CONTEMPORARYEDUCATIONAL PSYCHOLOGY3, 136- 143 (1978)
The Encoding Versus the External Storage Hypothesis in Note Taking JOHN P. RICKARDS AND FRANKFRIEDMAN Purdue CJ/~iwmit~ College students (N = 85) read a passage in which each sentence had been normatively assessed as to its importance to the overall meaning of the passage. Students expecting an essay examination took nntes on sentences of higher structural importance than those anticipating a multiple choice test, even though there was no difference in the number of notes taken or in total test performance. The students took notes on 31% of the passage sentences and such notes were of high structural importance value. Most importantly, note taking seemed to serve as both an encoding device and as an external storage mechanism, with the latter being the more important function. The external storage function not only led to enhanced recall of the notes, but also facilitated the reconstruction of other parts of the passage.
Di Vesta and Gray (1972) have proposed that note taking functions in either or both of two ways, as a technique that enhances “encoding” of passage material and/or as a means of storing material externally (“external storage” hypothesis). The encoding hypothesis suggests that the act of taking notes results in a transformation of passage material. The precise nature of the transformation has not been fully specified, but it likely involves some processing beyond verbatim learning, such as organizing information and/or “sifting out relevant material” [Di Vesta & Gray, 1973, p. 131. The major implication of the encoding hypothesis is that the mere act of taking notes without any opportunity to review such notes will enhance performance. The external storage hypothesis, on the other hand, indicates that notes are taken to store passage information likely in verbatim fashion for later use for recall purposes. Hence, the external storage idea suggests that note taking per se is not facilitative of recall, unless learners are given the opportunity to review their notes prior to recall. In the experiments conducted by Di Vesta and Gray (1972, 1973) they found that a group which took notes (note take condition) was invariably superior in passage recall to a group which did not take notes (no notes condition), thereby providing support for the encoding hypothesis. However, these researchers did not find any support for the external storage hypothesis, since no advantage over a note-taking group was found for a group which took notes and also reviewed them (note have condition). Send reprint requests to Dr. John P. Rickards, sity, West Lafayette. Indiana 47907. 136 0361-476X/78/0032 -0136$02.00/O Copyright 0 1978 by Academic Press. Inc All rights of reproduction in any form reserved.
Department
of Education,
Purdue Univer-
ESCODING
VS
EXTERSAL
STORAGE
IN
SOTE
‘TAKING
137
In a subsequent experiment, Fisher and Harris (1973) reported results which suggested that notes serve both an encoding function and an external storage function, at least on a test of immediate recall. In their experiment, the note have condition exceeded the note take condition which, in turn, was superior to a no notes condition. On the delayed test, however, only the note have condition was superior to the no notes group. Carter and Van Matre (1975) argued that the Fisher and Harris study did not represent a “strong test” of the external storage hypothesis, since it involved conditions in which the notes review period occurred immediately after the passage rather than just before testing on a long term retention test. They further argued that this latter-mentioned experimental arrangement more closely approximated the typical use of notes by students. Using this arrangement for their note have condition, Carter and Van Matre found that the note have group recalled more than the note take group or the no notes groups, including a no notes group which mentally reviewed the passage prior to the criterion test. From these results, they reasoned that in order to benefit from note taking, students must have them for review just before the test. That is, Carter and Van Matre’s results supported the external storage hypothesis, but not the encoding hypothesis. In all previous experiments, including the Carter and Van Matre experiment, no separation was made between recall of material from the notes and recall of material not from the notes (nonnote recall). Separate analyses of this sort seemed important because the external storage idea suggests two possible functions for notes: a r-elzrcrrsnl function, whereby enhanced recall is only due to recall of material from the notes just reviewed. In this case, a note have condition would be superior to a note take condition in recall of notes, but not in nonnote recall; alternatively, areconstruction function suggests that the recall of notes allows learners to reconstruct parts of the passage on which no notes were previously taken. Consequently, a note have condition would exceed a note take condition in both notes recall and nonnote recall. Central to an understanding of note taking behavior is an analysis of the qrrcr1it.vor kind of notes students take. Yet, no study to date has examined this important issue. Johnson (1970) has devised a method for calibrating text sentences in terms of their “structural importance.” This method was developed via a normative assessment procedure which involved asking readers to rate each text sentence as to its importance to the overall meaning or semantic content of the passage being presented. Sentences rated as high in structural importance were the more abstract, general statements (in many instances. topic sentences), whereas sentences rated as low in structural
138
RICKARDS
AND
FRIEDMAh’
importance were often statements representing specific examples and illustrations. Using this method of text analysis, we were able to assessthe quality of the notes taken. This was especially important in view of our interest in the effect of test mode expectancy on note taking behavior. Previous researchers (Hakstian, 1971; Weener, 1971) have found no difference in performance between learners expecting a multiple choice test and those anticipating an essay examination. But, in each of these studies, the only dependent measure employed was the total amount of notes taken. While quantity of notes taken may not vary as a function of test mode expectancy, the quality of such notes may. The present study was designed, in part, to examine this point. METHOD
Subjects The subjects were 85 college students enrolled in an introductory psychology course. Although participation in the experiment was voluntary, the students did receive credit toward their course requirement for such participation. One person from one condition and two people from two other conditions failed to take the delayed retention tests. Thus, we had unequal cell frequencies which necessitated our using unweighted means in our analyses.
Design The design was a 3 x 2 factorial between subjects design. Groups of students were randomly assigned to one of three overall conditions: take notes and review the notes before a test on the material: take notes and mentally review the material before the test; or. not take notes and mentally review the material before the test. These groups were randomly divided and half were told they would be given an essay test over the material, and the other half were told they would be given a multiple choice test. Thus. the two between subjects factors were study strategy condition (note have, note take. and no notes/mental review) and test mode expectancy (essay and multiple choice). In examining notes recall and nonnote recall. the design exployed was a 2 x 2 factorial between subjects design with the no notes/mental review condition deleted from the study strategy factor.
Materials The instructional material was an 80-sentence passage entitled, “Evolution of the Brain.” (SI) of each sentence of the text were Normative ratings of the “Structural Importance” obtained from Johnson. These assessments were derived from a procedure in which 43 college students were assigned to one of three subgroups having the task of eliminating either 1%.M, or +5 of the sentences that were least important to the overall thematic content of the passage. A count of the number of times that a sentence was judged essential to the semantic content of the passage provided a measure of the structural importance of that sentence. A completion test (cued recall) was developed by taking a high rated (on the average upper 80%) and a low rated (on the average lower 20%) structural importance sentence from each of the 16 paragraphs of the passage and deleting an essential word or phrase. The completion test consisted of 16 low Sl fill-ins on the first page and 16 high SI fill-ins on the second page. In taking the completion test. once a subject turned to the second page, (s)he was not allowed to return to the first page.
ENCODING
VS EXTERNAL
STORAGE
IN
NOTE
TAKING
139
Procedure The experiment took place in an ordinary classroom. Upon arrival, students in each of the study strategy conditions were randomly divided into two groups. The instructions given to the students informed them that the purpose of the experiment was to examine the effects of note taking on the learning of written material. The students in the no notes condition received instructions which did not mention note taking. All students were then told that a test would be given 1 week later covering the reading material. One group was told to expect an essay test, and the other to expect a multiple choice test. The students were urged to take notes as if they were reading an article in the library as a course assignment. The three page article was then handed out and the students were instructed to raise their hands when they completed the article. When % of the students in each group were finished, they were dismissed and the others were allowed to leave whenever they finished. All participants returned I week later. At that time, students in the note have condition were given back their notes and were instructed to review them for IO minutes in preparation for the upcoming test. The students in the note take and no notes/mental review conditions were told to mentally review the article for IO minutes. After the review period, the students were requested to recall as much as they could of the passage. Recall was invariably completed within the IS-minute period. The students were then given the completion test with the instruction not to turn back to page I after turning to page 2. The completion tests were finished within I5 minutes.
Scoring All scoring was done in “blind” fashion. The notes taken were scored as to which sentences in the original passage the notes referred. Since the SI ratings of each of the 80 sentences were available, the structural importance of the notes taken could be computed. The completion test was scored using a lenient criterion. The answer did not have to be verbatim to be correct but must have been synonymous with that stated in the text. The free recall protocols were scored according to a method developed by Cofer (1941) and later modified and used by Howe (1970) and by Rickards and August (1975). Basically. this involved reducing each text sentence to the minimum number of words possible without losing the essential meaning of the sentence. To receive credit. a student’s protocol must have contained the designated words or their semantic equivalents. For example. one of the text sentences read, “In general, the higher the animal’s evolutionary status. the more wrinkled its cerebral cortex.” The key words for this sentence were: “Higher evolutionary status, more wrinkled cortex.” Variation in phrasing and synonyms were considered correct if they did not alter the meaning of a given sentence.
RESULTS
Notes Students took notes on 3 1% of the sentences of the passage (X = 25.47 out of 80 sentences). The sentences on which notes were taken (X = 23.24) were of significantly higher SI value than the SI value of the entire passage (2 = 19.35), z = 2.95, p < .002. While test mode expectancy did not affect the number of sentences on which notes were taken (Essay, X = 25.83, Multiple Choice, X = 25.1 I), the group expecting an essay test (X = 24.2) took notes of higher SI value than the group expecting a multiple choice test (X = 22.3), F (1, 52) = 4.00, p < .05.
140
RICKARDS
AND
FRIEDMAN
Free Recall
The 3 x 2 unweighted means analysis of variance for total free recall demonstrated a significant main effect due to study strategy, F (2, 79) = 54.53, p < .OOl. No other effects were significant in this analysis. Simple effects analysis using the Newman-Keuls procedure for unequal groups (Winer, 1971, pp. 601-603) revealed that the note have group (2 = 13.81) yielded more recall (p < .05) than the note take group (X = 5.20) or the no notes/mental review group (X = 4.13), the latter two of which, in turn, did not differ from each other. The 2 x 2 unweighted means analysis of variance for notes free recall yielded a marginally significant main effect for test mode expectancy, F (I, 51) = 3.38, p < .07, with the essay test expectancy gr_oup(X = 8.13) exceeding the multiple choice test expectancy group (X = 6.17). The above analysis also demonstrated a significant main effect for study strategy, F (1, 51) = 62.16, p < .OOl, such that the note have group (X = 11.35) was superior to the note take group (X = 2.95). The interaction effect was not significant. The 2 x 2 unweighted means analysis of variance for nonnote free recall did not reveal any significant effects. Cued Recall (Completion
Test)
The 3 x 2 unweighted means analysis of variance for total cued recall showed a significant main effect for study strategy, F (2, 79) = 21.23, p < ,001. No other effects were significant. Simple effects analysis employing the Newman-Keuls method for unequal groups demonstrated that the note have condition (X = 17.65) was superior (p < .OS)to both the note take condition (X = 13.35) and the no note/mental review condition (X = 10.50). Moreover, the note take group recalled more than (p < .05) the no notes/mental review group. Following this overall analysis, examination was made of notes recall and nonnote recall for low and for high SI cued recall items. The 2 x 2 unweighted means analysis of variance for cued recall of low SI items that had been in the notes yeilded a significant effect for study strategy, F (1,51) = 3.85, p < .05. More specifically, students expecting a multiple choice test (2 = 2.38) recalled more low SI items from their notes than did students expecting an essay examination (X = 1.55). No other effects were significant in this analysis. In the 2 x 2 unweighted means analysis of variance for cued recall of high SI items that had been contained in the notes, there was one marginally significant effect due to test mode expectancy, F (1, 5 1) = 3.15, p < .07, and another due to study strategy, F (1, 51) = 3.75, p < .055. The effect due to test mode expectancy suggested that students expecting an essay examination (X = 5.11) recalled more high SI items from their notes
ENCODING
VS EXTERh’AL
STORAGE
IS
NOl-E
l’.-ZKISG
141
than students anticipating a multiple choice test (2 = 4.10). The direction of this effect is the opposite of that reported above for cued recall of low SI items from the notes. Given that the essay group took notes of higher SI value than the multiple choice group, it is quite likely that at the outset the essay group had more high SI items in their notes, and the multiple choice group had more low SI items in their notes. Hence, these results were not unexpected. The effect due to study strategy indicated that the note have group (2 = 5.15) recalled more high SI items from their notes than did the note take condition (2 = 4.05). The 2 x 2 unweighted means analysis of variance for cued recall of low SI items not contained in the notes revealed a significant effect for study strategy, F (1, 51) = 12.40, p < .OOl. This important result indicated that those who took notes and also reviewed them (2 = 6.73) recalled more low SI items not in their notes than those who only took notes (.? = 4.60). No other significant effects were produced by this analysis. The 2 x 2 unweighted means analysis of variance for cued recall of high SI items not contained in the notes failed to yield any significant effects. DISCUSSION Essay versus Multiple
Choice Test Expectant)
The results of the present experiment confirm those of previous experiments (Hakstian, 1971; Weener, 1971) in finding no effect for test mode expectancy on the number of notes taken or total test performance. However, the present study suggests that qlrality of performance does vary as a function of what type of test is anticipated. That is, those expecting an essay test chose for note taking sentences of greater importance to the overall meaning of the passage than those expecting a multiple choice test. Also, while the total number of cued recall items from the notes was unaffected by variation in test mode expectancy, the essay group recalled more high SI items than the multiple choice group, and vice versa for low SI items. Assuming that essay test expectancy created a greater focus on high level, idea-oriented passage material, perhaps dependent variables testing applications or inferences would have yielded yet other differences in performance between these two groups. The Rc~cotlstruc,tiotl Hypothesis
Coufirtued
The present results support those of Carter and Van Matre (1975) in terms of total free recall of passage material. That is, both studies suggest that students who take and have notes for review will outperform those who only take notes or students who don’t take any notes but do mentally review the passage prior to recall. Separation of total recall into notes recall and nonnote recall showed that the only advantage of note having over note taking was in the recall of
142
RICKARDS
AND
FRIEDMAli
material contained in the notes themselves. It could be, however, that nonnote recall was subject here to a floor effect. A maximum of only 3 items was recalled in this measure. This was likely due to the length of the retention interval (1 week), the difficulty of the material (brain anatomy), and the type of test given (free recall, hence no cues). In another performance measure which did employ cues (the completion test), the results were somewhat different. On this measure, both the note have group and the note take group exceeded the no notes/mental review group in over all recall, thereby suggesting that some advantage was accured by simply taking notes. These results supported the encoding view of note taking (Di Vesta & Gray, 1972). Moreover, they were pertinent to the external storage view as well (see also, Fisher & Harris, 1973). While one interpretation of the external storage idea is that learners will only recall material contained in their notes (the rehearsal hypothesis), another possible interpretation is that the notes recalled will allow learners to reconstruct other passage material hypothesis). (the reconstruction Since the note have group was superior to the note take group in recall of high SI completion test items contained in the notes and low SI items nut contained in the notes, these results support the reconstruction hypothesis. More specifically, it appears that recall of high level material from the notes enabled learners to recall lower level passage material that had not been previously recorded in their notes. A similar pattern of results emerged with regard to another study strategy, underlining. Rickards and August (1975) found that readers chose for underlining sentences of relatively high structural importance, at least when asked to underline only one sentence per paragraph. Moreover, the high-level underlined material seemed to serve an assimilative function for material that had not been underlined. Further research might well explore the nature of the assimilative process associated with each of these strategies for enhancing text comprehension. Such research might provide insight into the assimilation process itself as well as augment our knowledge of the psychology of instruction. REFERENCES CAKIER, J. F., & v.4~ MATRE, N. H. Note taking versus note having.JorrvntllNf’Edrr~Llfionrrl P.~yc/w/og~, 1975. 67, 900-904. COFER, C. N. A comparison of logical and verbatim learning of prose passages of different lengths. Amrriccrn Jorwnul of‘ Psychology, 1941. 54, I-21. DI VESTA, F. J., & GRAY, S. G. Listening and note-taking. Jortr-no/ ofEd~crtiona/ P~ychdogy, 1972, 63, 8- 14. Dr VLSIA, F. J., & GRAM, S. G. Listening and note-taking: II. Immediate and delayed recall as functions of variations in thematic continuity, note-taking, and length of listeningreview intervals. Jor~rncrl of‘Edrtcationa/ Psychology. 1973, 64, 278-287. FISHER, J. L.. & HARRIS, M. B. Effect of note-taking and review on recall. Jor~mol oj Edu~~crtiontrl Psyc~ho/o,~~. 1973, 65, 321 -325.
ENCODING
VS EXTERNAL
STORAGE
IN
NOTE
143
TAKING
A. R. The effects of type of examination anticipated on test preparation and performance. Jo~trrral ofEducrrrionc~/ Reserrrch. 1971, 64, 319-324. HOWE, M. J. A. Repeated presentation and recall of meaningful prose. Jolrntnl r?f‘Ed~crrtional Psychalo,qy, 1970, 61, 214-219. JOHNSON. R. E. Recall of prose as a function of the structural importance of the linguistic units. Journrrl of Vrrhrrl Lrcrrning nnd Verbal Bchcr\,ior. 1970, 9, 12-20. RICKARDS, J. P.. & Auc;rst , G. J. Generative underlining strategies in prose recall. Jorr,ncr/ HAKSTIAN,
of Educc~tionrrl
Psycho/og.v,
1975, 67, 860-865.
P. D. The effects of recall mode and recall interval expectancies on note-taking and recall. In F. J. Di Vesta, N. M. Sanders, C. B. Schultz, & P. D. Weener (Eds.),
WEENER,
Instructional instructiorl. WINER,
Strutrgies:
Multil~clrirrhle
studies
oj’ psychologic~al
processes
relotrd
to
ARPA Semi-Annual Report (Order No. 1269), 1971. Pp. 59-73. B. J. Stutisfic,cr/ principle.7 in rxperimc,n/rr/ design. New York: McGraw-Hill, 1971.