished in simple rote fashion as formerly thought. This matter is related to the classical controversy in psy- chology concerning the nominal vs. effective stimulus. The nominal stimulus refers to the measured.properties of the stimulus material as presented. The effective stimulus refers to the psychological consequences of stimulation and cannot be directly Observed. How the two are related is complex and difficult to determine, but investigations of this relationship are crucial in developing com- prehensive models of behavior. This is particularly true in learn- ing verbal materials because of the complete degree to which effec- tive stimulation depends on the activities of the experimental subject (Rothkopf, 1965). Miller (1956) has written extensively on coding processes used by 83 in transforming the nominal into the effective stimulus. Underwood and Keppel (1963) call this process encoding. They - 7 - maintain that there is a transposition process in memory which may involve associational aids or mneumonic devices. -In addition, if a subject in fact encodes the material he must necessarily decode it to produce the original stimulus. More specifically, the effec- tiveness of a person's encoding-decoding procedures will be ulti- mately related to learning. Thus, in opposition to the noncogni- tive argument of investigators such as Thorndike, these cognitive theorists support the position that conscious recognition of the relational aspects of the material to be learned is directly re- lated to performance. They furthermore state that, "If it were not for an unreasonable prejudice against questioning subjects psy- chologists would long ago have discovered...that one can usually understand a person's behavior much better (Farber, 1963). Several experimenters have recently discussed the feasibility of questioning Ss and using their verbal reports to study how peOple learn (Epstein, Rock, & Zuckerman, 1960; Miller, Galanter, & Pribram, 1960; Underwood & Schulz, 1960; Farber, 1963; Eagle & Leiter, 196A). Generally these investigators are in agreement that verbal reports may provide valid information on how peOple learn, but relatively few systematic and extensive investigations have been undertaken to support their conjectures. Some recent research by Martin, Boersma, & Cox (1965) has attempted to systematically analyze and classify Ss verbal reports. These experimenters gave 38 a PA learning task after which they were to state in writing how they had attempted to make each association. In brief, Martin et a1. were able to classify verbal reports into seven - 8 - categories: (1) No association, (2) Repetition, (3) Single letter cue, (h) Multiple letter cue, (5) Word formation, (6) Superordi- nate, and (7) Syntactical. These categories are rank ordered along an apparent continuum of cue complexity. A description of the classification scheme is given in Table 1. Plotting mean correct responses for each pair as a function of strategy level resulted in a monotonically increasing relationship from strategy levels one through seven. In addition, each subject was assigned a total strategy level score based on the sum of his strategy ratings for all eight pairs. A Spearman rank order correlation coefficient between total strategy level score and total correct responses on the task was .62 indicating the higher the strategy level the greater the number correct. They found this scheme to be very re- liable and to have predictive value. 'With this approach, Martin et a1. (1965) have shown that it is possible to classify a widely varying collection of idiosyncratic verbal reports into relatively few categories. Furthermore, rank ordering these categories made possible a certain level of quantification which yielded a positive relationship between verbal reports and performance. Since actively searching for cues, or coding the material, has been shown to be part of the learning process it may be reasoned that giving strategy aids would effectively reduce the time re— quired to learn. As expected, such studies have shown that Lnen 33 are given instructions specifically designed to increase the probability of mediational links in PA learning, or when the links themselves are provided by E, the results are often striking. In - 9 - TABLE 1 Classification and Rank Order of Associative Strategies Associative Type of Cue Subject Example of Vérbal Strategy Reported Using Report 1. No S was not able to state Sagrole-Polef: "Don't Reported how he managed to make know how I learned Associations the association. this pair." 2. Repetition S reported rehearsing Volvap-Nares: "Just the pair. kept repeating these words to myself." 3. Single S reported using a sin- Tarop-Gojey: "Noticed Letter gle letter in each of that each word con- Cues the paralogs in making tained an 0." the association. h. Multiple S reported using mul- Sagrole-Polef: "Each Letter tiple letters in each word contains an OLE." Cues of the paralogs. 5. Werd S reported that an Meardon-Zumap: "The Formation actual word was em- word EAR is contained bedded in one or both in Meardon and learned of the paralogs and that EAR goes with made use of these words Zumap." in making the associa- tion. 6. Super- S reported selecting Sagrole-Polef: ordinate elements from each of "Sagrole begins with g the two paralogs that and Polef with 2 - had some relationship thought of State to each other. ‘Police." 7. Syntactical S reported selecting Rennet-Quipson: elements from each of the two paralogs and embedding these ele- ments into a sentence, phrase, or clause. "Changed Rennet to Bennet and saw Quips in Quipson - thought Bennet Cerf Quips on TV." -10... every case the number of trials, number of errors, or time to cri- terion were significantly reduced, both for normal and retarded Ss (Spiker, 1960; Cramer, 1962; Jensen & Rohwer, 1963; Kitao, 1963). Merely naming the stimulus items has produced facilitation of learning (Jensen, 1963). This effect has also been shown to occur when only one word provides the mediating link (Epstein, Rock, & Zuckerman, 1960; Glanzer, 1962). But studies in which the experi- menter provides the link in the form.of a syntactical or thematic strategy in general may be expected to have the greatest facili- tating effect (Jensen & Rohwer, 1963 ; Bruner & Oliver, 1963; Davidson, 196A; Martin, Boersma, & Cox, 1965). The facilitating effect of’mediational cues has also been illustrated in long term retention. Wallace, Turner, & Perkins (1957) showed that when Ss were instructed to form a visual image connecting two words on a PA task they could remember up to 700 pairs with 95% accuracy. In general these studies illustrate the increase in learning and retention when Ss are provided associational cues or are in- structed to provide their own. But no studies could be found which systematically investigated these effects with respect to age. More specifically, no studies were revealed which examined the effects of giving the same syntactical mediational cues at various age levels. No studies were disclosed which investigated the extent to which mediational cues should be given. Providing aids on only .part of the material to be learned may provide the subject with enough infbrmation to formulate his own on the unaided material. _ 11 _ Giving aids on all the material may be too much to assimilate and interfere with maximum performance. On the other hand, any trans- fer from the aided items may not be as beneficial as if E provided cues on all the material to be learned. These effects also have not been studied with respect to slow and fast learners. Studies comparing the mediational activities of slow and fast learners seldom show consistent or clear results (Scheible, 195%; Underwood & Schulz, 1960). An investigation of these problems utilizing the Martin et a1. (1965) strategy classi- fication scheme may contribute valuable information to the study of the learning process. The present study was designed to determine whether the methods for collecting strategy information from college students employed by Martin et a1. (1965) could be utilized for hth, 6th, and 8th graders. More specifically, it was designed to determine whether children in elementary grades could effectively verbalize about cues which they used to form associations. If so, it may be possible to employ Martin's classification system to study the for- mation of associative strategies developmentally. It was reasoned that one important factor involved in a person's increasing learn- ing ability may be an increase in the use of high level strategies. It was also of interest to determine if there were any sex differ- ences in performance and/or strategies used and to investigate such differences develOpmentally. Another purpose of this study was to investigate whether stra- tegy aids given by E would facilitate perfbrmance of Ss in the -12.. elementary grades. Furthermore, if provision were made for aiding a different number of items for the various groups, it might be expected that the more items given a strategy aid the better the ' performance of the group. In addition, Ss may be expected to per- fbrm.better on the unaided items within a list which has other items aided. It was reasoned that providing strategy cues for some items within a list may help 85 to formulate their own strategies on the unaided items. This in turn may facilitate perfbrmance on these items relative to a control group which had the same pairs, but none of which were given strategy aids. Another aspect of this experiment which has been little inves- tigated is the relationship between strategies employed during learning and long term retention of the material. It was reasoned that given strategy aids would increase retention on the aided pairs, and perhaps on those unaided pairs within a list which con- tained aided items. For the present study it was also conjectured that one of the differences between fast and slow learners may be that the latter have not developed the habit of searching for cues in material to be learned. On this grounds one might expect that the administra- tion of strategy aids by E would facilitate the perfbrmance of slow learners more than the fast learners. DESIGN AND METHODOLOGY Subjects Two hundred forty-two students, 130 males and 112 females, served as Ss in this experiment. The students were from three hth, 6th, and 8th grade classes in a suburban community. subjects were tested in their classrooms as a group. Table 2 presents character- istics of the classes at each grade level. Subjects were elimi- nated from the analysis if in the Opinion of the teacher they had severe emotional problems or were retarded. All Ss were naive with respect to paired-associate (PA) learning tasks. Materials Sixteen low meaningful (L-M) and sixteen high meaningful (H-M) disyllables were selected from Noble's (1952) list. The L-M items were used as stimuli and H-M items as responses. These low-high pairs were divided into two lists of eight pairs, one designated the practice task list and the other the criterion task list. An effort was made during the construction of pairs, and during the assigning of pairs to lists, to avoid any obvious association be- tween the items in a pair or among the pairs of a list. The mean value of the stimulus and response items were, respectively, 1.91 (s2 = .58) and 7.hh (s2 = 1.93) for the practice task items and 1.23 (82 = .02) and 7.h8 (s2 = 1.h1) for the criterion task items. The criterion list is presented in Table 3. Each stimulus- response pair and each stimulus (without the response) was photo- graphed on separate slides for visual presentation. A Kodac 700 - 13 - - 1h - wnwwm m :0 ofl< wamuw Hz. m.ma NH ma mN mnfiwo : so pee :umnw N:. >.ma 2H m NN w Hoopooo 0.0 mm. m.mH m ma wN mpfiwm m :0 pw¢ wnmnw mN. >.HH :H :H mN mafiom : so owe :nmum :N. w.HH Ha ma wN w Houpooo one >2. m.HH ma NH mN mpflwm w co ow< mums: mN. m.m ma HN :m mpfiom : co pfi< :umn: mm. w.m NH NH :N : 85:8 0.: mm . N. .m 3 3 pm scepfiocoo pmmfipmona m ow¢ owe Geog moHsEmh moawz o Ho>on N compo mommwao mowz cop Mom accepwcwfimom pomSpooaa pow.moflpmwnmpowmono eageom N mgm¢9 - 15 - TABLE 3 Paired-Associate List and Strategy Aids for Criterion Task Criterion Task Strategy-Aid NEGLAN MEARDON SAGROLE VOLVAP LATUK BODKIN TAROP ZUMAP LEADER INSECT MONEY JEWEL OFFICE WAGON 'DINNER KENNEL "Negro leader" "Meadow insect" "Role of money" "valuable jewel" "Late to office" "Book in wagon" "Tar for dinner" "Zoos have kennels" - 16 - carousel projector with a Lafayette T-2K automatic timer was used for presentation of the slides. Procedure Measures on four separate tasks were obtained from each indi- vidual: (1) a practice task, (2) a criterion task, (3) a strategy report task, and (h) a retention task. The practice task.was ad- ministered 2h hours before the criterion task. Strategy infor- mation was collected immediately after the criterion task, while retention data were obtained A8 hours later. All Ss were naive with respect to the testing schedule. The practice task was given to insure that everyone understood the nature of the criterion task and to control for warm-up and learning to learn effects, which have been shown to confound the results of similar studies (e.g. Mattson, 1965). Another purpose of the practice task was to assess comparability of groups. This was necessary to insure that differences on the criterion task could be attributed to treatments and not to an initial difference in the ability of the groups. The task also served as an instru- ment for defining slow and fast learners for the criterion task. Thus E was able to determine the differential effect of treatments on slow and fast learners. Finally, the practice task gave 85 an Opportunity to become acquainted with the concept of strategies before the collection of strategy information. A pilot study showed that hth grade students had difficulty interpreting the question, "What tricks, if any, did you use while trying to learn - 17 - these pairs?" Consequently, a great deal of individual attention was required if 83 were to adequately express in writing how they had attempted to form an association. It was reasoned that ac- quainting $5 with the concept of strategies or tricks would reduce the amount of individual attention required and increase the va- lidity and reliability of their reports on the criterion task. At the beginning of the practice task Ss were given test book- lets and instructed to fill in the face sheet. Immediately follow- ing the face sheet a sample test page appeared. The E read the instructions and used this sample to help explain the task. Sub- jects were told to study the pairs of items as they appeared on the screen during each learning trial, and on test trials to circle the word in their test booklet they thought went with the stimulus item. Complete instructions are given in Appendix A. For the practice task, learning trials were presented at a A second rate with a 5 second intertrial interval. For the test trials, however, the timer was switched to manual control so that all Ss had suffi- cient time to respond. Exposure time for each test item was ap— proximately 10 seconds. Four learning trials were alternated with four test trials. Each of the eight pairs were presented sepa- rately for learning trials, as were the individual stimulus items for test trials. Duplicate copies of each slide enabled E to ran- domize slides on every learning and test trial and thus avoid pos- sible serial position effects. A recognition procedure was used for test trials. The test booklets contained 32 pages with the eight responses randomly - 18 - presented on each page. Odd numbered pages were on yellow paper, and even numbered pages on white paper so that E could call out the number and color of the page for every test item. A quick glance around the room insured that every subject was on the right page. At the conclusion of the task 83 were shown each pair sepa- rately and asked if they had used any tricks while attempting to learn the pairs. Three different students were selected to respond fOr each item. This procedure acquainted $5 with the concept of strategies for the purpose of improving the collection of strategy information on the criterion task. The criterion task was presented 2% hours after the practice task. The procedure was similar to that of the practice task with the following exceptions: (l) a different PA list was used, (2) subjects were run for five trials, (3) exposure time per item indi- cated that the task was somewhat easier than expected, and (h) spe- cific instructions were given for the respective treatment condi- tions. Criterion treatments were randomly assigned to the three classes at each grade level. Table 2 presents the respective treatment assignments. No strategy aids were given for the control treatment. Thus, performance of these groups provided a standard against which to judge treatment effects. For the E-h treatment, however, E gave strategy aids on A of the 8 pairs. The aided pairs are the first h pairs presented in Table 3. Aid was given simply by having E verbally state the strategy for each pair on learning trials one and two. For the E-8 treatment, aid was presented - 19 - verbally on all eight pairs on trials one and two. These strategy aids are presented in Table 3. The aids employed in this study were selected from those shown to be effective in the pilot study. Instructions for each treatment are presented in Appendix B. In brief, the control groups were told that they were going to do the same kind of task they did the day before, but with differ- ent pairs. The E—h groups were told the same thing. In addition, they were told E would give them a trick on h of the 8 pairs which might help them learn those pairs. The E-8 groups were instructed similarly except that E said he would give a trick for all 8 pairs. 33 were told they could use these tricks or not as they pleased. A recognition procedure was employed for the criterion task. The test booklets contained ho pages with the 8 responses ran- domized on each page to avoid possible serial position effects. Again yellow and white pages were alternated to aid in the adminis- tration of the task. After completion of the criterion task, E passed out a booklet which had each pair printed on a separate page. 85 were instructed to write down, to the best of their ability, how they had attempted to form.each association. If they used a trick they were to state what it was. If Ss in the aided groups used E's tricks they were to write these down as given. Subjects were given approximately 90 seconds per pair to report a strategy. Complete instructions for this task are presented in Appendix C. Using a recall method, retention data were collected h8 hours after the criterion task. 83 had not been told there would be a -20- retention test on the criterion pairs. Each stimulus word was pre- sented separately on a blackboard for 15 seconds. Subjects were instructed to write down both the stimulus and the word they thought went with it. Complete instructions for the retention task are presented in Appendix D. Three experimenters conducted the testing for all experimental procedures. The pilot study revealed that this number could ade- quately handle the group testing procedures. RESULTS Developmental Analysis To investigate performance on the PA task deve10pmentally, several analyses were performed. The hth grade control (h-C), 6th grade control (6-C), and 8th grade control (8-C) groups in the cri- terion task were selected for study, since these groups were not confounded by treatment effects. Figure 1 presents the criterion task acquisition curves for these groups. Mean total correct re- sponses and variances for h-C, 6-C, and 8—C groups were, respec- tively, 20.93 (s.2 = 75.23), 27.13 (s2 = 67.51), and 29.01; ($2 = 50.63). A.one-way analysis of variance among total correct responses for the three control groups yielded a statistically sig- nificant F ratio (F = 7.812; df = 2,80; p < .01). Individual com- parisons showed that 8-C and 6-C were significantly different than h-C (p < .01), but not significantly different from each other. In short, ability to perform on the task increases at a decreasing rate among elementary school children at these ages. To investigate the hypothesis that associative strategies may be related to this increase in performance with age, four judges independently rated the verbal reports of the h-C group. Each sub- ject was then assigned a total strategy score based on the sum of strategy ranks for all 8 pairs. The Kendall coefficient of concor- dance (W) among all four raters on total strategy score was .98 (see Siegel, 1956, pp. 229-238). Since inter-rater reliability was extremely high only one rater was selected to rate 6th and 8th - 21 - MEAN CORRECT RESPONSES Fig. 1. Acquisition groups on criterion task. ,0 0’. ........._. 4 — c 3 4 ' 5 TRIALS h— curves for 4, 6, and 8th grade control - 23 - grade verbal reports. Total strategy level scores were subse- quently computed for Ss in each of these groups. Table h shows percent frequency of strategy level use for each control group. It reveals an increase in high level strategies as age increases. Mean total strategy scores and variances for h-C, 6-C, and 8-C groups were, respectively, 28.h8 (s2 = 170.33), 39.96 (s2 = 52.63), and h3.36 ($2 = 116.37). A one-way analysis of vari- ance on total strategy scores for these groups yielded a signifi- cant F ratio (F = l8.h58; df = 2,80; p < .01). Individual compari- sons revealed that 8-C and 6-C were significantly different than h—C (p < .01), but were not significantly different from.each other. Thus, these data show that frequency of high level stra- tegies and total strategy scores increase with an increase in age. To assess the relationship between strategy used and perfor- mance on individual pairs, mean correct responses for each strategy level were computed. These means are plotted for each group in Figure 2. In general, the figure shows the higher the strategy level the better the performance. In order to determine whether this positive relationship held for total strategy scores, Spearman rank order correlations between total strategy scores and total number correct responses were computed. The correlations for h—C, 6-C, and 8-C were, respectively, .5h, .61, and .h5 (p‘< .01). In short, these results indicate that a significant amount of variance in Ss performance on the PA task can be accounted for by their ver- bal reports. - 2h - TABLE h Percent Frequency of Strategy Level Use for h, 6, and 8th Grade Control Groups Strategy Classification Grade Level 1 2 3 h 5 6 7 h 29 19 7 9 3 1h 19 6 h 19 5 9 6 2h 33 8 5 6 11 3 6 35 3h _ 25 - .xmmu downouwuo so maaouw Houucoo How zwmumuum hp momcommop uoouuou amass: amwz .N .mHm zo_._.om._.om._.mH xwoumuum mo coHuocsm m mm COwudmumu co mmmcommou uomuuoo ucwoumm .m .wwm zo_._.092um _ A o. n v m m _ > \. v \ I . x a 1» media l......| \ a, x I 3880 8251...... \ , t \ ’ 0 $25 \ x \x ‘\ \\. I: a o s \.\. .I.’¢ .\ b ~ . .a \ \ o\ la . \ ~ ~.\ ’li‘ \\\ s\ I ‘V \. a. \V .f \x \ I . ~ \ ./. .\~ . I. x x .\ Io ~ .\\\\ \b/l . x .V ., I s I I, ‘ é ON on O¢ on ow Oh Om Om NOLLNBISH - 1038800 1N3083d - 32 - the relatively small frequency of strategy use at these levels, and to the restricted range of retention scores. A one-way analysis of variance on total retention scores for the three control groups yielded a significant F ratio (F = 7.213; df = 2,80; p <:.Ol). Individual comparisons revealed that 6-C was significantly different than h-C (p 4:.05), 8-C was significantly different than h-c (p < .01), but 8.0 and 6-0 were not signifi- cantly different from each other. Mean correct responses and vari- ances on the retention task for groups h-C, 6-C, and 8-C were, re- spectively, 3.93 ($2 = 6.76), 5.50 (s2 = h.70), and 6.29 (s2 = 7.99). The 8-0 'group retained the most, while 6-0 retained more than h-C. Rank order correlations were also computed between total strategy scores and total correct responses on retention. The correlations for groups h-C, 6-C, and 8-C were, respectively, .57 (p <.Ol), .59 (p < .01), and .32 (p < .05). Thus, it appears that a significant amount of variance in retention of PA.material can be accounted for by verbal reports on acquisition. To summarize, the results of the develOpmental analysis showed that elementary school children could effectively verbalize about cues which they used to form.associations. In addition, the data revealed that acquisition and retention scores increase with age as do total strategy scores, and that better performers used higher level strategies at each grade level. Moreover, females performed better than males at every grade level and used consistently higher level strategies on the PA task. -33.. Analysis of Treatment Effects Another aspect of this study was to determine whether strategy aids given by E would facilitate performance of Ss in the elementary grades. To establish initial comparability of groups at each grade level, a grade level by treatment analysis of variance on total correct responses on the practice task was performed. (Treatment designation refers to that which the group received on the criterion task.) This analysis, summarized in Table 9, revealed a significant grade level effect (p <5.01) and an insignificant treatment effect. There was, however, a significant interaction (p <1.05). Table 10 presents mean correct responses and variances for all groups on the practice task. Examination of this table shows the performance of the hth grade E-h group (h-E-h) was superior to the other two hth grade groups on the practice task. A.one-way analysis of variance and individual comparisons on performance at the hth grade level es- tablished this group as the source of interaction. At this level the E-h group was significantly different than both the C and E-8 groups (p <:.01). A geometric interpretation of this interaction is presented in Figure h. Thus, with the exception of group h-E-h, all groups proved to be comparable at each grade level. To assess the effects of giving different ambunts of aid, a grade level by treatment analysis of variance on total correct re- sponses on the criterion task was performed. This analysis, sump marized in Table 11, showed that both main effects were significant .(p < .01). The interaction was also significant (p < .01), which may be attributed to the initial superiority of the h-E-h group. - 3h - TABLE 9 Analysis of Variance on Practice Task Scores Source DF F Grade Level 2 l9.55** Treatment 2 1.79 Grade Level X Treatment h 2.81* Residual ms 233 (h2.53) * p <:.05 **'p <:.Ol Mean Correct Responses and Variances for Practice Task - 35 - TABLE 10 Treatment Condition Grade Level 0 E-h E-8 h x, 16.52 22.h2 16.76 8 33-95 h2.h3 h8.31 6 352 20. L6 20. 50 21.75 s Minot 52. 1+2 71.01 8 362 25.18 21+.77 2h.80 s 2h.97 37.90 23.75 .xmmu mowuomnd o£u so cowuwpcoo ucmEummuu fiuwa Ho>mH opmum mo cowuomuoucw mo cowumuouauoucw owuumEomw ¢ .¢ .wfim mF2m2FoH opmuw mo COwuomuoucw wo newumuouduoucw owuumaoow < mkzm Ekdmmk - ho - at mo