'a I .. 1.11 . .~' ..~ . «.3 , c.4124“ “J”. 1. “u". \f'u C“ .1! §3' 1931“." 1.11.1315! 13.3”“ _.’ ,'5 5' m 1:] .p ‘ 11‘1- W“ “5'11": “1:7 13,1 1’1”?“ ‘ A.» ”11‘1” «1,111 “1:1 3:11-1:17" N BF. 1: 4““1‘ . q 3 "111151“.‘3‘1 _; .. A ‘5;th q 1-“ ”111411. ‘; 33371311 ."r 1 1, 1 ‘ 4V9 Wfléz': J": 7:: v €411: Ju ~1. H '1‘ ‘DV f4 1 ((14:11 1. 1 ,.. “L‘éylnk “3‘31," . / . «1315. ,1 5:1 1, . I . 111:1") ‘ f‘ ‘y ., , , 1.. 111:6 .1. .. 1., I!" .1 ' ”’1'11’1 11‘; l n “HY/‘3‘" ' ,. “‘1-f'T-. - .,‘I - .. . . . . ,. r ”W (an-411.1% V ;..1‘41 3 ~ 11. 1 , .,,. , . .. » .. .1- .. 1.717.411 .1 1 . . 1 - JG 1 . 1 11.62"; h 1 fglef,‘.r“,.':1 , A, ‘ :5 . 1 ' 51.1.3111“: J 1 r .. - J..,I.£.‘i, n.” l 53 21’ 1 1.... , ",4, .- 51’1"“qu”' 71.1.4.» ,'."i @1351..." 31.315 "W“ *""-9r2{r—:‘I {£3113 ' ’3' r KL": .5 ,xS-fg; 4; .,. '1'" «in»; «.4... rr ,'S'L .. A- n ur’ ‘: 1 . ,'. . , , . 1 111 \ nr‘. 1 - H111.) ,' 1- . .-’.u"./.’.1 .. , 1 5" y a 1' 1'1" .yx:r.:;r~~1- |il‘0u'l‘ 2113111 1')“ 9-,: rwb‘v ’1’! ‘11 I .5 ;. *2: r . '1’»- uni-u 339311314521; Malaria?“ $1.139!" 1' "t'r 3A!- ”'01! "fl'bfl'm" . ”11;”?!- r- : I. V " 'l V VI ID rt] ' .,,,,.4.,..1.‘ 1 1. .1 . . .1 1g ...'.11.. . z . ' 1 “1:11.1511."Varmzmrmf .11 1.”;1.’ 1.... 21:11 ,'. “‘1 nm 1' ‘ m i 1 . v 11‘"- vn ‘3‘ v M4 .1 111-1119;: 1 VI- (vu- 1.’"1—‘1m..“.'1- 1:." 1111.11,, 1'13, , , 1M . m: I 1111' .- aurimfizmi r" m! 11H _ ”tr”. $1,»; v {Bryn-fun: 7;" 111'. ‘vu’n ' ,1“..— m’y’ 0': 3..- $5 > '7‘”. NULL; “T 'L‘.. ’ , , THESIS This is to certifg that the thesis entitled A COMPARATIVE STUDY OF PROGRAMMED INSTRUCTION AND VIDEO-TAPED LECTURES IN PUBLIC SPEAKING presented by Philip P. Amato has been accepted towards fulfillment of the requirements for Ph. D. degree in Speech Major professor Date August 13, 1963 0-169 LIBRARY Michigan State University ABSTRACT A COMPARATIVE STUDY OF PROGRAMED INSTRUCTION AND VIDEO-TAPED LECTURES IN PUBLIC SPEAKING by Philip P. Amato The increased enrollment in Speech classes, particu- larly in the basic theory-performance courses, has created a unique problem for speech teachers which demands serious attention. The basic problem is this: the larger the class the less the Opportunity for individual attention. This problem is compounded by the inherent limitations of the lecture method in presenting factual material (passive listening, active listening with misunderstanding, etc.) Recent deve10pments in the area of programed instruction indicate its potential utility as a cogent method of presenting the theoretical aspects of speech courses. The primary purpose of this study was to measure the relative effectiveness of programed instruction and video- taped lectures as methods of presenting public Speaking lecture material. A 45 minute lecture on Outlining and Speech Organization was recorded on video tape. Two forms of a 62-frame program were constructed using the identical material as that of the video-taped lecture. One form of the program (constructed reSponse program) consisted of Philip P. Amato frames in which the critical words in questions or state— ments were deleted, and in reSponse to-which the—student was required to construct a written answer. -The other form (reading program) consisted of the same questions and statements as the constructed reSponse~program, but with the critical words underlined. In this form the student simply read each frame. The video—taped lecture and the programs were pre- sented to 223 subjects enrolled in the basic public Speaking course at Michigan State University. The subjects were assigned at random to one of seven groups (Six experimental and one control). The control and experimental conditions are described below: TV: Subjects in this condition viewed the video- taped lecture. CR: Subjects in this condition worked through the constructed reSponse program. R: Subjects in this condition worked through the reading program. TVC: Subjects viewed the Video-taped lecture and then worked through the constructed reSponse program. TVR: Subjects Viewed the video—taped lecture and then worked through the reading program. C-R: Subjects first worked through the constructed response program and then the reading program. C: Subjects in the control group were given the posttest. Subjects using the programs were instructed to in- dicate the time they began and finished working on them. Philip-P-~Ameto Following the completion of the learning-task3~subjects in those groups that used the programs were also given an 8-item questionnaire to fill—out anonymously. Three measures were recorded: (1) a posttest score—-a measure of amount learned; (2) a time score-—the amount of time subjects took to learn; and (3) an attitude measure-~the reaction to the learning task. Three analyses of variance and two Scheffe tests were computed to evaluate the significance of the differ- ences among the posttest means of the seven groups. Critical ratios were computed to test significant differ— ences among time scores. Questionnaire reSponses were not tested statistically but reported as percentages. The results of this study tend to indicate that: l. Programed instruction is a more effective and efficient method of presenting public Speaking material than video-taped lectures. 2. Reading programs are as effective as, and far more efficient than, constructed reSponse programs in presenting public speaking material. 3. Combinations of programed instruction and video- taped lectures, while taking much more time, do not increase the total amount of learning beyond that acquired when programs are used alone. A. Although students feel that programed instruction is a more effective and easier method of learning than video—taped lectures, they do not prefer it as the sole source of instruction. 5. Students have a strong preference for using both programed instruction and video—taped lectures as methods of presenting public Speaking material. Philip P. Amato This study also includes a brief history of the development of programed instruction; the three most dominant theories of programing; and a review of part of the experimental literature. Implications and the potential utility of programed instruction as an auxiliary teaching device in selected areas of speech education are also discussed. A COMPARATIVE STUDY OF PROGRAMED INSTRUCTION AND VIDEO-TAPED LECTURES IN PUBLIC SPEAKING By Philip P. Amato A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Speech 1963 Cepyright by PHILIP P. AMATO '1?Q+ ACKNOWLEDGMENTS The author wishes to express his deep appreciation to all those who assisted him throughout his whole program of study and the preparation of this thesis. He is expecially indebted to Dr. Frederick G. Alexander who served as his major professor. The author also wishes to express his sincere graditude to the other members of his Guidance Committee-~Dr. Kenneth G. Hance, Dr. David R. Krathwohl, Dr. Walter Stellwagon, and Dr. Gordon L. Thomas-‘for their interest, guidance, and academic stimulation. Others whom the author wishes to acknowledge are: Dr. David C. Ralph and the Speech 101 staff for their help in conducting the experiment; and Don Freeman and Martha Otis for their helpful suggestions and assistance. Finally, this thesis is dedicated to the author's parents, Mr. and Mrs. Vincent J. Amato, for their warm encouragement, unfailing assistance, and inspiration. iii TABLE OF CONTENTS ACKNOWLEDGMENTS . O C O O O O O O O O O O O 9 O O O 0 LIST OF TABLES LIST OF APPENDICES. . . . . . . . . . . . . . Chapter I. AN OVERVIEW OF THE STUDY . . . . . . . . . . Introduction Purpose of the Study Limitations Justification Organization II. PROGRAMED INSTRUCTION. Some Early Developments in Self-Teaching Devices Programing and Presentation Devices A Review of Selected Experimental Studies III. THE RELATIVE EFFECTIVENESS OF SIX METHODS OF PRESENTING PUBLIC SPEAKING MATERIAL. . . Statement of the Problem Review of the Literature Hypotheses Method Results Discussion Conclusions IV. THE IMPLICATIONS AND POTENTIAL UTILITY OF PRO— GRAMED INSTRUCTION IN SPEECH EDUCATION The Use of P.I. in the Basic Speech Course The Use of P.I. in Speech Science and Cor- rection The Use of P.I. in other Areas of Speech The Need for Future Research The Future of P.I. in Speech Education Conclusion iv Page iii vi vii ll 47 93 Page BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . 104 APPENDICES . . . . . . . . . . . . . . . . . . . . . llO Table 10. 11. 12. 13. 14. LIST OF TABLES A Set of Frames Illustrating the Principles of Linear Programing . .-- . . . . A Set of Frames Illustrating the Intrinsic or Branching Method as Used in a Scrambled Text— book . . . . . . . . . . . . . . . . . . . A Breakdown of Presentation—Mode Abbreviations Means and Variances of Posttest Scores Analysis of Variance for Groups: TV, CR, and R. . Analysis of Variance for Groups: TVC, TVR, and C-R. . . . . . . . . . . . . Analysis of Variance for all Seven Groups. Scheffé Test of Difference for Groups: TV, CR, and R. . . . . . . . . . . . . . Scheffé Test of Differences for all Seven Groups Completion Time (Minutes) for Groups Using Program A. . . . . . . . . . . . . . Completion Time (Minutes) for Groups Using Program B. Completion Time (Minutes) for all Experimental Groups Summary of Questionnaire Responses for Groups that Used Programs Posttest Scores, Totals, Means, Variances, and Standard Deviations for the Control and Ex- perimental Groups. vi Page 19 23 68 76 77 77 78 79 8O 81 82 82 85 152 LIST OF APPENDICES Appendix Page A. Program A . . . . . . . . . . . . . . . . . . 110 B. Program B . . . . . . . . . . . . . . . . . . 136 C. Posttest. . . . . . . . . . . . . . . . . . . 139 D. Questionnaire . . . . . . . . . . . . . . . . 148 E. Posttest Raw Score Data . . . . . . . . . . . 151 F. Computational Formulas. - . . . . . . . . . . 153 vii CHAPTER I AN OVERVIEW OF THE STUDY Introduction Today, more than ever before, communication and the important role it plays in every day life is seriously being considered and studied. The importance of effective Speech, for example, is recognized not only by colleges and universities, but also by business and industry. Approximately one half of the colleges and universities in the United States require that every student take at least one Speech course, usually voice and diction or public speaking.1 Many business and industrial concerns demand or at least encourage executives and supervisors to take a course in business and professional Speaking. The impact of the student pOpulation growth, which has already over-taxed primary and Secondary educational facilities and manpower, is also beginning to have its effect on speech education. The increase in enrollment of Speech classes, particularly in the basic courses, has created a unique problem for speech teachers which demands 1E. c. Buehler and w. A. Linkugel, Speech: A First Course (New York: Harper & Brothers, 19627j'19. 1 serious attention. In the basic theory-performance type course, such as voice and diction or public Speaking, the amount of time each student has for performance is usually pr0portional to the number of students in the class. As the size of the class grows,the probability of students receiving sufficient time to perform becomes less and less. In addition to textbooks, there are two dominant methods of presenting factual material in the basic Speech course. They are: (l) the classroom lecture—discussion approach; and (2) the lecture hall approach in which all students enrolled in the course listen to the same lecture. Both methods have a number of individual advantages and disadvantages which have been frequently discussed and de- bated. Regardless of the approach, the problem of student reception still exists. In both'situations, one student may be entirely active, another may be entirely passive. Even the most careful organization of material is ineffec— tive when the student is inattentive. Moreover, even though a student may be attentive to the material being presented, he seldom receives immediate information as to the correctness of his response and consequently may form misconceptions.’ The student is also unable to proceed at his own rate. In fact, because of the mass entrance of stu— dents into the classroom, the teacher is often forced to lecture to the average student. The students at the extremes seldom receive the attention required to facilitate learning; the brighter student is bored while the slower student struggles. The limitations of the lecture method often force an instructor to sacrifice recitation meetings in order to restate or explain material which the student failed to graSp during the lecture. The more time taken to clarify material covered in the lectures, the less time available for recitation. There is, therefore, a growing need for ways and means of accommodating large numbers of students in speech courses without destroying the small class—individual attention situation which many Speech teachers believe is necessary in Speech education. The closed-circuit television presentation of material represents an attempt to cope with the problem of increased enrollment and in many ways has alleviated some of the prob- lems facing the lecturer in the large lecture hall. For example, it has done away with such frequent problems as poor acoustics and lighting which often make it difficult for the students to hear what is being said and see the visual aids being used. Closed—circuit television, however, does not provide each student with individual instruction, and as in any situation where students are not instructed individually, the lecturer is faced with the problem of stu- dent reception. This brings on the problems of passive listening or active listening with misunderstanding, and with them the problems of using recitation meetings to clarify lecture-covered material. Hence, the major problem facing the lecturer still exists. Finally, very few schools have the facilities or budgets for closed—circuit television systems. Two possible solutions to the problem are: (l) assign a private tutor to each student; and (2) eliminate the lectures and theoretical aspects of the course and use that time for practice and performance. The first solution is impractical from an economic standpoint, the second from an academic standpoint. Without a clear understanding of Speech principles the student will profit little from his Speaking experiences. Recently much attention has been focused on the devel- opment of teaching machines and programing techniques as a method of self-instruction. Programed instruction is a process of presenting to the student a body of material through small sequential steps or frames in the form of ques- tions or statements. Depending upon the type of program or format, the student may overtly respond to each frame by constructing or selecting a correct answer or he may covertly respond either by mentally composing a correct answer, or by Simply reading each frame as a complete statement. In the reading format the correct answers are supplied within the frame, and the student is not required to reSpond. The potential utility of this new method of auto— instruction is currently being investigated and mounting reports of successful experimentation indicate that programed instructional methods are highly effective and efficient. Harm believes that programed instruction may offer a means of achieving the ”optimal ratio of theory to practice.”2 According to Harm, programed learning appears most suitable for exactly those kinds of material teachers grow weary of explaining term after term but which the student seems unable to get from the text with- out instructor interpretation.3 Since programed learning is an individua1_tutor type method based on a one-to-one (program to student) relationship, it represents a means of COping with large numbers of students without the problems of the typical lecture situation. Perhaps it is time for the speech teacher to examine closely these new methods of instruction and determine what possible role they may play in the future of Speech education. Purpose of This Study In view of the increase in enrollment of Speech classes and the problems facing the lecturer, either in person or on television, it seemed worthwhile to consider programed instruc— tion as an alternative method of presenting lecture material and to subject it to experimentation. Consquently, the primary purpose of this study was to ascertain objectively and quantitatively the relative effectiveness of presenting 2L. A. Harm, "Programed Learning for the Field of Speech," Speech Teacher, 10 (1961), 219. 3Ibid. —-—-—.— lecture material through programed instructional devices compared to conventional methods of presentation. While there are many ways of approaching this problem, the ex- periment conducted as part of this study was designed to investigate methods of presenting public speaking materials and focused on: (1) developing a program on Outlining and Speech Organization; (2) comparing two forms of this pro- gram and atelevised video-taped lecture on the identical material; (3) comparing combinations of two of these three modes of presentation (video-taped lecture and two forms of a program); and (A) comparing all six modes of presen- tation, i.e., the three basic modes of presentation and combinations of two of these modes. A second purpose of this study was to present a brief history of the development of programed instruction, the three most dominant theories of programing, and a review of part of the experimental literature. A third and final purpose of this study was to con— sider and discuss the implications and potential utility of programed instruction as an auxiliary teaching device in selected areas of Speech education, particularly the basic theory—performance course. Limitations Three major limitations were imposed on this study. First, although programed instruction is a relatively new .field of inquiry, a vast amount of literature has appeared, which makes it difficult to consider its complete depth and scope. This study considered only the major historical deveIOpments, the three dominant theories of programing, and a selected portion of the experimental literature. The second major limitation placed on this study was in the general nature of the experiment conducted. Only one aSpect of public speaking was considered~- organization and outlining a Speech. The programs used were based on only one major programing theory and limited to 62 frames. Only the televised video-taped lecture, and two forms of a program, along with combinations of two of these three basic modes of presentation, were compared. The ”in person” lecture hall, "live" television, and classroom lecture—discussion methods of presentation were not considered. The questionnaire used to measure student attitude toward programed instruction was limited in depth and SCOpe and did not readily lend itself to statistical analysis. Hence, only percentages of student responses were reported. Finally, Since the students continually applied the information learned in later classroom work, no test of retention over a long period of time was given. A third major limitation imposed on this study was in the discussion of the implications and potential utility of programed instruction in Speech education. Only selected areas, particularly the basic theory-performance courses, were considered. Justification Because the recent interest in speech training and the increased enrollment in basic Speech courses threaten to destroy the small class-individual attention situation, there is merit in studies that seek ways of coping with this growing problem. And since the results of experimen- tal studies have amassed an impressive array of evidence supporting programed instruction as a potentially effective and efficient method of presenting lecture material, there is merit in studies that seek to determine whether these findings are transferable to areas of Speech. To date, there has been little research conducted which has attempted to determine the potential utility and feasibility of programed instruction in Speech education. A Studies by Hillis and Holland5 used tape recorders “J. Hillis, A research project in progress Since 1960 which is investigating a tape recorder—workbook type program in applied phonetics at Michigan State University (Department of Speech). 5A. Holland, "The DeveIOpment and Evaluation of Teaching Machine Procedures for Increasing Auditory Discrimination Skill in Children With Articulation Disorders" (unpublished Ph. D. dissertation, Department of Speech, University of Pittsburgh, 1960). and non-continuous type programs. The experiment reported in this study used a continuous type program and programed booklet as a vehicle of presentation. Finally, there have been no studies reported which compared television (”live" or video tape) and programed instruction as modes of presentation. Organization In alignment with the three major purposes of this study, the organization was divided into three major areas: (1) an overview of programed instruction; (2) an experiment testing the relative effectiveness of methods of presenting public speaking lecture material through televeision and programed instructional devices; and (3) a discussion the implications of programed instruction in Speech education. The chapters of this study follow this general plan of organization. Chapter I endeavors to set forth a basic overview of the study. Chapter II presents a brief picture of the historical develOpment of programed instruc— tion; a discussion of the three dominant theories of programing; and a review of part of the experimental literature. Chapter III consists of a report of the experi— ment, including: a statement of the problem; a review of the literature related to the problem; a listing of the hypotheses tested; a description of the subjects, materials, and procedure followed; and a discussion of results; and 10 conclusions drawn from the data. The final chapter, Chapter IV, includes a discussion of the implications and potential utility of programed instruction in selected areas of Speech education. CHAPTER II PROGRAMED INSTRUCTION Most of the major develOpments in the field of programed instruction have occurred within the past ten years.1 The earlier self—teaching devices were conceived as teaching aids in conjunction with standard teaching methods. The first teaching machine, for example, was deveIOped and patented in 1866 and was designed to aid in teaching spelling.2 Today, programed instruction is generally conceived as a sole source of instruction, i.e., a method of instruction without the aid of other standard teaching devices (textbooks, lecture—discussion, etc.). Some Early DevelOpments in Self—Teaching Devices In 1915, Sidney L. Pressey began a series of studies designed to measure the effectiveness of self-testing devices which provided immediate confirmation of reSponseS. In his early studies Pressey used a mechanical device 1For a comprehensive picture of the history of pro- gramed instruction see Teaching Machines and Programed Learning: A Source Book, ed, A. A. Lumsdaine and R. Glaser (Washington, D. C.: National Education Association, 1960), 5-23. This book also contains numerous articles on pro- graming theories, techniques, and experimental studies. 2I. Mellan, "Teaching and Educational Inventions,” Teaching Machines and Programed Learning: A Source Book, pp. 265974. This device did not provide immediate feedback concerning the correctness of the student‘s reSponse. 11 12 3 called the ”Drum Tutor" which is about the size of a typewriter and consists of a window which diSplayS a multiple choice item and four keys which represent each of the four alternatives. A sequence of questions (usually 30) is inserted into the machine with the first question visible to the student. The student reads the question and reSponds by depressing a key. If the reSponse is correct, the next question in the sequence appears in the display window; if incorrect, the machine does not advance the next question and the student must try again. These studies revealed that the ”Drum Tutor,” in addition to serving as a mechanical tester, produced significant increments in learning. A more thorough investigation of self-testing devices as potential self-teaching instruments was made by Pressey and his associates using a punchboard type device.4 The punchboard device consists of two 3" x 5" thin punch~ boards, the tOp one of which contains a series of numbered rows of holes. An answer sheet is placed between the two boards; the multiple choice test items are presented on separate reusable sheets. After reading a question the 3S. L. Pressey, ”A Simple Device Which Gives Tests and Scores-—and Teaches," School and Society, 23 (1926), 373-76. “S. L. Pressey, ”DevelOpment and Appraisal of Devices Providing Immediate Automatic Scoring of Objective Tests and Concomitant Self~Instruction," Journal of Psychology, 29 (1950), 417-A7. 13 student responds by inserting his pencil into the hole corrSSponding to the alternative on the test sheet. If his choice is correct, the whole pencil goes through the paper; if incorrect, only the point of the pencil goes through the sheet and the student must try again. These studies also found that in addition to testing, the punch- board device produced Significant increments in learning. These early devices did not receive any wideSpread acceptance or attention. While successful in producing significant increments in learning they were not programed instructional devices in the true sense of the term. The material presented to the student was not a program but a test based on subject matter taught yia standard teaching methods such as lectures, discussion, and textbook readings. The tests were not designed to produce an organized change in learning behavior; on the contrary, the learning that took place was incidental to the testing and principally the result of reSponse feed- back. The idea of a teaching device designed to serve as a sole source of instruction began to receive considerable attention after 1954. In that year, B. F. Skinner reintroduced the concept of automated instruction in a provocative paper presented during a psychology conference at the University of Pittsburgh.5 He reported the develOpment 5B. F. Skinner, ”The Science of Learning and the Art of Teaching,” Harvard Educational Review, 24 (1954), 86-97, and feasibility of a mechanical device designed to present a continuous discourse program which would serve as a sole source of instruction rather than as a testing device that produced learning as a byproduct. Skinner developed a small box-like device about the size of a portable typewriter. The top surface of the Skinner teaching machine contains a display window through which a question or problem may be seen. The problem is printed on a paper disk and consists of questions or statements in which one or two words are missing. The student reads the problem and responds by writing in the provided answer space to the right of the display window. He then raises a lever and the correct answer is exposed which the student compares with his own answer. After indicating with an appropriate movement of the lever whether his answer was correct or incorrect, the student reads the next item in the program which appears in the display window. At the completion of the program, incorrect items are repeated. Skinner and his associate, James G. Holland, found that the program and teaching device were capable of producing learning without the use of standard teaching 6 methods. 6J. G. Holland, ”Teaching Machines; An Application of Principles From the Laboratory,” Programed Learning: Theory and Research, ed. W. I. Smith and J. W. Moore (New York: D. Van Nostrand Company, Inc., 1962), pp.34—48. 15 Since 1954, self-teaching methodology has developed into what is now commonly referred to as programed instruc- tion. In programed instruction the material to be learned is broken down into a series of small steps. Depending upon the program format and presentation mode, these steps may be written in the form of a multiple choice item or an incomplete question or statement. In either case, the student is required to make an overt reSponse. The presentation process, depending upon the complexity of the device, may prompt or hint, reveal the correct answer immediately after the student responds, branch the student off to a remedial phase of the subject matter, and keep score for the complete series. The vehicle of presentation may range from a machine to a Specially designed book. Programing and Presentation Devices Programing Theories.--The power of programed instruc— tion lies not in the presentation device or machine but in the program. At the present time there are three dominant schools of programing from which stem a number of programing techniques or systems. These three programing theories are known as ”linear,” ”intrinsic” or ”branching,” and ”multiple choice." Linear PrOgraming.-—This theory of programing was developed by B. F. Skinner, and the learning model is basically a conditioning model. Of the three theories, 16 it is the most popular. In 1960, for example, 62 of the 81 programs surveyed were of this type.7 Much of this popularity, however, seems to stem from the ease with which this type of program can be constructed and presented. Linear programing is based on the concept of operant conditioning, which differs from classical conditioning. Briefly, operant conditioning postulates that each time a response is reinforced in the presence of a certain cue or stimulus the probability of the recurrence of that response in the presence of the same stimulus is increased. Once a reinforcer (which is contingent upon a reSponse) gains control over a form of behavior, that behavior can parsimoniously be maintained by an intermittent reinforce- ment schedule.8 Through a serious consideration of Thordike's Law of Effect9 and basic principles of Operant conditioning, 7J. W. Rigney and E. B. Fry, Current Teaching-Machine Programs and Programing Techniques, Supplement 3, Audio— Visual Communication Review, 9. 8For a more concise picture of Skinner‘s theory of learning and Operant conditioning see E. R. Hilgard, Theories of Learning (New York: Appleton—Century—Crofts, Inc., 1948), 80—109. 9According to this law, we tend to repeat those responseswhich are followed by a rewarding state of affairs and avoid those which are accompanied or followed by an unhappy or annoying state of affairs. E. L. Thorndike, Educational Psychology (New York: Columbia University Press, 1921, II, 4. l7 Skinner deveIOped a system of programing which employs four major principles: (1) clarity and simplicity; (2) reinforcement; (3) gradual progression; and (4) fading or vanishing.lO The principle of clarity and simplicity, according to Skinner, suggests that a good program is one which makes it almost impossible for the student to make an error. The concepts or principles to be learned are atomized, constructed into Simple statements, and presented with some degree of redundancy. The linear method usually requires that the student construct his own response; this is usually done by exposing to the student a question or problem which contains a blankis) for one or two missing words. This method also avoids using multiple choice items so that the learning process is not one of trial and error. In fact, little provision is made for errors since they are considered irrelevant to learning any may hinder its process. Hence linear programs, if properly constructed, are refined to the point where very few errors occur. In addition to the atomization of material, the program employs prompting, cueing, or other suggesting devices. loR. Glaser, Principles and Problems in the Preparation of Programed Learning Sequences, A Report Prepared Under COOperative Research for the United States Office of Education (Pittsburgh: University of Pittsburgh, 1960), 4—8. 18 The second principle of linear programing is rein- forcement. According to this principle, a desired change in behavior, defined as learning, can best be brought about by rewarding or reinforcing the desired behavior. The change is acquired as a result of ”contingencies of reinforcement."11 Reinforcement does not follow unless the conditioned reSponse appears; in other words, the reinforcement is contingent upon the response. In tradi- tional arrangements, such as in the classroom,reinforce- ment might have to wait until homework is corrected or test papers returned. By this time, the reinforcement has lost a good deal of its potency. In contrast, the linear program theorist argues that programed instruction provides immediate feedback, a higher rate of reinforcement, and active student participation in the learning process. This is by far one of the most critical features of this new method of instruction. The principle of gradual progression is concerned with getting the student from a basic point (initial repertoire) to a higher or complex point (terminal repertoire) through a series of finite steps. In working from the initial to the terminal repertoire, each and every small change in behavior which is in the direction of the terminal repertoire is reinforced. This principle llSkinner, op. cit., 86. 19 runs throughout the program and serves to make the student correct as often as possible. (See Table 1.) TABLE 1 A SET OF FRAMES ILLUSTRATING THE PRINCIPLES OF LINEAR PROGRAMING Answer Blank and Correct Frame ReSponse l. Plosives are exploding sounds. When we build up pressure in out mouths and suddenly release it we produce a Sound. plosive or exploding 2. In order to make the sound [p] we must bring our lips together build up in our mouths, and suddenly release it. air or pressure 3. The sound [0] 15.29: a ________ * because there is no pressure built up in the mouth. Try iti plosive 4. Which sound is not a plosive? [p] [f] lb] {d1 r__l 4 It 3; L..J *This is one method of ”prompting” or ”cueing.” Each Space represents a letter in the correct answer. The final principle, fading or vanishing, is one which is concerned with the withdrawal of stimulus support. It involves a weaning process which takes the student away from any dependence upon the program (such as the ”prompting” techniques used in Table 1) which he may have deveIOped. Built into the program is a gradual process of removing prompts and cues so that by the time the student has completed the lesson he reSponds to the material as a stimulus rather than to built—in prompts or cues. Hence it may be said that the linear program requires at least four basic features. (1) Program items must be relatively small in terms of the amount of information they impart to the learner. (2) Each item must be presented in a logical sequence. (3) The learner must overtly respond to each item. (4) The learner must receive immediate feedback or knowledge of results. The linear programing model has been the subject of much investigation. In contrast to the requirements stated above, many studies have found, for example, that there is no significant difference in criterion perform— ance between programs requiring overt responses and programs requiring covert reSponses. A further discussion of experimental findings will be presented in the latter part of this chapter. Intrinsic Programing.--The second major programing theory was deveIOped by Norman A. Crowder, who views the teaching of human beings as essentially a communication 2 process.1 This approach to programing makes no 12N. A. Crowder, ”Automatic Tutoring by Means of Intrinsic Programing,” Automatic Teaching: The State of the Art, ed. E. Galanter (New York: John Wiley and Sons, Inc., 1959), 109-116. 21 presumptions as to how to set up conditions under which efficient learning takes place and uses feedback to control the communication process. Therefore, the program frames or items in intrinsic programs are not restricted to as rigid a model as that proposed by Skinner. Furthermore, in this type of programing knowledge of results is primar- ily used to determine whether the communication was success— ful rather than as a method of reinforcement. Intrinsic programing is more commonly known as ' since it employs a branching procedure ”branching,’ which involves Shifting the difficulty of the material presented to a lower level when the student may have trouble responding to a particular item. It may also work the other way; that is, the material may be shifted to a higher level when the student is reSponding very rapidly and correctly. To implement these shifts either a machine or a Specially designed (”scrambled") book is required. In the ”scrambled” book format each step of the program is presented on a different page. The material (program steps) is scrambled throughout the book to prevent the student from merely reading through the program. Thus, step one may be on page 1,step two on page 29, and so on. (See Table 2.) The student is given a short discussion of the material to be learned, followed by a multiple choice question designed to test him on the 22 material just discussed. Each alternative leads the student to a different page in the book (in the machine the pages are usually represented by a frame on a micro— film and alternative selecting is done by pushing apprOpriate buttons). If the student selects the correct alternative, he is given additional information and a new problem; if he selects a wrong alternative, he is told why he is wrong and referred back to the problem where he selects another alternative. The shifting of material is contingent upon the student's reSponses and for this reason the intrinsic or branching method is sometimes referred to as ”adaptive programing.” In Table 2, choosing "factors" on the first try permits the student to bypass two remedial steps. The choosing of a wrong alternative leads the student to a remedial level. Thus, the shifting of material difficulty is dictated by the student's reSponse. Intrinsic programing has certain obvious advantages in c0ping with individual differences. Unlike the linear program, every student is not required to read every step. The one main objection to this method is its prodigious use of time and materials. For example, a sequence of fifty questions with only two alternatives would require fifty extra pages in a text or frames in a machine just to carry the wrong answers. In a sequence of fifty questions with the usual four alternatives,one hundred and fifty 23 TABLE 2 A SET OF FRAMES ILLUSTRATING THE INTRINSIC OR BRANCHING METHOD AS USED IN A.SCRAMBLEDTEXTBOOK‘13= Page 1 In the multiplication 3 x 4 = 12 is called the product and the numbers 3 and 4 are called the Page 15 quotients Page 29 factors Page 43 powers [If the student chooses the alternative "powers” he turns to Page 43.] Page 43 Your answer was: 'powers.’ We'll get to the powers of numbers pretty soon, but we're not there yet. The numbers that are multiplied together to form a product are called 'factors,‘ 'powers.’ Now return to Page 1 and choose the right answer. [The student returns to Page 1 and now chooses the word "factors" which directs him to Page 29. Note: If the student chooses “factors” the first time, he bypasses Pages 15 and 43, which are remedial steps.) Page 29 Your answer was: ifactors.‘ You are correct. The numbers which are multiplied together to form a product are called 'factors.‘ Thus in multiplication '3 x 4 — 12‘ the numbers 3 and 4 are the factors, 12 is the product. Is it possible for the same number (same quantity, that is) to be used as a factor more than once in forming a product? Page 59 Yes Page 71 No l3Ibid., 15—17. 24 extra pages or frames would be necessary to accommodate the wrong alternatives. The mechanical devices employing micro—film are capable of large programs of two or three hundred items but these machines are quite expensive. Another factor to be considered in this type of programing is the inherent difficulty in constructing good multiple choice items. Multiple Choice Programing.——The third type of programing theory was developed by Sidney L. Pressey. As stated earlier, Pressey views the program as an auxiliary teaching deVice rather than as a sole source of instruc— tion. The steps in the program are cast in the form of multiple choice items, but unlike the intrinsic method a wrong alternative does not direct the student to remedial areas. If he chooses a wrong alternative the student is either instructed to choose another or given the correct answer and instructed to continue along in the program. Thus, in item sequencing this approach is linear. Fry discusses a number of differences between linear and multiple choice programing;14 the major difference between these two methods of programing is in the types of questions asked and response modes used (multiple choice vs. constructed response). The current trend in ”styles of l4E. Fry, ”Teaching Machine Dichotomy: Skinner vs. Pressey,” Programed Learning: Theory and Research, 81-86. 25 / " however, seems to be in the direction of programing, composite programing--particularly the combination of multiple choice and linear methods; the branching or 15 intrinsic method is seldom used. Programing Techniques.—-In addition to the three programing theories and their respective programing techni— ques, a number of other systems have emerged.16 In the ”Ruleg” system, for example, the subject matter to be programed is classified into two classes of statements-— 17 The program frames or items are rules and examples. usually written as either incomplete rules or incomplete examples to which the student must respond by filling—in 15Programs, '62: A Guide to Programed Instructional Materials (New York: The Center for Programed Instruction, In6., 1962), xi—xii. 16The individual theories and techniques of programing are scattered throughout the literature. There are, however, a number of sources which present selected articles by noted theorists and critics, particularly Lumsdaine and Glaser (op. cit.) and Galanter (op. cit.). Summaries of the major programing theories and systems of programing are found in Current Teaching—Machine Programs and Programing Techniques (Rigney and Fry, op. cit.), ’14419, and in Lawrence Stolurow, Teaching by Machine (Washington, D. C. U. S. Department of Health, Education, and Welfare, Office of Education, 1961), 17-50, 17J. L. Evans, R. Glaser, and L. E. Homme, The RULEG System for the Construction of Programed Verbal Sequences (Pittsburgh: De artment of Psychology, University of Pittsburgh, August 1960 . A discussion of this system Egg aéso be found in Lumsdaine and Glaser (op. cit., —9 ). 26 the answers called for in the blank(s). This system of programing follows a linear format. There are at least ten different systems of programing, but they are simply variations of the three major theories. Presentation Devices.——The two most pOpular methods of presenting programs are the machine (ranging from a simple hand—operating piece of hardware to computers) and programed textbook. Today, there are over 80 machines on the commercial market and almost as many in the research laboratory.18 There are also two major types of programed textbooks. Although they represent different designs and degrees of complexity, these devices contain at least four basic 19 (1) Presentation Function: A features or functions. question or problem is presented to the student in reSponse to which he must construct or indicate an answer. (2) Comparator—Feedback Function: The student is informed whether his response is correct or incorrect. (3) Program- ing Function: The sequence in which the items are presented is controlled by the program or the student's response 18J. D. Finn and D. G. Perrin, Teaching Machines and Programed.Learning: A Survey of the Industry—1962 Washington, D. C”: Department of Health, Education, and Welfare, Office of Education, 1962), 23. This publication includes a complete directory of machines and programs currently found on the commercial market. 19B. F. Skinner, ”Teaching Machines,” Science, 128 (October, 1958), 969-77. t\) '\l (as in the intrinsic or branching method). (4) Selective~ Pacing Function: The student proceeds at his own rate; the timing of the questions and answers is usually under the control of the student.20 There are basically two types of programed textbooks: (l) the linear or constructed response type; and (2) the scrambled book, which was discussed earlier. Both types of books represent two of the three major approaches to programing (linear and intrinsic or branching). In the lin— ear or constructed reSponse book, questions or statements contain one or more blanks which represent the critical word(s) in the frame or item. The student responds by constructing a written answer. Depending upon the textbook format, the student either turns the page (horizontal for- mat) or reads down to the left or right of the next item (vertical format) to ascertain the correctness of his reSponse. After comparing his reSponse to the correct answer in the text, the student reads the next item in the sequence. A Review of Selected Experimental Studies Since 1954 a sizeable number of studies have been conducted to test the effectiveness and efficiency of programed instruction. On the whole, these studies indicate that groups taught by means of programed instruc~ tion tend to score significantly better on criterion 20A more detailed treatment of the functions of certain machines may be found in Teaching by Machine (Stolurow, pp. 013;), 17-50. 28 performance measures than groups taught by conventional methods. The criteria generally used in these studies are: (1) time necessary to complete learning task; and (2) achievement tests. The majority of recent studies have investigated the relative merits of different principles of programing, particularly those set down by Skinner in the linear method. The remainder of this chapter is devoted to studies of this nature. It is hoped that such a review of the experimental literature will provide the reader with a clearer picture of programing principles in relationship to their behavior under experimental conditions. Punchboard Studies.-~The early research relating to self-teaching devices was initiated by Pressey and was mainly concerned with the potential utility of multiple choice testing devices (”Drum Tutor," "punchboards"), which provide immediate knowledge of results, as instruc- tional tools to supplement regular classroom instruction. Studies by Pressey,21 Little,22 Briggs,23 Angell,gn 21Pressey, ”Development and Appraisal of Devices Pro— viding Immediate Scoring of Objective Tests and Concomitant Self—Instruction,” Journal of Psychology, op. cit., 417-47. 22J. K. Little, ”Results of Use of Machines for Testing and for Drill Upon Learning Educational Psychology,“ Journal of Experimental Education, 3 (1934), 45-49. 23L. J. Briggs, ”The DevelOpment and Appraisal of Special Procedures for Superior Students and an Analysis of the Effects of *Knowledge of Results’,” Abstract of Doctoral Dissertations, 58 (1949), 41—49. 24G. W. Angell, ”Effects of Immediate Knowledge of Quiz Results on Final Examination Scores in Freshman Chem- istry,” Journal of Educational Research, 42 (1949), 391—94. 29 25 26 Jensen, Jones, and others offer an impressive array of evidence showing that the punchboard type test can produce significant increments in learning and that students using the device score higher on criterion performance than students not using the device. These studies, however, did not attract any widespread use of this method of auto- instruction. The Program as the Sole Source of lnstruction.--Skinner reintroduced the concept of automated instruction in 1954. While the Pressey punchboard approach was interested in the teaching machine as an auxiliary teaching device, Skinner approached it as a possible sole source of instruc- tion. A number of studies were first conducted to test the feasibility of this point of view. Studies by Porter,27 28 Home and Glaser, Blyth,29 along with a number of others, showed that machines can teach on their own and that 25B. T. Jensen, ”An Independent Study Laboratory Using a Self-Scoring Test,” Journal of Educational Research, 42 (1949), 134—47. 26R. S. Jones, "Integration of Instruction With Self- Scoring Measuring Procedures,” Abstract of Doctoral Dissertations, 65 (1954), 157-b5. I . 27D. Porter, ”Some Effects of Year-Long Teaching Machines,” Automatic Teaching: The State of The Art, 84—90. 28L. E. Homme and R. Glaser, ”Relationships Between the Programed Textbook and Teaching Machines,“ Ibid., 103-7. 29J. Blyth, ”Teaching Machines and Human Beings,” Educational Record, 41 (19b0), 116-26. 3O programed instruction material, using either machines or programed textbooks, produces significantly better criterion performance than non-programed presentation. Machine Versus Programed Textbook.--Studies comparing the machine and programed textbook modes of presentation indicate that there is no significant difference in 30 achievement between the two modes. Eigen, et al. found that there is no significant difference in mastery between machine, horizontal text, or vertical text.31 Because of the inherent problems associated with any mechanical device, these studies generally show a signifi- cant saving in time within those groups utilizing the programed textbook. The results of these studies suggest that programs may be presented either by machine or pro- gramed textbook with the same degree of effectiveness and that the programed textbook is the more efficient of the two modes in terms of time necessary to complete the program. A recent survey shows that 93.4 per cent of the available programs can be obtained in programed textbook format.32 30L. s. Goldstein and L. G. Gotkin report and summar- ize eight studies which have compared machine and programed textbook presentation of self-instructional materials. All of the studies dealt with the Skinner linear-type pro- gram. No significant differences between modes emerged and in five studies, significant saving in time was effected with use of the programed textbook. (”A Review of Research: Teaching Machines vs. Programed Textbooks as Presentation_ Modes,” Journal of Programed Instruction, 1 [1962] , 20-36.) 31Eigen, et. al., ”A Comparison of Three Modes of Pre- senting a Programed Instruction Sequence, ”Journal of Educational Research, 55 (1962), 453—bO. 32Programs, '362, xiii. 31 Intrinsic or Branching Versus Linear Programing.-- A number of studies have been concerned with the relative effectiveness and efficiency of these two programing theories. In the linear technique prOposed by Skinner, a fixed sequence of items is displayed and the learner is required to go through all the items. The branching method, in contrast, uses a variable sequence of items, allowing the learner to branch to remedial material or bypass to more difficult material. Coulson and Silberman presented a program on elemen- tary psychology to college students through a simulated teaching machine.33 The results showed no significant difference on the criterion test scores between the groups using linear and those using branching programs. The branching group, however, learned the material in much less time than the linear group. In a later study by Silberman, et. al., a logic program was presented to groups of high school students through a computer—based teaching machine.31L The researchers found that the branching or intrinsic group did no better 33J. E. Coulson and H. F. Silberman, ”Effects of Three Variables in a Teaching Machine,” Journal of Educational Psychology, 51 (1960), 135-43. 34H. F. Silberman, et. al., loc. cit. on criterion performance than the linear group, nor did it take less time for the branching group to learn the material. Two later studies by Campbell also failed to show the superiority of one method over the other.35 Thus the Skinnerian assumption that programs must be presented in linear formats does not seem to hold up under experimenta- tion. 0n the other hand, there is little evidence suggesting that intrinsic programing is more effective or efficient than linear. It is obvious that more investigation is necessary before further assumptions on either type of programing can be made. Small Versus Large Steps.--Another Skinnerian and linear programing assumption maintains that the program must contain a large number of small steps. The results of studies comparing small versus large step programs do not appear to support this thesis. Two early studies, a one by Coulson and Silberman,3 another by Evans, Glaser, . 7 . . . -1 . . . and Homme,3‘ suggest that smaller steps (more items to 35V. N. Campbell, Adjusting Self-Instruction Programs to Individual Differences: Studies in Cueing, Responding, and Bypassing (San Mateo, Calif.: American Institute for Research, 1961). 36 Coulson and Silberman, loc. cit. 37J. L. Evans, R. Glaser, and L. E. Homme, ”An Investigation of ’Teaching Machine’ Variables Using Learning Programs in Symbolic Logic,” Journal of Educational Research, 55 (1962), 433-52. 33 cover the same subject matter) produce significantly better performance on immediate and delayed criterion tests. 0n the other hand, recent studies by Shay,38 40 Smith and Moore,39 Briggs, et. al., have found no statistically significant differences between the small and large groups on criterion performance. However, the inter-study variance in the research parameters in the above mentioned studies, as in most of the experiments reported, makes it difficult to assess the full signifi- cance of the experimental findings. For example, the two earlier studies presented programs in elementary psychology and symbolic logic to college students, while the latter studies employed programs in spelling, Roman Numerals, and the structure and function of the U. N., and presented them to elementary school children. The Shay study defined an item step size in terms of the ”difficulty of giving the correct answer" and the criterion measure was the number of 380. B. Shay, ”Relationship of Intelligence to Step Size on a Teaching Machine Program,” Journal of Educational Research, 52 (1961), 98-103. 39Smith and Moore, ”Size of Step and Cueing,H Programmed Learning: Theory and Research, 202-6. AOBriggs, et. al., ”Experimental Results Regarding Form of Response, Size of Step, and Individual Differences in Automated Programs,” Programed Learning and Computer- Based Instruction, ed. J. E. Coulson (New York: John Wiley and Sons, 1962), 86-98. 34 errors on the program. In the Smith and Moore study, pictorial and non-pictorial cue programs were used and the difference between a small step and a large step was three frames.l‘Ll In view of these interstudy experimental variations and in light of the contradictory findings, it would seem that further research is required before this programing assumption can be discarded or assumed. Scrambled Versus Ordered Sequence.--The a priori assumption by Skinner that the optimum program is one in which the items are presented in a logical sequence has also been the focus of recent investigation. In a previous study, according to Roe, Case, and Roe, a student failed to read the instructions in his programed text and worked through the program by reading down each page rather than moving along the horizontal divisions from page to page.42 In Spite of this procedure, the student scored high on the criterion test. 4iGoldstein ”wonders if a difference in step size of only two or three frames per word can reasonable be expected to produce any significant effects. Also question- able is the assumption that the presence or absence of a pictorial cue presented only once for each new word is sufficient to differentiate between programs. . .Failure to control for 'cheating‘ could obscure a significant rela- tion between error rate and step size on cueing. An experiment which used *more difficult‘ subject matter, had greater range in step size and cueing, and controlled for 'cheating' might give results different from those of this study.” (”Recent Research,” Programed Instruction, 2 [December, 1962], 6.) 42V. K. Roe, R. W. Case, and A. Roe, ”Automated Teaching Methods Using Linear Programs,” Programed Instruction, 1 (October, 1961); 7. L10 \J“! In 1962 V. K. Roe conducted an experiment to deter- mine whether sequential ordering of frames in a program is better than random ordering of the same items.43 She used a 7l-frame, multiple choice program designed to teach elementary probability. The results indicated that the sequence of items had no significant effect on variables that were measured. The researcher also found that prior mathematical aptitude did have a significant effect on both the error—score and the criterion test score. Although the assumption that sequential ordering of frames in a program is better than random ordering of the same items has been considered axiomatic by many theorists, the above study, along with those of Gavurin and Donahue, 45 and Levin and Baker, does not seem to support this principle. 0n the other hand, the scrambled format used in these studies may be viewed as a series of ordered sequences in which the student pigeonholes in his mind the frames relating to specific concepts. This idea may be posited on the basis of the small number of frames 43V. K. Roe, HScrambled vs. Ordered Sequence in Auto-Instructional Programs,” Journal of Educational Research, 52 (1961), 98-103. 44 . E. I. Gavurin, and V. M Donahue, ”Logical Sequence and Random Sequence,” Automated Teaching Bulletin, 1 (1961), 3-9. u 5G. R. Levin and B. L. Baker, ”Item Scrambling in a Self-Instructional Program," Programed Instruction, 1 (April, 1962), A. 36 and concepts generally employed in these studies. It appears that studies employing large numbers of frames and concepts should be conducted before final judgment can be passed on the assumption of ordered sequencing of program frames. Immediate Versus Delayed Knowledge of Results.-- Another Skinnerian assumption and one held by most prOponents of programed instruction is that the program must provide the learner with immediate feedback or confirmation of results concerning the correctness of his response to each item. Results of the early ”punchboard” studies indicate the necessity for such feedback. Little found performance measures for groups which received immediate knowledge of results superior to those groups whose reSponses were scored and returned the next day.“6 Briggs attempted to verify the effectiveness of knowledge of results.247 While Little demonstrated the importance of immediate feedback by using a machine and non-prOgramed set of multiple choice items, Briggs used the punchboard device (also employing a non-programed set of multiple choice questions). He found that the 46Little,_loc. cit. A7Briggs, loc. cit. punchboard used in the regular manner, i.e., working each item until a correct answer is found, was superior to the punchboard group in which the student was allowed only one try at each item, and was then informed as to the correctness of that one try. Both modes of knowledge of results were superior to the regular test procedure in which no knowledge of results was given. Angell conducted an experiment to determine the effects of immediate and delayed knowledge of quiz results on three types of learning in freshman chemistry: facts and prin- ciples, application of facts and principles in non—quantita- tive problems, and application of facts and principles in quantitative problems.’48 Students in the experimental group used punchboards on three hour-long quizzes. The control group used machine-scored (IBM) answer sheets and received no information concerning the quiz results until the next recitation section meeting. The criterion of improvement used was scores on the final examination of the course. Angell found that the experimental group did significantly better than the control group at the .01 level. The necessity for immediate feedback or knowledge of results for non-programed multiple choice items seems to 248Angell, loc. cit. 38 be well documented. Whether such immediate confirmation of results is critical where the program is the sole source of instruction is an experimental question which has recently received attention. Meyer presented a lO-lesson program designed to teach Latin prefixes in English words through a programed textbook to 44 eighth grade students.z1L9 An immediate— knowledge—of-results group was compared to another which received confirmation after a 24-hour delay. Meyer found that the delayed~knowledge~of-results group made almost twice as many errors on the program and scored lower on the criterion test than the group receiving immediate confirmation of responses. In a study investigating certain programing characteristics, Evans, Glaser, and Homme considered the question of immediate versus delayed feedback in terms of . .. . _ I:0 minutes rather than hours or days.” The researchers employed a program in symbolic logic in which most items required more than one reSponse. The subjects were not allowed to check their reSponses until they had completed 498. E. Meyer, ”A Test of the Principles of iActivity,* 'Immediate Reinforcement,“ and 3Guidance' As Instrumented by Skinner‘s Teaching Machine,” Dissertation Abstracts, 20 12 (1960), 4729—30. 50Evans, Glaser, and Homme, ”An Investigation of 'Teaching Machine"Variables Using Learning Programs in Symbolic Logic,” Journal of Educational Research, 55 (1962), 433-52. 39 all reSponses to that particular item. This procedure delayed the feedback between reSponse and response confirmation; the delay averaged about two minutes, with a range of thirty seconds to five minutes. The immediate and delayed groups were also compared to a group making no response (items in the form of complete sentences). The results showed that delay of confirmation or knowledge of results from 30 seconds to five minutes resulted in no significant performance decrement. In fact, this study indicated that perhaps the response (and therefore feed— back) is not necessary since the no-response group (read complete statements) scored as well on criterion per- formance as the immediate and delayed feedback groups. A complete discussion of types of responses will be presented later. A recent study by Ripple used a l34~frame linear programed textbook of the constructed reSponse type to teach the background, basis, and techniques of programed instruction to college students (N2240).51 Of the various conditions employed, two are relevant to this discussion. In one condition the constructed responses of the subjects were immediately confirmed; in a second condition the subjects responded but received no feedback. 51'R. E. Ripple, ”A Comparison of the Effectiveness of a Programed Text With Three Other Methods of Presentation,” PrOgramed Instruction, 2 (May, 1963), 6. A0 A 50-item (25 multiple choice and 25 completion items) served as the criterion measure; it was administered two days after training and again ten days later. No significant differences emerged between the two treat— ments. An interesting sidelight to this study was that the error rate of the group receiving feedback was ”conspicuously” less than the groups which did not receive feedback. Results of other studies also seem to suggest that there is no significant difference between groups with and without knowledge of results.52 Goldstein points out that the effectiveness of immediate confirmation of results may depend upon the program used, since Hsome programs, particularly those employing conversational chaining have "53 the answer imbedded in succeeding frames. In such a program, deleting words and giving the answer in separate answer frames has little value. Perhaps obvious program items or frames should not require a reSponse and the subsequent confirmation of results. Results of recent studies investigating overt and covert responding, and reading prOgrams (a linear program without deleted words-— 52Smith and Moore, loc. cit. J. F. Peldusen and A. Birt, ”A Study of Nine Methods of Presentation of Programed Learning Material,” Journal of Educational Research, 55 (1961), 460—71. 53Goldstein, ”Recent Research," Programed Instruction, 2 (December, 1962), 6. in frames are in the form of complete statements which the student reads) suggest that good conversational chaining programs are capable of teaching without the use of reSponse-making or knowledge of results. A full discussion of the research findings on this subject will be presented in Chapter III. The ReSponse Mode.—-There has been considerable discussion in the literature of the relative merits of the different response modes.54 Moreover, a considerable body of research has emerged attempting to evaluate the effectiveness and efficiency of these modes. Skinner requires that the student actively reSpond to an item by constructing a written reSponse° Pressey and Crowder (with modifications) suggest that picking a multiple choice alternative issufficient. A third type of response is offered by Evans, Glaser, and Homme which requires that the subject mentally compose the reSponse called for in the program item without recording it; this type of responding is called ”implicit” or "covert” 55 responding. SALD. Porter, ”A Critical Review of a Portion of the Literature on Teaching Devices, ” Harvard Educational Review, 27 (1957), 126-47. W. Deterline, "Response Mode: Different Effect or Different Purpose?” AID, I (September, 1961), 45, R. s. Hatch, "More On the Respdnse Mode Controversey,” AID, 1 (December, 1961), 4—5. 55Evans, Glaser, and Homme, ”A Preliminary Investi- gation of Variations in the Properties of Verbal Sequences of the 'Teaching Machine‘ Type,” Teaching Machines and Programed Learning: A Source Book, 486—96. 42 Research in the area of reSponding may be grouped into five areas of investigations (1) multiple choice versus written or constructed reSponse; (2) overt versus covert reSponse; (3) overt versus reading; (4) covert versus reading; and (5) overt versus covert versus reading. The last three areas will be discussed in Chapter III. Constructed Versus Multiple Choice.—-Of the four studies reported here, two found that constructed response programs produced more learning than multiple choice programs when the criterion measure was a constructed response type test. The other two found no significant difference. Fry used a non—continuousediscourse type program designed to teach Spanish words and phrases to I50 ninth grade students.56 Two post~tests, one given immediately after traininge-the other given two days later, served as criterion measures. Both tests consisted of equal numbers of multiple choice items and constructed reSponse items. ReSponses to the multiple choice items all approximated the maximum possible score and hence did not reflect any significant difference. On the other hand, the constructed response items showed significant results 56E. Fry, ”Teaching Machines: An Investigation of Constructed Versus Multiple~Choice Methods of Response,” Automated Teaching Bulletin, 1 (1959), 11—12. 43 favoring the constructed response mode of training. The results of this study also showed that in the condition where time was allowed to vary, the constructed reSponse program took significantly longer than the multiple choice group. Coulson and Silberman used a program on elementary psychology and presented it to college students through a simulated teaching machine.-57 They found that the multiple choice response mode took significantly less time than the constructed response mode during training. No significant difference was obtained between reSponse modes on the criterion tests for the branching or intrinsic procedure. When a linear procedure was used, the constructed response mode was superior to the multiple choice mode on a lQ-item constructed response test. Using a 192~frame program on elementary probability, Roe, et. al., compared the constructed response and multiple choice modes of responding.58 The program was presented through teaching machines allowing students to proceed at their own rate of reSponding. The researchers found no achievement differences (20«item posttest); but they did find the usual time saving of multiple choice responding over written reSponding. 57Coulson and Silberman, loc. cit. 58A. Roe. et. al,, "Automated Teachin Methods Using Linear Programs,"*Programed Instruction, 1 May, 1961), 6. 44 Evans, Glaser, and Homme presented a 72—frame program in symbolic logic to 60 college students.59 No significant difference was found between the two modes or response on both the immediate and one-week delayed posttests (three l50—item tests). As in the case of the other three studies reported, the multiple choice format required less time to complete the program. Overt Versus Covert ReSponding.-—Of the three studies reported below, one found a significant difference favoring overt responding while the other two found no significant difference between the two reSponse modes. Evans, Glaser, and Homme compared overt (written) and covert (implicit) responding using a ”Fundamentals of Music” program.60 The authors found that the covert group took less time to complete the program and did better, though not significantly so, on the posttest. Cummings and Goldstein, in a more recent study, presented a 119—frame program (Diagnosis of Myocardial 61 Two groups Infraction) to 63 college students. were compared: overt (written) responding and covert 59Evans, Glaser, and Homme, ”An Investigation of 'Teaching Machine' Variables Using Learning Programs in Symbolic Logic.” Journal of Educational Research, 55 (1962) 433—52. 6OEvans, Glaser, and Homme, ”A Preliminary Investiga- tion of Variations in The Properties of Verbal Sequences of The 'Teaching Machine2 Type,” Teaching Machines and Pro- gramed Learning: A Source Book, 486-96. 51A. Cummings, and L. s. Goldstein, ”The Effects of Overt and Covert Responding on Two Kinds of Learning Task," Programed Instruction, 2 (April, 1963), 7. 45 (mentally composing) responding. The results of the immediate and ten-day posttests reveal that the overt group achieved significantly higher scores on the pictoral and verbal sections of the tests, but that the differences for the verbal sections were not as great as the pictoral sections. On the performance criterion, however, the covert group was far superior to the overt group (the overt group required a mean time of 98.8 minutes to complete the program while the covert group required only 50.3 minutes). Keislar and McNeil used a 432—frame program on kinetic molecular theory and found no differences between overt and covert responding.62 The subjects, 300 primary grade school children, were assigned to four groups. The two groups of interest to this discussion were taught individually through a teaching-machine device consisting of colored slides and a tape recorder commentary. The overt group was required to answer by pressing a button to indicate choice of alternatives; the covert group watched and listened to the commentary without physically reSponding. No significant difference emerged between the response modes on paper-and-pencil tests and individual interviews which measured retention and transfer concepts. 62E. R. Keislar, and J. D. McNeil, "A Comparison of Two Response Modes in An Autoinstructional Program With Children in The Primary Grades,” Journal of Educational Psychology, 53 (1962), 127~31. 46 Discussion.—-It is rather difficult to summarize and assess the research findings of the studies reported. This is due mainly to the large variance in the research para- meters, such as: sample size, program format, research design, conditions, etc. Nevertheless, some tentative conclusions may be drawn from the results of these studies.63 The first is that both linear and intrinsic programs are equally effective in producing learning andtfimfig in terms of time necessary to learn, the intrinsic or branching program appears to be the more efficient of the two. Secondly, in terms of programing principles, the linear model does not maintain a position of superiority over other systems and techniques. This does not mean that linear programing principles are inferion but rather that there are a number of methods of deveIOping good programs. Perhaps this accounts for the recent trend in program styles. As noted earlier, the trend is a departure from pure linear or intrinsic styles to composite programs which employ combinations of different programing principles. Finally, it is obvious that more research is necessary to further establish successful methods of programing. 63By and large, these studies generally used a control group taught by conventional teaching methods, usually the lecture-discussion technique. The results offer an im— pressive array of evidence indicating the superiority of the programed instruction groups. CHAPTER III THE RELATIVE EFFECTIVENESS OF SIX METHODS OF PRESENTING PUBLIC SPEAKING MATERIAL The experiment reported in this chapter developed out of: (1) a need for methods of coping with the growing enrollment in speech courses, particularly basic theoryu performance courses: and (2) the potential utility of programed instruction as a method of presenting lecture material. The experiment represents an attempt to inves- tigate the relative effectiveness of programed instruction as a method of presenting public speaking material. Statement of the Problem At the present time .here is no evidence which < justifies the practice of re inting lecture material "C U.) (T\ by means of vide—taped lectures rather than programed instruction. In the area of speech education, there is no evidence which justifies the use of any lecture method over programmed instruction. One of the que tions asked U) in the experiment reported here is whether there is a significant difference in learning and time needed to learn when the same lecture material is presented by televised video~taped lectures or programed instruction. The effective difference between these two methods is that in the video—taped lecture presentation the rate 47 48 of material is not controlled by the student, nor is he required to respond actively to the material; in the programed instruction method the student proceeds at his own rate of speed and must either actively reSpond to the material or read it. A second question considered in the experiment is whether it is important for the learner to make a reSponse during the learning process. Recent studies in the area of programed instruction have attempted to de— termine the relative effectiveness of different methods of reSponding to programed material. The three modes of response generally compared are: (l) constructed response (writing out an answer); (2) multiple choice (choosing an alternative); and (3) implicit or covert response (mentally composing an answer). Some program theorists, for example Skinner, insist that the student must actively reSpond by constructing a written answer. Results of studies, however, have generally shown that multiple choice and covert responding do not affect the quality of learning. In fact, multiple choice and covert reSponding generally increase the efficiency of the learning performance.2 A number of studies comparing lSupra, Chapter II, 14. 2Supra, Chapter II, 34-38. 49 different reSponse modes often include a "reading group." The subjects in the reading group are instructed to read the same programed material as the response groups but with the answer already written in the blanks. Hence a ”reading program” is one in which the student reads with- out responding. The results of these studies generally show that the reading group is as effective as, and far more efficient than, any other procedure. These findings suggest that student responding may not always be necessary in certain types of programs. This question was also considered in the experiment. Three other basic questions were considered; two of them generate from the two questions stated above. One considers the question of time needed to learn in the three modes of presentation (video—taped lecture, constructed reSponse program, and reading program); the other is related to student attitude toward these modes, specifically televised taped lectures and programed instruction. The third question stems from all the others and asks whether there is a significant difference in learning, time needed to perform the learning task, and student attitude when combinations of two presentation modes are used; a video-taped lecture and a constructed response program, a video—taped lecture and a reading program, or a constructed reSponse program and a reading program. These questions generate a number of specific hypotheses which will be stated following a review of the literature. 50 Review of the Literature There is a general framework within which the questions stated above can be viewed. Although there have been no studies reported which compared programed instruc— tion and video—taped lectures, the effect of programed instruction on learning achievement and learning efficiency (time needed to learn) can be examined by a review of the literature which compared it to other lecture presentation modes.3 These studies generally reveal the superiority of the programed instruction method. The present study developed, in part, out of the need for research which compares programed instruction and video—taped lectures as modes of presentation. There have been a number of studies which investigated the relative effectiveness of mode of response and reading programs, and student attitude toward programed instruc- tion. These studies are reported below. Response Mode Versus Reading Programs.——There have been a number of studies which investigated the relative effectiveness of program groups in which the student was required to respond (overtly or covertly) to program frames,and groups that simply read the program frames as complete sentences (correct responses supplied). The 3Supra, Chapter II, 23—38. 51 overall results of these studies suggest the potential superiority of the reading program for certain types of material. The comparison of covert responding and reading programs in the experiment reported in this chapter grew out of the results of the research cited below. The twelve studies reported here may be grouped into three areas of investigation; (1) overt versus reading (and listening); (2) covert versus reading; and (3) overt versus covert versus reading. Overt Responding Versus Reading.—-Six studies were conducted comparing these two types of programs. Four indicate no significant difference between the two treatments. A fifth study indicates the superiority of overt responding with groups ”below median MA.” A sixth study found significant differences favoring overt responding but the type of reading group was not clearly defined. In all cases, the reading groups took significantly less time to complete the program. Evans, Glaser, and Homme used a 72~frame program in symbolic logic and presented it to sixty undergraduates to study the effects of differences in two overt response modes (written and multiple choice) and reading (the correct answer just below the material but within the frame).l‘L The reading group, while taking less time on “Evans, Glaser, and Homme, An Investigation of 'Teaching Machine' Variables Using Learning Programs in Symbolic Logic, Journal of Educational Research, 55 (1962), 433—52. the program, took the longest mean time for completion on the three l5—item performance tests. This difference, however, disappeared over the one week retention interval. In no case were significant achievement differences found, either on immediate or delayed posttests. Silberman, et. al., using a logic program with high school students through a computer-based teaching machine, compared written responding, reading frames in statement form, and reading papargaphs.5 Results showed that the reading groups took less time to complete the program. No achievement differences among the treat- ments were found. Gropper and Lumsdaine presented a programed lesson on body chemistry to 150 high school students through 6 television. The overt group was required to construct reSponses on a work sheet completing sentences presented on the TV screen, while an identical lesson in which the blanks were already filled—in and which the instructor read twice while the students read the sentences silently was presented to the reading group. An achievement test administered the day after the experiment and again approximately two weeks later served as the criterion 5H. F. Silberman, et. al., ”Fixed Sequence Versus Branching Auto—Instructional Methods,” Journal of Educational Psychology, 52 (1961), 166—72. 6G. L. Cropper and A. A. Lumsdaine, ”An lnvestigation" of the Role of Selected Variables in Program TV Instruction, Audio—Visual Communication Review, 9 (November—December, 1961), A48—56. measure. No significant differences emerged on either the immediate or delayed posttests between the two groups. Roe, et. a1. presented a l92~frame program on elementary probability to 186 freshman engineering students (four ability levels) through programed textbooks (horizontal format).7 The overt group constructed responsescalled for by the blanks in each item. In the reading group the blanks were filled—in with the correct answer. A 20—item constructed response test was given immediately after training. The reading group took less time than the students who constructed answers. The students in the lower quartiles did not score as high as the students in the upper quartiles; however, there is no indication that any one of the methods was better for a particular aptitude level, on the basis of either time or achievement performance measures. Using the same subject matter and type of students as Keislar and McNeil8 but with vocalization as a form 7Roe, et. al., loc. cit. In the horizontal format each page is divided into a number of panels in which the program frames appear. The learner begins at the tOp panel of page one, reads the frame, and constructs a response or chooses an alternative. After responding he turns the page and compares his answer with the correct response, which is to the left of the tOp panel containing frame two. When he gets to the last page, going from page to page using the top panel, he returns to page one and begins reading the second panel on each page. He continues this procedure until he completes the whole program. 8 Keislar and McNeil, loc. cit. 54 of overt response, Wittrock presented the kinetic molecular theory program to 80 first grade students through colored slides and tape recorder commentary.9 For the overt response group the program was structured in completion-item form with the last word of most of the sentences to be supplied audibly by the subject during a five second pause (indicated by an audible tap of a bell) in the taped commentary. After the pauses, the taped commentary presented the correct response. In the reading and listening group the slides and tapes were identical to the ones used by the overt group; the five second pause and instruction for overt responding was omitted. Subjects were divided into two individually matched groups of 40 subjects according to mental age (MA) and, as nearly as possible, according to chronological age (CA), and sex. The subjects were presented the programs in groups of ten or less on eight consecutive school days. Immediately following this group- setting instruction period, a ten minute individual standardized interview, and written (multiple choice) group test was administered. One year later the written, multiple choice test was again given to the subjects. Results show that the reading-listening group (120.2 minutes) took less time than the overt group (165.0 minutes ) to complete the program. On the immediate 9M. C. Wittrock, ”Response Mode in the Programing of Kinetic Molecular Theory Concepts,H Journal of Educational Psychology. 54 (1953). 89—93. 55 posttest the variance due to interaction between response mode and MA was statistically significant at the .05 level. The overt response mode was more effective with the below median MA subjects than it was with the above median MA subjects. On the one—year—later retention test, no significant interaction between MA and response mode emerged. The Ripple study mentioned earlier also compared an overt (written) response group and a reading group. He found that the overt group differed significantly from the reading condition on criterion posttest measures (immediate and delayed). The author concluded that active involvement required by the overt programed condition contributed to the increase of learning. However, as pointed out by Silverman in his review of the study, to evaluate this study, one must have answers to certain questions about the reading. . .was the material arranged in discrete frames or was it put in paragraph form? How long did it take the subjects to read the material? Were they allowed to read and reread the material until the time was up?11 Covert Versus Reading.~-Of the two studies on this topic, one failed to reveal any significant differences between covert responding and reading; the second found that the reading group did significantly better on the posttest. lORipple, loc. cit. H llR. A. Silverman, ”Recent Research, Programed Instruction, 2 (1963), 6. 56 The Keislar and McNeil study mentioned earlier presented slides and taped commentary to two groups.12 Both groups received instruction in group setting in which the program and commentary were presented to the subjects in a classroom (12 to 15 subjects at a time). The slides were projected on a large screen (correct answer identified by a green sticker) while the commentary was played on a tape recorder. Subjects in one group received the program in question form as in the teaching machine group; in the other group, all questions were rephrased and presented as statements. No difference in the effect of these classroom presentations on subject mastery was found.13 Silberman, et. al. presented a 6l—frame logic program to high school students through a computer— based teaching machine.14 The authors found that the l2Keislar and McNeil. loc. cit. 13In comparing these two groups with the two using individual instruction via the teaching machine, a highly significant variance (E) was found (control group included). The two individually taught teaching machine groups differed significantly from the two group—setting instruction groups. The authors concluded that ”it appears that individual instruction, whether requiring an overt response or not, was superior to group instruction in which the children observed and listened in a group setting to the same program as an audiostrip film.’ (Ibid., 130 . 14Silberman, et. al., loc.cit. 57 reading group did significantly better than the covert group on the posttest; however, no significant difference in time to complete the program emerged. Both groups took about the same time to work through the program. Overt Versus Covert Versus Reading.—-The four studies below report seven experiments comparing these two response modes with reading groups. Four experiments obtained no significant difference between the treatments. One experiment indicated the superiority of overt respond- ind over covert responding and reading. The results of a sixth experiment favor the reading group at the ”lowest difficulty level” and the overt group at the ”intermediate difficulty level” (both about the same at the highest level). A seventh experiment found the reading group superior to overt and covert responding modes. Goldbeck, Campbell, and Llewellyn investigated four conditions: overt (written), covert (mentally composing responses), Optional overt (subject may or may not compose a written response), and reading.15 A 32- frame program on light was presented to 62 eight grade students. On the immediate posttest the authors found 15E. A. Goldbeck, v. M. Campbell, and J. E. Llewellyn, Further Experimental Evidence on Response Modes in Automated Instruction, Technical Report No. 3 (Santa Barbara, California: American Institute for Research, December, 1960). 58 no significant differences among the means of the four groups; the mean for the reading group was slightly higher. The reading group was also significantly superior in terms of time spent on the criterion test, the reading group took the longest time to complete the items while the covert group required the shortest. This finding is similar to the Evans, Glaser, and Homme reported results concerning time spent on the criterion test.l6 Silverman and Alter conducted three experiments to compare modes of response.17 In the first experiment, spoken, written, and written and spoken responding (combined) were compared with one another and a reading group. The total posttest was composed of eight constructed response items and fourteen multiple choice items. As in the Fry18 and Coulson and Silberman19 studies, performance on the multiple choice and constructed response portions of the test were considered separately. Results of this study show no significant difference between the experimental groups and unlike previous studies (Fry; Coulson and l6 Evans, Glaser, and Homme, loc. cit. l7R. Silverman, and M. Alter, ”The Response in Programed Instruction,” Journal of Programed Instruction, 1 (1962). 55—78. 18Fry, loc. cit. 19Coulson and Silberman, loc. cit. 59 Silberman) overt responding was not superior to covert responding on either the constructed or multiple choice subportions of the test. Experiment II compared written responding with reading under conditions of external and self-pacing. Results indicate that neither the main effects nor the interaction were significant on the total test and on the multiple choice subportion of the test the reading group was superior to written responding. Experiment III also compared written responding with reading using a programed text and machine with a program designed to teach binary numbers. Results of this third study reveal no significant difference either in the main effects or the interaction. Goldbeck and Campbell20 reported a related finding to Wittrock.21 The authors used a 35—frame discrete discourse program on ”men of history, geography, etc.," with 63 seventh—graders. The three groups studied were: overt (constructed written responses), covert (mentally composing responses), and reading (completed frames with the critical word underlined). The subjects were assigned to three difficulty levels. They worked at their own pace and were given a 35—item, constructed response criterion test at the completion of the program. Results 20R. A. Goldbeck and v. N. Campbell, ”The Effects of Response Mode and Response Difficulty on Programed Learning,” Journal of Educational Psychology, 53 (1962),.110-18. 21Wittrock, loc. cit. 60 show an interaction between response mode and difficulty level. The overt response groups performed below the covert response groups at the lowest difficulty level and about the same at the high difficulty level. In terms of training time and achievement, the reading groups were most efficient and overt responding least efficient. A second study by the same authors employed a 32-frame continuous discourse program about light. An Option mode which permitted the subject to write his response if he was sure of it or omit it if not was also employed. This study failed to reveal signifi— cant achievement differences on an immediate posttest between the groups. However, on a ten-week retention test (same as immediate posttest—~number of items not reported) the groups did differ significantly with the reading group performing best at the .05 level. On both the immediate and ten-week retention test, the covert group scored lowest. This study differs from others comparing overt, covert, and reading groups. 23 However, it has been severly questioned by Cummings. 22Goldbeck and Campbell, loc. cit. 23A. Cummings, ”Recent Research,” Prpgramed Instruction, 2 (September-October, 1962):'4-61 61 A study by Krumboltz and Weisman found that the overt responding mode increased delayed retention.24 The authors investigated the effects of response mode on threesroupS: overt (written), covert (mental), and reading. The program, a 177—frame programed textbook, was designed to teach fundamentals of educational test interpretation. Two fifty—item completion tests served as post- and retention tests. The authors found that the written response group took 10 per cent more time than the other two groups. On the first posttest no significant difference on the three groups was found; however, on the two—week retention test, the written response group proved significantly superior to theother two groups. This result is in contrast with the findings reported by Goldbeck and Campbell,25 and Evans, Glaser, and Homme.26 Student Attitude Toward Programed Instruction.-—The inclusion of a student questionnaire in the experiment reported in this chapter developed from the results of the studies reported below. At the present time there have 2hrJ. D. Krumboltz and R. G. Weisman, ”The Effect of Overt Versus Covert Responding to Programed Instruction on Immediate and Delayed Retention,” Journal of Educational Psychology, 56 (1962), 89w92. 25Goldbeck and Campbell, loc. cit. 26Evans, Glaser, and Homme, loc. cit. 62 been few studies which have considered the question of student attitude toward programed instruction. Within the six studies reported here, programed instruction is generally compared to the lecture-discussion method. The results of these studies reveal a favorable reaction to programed instruction over other methods of learning. One of the first surveys concerned with student reaction or attitude toward programed instruction was conducted by Skinner and Holland.27 They surveyed (using a questionnaire) 187 students who worked through an introductory psychology program as part of a second— level general education course. The program was presented to the students through Skinner teaching machines.28 Responses to the five questionnaire items reported indicate a positive attitude toward programed instruction. In general, students felt that: (l) the program contributed to a better understanding of the course—textbook; (2) the program taught them more than the text; (3) if the pro- gram was not used they would have gotten less out of the course; and (4) if they were to take another intro— ductory course in a science or similar field, programs 27B. F. Skinner and J. G. Holland, ”The Use of Teaching Machines in College Instruction,H Teaching Machines and Programed Learning: A Source Book, ed. Lumsdaine and Glaser, 159—72. 28Supra, Chapter II, 11—12. 63 should be used for part of the course. In a more recent studyjNaumann reported that the attitudinal reactions of the 44 college students who worked through the first third of the Holland-Skinner psychology program were rather favorable and generally supported the correSponding data obtained with students in the Skinner and Holland study.29 Feldusen presented a 37—frame linear program (constructed response) on seven basic concepts of programed instruction to 129 college students enrolled in a general psychology course.30 Immediately after finishing the program the student was given a questionnaire. The question— naire consisted of both structured and open—end questions (number of items not reported). The questions were general and ranged from how effective the student felt the program was as a teaching device (that is, programs in general), to open-end questions asking for advantages and disadvantages of programed instruction. The results of this survey indicate a highly favorable reaction to self-instructional devices. Ninety-three per cent of the students felt that the program would be more or very much more effective than a textbook; 92.3 per cent also favored a combination of programs, textbooks, and lectures. A number of factors cast 29T. F. Naumann, ”Laboratory Experience in Educational Psychology,” Journal of Programed Instruction, 1 (1962), 9-18. 30J. A. Feldusen, ”Reactions of College Students to a Self—Instructional Teaching Device and Programed Instruction,I AID, 1 (August, 1961), 37-38. 1 64 some doubt on the validity of some of the results of this survey. Many responses to items on the questionnaire were probably prompted by information presented in the program. For example, the program teaches the student the advan- tages of programed instruction. Later, in the question- naire, the student is asked to list or discuss the advantages of programed learning. Thus, the responses to these items may be an indication of how well the program taught the student rather than how favorably the student was reacting to programed instruction. Furthermore, on the basis of this limited program (37-frame) the students were asked to generalize their opinions of programed instruction to rather broad areas (comparing it to other methods of learning, how it should be used, etc.). Geis and Peyser report a two year survey which was conducted at Hamilton College to ascertain, periodically, student reaction to a 7,000—frame program used in an introductory psychology course.31 Four periodic surveys were conducted between the academic years; 1960—1961, and 1961—1962. An average of 80 students per survey were given questionnaires (number of items not reported). The reSponses to three questions were reported; two are relevant to this discussion. In response to the open-end 31G. L. Geis and C. s. Peyser, ”A Survey of Reactions of College Students to a Program in Introductory Psychology,‘ AID, 2 (January, 1963), 208—10. 65 questions: ”Comment briefly upon this question: Do you think the program material has helped (or hurt) you in this course?” 90 per cent felt that it helped, none felt that it hurt. In reSponse to the question: ”Do you think that the time spent on the program might have been spent on some other activity in the course (e.g. lectures, outside reading, etc.)?" 59 per cent said no, 14 per cent said yes, 24 per cent said partly, and 3 per cent said equally. The most common neg- ative reaction to the program was that it was "dull, monoton— our, or unchallenging." No questions which attempted to compare the program with another specific mode of presen— tation were reported. Hughes and McNamara presented a 7l9—frame program on the IBM 7070 Data Processing System to 112 computer maintenance trainees.32 The control groups were taught by conventional classroom methods (lecture—discussion) and the instruction period covered four mornings and totaled 15 hours. The experimental groups were given programed texts and learning time was Spread over three days and totaled 11 hours. Students working on the program, however, were not permitted to proceed at their own speed. After completing the posttest, the experimental groups were asked to fillout a questionnaire (8—items) anonymously. 32J. L. Hughes and W. J. McNamara, ”A Comparative Study of Programed and Conventional Instruction in Industry,“ Journal of Applied Psychology, 45 (1961), 225—31. 66 The results of the experiment showed significant gains in learning and reduction of training time for the experimental groups using programed instruction. Results of the questionnaire indicate that the subjects were highly favorable to programed learning. Of the 70 subjects responding: 87 per cent like programed instruction more than conventional instruction; 83 per cent said that they prefer using it in future training courses; 79 per cent felt that the programed instruction method taught the material much more effectively than the regular class- room method (the other 21 per cent felt that it taught the material ”somewhat more” effectively than the regular classroom method); and 60 per cent felt that the programed instruction method required less home study than the regular classroom method. Three items in the questionnaire were of the Open-end type and asked for additional reactions to the program. Negative comments were few and were directed to the: (1) amount of repetition and written responses required; (2) absence of an instructor and class discussion; (3) amount of time allotted for the study of materials; and (4) failure of programed textbooks to pro- vide summaries or outlines of topics. In a more recent study, Hughes measured the effect of change in programed text format and reduction in class— room presentation time (from 11 to 8 hours) on learning achievement and attitude on the same type of subjects 67 33 (maintenance engineering trainees). The formats used were constructed response and covert response programs. 199 subjects were used over a period of six months. The results of the 8—item questionnaire indicate an "increased negative trainee reaction to programed instruction as the study progressed, but it did not attain a high level."34 It is interesting to note that the covert responding groups reacted less negatively to programed instruction than did overt responding groups. Summary.~—To date, no studies comparing programed instruction and video—taped lectures as methods of pre— senting lecture material have been reported. There have been, however, a number of studies comparing programed instruction and other methods. The results of these studies show that programed instruction is the more effective and efficient mode of presenting lecture material; they also suggest that the same findings may emerge when programed instruction is compared to the video—taped lecture method. The results of studies investigating response modes and reading programs consistently show that the reading program is as effective, and far more efficient, than any 33J. L. Hughes, ”Effective of Changes in Programed Text Format and Reduction in Classroom Time on the Achievement and Attitude of Industrial Trainees,H Journal of Programed Instruction, 1 (1962), A3-5A. 3L‘lbid., 43. 68 other type of program. Finally, the results of attitude surveys reveal that students respond favorable to programed instruction. It is from the foregoing literature that the experiment reported in this chapter emerged. Hypotheses The questions considered in this experiment generate a number of specific hypotheses. The rationale for each hypothesis is derived from the literature. In stating the hypotheses and in future references the presentation modes are designated by abbreviations. These abbreviations and their respective presentation mode represent the six experimental conditions employed in this study and are presented in Table 3. A seventh condition or group served as a control and is designated as C. TABLE 3 A BREAKDOWN OE PRESENTATION—MODE ABBREVIATIONS Abbreviation Presentation Mode TV video~taped lecture CR constructed response program (Program A) R reading program (Program B) TVC video—taped lecture and constructed response program TVR video—taped lecture and reading program C—R constructed response program and reading program 69 Six null hypotheses were formulated for The first three are concerned with learning ac testing. hievement (as measured by a posttest); the last three are concerned with the amount of time needed to learn. H1: There is no significant difference amount of learning among the three modes of presentation (TV, CR, and H2: There is no significant difference in the basic R). in the amount of learning among combinations of two of the basic modes of presentation and C—R). H3: There is no significant difference amount of learning among the three modes of presentation and combinati (TVC, TVR, in the basic ons of two of the basic modes (TV, CR, R, TVC, and C—R). H4: The mean time needed to complete the constructed response program (Program A) is equal to 45 minutes (time needed to video—taped lecture), against the a P'A 4.45 minutes. view the lternative: H5: The mean time needed to complete the reading program (Program B) is equal to 45 minutes, against the alternativezjuB 4 45 minutes. H6: The mean time needed to complete th e constructed response program (Program A) is equal to the mean time needed to co reading program (Program B), agains alternativezlp A £le. Method Subjects.~—A total of 223 undergraduate were used as subjects in this experiment. The were members of a one—quarter, 3 credit course public speaking (Speech 101) at Michigan State The majority of the subjects were freshmen and mplete the t the students students in basic University. sophomores. 70 Prior to the experiment, the subjects had been exposed to three video-taped lectures (during the one hour weekly lecture period) presented over a closed—circuit television system and three weeks of recitation-section meetings (three hours per week). Materials.——The materials used in the study included a video—taped lecture, a constructed response linear program, a reading program, a posttest, and a questionnaire. Video—Taped Lecture.——A 45 minute lecture on the principles and methods of outlining and organizing a speech was recorded on a standard two~inch video—tape. During the experiment the taped lecture was presented over a closed-circuit television system by an Ampex—lOOOC Video Tape Recorder. Constructed Response Linear Program.—-A 43-frame linear program based on the identical material used in the video—taped lecture was developed and tested. The frames were typed on a 5 x 8—inch index cards with the correct responses on the back. The program was presented to five volunteer undergraduate college students; none of the subjects had any previous public speaking training. Each subject wrote his responses on a separate answer sheet and then compared them with the answers on the back of the card. Subjects were instructed to note any errors in the program or reactions concerning the frames. The error rate was 71 calculated for each subject. It varied from 4 to 19 per cent, with an average error rate of 11 per cent. The initial program was revised on the basis of this testing and increased to 55 frames. The revised program was presented to seven other subjects, non of whom had any previous public speaking training. The testing procedure was the same as that used with the initial program. At the completion of the program the subjects were given a 25~item completion test based on the more salient points covered in the program. The mean score for the seven subjects was 22. The error rate was less than 6 per cent. It was felt that a low percentage of errors was necessary because of the limited.size of the program. The program was again revised and increased to 62 frames. The revised 62-frame program was stenciled and dup- licated in two forms. In the first form, Program A, the critical word(s) in the frame is deleted and the student is required to construct a written reSponse on a separate answer sheet. (Appendix A) The correct answer is located to the left of the succeeding frame. The booklet followed a vertical format34 and served as the constructed response pro- gram. 32+The frames proceed vertically from the tOp to the bottom of each page——frame one is followed by frame two, and so on. The answer to each frame appears in the left or right hand margin of the succeeding frame. The student uses a card or sheet of paper to mask the page—~exposing only the frame he is working on. After responding to the frame, he slides the mask down the page to expose the correct answer and the next frame in the sequence. 72 Reading Program.--In the second form of the revised 62-frame program, Program B, the frames were identical to Program A except that the critical word(s) was underlined rather than deleted. (Appendix B) Hence programs A and B were identical in content structure except that in the latter the student reads each frame without constructing a response (either in writing or mentally). Program B also followed a vertical format but no answer appears in the left hand margin of the succeeding frame because no reSponse is required. Posttest.-—A forty—five item (5 alternative) multiple choice test was constructed to measure factual understanding of outlining and organization in public speaking. (Appendix C) The Split-half reliability coefficient (even item scores compared to odd item scores) as computed by the Spearman Brown prophecy formula was .965. (See Appendix F for computational formula.) Questionnaire.-—An eight—item questionnaire was con- structed to measure student attitude toward programed instruction when compared to the regular televised lecture. (Appendix D) The questionnaire consisted of five structured and three Open-end items based on a questionnaire developed by J. L. Hughes.35 Each of the five alternative items consisted of a five—point descriptive scale. The open—end items provided for additional positive and negative reSponses. 73 Procedure.-~Subjects were assigned at random to one of seven groups——six experimental and one control. The experimental conditions for these groups are discussed below. Experimental Condition I (TV, N=30): Subjects viewed the video—taped lecture (45 minutes). The subjects were not told that they were part of an experiment. Experimental Condition 11 (CR, N=37): Subjects in this condition were given a COpy of Program A (constructed reSponse) and told that they were part of a study designed to investigate new methods of presenting lecture material. Experimental Condition III (R, N=32): Subjects were given a COpy of Program B (reading) and the same explanation of the experiment as in Condition II. Experimental Condition IV (TVC, N=31): Subjects first viewed the video—taped lecture. After the lecture each subject was given a copy of Program A and told that the experiment was aprt of a study to determine new methods of presenting lecture material. Experimental Condition v (TVR, N=3l): Subjects first viewed the video-taped lecture. Following the lecture each subject was given a copy of Program B and an explanation of the study. Experimental Condition VI (C-R, N=32): Subjects were first given a copy of Program A and an explanation of the experiment. After finishing and returning Program A, each subject was given a cOpy of Program B to work through. 74 Control Condition (C, N=30): Subjects in this condition were given only the posttest. On the evening of the regularly scheduled televised lecture period, the seven groups met in seven different rooms. Two television sets were in those rooms where the groups watching the video—taped lecture were located. A faculty supervisor was assigned to each room but restricted as much as possible (turning on the television sets, passing out the programs, etc.). They remained at the front or rear of the romn and answered questions as briefly as possible. Subjects working through the programs were instructed to note the time they began and finished working through the program and to hand them in to the instructor as soon as they were through. Immediately following the learning task (televised lecture, and/or program), each subject was given the post— test. The subjects were told that the results of the test would not affect their grade in any way. No test of retention was given because subjects continually applied the material learned in later classroom work. At the completion of the posttest those groups that worked through either form of the program or both (all groups but TV, and Control) were given the questionnaire. Three measures were recorded: (1) a posttest score—— a measure of amount learned; (2) a time score—-the amount of time the subjects took to complete the learning task; and 75 (3) an attitude measure——a measure of reaction to the learning method. Results Posttest Data.——For greater ease in analyzing the data the number of subjects in each group was reduced to 30. Where the number of individuals in any given group (n1) exceeded 30, n — 30 subjects were discarded using i a table of random numbers. Hence, of the original 223 subjects, the scores of 13 were discarded. It is doubtful that the retention of scores for these subjects in the computations would have significantly altered the results. Table 4 presents the posttest—score means and variance for all seven groups. The mean posttest scores for the control and experimental groups ranged from 15.30 to 37.27 out of a possible 45. (See Appendix E for complete listing of raw scores within each group.) The variance for the control and experimental groups ranged from 8.20 to 14.91. The homogeneity of variance among all groups was tested by Hartley's Max-Min test.36 The resulting F—max ratio was equal to 1.817. The probability of F-max exceeding 3.02, given 7 groups with 30 individuals in each group, was equal to .05. The hypothesis that there is no significant differences among the variances 6 3 H. O. Hartley, ”The Maximum F. Ratio as a Short—cut Test for Heterogeneity of Variance,” Biometrika, 37 (1950), 308-312. 76 of the 7 groups (06:: .05) was accepted. Therefore, the assumption of homogeneity of variance which underlies the analyses of variance used in this study was probably not violated. TABLE 4 MEANS AND VARIANCES OF POSTTEST SCORES TV CR R TVC TVR C-R C Mean 26.60 36.70 36.53 36.67 36.70 37.27 15.30 S2 9.08 10.98 8.33 9.40 17.15 8.20 14.91 Analyses of Variance.--In order to evaluate the significance of the differences suggested by Table 4, three 37 The significance analyses of variance were undertaken. level of .05 was used in all cases. The results of these analyses are presented in Table 5, 6, and 7. (See Appendix F for computational formula.) Table 5 represents the test of H1: There is no .significant difference in the amount of learning among the three basic modes of presentation (TV, CR, and R). The F2_87.needed for significance (CT = .05) is equal to approximately 3.10. Therefore, the first hypothesis was rejected at the .05 level of significance. 37W. J. Dixon and F. J. Massey, Introduction to Statistical Analysis (New York: McGraw—Hill, Inc., 1957) 145—152. 77 TABLE 5 ANALYSIS OF VARIANCE FOR GROUPS: TV, OR, AND R Sum of Mean Source df Squares Square F Between groups 2 2007.09 1003.55 106.08 Within groups 87 822.97 9.46 Total 89 2830.06 Table 6 represents the test of H2: There is no significant difference in the amount of learning among combinations of two of the basic modes or presentation (TVC, TVR, and 0—H). The F needed for significance 2—87 (0( = .05) is equal to approximately 3.10. Therefore, the second hypothesis was accepted. TABLE 6 ANALYSIS OF VARIANCE FOR GROUPS: TVE, TVR, AND C—R Sum of Mean Source df Squares Squares F Between groups 2 43.39 21.75 2.14 Within groups 87 884.17 10.16 Total 89 927.66 Table 7 represents the test of H3: There is no significant difference in the amount of learning among the 78 three basic modes of presentation, combinations of two of the basic modes, and a control group (TV, CR, R, TVC, TVR, 0—H, and C). The F6—2O3 needed for significance (0( = .05) is equal to approximately 2.14. Therefore, the third hypothesis was rejected at the .05 level of significance. TABLE 7 ANALYSIS OF VARIANCE FOR ALL SEVEN GROUPS Sum of Mean Source df Squares Square F Between groups 6 12692.19 2115.37 200.70 Within groups 203 2139.43 10.54 Total 209 14831.62 Data Snooping.——In order to test for significance of difference among the means of any two groups, the Scheffé test was employed whenever the differences among means were shown to be significant by an analysis of variance. (See Table 5 and 7.) The Scheffé method was used for two reasons: (1) ”it is closely linked to the F—test and requires only the F table”; and (2) ”the frequently advocated Duncan p§y_ multiple range test is currently under suspicion by mathe- n38 matical statisticians. The results of these tests are presented in Table 8 and 9. (See Appendix F for computational formula.) 38Q. McNemar, Psychological Statistics (New York: John Wiley and Sons, Inc., 1949), 286-87. 79 Table 8 represents the Scheffé test of differences for the three basic modes of response groups (TV, CR, and R). Each cell in this table presents the mean for the column group under consideration minus the mean of the correSponding row group. For example, the cell correSponding to the column CR and row TV, which is equal to 10.10, was found by subtracting the mean of TV (26.60) from the mean of CR (36.70). (See table 4.) According to the results of the Scheffé test, when absolute values in the cells are equal to, or greater than, 6.25 (OC =.O5), the hypothesis that the two means under consideration are equal should be rejected. These values are indicated by asterisks. TABLE 8 SCHEFFE TEST OF DIFFERENCES FOR GROUPS: TV, OR, AND R Columns Row TV CR R TV ___ 10.10* 9.93* CR —.17 Table 9 represents the test of difference for all seven groups (TV, CR, R, TVC, TVR, C~R, and C). As in Table 8, the row means have been subtracted from the column means. According to the results of the Scheffé test for these seven groups, absolute values equal to, or greater 80 than, 9.50 (°( = .05), result in the rejection of the hypothesis that the two means being considered are equal. These values are indicated by asterisks. TABLE 9 SCHEFFE TEST OF DIFFERENCES FOR ALL SEVEN GROUPS Row TV CR R TVC TVR C-R 0 TV _ 10.10* 10.07* 10.07* 10.10* 10.67* —11.30* CR _ —.17 —.03 .00 .57 -21.40* R __ .14 .17 .17 41.23.; TVC _ .03 .60 ' —21.27* TVR __ .57 -21.40* C—R __ —21.97* C Time Data.——The time required to complete the learning task in the control and experimental conditions was recorded. In those conditions where more than one task was employed (TVC, TVR, and 0—H), separate time scores were also recorded for each task. On each program form, the subject was asked to indicate the time when he began and ended working on the program. These time scores varied from subject to subject and from condition to condition. The time required to view the video—taped lecture was 45 minutes. A breakdown of experimental condition, task, and time needed to learn is presented in Table 10, 11, and 12. 81 Table 10 presents a listing of total time, mean scores, and variance for groups that worked through Program A (constructed reSponse). The time scores in this table represent only the time needed to work through Program A in the condition in which it was used. Thus, the mean time for working through Program A in condition C—R was 39.83. TABLE 1O COMPLETION TIME (MINUTES) FOR GROUPS USING PROGRAM A Condition Task Tifigl TIE: S2 CR A 1207 40.23 C-R A 1195 39.83 TvC A 1076 35.87 Total A 3478 38.64 45.87 Table 11 presents a listing of total time, mean scores, and variance for groups that worked through Program B (reading). Table 12 presents a listing of task(s) and total time needed to learn in all six experimental conditions. The standard deviation of the sample mean for Program A (constructed response) was found to be equal to 4.84 02 ( § ), while the correSponding value for Program B (reading) equaled 1.13. H4 (The mean time needed to complete Program A is equal to 45 minutes, against the alternative: p.3145 A fl 82 TABLE 11 COMPLETION TIME (MINUTES) FOR GROUPS USING PROGRAM.B. Total Mean Condition Task Time Time ~32 R B 500 16.67 C—R B 511 17.03 TVR B 398 13.26 Total B 1409 15.66 .10.70 TABLE 12 MMHEHDNTME(umwms)mmlmLEmmmmmwm.mwms Condition Task Time Total Time Tv TV 45 45 OR A 40.23 40.23 R B 16.67 16 67 TVC TV + A 45 + 35.87 81.21 TVR TV + B 45 + 13.26 58.26 C—R A + B 39.83 + 17.03 57.26 minutes.) and H5 (The mean time needed to complete Program B is equal to 45 minutes, against the alternative: P-B‘445 minutes.) were then tested (one-tailed) in the following manner.39 (See Appendix F for computational procedure.) 39Dixon and Massey, op. cit. 81—91. 83 The mean time needed to complete Program A and that needed to complete Program B were each subtracted from 45 minutes (time needed to View the video—taped lecture). These values were in turn divided by the corresponding sample— mean standard deviations. The resulting values or critical ratios were found to equal 1.32 for Program A and —26.01 for Program B. When a critical ratio is found to be equal to, or less than, —l.65 the hypothesis that }lo is equal to a constant should be rejected at the .05 level of significance. In order to test H6 (The mean time needed to complete Program A is equal to the mean time needed to complete Program B, against the alternative: }1A 74}lB.) a t—test 40 (See Appendix F for computational formula.) was employed. The value for t was found to equal 23.7. Given 178 (N1 + N2 —2) degrees of freedom, when t is found to be equal to, or greater than, approximately 1.97 (O( = .05), the hypothesis that the two means are equal should be rejected. Therefore, H6 was rejected at the .05 level of significance. Questionnaire Data.——After completing the posttest, all groups that worked through either Program A or Program B or both, were given an 8—item questionnaire (CR, R, TVC, TVR, and C—R). Although the subject responded to the ”01bid., 121—22. 84 questionnaire anonymously, he was asked to Specify the form of program he used and the room in which he was located. This provided a means of measuring attitude by treatment. The data were recorded on IBM Punch Cards and grouped by means of an IBM card—sorter machine. Table 13 presents a summary of percentages of questionnaire responses on the first five items. This table includes the responses of the 13 subjects that were discarded. The last three items on the questionnaire were open— end items and called for additional positive and negative reaction to programed instruction as compared to the regular video—taped lectures. The percentages of responses to these items was 100. The most frequent favorable response (84 per cent) to programed instruction by the CR (construc- ted response) and R (reading) groups was that it was "repetitious." On the other hand, the most frequent negative response (79 per cent) by the TVC, TVR, and 0—H groups was that the program was ”too repetitious." Discussion Effect on Learning Achievement.——Following the computation of three analyses of variance, H1 and H3 were rejected, while H2 was accepted. The F of H1 and H3 was found to equal 106.08 and 200.70 respectively which was well beyond the .05 level of significance. 85 SN 0: wN MN NN :N mmOH floss .m :N Hm mN m: mm Nm mmoH panamsom .2 mm 0H mm mm em em damn one nsons .m NH MH MH @ 0 NH egos pmzswEow .N N o w m m o 090E Sosfi .H aHm means pasta on as was nasoannap son Amv mH mH OH 0H NH 3H ohoE nosE .m wN mm mN mN wN SN whoa passwEom .: mH mN NH NH NH HH came one pzonm .m MN NH MN mN NH Nm mmoH steamsom .N NH 0 mN m NN mH mmoH 3058 .H ascents Hm one mesa sos on son ANV cm em mm mm mm em osanoouno once sass .m m: >3 mm Nm :3 w: wsoE pwnzoEom .: mH ma ma ma mH HH case the peeps .m w 0 MH 0 m HH mmOH pmzzosom .N m m MH m m w w>HpowMNC mmmH .H deHhmme one 50% unwrap Ewnmopg one mm: HHoz 30m AHV HoH soocam ca monzpowH >9 pstwmh ow pmamaaoo Hence m-o m>e use m mo *snomwsmo omsonom Mo moNNQQmosom CHwom use COHpmofia m2w£ anomopmo OHmom one CH mpfloEOpmpm can mGOHpmosv one “woman m>¢m oB* 5m 3m mm ON 3m s3 hemmed szconpm .m H3 >3 N3 N3 H3 SN cosmsoeosg mEom .3 w O OH O OH m memo p_soe .m OH OH MH w m 3H CoHpooon wEom .N 3 O O OH O m poompo hchospm .H spews sh ens Hm neon can op mxHH 50% OHsos “endpsw esp CH Amy OH O MH MH O HH seemed kHNCospm .m Hm 33 OH MN 3m Nm coconoeong oEom .3 NH NN m MH MH HH whwo p_cop .m mm OH N3 mm mm mm soanooHno pson .N NH 0 MN NN OH 3H poonpo HHNSOSpm .H a>e no essence Hm on: on mxHH soh pHsoz «endpse one CH A3V “HOH cocoon ca mosspooH >9 AMHSNOR one op OmanEoO Hence mic m>e o>e. m mo *snowtewo mmdogmom No mmwwpcoohwm UHwom one :oHmeSO Apossascoov mH mqm Where: SW2 2 within sum of squares mi = number of individuals in ”1th” group m.j = number of individuals in "3th” group / 2 SD = SD (2) K = 1/’ (G — 1) - (FG_1, N—G ’qf) Where: G = number of groups N—G = F required for the «I level of significance for n1 = G — l + n2 = mg — G degrees of freedom Where: D = difference between group means needed for significance 2Q. McNemar, Psychological Statistics (New York: John Wiley and Sons, Inc., 1949), 286—87. 156 ONE—TAILED TEST3 Given: H: 110 = C H: )10 C If C.R.:E_ Zl_l/2,c<:then reject H Where: (1) Zl—l/2 9C may be found from any table of 'the cumulative normal distribution (2) CR: i‘2_c 5 Where: X = sample mean C = hypothesized mean (a constant) 2: NZX2 — (Z102 ./ N (N — 1) 3Dixon and Massey, op. cit., 81—91. 157 t—test Given: H: )11 = )12 $141542 If [t]_<_ tl-l/2OC(N1 + N2 - 2), then reject H tl-LflB“C(RH_I'N2 — 2) = value of t at level of significance with N = N — 2 degrees of freedom as found from any standard t—table t _ i1 - X2 Sp-7/ l/Nl + l/N2 Where: X1 2 sample mean of first group X2 2 sample mean of second group 2 (2x11)2 2 (2X21) S2_ 2X112— N1 +2X21 _ N2 p _ N1 + N2 — 2 _ 2 Nl = number of individuals in first group N2 = number of individuals in second group 4 Ibid., 121-22. 158 SPEARMAN-BROWN (SPLIT HALF)5 The Spearman-Brown prophecy formula is given by the formula: I’22 = 2 r13 1+I‘ij Where: N SLXY' — :EIX.: Y w/(Nzx2 — (EX)2 (NS-Y2 - (z 17)?) riJ — 5V. H. Noll, Introduction to Educational Measurement (Boston: Houghton Mifflin Co., l957), 70-71. ROOM USE (mu ROOM USE ONLY IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII lllIIlllfllllljlflllglllllll[11411111113111illlll