I!“§§-.,"-""~‘ 1, 1m: “ 53 if: CO W31 {éUTEOH C2;- i'ER‘iTELL mun” : (‘2.5 ‘.$‘.'k: "acme? MC.:::L&.:21‘ u ‘2 {5"} REASG. {LN-3 :43 ‘PSY f é‘fOLGGY r 5’1 . 7' hmfi Thesis {ear rho Dogma of M. A. CHEGAN STAT" COLL LEGE Vici‘er D. Sama W52 v22 51:! U IRES/s 0-169 Date This is to certilg that the thesis entitled J’ALA -; 4h72Lli‘ use i0 11 sexy“; JC 14* r}, _ ct: :3, 7;:‘0513“ in ipyé}olg y presented bl] up“ er In A J1 has been accepted towards fulfillment of the requirements for " - (3".3131L‘ ' ‘I I‘ ' degree in Major profee CONTRIBUTION OF INTELLIGENCE TO THE LEARNING OF VOCAIBULARY. FACTS AND MONING IN PSYCHOLOGY By Victor D. Senna. A THESIS Submitted to the School of Graduate Studies at Michigan State College of Agriculture and Applied Science in partial fulfillment of the rOQuiremente for the degree of MASTER OF ARTS Department of Psychology 1952 TH. Fa: ACKNOfiLEDGMENT Grateful acknowledgment is made to Doctor Donald M. Johnson for hie advice and neietance in the course of this research ****#*#******** ¢*******#**** ********##* ********* **#**** ***#* ‘1‘ t mm 01' CONTENTS Zen InmchIoNOOOOOQOO0.000.000.0000.0.000.000.00000000 eeeeee o... 1 “W or mPROBmOOOOOOOOOOO...0000.00.00.00.0.00.0.0... 1“ 80mm or MUSE I! “IS RESEABWOOOOOOOOOOOOOOOOOeeeeeeeeo 15 DECRIPTIOI 01' m MEeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeoeeo. 16 VOCABUI-ABYH............... eo00000000000000000000000000. 16 Immeoeoeoeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 17 MOMNGooeeeoeeeeeeooeeeeo.eonOOOoeeeeooeeeoeeeeeeeoe. 17 “ERICA! COUNCIL OF NATION PSYCHOLOGICAL “WIDE” 17 WSOOOOOOOOOOOOOO00.000000000000000.0000000000000000000000. 21 STATISTICAL TMIQUESOOOOOOOOO0000000000000000000000000. a guns I] GWOOOOOOOOOOOOOOOeoooeeeeeeeeeeeeeeoaeeoeo. 21 GWIONS 3mm Infmmnm MTSeeeeeeeeeeon 23 INIERGORBEA'I‘ION 3mm THE SUB-EST5................... 29 CORRELATIOIS WITH THE 1.0.1. PSIGHOLOGICAL ELIMINATION... 31 CORRELATION 01‘ m CHIN SCORES I113 TE] A.C.l............ 35 MSOUSSIOUQOOOO0.000.000.0000.000000000IOOOOOOOOOIOOO0000...... 39 8mmcomLUSIONSOOOOOOOO0.0.0.0....OOOOOOOOOOOOOOOOIO... “ BIBummneeeeeeeeeeeeeeeeeeeeeeoeeeeeeeeeeeeeeeeeeeeeeeeeeeo b5 LIST OF TABLES TABLE PAGE I. SUMMARY OF RESEARCH CORRELATING GAIN SCORES WITH mmLIGmcnOOOOCOOOO0000......O...0......0000000000000 u II. CHANGE OCCURRIHG IN CORRmnATIONS BETIBEN GRADES AND INTELLIGENCE WHEN RATE OF PEN-HESS OF COURSE IS CHANGED 13 III. COEFTICIENTS OF RELIABILITY OF INITIAL AND FINAL TESTS. 18 IV. MEANS. STANDARD DEVIATIONS. DIFFERENCES BETEEN MEANS AND CRITICAL RATIOS OF FIRE—TEST AND POST-TESTS. . . . . . . . . 22 V. COEFFICIENT OF CORRELATION BETWEEN THE INITIAL AND BM smmTSQOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOCOCOOOOO 28 VI. INTERCORRELAI'IONS OF INITIAL AND FINAL TESTS........... 30 VII. CORREATIONS BETWEEN THE THREE SUB-TESTS AND THE A0003. Tom ”D LIMUISTIG SCWES..................... 32 VIII. COREELAHONS BETWEEN GAIN SCORES IN PSYCHOLOGY ADD INTELLIGENCEOOOOOOOOOOOOOIOOOOOOOOOOOOOOOOOQOO00.9.00. 36 FIGURE I. II. III. IV. LIST OF FIGURES GRAPH SHOWING ACHIEVIMM ON INITIAL AND FINAL SUB- TETS m 'wnmroooooooooeeeeeeeoe00000000000090 @APH SHOWING ACHIEVED!!! ON INITIAL AND FINAL SUB- st IN FACTSOOOOCOOOOOOOOO0.0.0.0000.000.000.000. GRAPH SHOWING ACHIEVDIEHT OH INITIAL AND FINAL SUB- M! In momGOOOOOOOOOOeOOOeeeoeoooeoooooeeeeoe WE SHOVIHG ACHIM OH INITIAL AND FINAL TOTAL mT In PSYCHOWIOOOOOeeeeeeeeeeeeeeoeeeoneeeeeeo PAGE 21! 25 INTRODUCTION The statement has freQuently been made that intelligence is the ability to learn. but recently this statement has been Questioned by several investigators. The study reported here examines the evidence obtained by correlating scores on a standard intelligence test vith different measures of improvement in a college course in psychology. The argument that intelligence is the ability to learn is based on correlations between scores on intelligence tests and scores on achievement tests in many branches of knowledge. But obviously achieve- ment at any time depends not only on ability to learn but also on many other factors. such as previous achievement. motivation. persistence. opportunity to learn. etc. It is possible to control some of these factors. particularly opportunity to learn and previous achievement. by the use of improvement or gain scores. The subjects are tested before and after a learning period and the initial score is subtracted from the final score to get the gain score. Gain scores are then corre- 1ated with scores on an intelligence test. Gain scores have been obtained from two kinds of experiments. Some investigators have had their subjects. after taking the initial test. practice for a standard period of time in the laboratory. then take the final test. This type of research has used such tasks as addition. subtraction and cancellation. Others have studied improve- ment in regular college courses. -1- -z- A typical experiment of this type. which was similar to the one undertaken in this study. is by Carlson. Fisher and Young (12). Examina- tions based upon established facts and principles in psycholog were prepared and revised by the experimenters and given as a proficiency examination. mm examination was composed of multiple-choice and true-false statements. The same test was given at the beginning of the course and at the end. The experimenters made use of the scores of 118 students who had taken the Otis Intelligence test. They found that the correlation of intelligence with the pro-test was .556 while that with the post-test was .h29. This is what they write in connec- tion with this finding: While the scores on both pro—test and post-test correlate positively with scores on the Otis. the magnitude of the correlation is greater in the case of the pro-test than in the case of the post-test. It would appear that intelli- gence is more involved as a determiner of scores in the *pre-teet than it is in.the post-test. (13. p. 30) The next step was to correlate intelligence with the gain scores obtained by subtracting the pre-test score from the post-test score for every student. The correlation they obtained was -.039 which definitely indicates a lack of significant relation between gain scores and the Otis scores. They draw the following conclusion from their results: There are several possible interpretations of the results. One of these is that the proficiency test which we used was largely a memory test. and therefore did not measure increased understanding of principles. A further interpretation is that improvement following tuition is largely dependent upon non- intellectual factors. Among the more important of these. we would expect to find interest. motivation. opportunity to stu- dy. study habit. personality traits and emotional conflict at -3- the time of study and at the time of taking the examina- tions. These factors call for attention to the importance of non-intellectual processes in accounting for differential improvement in achievement. (12. p. 33) In spite of evidence of this kind. numerous authorities identify intelligence with the ability to learn. But with all these assertions. as Vbodrow says. “no one. so far as the writer is aware has marshalled the evidence in support of the view expressed in.these Quotations.“ (bl. p. lh9) On the other'hand numerous studies have appeared inter- mittently during the last three decades in.the psychological journals bringing out the fact that gain scores usually do not correlate with intelligence. The results of this kind of research up to l9h6 were summarized by Ioodrow (hi). The most pertinent results are displayed in Table I. ‘ However. among the similar studies undertaken. it has been found that in some the gain scores correlated with intelligence. Although these correlations are rather low. they are still significant. This would naturally throw a little doubt on rejection of the notion that intelligence is the ability to learn. it shall new review some of the studies which indicate that the relation between intelligence and gain scores is still an cpen Question. To the best knowledge of the writer. the latest work undertaken in this field was by Simrall (27). who tackled this problem in 191.7. From the popular notion that intelligence is the ability to learn. she drew a list of the implications that could be deduced from it. named imam Hoonoa 9.5." SB No 3... mo. on. .n.o.< boaoam flute 233333 an om. 3.: no... ooaooofiafldo: ma. nu. ma. eauaomno one» No. no. we. ooaouoznfigz .3... i... «H... node-2.88 :3 on. no. we. wag tam seam mm. mm. ma. c9333 mum.“ at: S S. i. 8. :25: . 333333 Snoop on .383. mo. 9. an. Ooaocoaom 83 .323 3H 3. R. 3. :8 3.333% .3on upcoucue .eao ecopuacha nomads made use mama omofioo on 3. an. «an: E .3... Stop...” 3:33. Susan 33 Honda :33 ion soon. .3.» one a a a eocowaaaepmn aaoae>edmo< enmeaauoawfl nogwuaaau my; mamoom Eda aznaammoo mega ho 535» H flnmaa unseen». omofioo m3 one; man. on“. :8 most?” gonzo announce Hama emoaaoo omH 300.: enna< awoaomoaem nowmuuhq 3... soaawdaeoflm coco mH.t you haoaea .emu nH.t moaaeHHeeceo mm. knees: enema dosed» 323: 2. 333336 oooo SS eweHHoo ooa Nu. newsman mound: omepmmm S. S. D. .d mm. m“... .3... :H on. no. i. .d .353 . «53.x S. on. n... .A A3. :33 or: no... mm. S. .d . ao.t am. mm. .A. ueoaw oedema 353.... no... am. an. .d 388m $3 ouofioo on no... em. on. .a .394 233 93 neoofiooo: Susan 38 Anon Hogs: 2.9 noon. a...» one u a a coconuaaeamu asesepeamo<_ wanesaaeann A QHDHHHBOO H HHMHH :3... H33 owed mm. mm. 320.3% 352.: moo... mo. 3. 33m omofloo OS 80... S. 3. .9104 .g bagpaoop mm? huoaonoham madam uuaovdpu mo... m3. mm. puma gamoouom mzma 382 %3 mm an. S. :8. :8 :3. 233m 223m RN. nan. $3.:on mmo... moo... fifiwfi NS. «9. .325 .524 08. mom. 23°.ch .524 :3“. 54. 529.30» 9.3 m3. was. 38 3‘33“ sauce: «3 33¢ .43 .fin 193$ 53 Hana 23%: :3 :3 use» on.» u u u ounowuaaopnu anoaohounod nuaoaauomnn codadauoo H sagas -7. She was able to destroy all these deductions with the data she obtained. However one of the shortcomings of this study is that no endeavour is made to explain the differences obtained in the correla- tions with the gain scbres. For the perceptual test. this correla~ tion was found to be -.O79. while with the spatial test it was found to be .277. The latter coefficient of correlation would be significant at the 1% level: nevertheless no comments are made. The difference between the two correlations is almost .35 which is not necessarily negligible. Since Simrall states that there is no relation between intelligence and gain scores. it should be expected.that all correla- tions found should be insignificant. The question which the author should.have asked herself is the reason for such disparity between the perceptual and spatial test. Connected with this difference found in the correlations of the gain scores. further data may be given that might be significant. Correlation of the initial and final scores with intelligence were found to be as follows: r with initial r with final Gain Spatial .60 .61 .28 Perceptual .59 .h9 -.08 In one case the correlation with intelligence remains the same, while in the other case the correlation drops. Would the difference in the gain scores correlations have any connection with this finding? All these Questions will be given some interpretation at the end of th‘. E papa re ~8- Another important point that could be raised in this experiment is whether the gain scores were the best the students could do. Sim- rall assumes that this is the case. This moot point is disposed of with the following lines: The instructions given to the subject were designed to stimu- late satisfactory motivating conditions. The results of the investigation indicate clearly that the behavior of the sub— Jects was relatively constant throughout the experiment (27. p. 32). Such a cursory treatment of an important factor is Questionable. No doubt the correlation between gain scores and intelligence is pretty low. but exceptions can be found. The first study in Table I. by Johnson (21). differs from all the other studies since he found that the improvement scores for mirror reading correlate higher with intelligence than with the final scores, .46 and .34 respectively. From his results. he draws the following conclusion which is opposite to the one reached by woodrow. Simrall and others: They show that there exists a fairly positive relation be- tween the ability to become efficient at learning to read inverted.print and intelligence as it is measured by the usual group test._ It is interesting to note that it is not the absolute amount which a person reads that is most impor- tant in this connection but rather it is the rapidity of learning. acQuiring new connections that is most closely related to mental ability (21. p. 541). He furthermore illustrates his point by comparing two curves constructed on the basis of the scores obtained by 30 students above the average in intelligence and 30 students below the average. The slope of the superior group was found to be more accentuated thus -9- showing that the superior group gained more though the initial dif- ference was small. Another study. by De Weerdt (13). was carried out with school children in the fifth grade. Reading (Chapman—Cook Speed Test) shows the greatest correlation.with intelligence. The gain score correla- tion reported is .56. the highest one recorded so far. For the sub- stitution and addition tests. the correlation of the gain score is slightly higher than the one obtained with the final scores. All the correlations for the gain scores were found to be positive except for the cancellation test. Taking the ratio of improvability of the ten highest scoring pupils and ten lowest scoring pupils. De Weerdt found that it is definitely higher for the more intelligent group. except the ratio for cancellation. Criticizing the inadeQuacies of these correlations. this is what he writes: A very few individuals within a group who achieve large incre- ments in comparison with their original low initial scores or those initially high who achieve small increments because of being near their physiological limit exert a disproporb tionate influence upon the results in the use of the correlan tion formula. With the additional factor of favorable and unfavorable attitudes toward certain tests which are like tasks previously met. the correlation formula as means of stating relationship incorporates more than the simple relap tionship between the actual content of the tests involved... It is evident then that the general test of intelligence does indicate capacity for improvement but the general test does not indicate how much improvement we may expect in a specific function. This at once suggests the practicability of measuring the capacity for improvement directly in the specific function rather than attacking it by inference through a general test (13. p. 557). One of the last studies in our summary table is the one carried out by WbOdTOV with fifth. sixth and seventh graders. Be. more than -10- any other'psychologist. studied extensively the problem of learning and its relation to intelligence with the aid of factor analysis. He no doubt gave good evidence that to identify intelligence with ability to learn as measured by our actual tests was Quite erroneous. However to the writer's point of view. he does not seem to have gone far enough in the development of his problem to insure complete and satisfactory treatment. It would be sufficient to comment here on woodrow's find- ings with the grade school children. In this study as in many others. he found that there is a “lack of significant relation.between change in score and I.Q. in grades beyond the fifth" (ho. p. 153). The writer computed the averages of the correlations in the three grades tested in this research and found the following: Correlations between gain scores and intelligence Fifth Grade .402 Sixth Grade .190 Seventh Grade .100 Apparently from the above results we can infer that the older the child gets. the less he uses his intelligence in class achievement. It can be assumed that by the time a child gets to the sixth and seventh grades he gets new distractions that do not exist in the lower grades. Girls might become interested in the way they look and boys would probably want to become football stars or train engineers. and this may cause such a big drop from .h02 to .100.- Woodrow does not elabor- ate on this difference but mentions it in passing in the following paragraph : *Italics are the present writer's -11- It should of course be kept in mind that the apparent deter- mining conditions of gains in school. that is. school atten- dance for a stated time in a given school class. although a more natural or life-like condition.than practice conducted in a.1aboratory is in reality. a complex of variables which are not kept constant for all pupils. Some variables which readily come to mind are irregularities in attendance at school. interest and liking for the subject. health. study habits. level of aspiration. various home influences and a complex of conditions not too well understood. determining what is rather vaguely called.motivation (#0. p. 156). Although.the next experiment which will be reviewed did not deal in gain scores. it would be interesting to mention it here on account of the contribution it can make to our'problem. Burtt. Chassell and Batch (11). three psychology instructors at Ohio State University. arb ranged to teach elementary psychology to classes selected on the basis of intelligence. Each instructor taught a class of high. medium or low intelligence and one or two control classes of heterogeneous intelligence. Correlations were calculated between intelligence and grades after the first semester and second semester as well. However. during the second semester a new variable was introduced for the high and medium intel- ligence classes. They were all Ipushed". that is. they were given longer assignments than usual. It was found that the most intelligent finished the course sooner. The lowbintelligence class maintained normal rate of progress. The theory implied.by the authors in.their investigation is that if each student is stimulated to do his best the correlation between intelligence and academic marks could be raised. Table II shows the results of the change in speed of instruction. As we can see the high intelligence group and medium one raised their correlation with intelligence from .32 to .73 and from .15 to .75 respectively while the controls did not show much change. The group -12. of low intelligence on the other hand.had their correlation drop from .67 to .22. The authors give various reasons for this fall. one being that during the first semester. the students were forced to work up to maximum intellectual capacity. Commenting on these results. this is what the authors have to say: The obvious conclusion is that from the pedagogical point of view nothing is gained by grouping students of superior intel- ligence for instructions under the conditions of this experiment unless they are forced to cover the material of the course at a more rapid.pace than average. Not only are they obviously able in.the latter case to cover ground more rapidly but the individual student comes near working at his maximum intellectual efficiency (11. p. 161). TABLE II CHANGES OCCURRING IN CORRELATIONS BETWEEN GRADES AND INTELLIGENCE WHEN RATE OF PROGRESS OF COURSE IS CHANGED (From Burtt. Chassell and Batch) (11) Groups First Semester Second Semester Selected high intelligence .32 .73 Control for high group .146 .hj Selected medium intelligence .15 .75 Control for medium group .38 .15 Selected low intelligence .67 .22 Control for low group .6“ .58 STATEMENT OF THE PROBLEM Now what does this array of research tell us? One thing that is Quite clear is that we do not have uniformity with these studies. Some achievements correlate high or low with intelligence initially and simi- larly at the final test. In some cases the correlation goes up and in othere it goes down. lurthermore correlations with the gain scores rs- nain relatitely high when the correlation with achievement improves in the final test. In other instances. it is nil. although both the corre- lations with achievement remain unchanged but high. All these differences naturally need explanation. Although we have found in most studies that intelligence is not the ability to learn. it would not be wise to con- sider the matter settled since in certain instances we have obtained opposite evidence. The Question arises. therefore. whether the same results would be obtained if different kinds of tests were used. for example. reasoning in psychology. To answer’these numerous Questions. the present study has been set up by giving separate tests of different aspects of the same course in order to see to what extent this differentiation affects the correla- tions with intelligence. -l#— SOURCES 0]" DATA USED IN THIS RESEARCH The data of the present students were obtained.in.part from an investigation carried out by Smith and.Johnson (unpublished monograph) * of the Department of’Psychology at Michigan State College. The purpose of their study was primarily to evaluate the effect of democratic teach, ing procedure on students' attitudes and achievements. I One hundred and sixty-two students. mostly sophomores. were given an achievement test during the first two periods of the term. Then.they were divided and matched according to the experimental design reQuired by such a research. .Lt the end of the course. the students were again tested with an eQuivalent test of psychology. There were therefore pro-test and.post-test scores in elementary’psychology available for the purpose of this research. The 162 students were divided into four classes. and each instructor taught two classes using the democratic procedure in one and the lecture method for the other. Since these dif- ferent methods of teaching produced only small effects upon the scores. the gains being approximately eQual for the four groups. the data available can therefore be used as coming from a single universe or pepulation. As pointed out earlier in this paper. research has indicated that gain in scores usually does not correlate significantly with intelligence. The novelty of the present study is that different kinds of achievement tests were used. In this way. correlations between intelligence and -15. * Will be published in Psych. Vono. 1953. -16— 'various kinds of achievements were obtained. It was pointed out in the introduction that there was no attempt in the studies made to explain differences obtained with different types of tests. nor why some gains correlated.higher with intelligence than others. Description of the Tests Out of a large pool of test items devised by Smith and Johnson. all multiple-choice items with five alternatives. three different tests were constructed. Each was made up of sixty items covering respectively: 1) vocabulary. 2) Factual knowledge. 3) Reasoning in psychological mat- ters. These three tests were divided in two forms of approximately eQual difficulty. Item analysis was carried out on the basis of scpres obtained previously from a similar*population of students. Thus one form was used.as the pro-test. and.the eQuivalent form was used as the post-test. Vocabulggz. The following is an example of the vocabulary items: SET 1) preparation 2) reaction 3) pathways h) choice 5) fixation. From Table III. we can see that this test when initially given had a reliability of .h6. which Jumped to .76 on the final test. The odd- even method was used. corrected for length by the Spearman43rown formula. Comparing the two reliabilities for the two forms. we can see that the post-tests show significantly greater reliabilities than the pro-tests. This naturally should be expected on account of the fact that many responses in the pre-test must have been chance responses. since stup dents had to rely upon inspiration based on their scanty contact with psychology in the past. M! ts. The Questions on factual knowledge depended largely on memory of specific facts which appeared in the textbooks. As an exam- ple. we can Quote one Question: Raising body temperature by electric current is a specific treatment for (l) Psychosomatic illness (2) Psychoneurosis (3) Brain syphilis (Lt) Manic-depressive (5) Paranoid schizo- phrenia. As shown by Table III. the reliability of the pre—test was .51 while it became .68 for the post-test. Reasgnizg. The third type of sub-test was made up of items which reQuired applications of learned principles to new problems. Naturally a certain amount of vocabulary and factual knowledge are assumed to be necessary to answer these Questions. An example of the reasoning item is as follows: Under which of these conditions will the sex drive in rats be stronger than hunger or thirst? (1) Give plenty of food and water (2) Place animal in obstruction box when young (3) De- prive animal of food. water and sex for two days (’4) Deprive animal of food and water for one day (5) Keep male and female together constantly. This particular test showed the greatest increase in reliability from the pre-test to the post-test. being .33 and .70 respectively. The American Council oLEducatign Psychological Examinatigg. The initial and final test scores were correlated with the total scores on the A.C.E. and with the linguistic part of the A.C.E. One hundred stu- dents' intelligence scores (83' sophomores. ll freshmen. 5 Juniors and 2 seniors) were used in this study out of the 162 of the original study. This was the number of A. 0.3. raw scores that were obtained from the Examiner's Office of Michigan State College. Since the data of these TABLE III COEFFICIENTS OF RELIABILITY OF INITIAL AND FINAL TESTS N = 162 (from Smi th 8: Johnson) Sub-Tests Pro-Test Post-Test r r Vocabulary . ’46 . 76 beta . 51 . 68 Reasoning . 33 . 7O -18- -19- tests are printed in booklet forms showing decile rankings. the writer had to refer to the original records in order to obtain original scores suitable for correlational purposes. Although the A.C.l. examinations were taken.by the students a.year or so before the actual experiment for the majority of them. the use of the same scores is Justified on account of the well-known reliabilities. .95 for the linguistic score and .97 for the total score. The reliabilities do not seem to change much after some time. Livesay (22. p. 67). after giving the same tests to seniors who had taken it four years earlier. found that the reliability was still as high as .88 for the total test. Regarding the validity of the A.C.E. scores. Berdie and others (8). have found that the range of correlations with grade-point average is from .25 to .66. Furthermore Garrett reports that the A.C.E. corre- lated.more closely with college average than did the scores of other intelligence tests (17. p. 129). Reverting to Berdie's survey covering thirteen colleges. universities and teachers' colleges. Michigan State College being one of them. it was computed by the writer that the average score obtained for the linguistic part of the A.C.E. was .67.2h. The average obtained for the present sample of one hundred.M.S.C. students was rather close. 66.79. In this study the total scores on the A.C.E. were correlated with the pre-tests and post-tests. and then the correlations were computed with the linguistic scores only. It was found. however. that the latter -20- could be used independently. Berdie and others point out in this con- nection: In a surprising number of cases the L scores yield almost as high or higher correlation than.the total score. .Even in biological sciences. wherein a superficial analysis would sug~ gest a nice balance of Quantitative and linguistic abilities the total scores predict little if any better than the lingu- istic scores (8. p. 811). RESULTS Statistical Technigue . There were for this study twelve sets of data. Four sets comprised the pro-tests. which included the total scores. Four other sets were made up of the post-tests. The last four sets were calculated by subtracting the former from the latter in order to obtain the gain scores for each sub-test and total. The pro-tests and post-tests were all intercorrelated and correlated with the A.C.E. linguistic scores as well as with the total scores. Gain scores were correlated only with the linguistic scores. There were in all twenty- eight correlations. To avoid any possibility of errors. every one was run twice. Gains in general. Before turning to an analysis of the data rela- tive to the main problem of this research. it would be appropriate to examine some general aspects of the results. Table IV gives the mean and standard deviation of each test. as well as the critical ratios be- tween the initial and final tests. we can see that the vocabulary shows the greatest amount of improvement while reasoning shows the least. although Quite significant. It can be inferred from these results that the extent of improvement as indicated by the critical ratios reflect the technical degree of the subject matter. The course in psychology has the greater effect on.the mastery of technical words than on the application of the concepts subsumed in their definitions. However. one criticism that could.be levelled at the inference drawn above is -21. TABLE IV MEANS. STANDARD DEVIATIONS. DIFEERENCES BETWEEN ms AND CRITICAL RATIOS OF PRE-TESTS AND POST-TESTS N = 100 Sub-Tests Mean of 8.1). Mean of 8.1). D. G.R. Pro-Test Post-Test Vocabulary 13. 31 3. ll 22. 72 3. 90 9. #1 18. 72 Facts 11.02 3.25 19.37 3.96 8.35 16.29 Reasoning 15. 66 2. 97 20. 83 3. 1+3 5. 27 ll. 71 Total 39. 89 6. 83 62. 92 8. 20 28. 03 20 . 10 ~22- -23- that the critical ratios may not have any comparative value since the pro-tests are not all of eQual difficulty. But still it is safe to assume that facts could be learned and greater improvement can.be made than could possibly be done with reasoning tests. Learning how to rea- son is very problematic. while there is no doubt about learning facts. The distributions of initial and final scores are represented in Figs. 1. 2 and 3. and for the total. Fig. h. It can easily be seen.that there is a greater overlapping in the reasoning tests than in the other sub-tests. Correlatiog:between inifiial and final_scores. Correlations com- puted between the initial and final test scores are shown in Table v. The coefficient of correlation between the pro-test and.post-test in reasoning is the highest. .90. Learning. therefore. had the least effect in disturbing the rankings of students in this sort of achievement. If .hO is corrected for attenuation. it is found that the estimated cor- relation is .83. which is obviously very high. Eacts hold second place wdth a correlation of .33. and lastly. vocabulary with a correlation of .114. this being insignificant at the 5% level. we can interpret this disparity. as already pointed out. to the fact that the study of psychology has the greater effect upon.the knowledge of’psychological terms. which does not necessarily imply a thorough understanding of the concepts. In the framework of the actual study. the learning of vocabup lary and facts depended more upon motivation or non-intellectual factors than upon intelligence. Reasoning depends more on intelligence both initially and finally. .swsl .uxO>..01 :00 cause s of. zsuiucas . II \. .r Sbflu 3o sins—L...“ 3— IIKVC .92 Ens-K K .0052 .3530: :00 :93... s of. resin-as 5.3; Cu 52...:Unncn .1019 62 5.0% .23! £20.30... :00 mus