’ llW/[lill/lllll/lll/ll/l/Il///I////I///I//I 93 105 J HEsla 00 5106 . This is to certify that the dissertation entitled . ~ presented by has been accepted towards fulfillment of the requirements for degree in W444 MM’LIAQ/ Major professor Date Z'X’f’fJ MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771 LKE-i-fislfi'fi' ! l Mififiiififim fiiftate . i Universality J K ._ —____ W MSU LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. '..s USING THREE FORMATS OF THE SEMANTIC DIFFERENTIAL TO DETERMINE THE MOST RELIABLE FORMAT OF THE TECHNIQUE AMONG KENYAN AND AMERICAN ELEMENTARY SCHOOL CHILDREN AND INVESTIGATE THE DIMENSIONALITY OF THEIR ATTITUDES TOWARD READING BY Frank Oyungu Ingule A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHI LOSOPHY Department of Counselling, Educational Psychology and Special Education Major: Measurement, Evaluation and Research Design 1983 {37—82/ ABSTRACT USING THREE FORMATS OF THE SEMANTIC DIFFERENTIAL TO DETERMINE THE MOST RELIABLE FORMAT OF THE TECHNIQUE AMONG KENYAN AND AMERICAN ELEMENTARY SCHOOL CHILDREN AND INVESTIGATE THE DIMENSIONALITY OF THEIR ATTITUDES TOWARD READING BY Frank Oyungu Ingule Three-point, five-point and seven-point formats of the Semantic Differential technique were used to rate the concept of READING by American and African fourth to sev- enth grade children. The purpose of the study was to identify the most reliable formats of the Semantic Differ- ential for the various grades and determine whether the technique identifies the same factors (dimensions) and factor structures among the American and African children selected for the study. The American subjects were selected from two schools in Lansing, Michigan while the African subjects were selected from two schools in Kenya (East Africa). A total of 448 American and African children participated in this study. The results showed that among the American subjects, five-point and seven-point formats were more reliable than the three-point format. The five-point format was more reliable in the African fourth and fifth grade than the three-point and seven-point formats. However, in sixth and seventh African grades, the five-point and seven-point formats were equally reliable but more reliable than the three-point format. The three-point format was consistently less reliable than the five-point and seven-point formats among both the African and American subjects. The factors identified among the American subjects, in their order of importance, were, Evaluation, Difficulty (Potency), and Usefulness (Activity). The same factors were identified as characterizing the attitudes of African subjects toward reading but the order of these factors were changed. Factor I among the African subjects was Usefulness (Activity), Factor II was Evaluation and Factor III was Difficulty (Potency). Results of the factor reliabilities showed that the most reliable factor among the American subjects was Eval- uation, followed by Difficulty (Potency), then Usefulness (Activity). The most reliable factor among the African subjects was Usefulness, followed by Evaluation and then Difficulty. The results of this research cannot be generalized beyond the subjects used because it is based on a fixed design model. The information on the dimensions is re- stricted to READING since this was the concept that was rated. To my wife, Eunice, whose love inspires the best in me. ACKNOWLEDGMENTS It is with grateful acknowledgment and appreciation for their many contributions in various stages of this dissertation and my graduate work: Professor William Mehrens, Professor, Measurement and Re- search Design, my advisor and the chairman of my disserta- tion committee, for his patience, availability and encour- agement throughout my graduate work at Michigan State University and for his vital assistance in various stages of this dissertation. I will always remember him as one of the best and most efficient scholars I have met in my life. Dr. Joe Byers, Professor, Professor of Educational Psychol- ogy, member of my committee for giving me an initial orien- tation in the use of computers for research and for his continued support and guidance throughout my studies at Michigan State University Dr. Richard Houang, Assistant Professor, Educational Psy- chology, member of my committee, for his determined efforts to introduce me to various quantitative and statistical aspects of research. He has been a valuable friend to me during studies at Michigan State University and has made substantial contributions to my development. Dr. David Dwyer, Professor, Linguistics, Anthropology and iii African Studies, member of my committee, for his encourage- ment and supportive ideas. He particularly helped me focus my research attention on my beloved people of Africa. My friend Dr. Paul Virts and Viola Virts, Dr. Gary Waldon and Ev Waldon, Dr. Tom English, Dr. Betta Mengistu, Dr. Lars Dunberg, Tom and Sue Hood, Donna Birkey, Reverend Charles Landon and Phylis Landon, for their support and encourage- ment during my graduate studies. Richard Skinner, superintendent, Lansing Christian Schools, John Omuyak, Elijah Mbalanya and Crispo Opanga for their assistance during the data gathering stages. Finally, to my wife, Eunice, my children, Sarah, Carol and Timoth, my father and mother, for their love and under- standing. iv TABLE OF CONTENTS LIST OF TABLES O O O O O O O O O O O O O O O 0 Chapter I. II. III. IV. THE P ROB LEM O O O O O O O O O O O O O 0 Introduction Semantic Differential Measurement of Reading Attitudes Problem Statement Purpose Importance of the Study REVIEW OF LITERATURE . . . . . . . . . . Introduction Semantic Differential - A Scaling Method Multi-Dimensional Advantage Number of Dimensions Pan-Cultural Factor Structure Dimensionality Issue Among Children Dimensionality of Reading Attitudes Evaluation of the Semantic Differential Reliability of the SD Technique Summary METHODOLOGY AND PROCEDURES . . . . . . . Construction of Scales Construction of Formats Construction of Directions Translation Administration Population Scoring Analysis Concept Reliability PRESENTATION AND ANALYSIS OF RESULTS . . Introduction Comparison of Reliabilities of Formats vii 14 49 61 Reason for Using Three Dimensions for Factor Analysis Reliable Formats for American Subjects Tests of Statistical Significance Summary Reliable Formats for African Subjects Tests of Statistical Significance Summary Information on Factors Factors from American Subjects' Ratings Summary Factors from African Subjects' Ratings Summary Factor Score Reliabilities Factor Score Reliabilities for American Subjects Summary Factor Score Reliabilities for African Subjects Summary V. CONCLUSIONS, RECOMMENDATIONS AND DISCUSSION Conclusions Recommendations Discussion APPENDIX A: EXAMPLE OF SCALES AND DIRECTIONS IN ENGLISH O O O O O C O C O O 0 APPENDIX B: EXAMPLE OF SCALES AND DIRECTIONS IN SWMILI O O C O O O O O O O 0 APPENDIX C: RESULTS FROM TESTS OF STATISTICAL SIGNIFICANCE . . . . . . . . . . APPENDIX D: CORRELATION MATRICES . . . . . . . . BIBLIOGMPHY O O O O O O O O O O O O O O O O O 0 vi 125 130 135 140 142 167 10. 11. 12. 13. 14. 15. 16. 17. ‘LIST OF TABLES The Sample . . . . . . . . . . . . . . Percentage of Variance Accounted for by the First Three Factors (Americans) Percentage of Variance Accounted for by the First Three Factors (Africans) Unstandardized Reliabilities for American Subjects . . . . . . . . . Standardized Reliabilities for American Subjects . . . . . . . . . . . . . . . Unstandardized Reliabilities for African Subjects . . . . . . . . . . . Standardized Reliabilities for African Subjects . . . . . . . . . . . . . Summary on Reliable Formats . . . . . . Extracted Factors from Ratings Using Five-point Format by American Fourth Grade Children . . . . . . . . Extracted Factors from Ratings Using Seven-point Format by American Fourth Grade Children . . . . . . . . Extracted Factors from Ratings Using Three-point Format by American Fifth Grade Children . . . . . . . . . Extracted Factors from Ratings Using Five-point Format by American Fifth Grade Children . . . . . . . . . Extracted Factors from Ratings Using Seven—point Format by American Fifth Grade Children . . . . . . . . . Extracted Factors from Ratings Using Three-point Format by American Sixth Grade Children . . . . . . . . . Extracted Factors from Ratings Using Five-point Format by American Sixth Grade Children . . . . . . . . . Extracted Factors from Ratings Using Seven-point Format by American Sixth Grade Children . . . . . . . . . Extracted Factors from Ratings Using Three-point Format by American Seventh Grade Children . . . . . . . . vii 56 65 66 66 67 70 71 73 76 78 79 81 82 84 85 87 89 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. Extracted Factors from Ratings Using Five-point Format by American Seventh grade Children . . . . . . Extracted Factors from Ratings Using Seven-point Format by American Seventh Grade Children . . . . . . Extracted Factors from Ratings Using Three-point format by American Fourth Grade Children . . . . . . Extracted Factors from Ratings Using Five-point Format by African Fourth Grade Children . . . . . . Extracted Factors from Ratings Using Three-point Format by African Fourth Grade Children . . . . . . Extracted Factors from Ratings Using Seven—point Format by African Fourth Grade Children . . . . . . Extracted Factors from Ratings Using Three-point Format by African Fifth Grade Children . . . . . . . Extracted Factors from Ratings Using Five-point Format by African Fifth Grade Children . . . . . . . Extracted Factors from Ratings Using Seven-point Format by African Fifth Grade Children . . . . . . . Extracted Factors from Ratings Using Three-point Format by African Sixth Grade Children . . . . . . . Extracted Factors from Ratings Using Five-point Format by African Sixth Grade Children . . . . . . . Extracted Factors from Ratings Using Seven-point Format by African Sixth grade Children . . . . . . . Extracted Factors from Ratings Using Three-point Format by African Seventh Grade Children . . . . . . Extracted Factors from Ratings Using Five-point Format by African Seventh Grade Children . . . . . . Extracted Factors from Ratings Using Seven-point Format by African Seventh Grade Children . . . . . . Summary of Factors Among African Subjects . . . . . . . . . . . . . Factor-Score Reliabilities from Five—point and Seven-point Formats by American Fourth Grade Subjects Factor Score Reliabilities from Three-point, Five-point and Seven- point Formats used by American Fifth Grade Subjects . . . . . . . viii 90 92 93 96 97 98 100 101 102 104 106 107 109 110 112 113 115 116 36. 37. 38. 39. 40. 41. American Sixth Grade Factor-Score Reliabilities American Seventh Grade Factor-Score Reliabilities African Fourth Grade Factor-Score Reliabilities African Fifth Grade Factor-Score Reliabilities African Sixth Grade Factor-Score Reliabilities African Seventh Grade Factor-Score Reliabilities ix 117 118 120 121 122 123 CHAPTER I THE PROBLEM Introduction This study attempts to investigate some psychometric properties of the Semantic Differential when the technique is used to scale reading attitudes of African and American elementary school children. The psychometric properties relate to determining the most reliable format for measur- ing the reading attitudes of children from various grades among the subjects taken from American and African cultural backgrounds. However, it is hOped that the study will not only provide some information on the most reliable SD for- mats for measuring reading attitudes but will also give some indication of the dimensionality (the underlying factors) of the attitudes among the African and American subjects. Actually, according to research findings by Leo M. Harrill (1971, p. 23), the Semantic Differential is one of the "most promising techniques" for analyzing reading attitudes. Semantic Differential The Semantic Differential Technique, as described by Osgood, Suci, and Tannenbaum (1957), was developed origin- ally for measurement of concepts. The respondent is re— quested to indicate the meaning he or she attaches to a 1 2 given concept by recording it on a rating scale (three, five, or seven-point) between carefully selected bipolar adjectives. Fishbein and Ajzen (1975, p. 74) make the following statements about the technique: Ascertaining an object's meaning was viewed as similar to playing the game of "Twenty Questions" with the respondent. Thus, to—identify the mean- ing of a given object, the respondent might be asked questions such as: "Is it hard or soft?" "Is it fast or slow?", etc. Thus, just as in the "Twenty Questions" game the ob- ject being sought is gradually identified, so the selection among pairs of common verbal Opposites should gradually establish the "meaning" of the concept. Many studies involving different concepts and diff- erent bipolar adjectives have repeatedly found three basic factors or dimensions. The first is called evaluation and is associated with scales such as good-bad, pleasant- unpleasant, beautiful-ugly, positive-negative. The second, potency, is associated with strong-weak, light-heavy, hard- soft, and the like. The third, activity, is characterized by scales such as active-passive, fast-slow, excitable-calm. These factors are mutually independent and usually account for most of the variance in the data. The dimensions (or factors) being investigated in this study are related to the subject of reading. These dimensions are, Evaluation, Usefulness, and Difficult. These factors are taken from a search of literature on the measurement of reading attitudes (e.g. Epstein, (1980), Osgood, (1969), Clarke, and NybergY' (1979)). In the use of Semantic Differential Technique, factor analysis of results from a number of scales is used to identify the number of dimensions which are being as- sessed. Thus if the rating on two scales (such as good—bad and clean-dirty) are highly correlated, these scales may be regarded as alternate measures of the same underlying dimension (or factor, or, in the case of the semantic diff- erential, meaning). Semantic Differential Technique has proved to be an effective tool for measuring concepts, attitudes and person- ality traits. There are a number of studies that show that the technique has reliable utility with regard to measuring self-concepts of students, measuring attitudes towards read- ing, identifying motivational factors in achievement, measuring students' attitudes to school subjects, for gath- ering information on effective instructional objectives, for assessing the effectiveness of some school services like counselling, etc. (Osgood, et a1. 1961; DiVesta, 1966; Donahoe, 1961, Lana, 1965; Chen, 1980; Sunal, 1975). There are other instruments such as Likert scales, Guttman cumulative technique, paired comparison, question- naires, interviews, and observation rating, which are used for the same purpose but the Semantic Differential is supe- rior because of its simplicity and potential to utilize more than one dimension (e.g. three dimensions). For instance, while an instrument like the Likert Scale can only utilize the Evaluative Dimension, the Semantic Differential can be 4 constructed in such a way that the Evaluative, Potency, and Activity, and even more dimensions can be used at the same administration. The Semantic Differential also compares very favor- ably with more sophisticated instruments like Thurstone Scales and Guttman Scallogram Analysis. For instance the correlation between the Semantic Differential and Thurstone Scales is as high as .90 (Osgood, 1961, p. 194). This means that whatever the Thurstone scales measure, the Semantic Differential can do just as well. However, while the Seman— tic Differential utilizes a multidimensional space, the Thurstone scales only utilize one dimension. This gives the Semantic Differential some superiority. Furthermore, the technique is relatively easy to administer and the D statis- tics developed by Osgood, C., George J. Suci, Percy H. Tannenbaun (1969) for analyzing the response data has an added advantage of simplicity. Research seems to show that despite the utility and simplicity of the technique there are some issues on which the user must ponder carefully before and while using the technique. For instance, research has shown that the fac- torial structures of subjects may differ according to the age of the subject. (Osgood, 1969). The Evaluative factor shapes earlier than the Potency and Activity factors. Hence although, the technique is potentially multidimensional, at an early age probably one dimension will be meaningful. Suci, (1952) has also shown in his research that the 5 factorial structures of high and low ethnocentrics differ. Bopp, (1955) shows in her research that factorial structures of normal people differ from that of schizophrenics. Measurement of Reading Attitudes In this study, the concept to be rated will be READING. Numerous studies on measuring attitude towards reading (Epstein, 1980; Alexander, 1976;- Athey, 1976;. Himmelfarb, 1974 , etc) suggest that the concept of READING has different meanings. This study focuses on the reading in the classroom, and in the general school situation like the library, and reading as a school subject. To ensure common understanding of the concept being rated for all the subjects, an exercise involving completion of a question- naire on reading is provided at the beginning of the testing for all the subjects. This exercise is aimed at providing some common definition of the concept under consideration. Problem Statement The Semantic Differential has been used in America, Canada and other western countries to generate a lot of in- formation on various aspects of school life. For instance, there has been some substantial research in which the in- strument has been used to identify the attitudes of students to school subjects in both elementary and high school (Nybergy, 1969;. Clarke, 1979; Epstein, ‘1980; Harrill, 1971; Keith, and Banning, .1968;- Davis, 1977 ). The findings from these studies are very interesting particu- 6 larly to one interested in the education of students similar to subjects sampled in these studies. Some revealing research has also been generated on the psychometric properties of the Semantic Differential in some of the above studies. This kind of research has pro- vided some information on the reliability of the technique (DiVesta, and Dick, 1966; Osgood, et a1. 1969 ), the stability of the factors (Clarke, 1979; Heise, 1969; Norman, 1959 ). Some interesting work has also been done on the developmental aspects of the Semantic Differential (DiVesta, .1966 ). However, the first question that arises is the generalizability of these research findings to non-Western subjects like African school children. Can the findings on the attitudes to school subjects and some of the psycho- metric properties of the Semantic Differential be general— ized to elementary school children from an African country like Kenya? The choice of the non-Western subjects for the study has been determined by the fact that I come from Africa and I would like to seek some information on how to use the technique within African subjects. Secondly, different formats of the Semantic Differ- ential are used in the research. Research seems to assume that the technique is a homogenous instrument. However, since there is more than one format of the Semantic Differ- ential, e.g., three-point, five-point, and seven-point scales, it is possible to postulate that these are various kinds of the same technique. The question that then arises is: What formats are most reliable for various subjects in different grades, age and even cultures? Osgood, (1961, p. 222) suggests that for (American) elementary grade children, the most reliable format is the five-point format. Does this equally apply to African chil- dren of the same grade? Additionally, it may be possible that for some grades a three-point format will provide more reliable measures while for others, a seven—point format will provide more reliable measures. The thrust of this argument can be put plainly. Which formats provide the most reliable measures of attitudes for various African and Amer- ican elementary grades? The third problem to be investigated in this re— search relates to the dimensionality issue. Dimensionality refers to the number of factors underlying the concept being rated. According to research (Osgood, 1964; Heise, 1979; Norman, 1963), the Semantic Differential reveals three underlying factors, namely, Evaluative, Potency, and Activ- ity (i.e. EPA dimensions). These dimensions are supposed to have some universality (Osgood, 1961). This study tested the universality of the EPA dimensions by using American and African subjects. To facilitate this comparison, the con- cept of READING was rated by the two groups. The comparison of the EPA dimensions was based on the educational levels within the African and American subjects. Therefore the third focus was determining whether the EPA dimensions are 8 maintained in African and American attitudes towards reading in various grades. Purpose The purpose of this study was therefore to determine four sets of things: 1. To determine the most reliable format on the Semantic Differential technique among the three-point, five- point and seven-point formats for American and African subjects. Concept reliability coefficients involving the computation of the reliability coefficients from the ratings on the combined eighteen scales on each format were to be used for comparison. 2. Determining factors (dimensions) that emerge at differ- ent grades in both the American and African subjects when the Semantic Differential is used to measure at- titudes toward READING. It was hoped that this would provide some information on the dimensionality of these attitudes in both sets of subjects. 3. Determining the order of any identified factors in terms of the most important factor, the second, and third factor at different grades and whether the order among the African subjects is different from the one among American subjects. 4. Determination of the factor(s) measured most reliably by the Semantic Differential technique among the Amer- ican and African subjects. Importance of the Study This study is important because it contributes to the study of the reliability of Semantic Differential tech- nique in two cultures (American and African), it provides guidance on how to use the technique in American and African elementary schools, the results have the potential of being generalized to other similar instruments and the results on the dimensionality of reading attitudes can be generalized to literacy programs in the third world. Contribution to Reliability of the Instrument At the heart of the evaluation of a measurement in- strument are two issues, namely, validity and reliability. Validity has to do with whether the instrument measures the construst it is supposed to measure. A number of studies exist that demonstrate that the Semantic Differential has high validity with respect to measuring attitudes personal- ity traits and concepts individuals have about given things (Sunal, _l975:_ Clarke, 1979;, Molholt, 1978;. Suci, 1952 ). This study will not concentrate on validity of the Semantic Differential but rather more attention will be given to the study of reliability of the technique. Reliability refers to the precision, or consistency with which an instrument measures. It implies the dependa- bility of the instrument for providing the measurements required. The question of reliability is important in social sciences because of existence of a number of factors that militate against a measuring instrument's consistency. In fact, while the validity is important, an instrument that provides inconsistent measures may actually not provide valid measures. This study is therefore based on an assump- tion that if an instrument that has a proven level of valid- ity can measure reliably when used in a given group of people, that instrument is appropriate for the group. Therefore the study will make a contribution in this sensitive area of reliability and, hence, appropriate— ness of various formats of the Semantic Differential 10 technique for various groups of people who differ in age, education, and even cultural background. Providing Guidance for Use of the Technique in Schools The study will further serve the purpose of provid- ing guidance on how to use Semantic Differential in the Elementary and High Schools. From the results of the study it will be possible for a teacher interested in using the Semantic Differential for instructional purposes to deter- mine the most appropriate format to use for a particular grade level. The results of this study will provide infor- mation on the format that has potential for yielding the most reliable measures for the relevant grade. In a school situation, the Semantic Differential has the following advantages (Epstein, .1980 , p. 39). 1. It allows subjects to respond anonymously, thereby in- creasing chances of receiving open and honest responses. 2. It gives subjects time to formulate reponses. 3. It is adaptable to large-scale assessment in that many people can be tested at one time. 4. It produces data which can be quantified and subsequent- ly analyzed by computer. Although the fourth advantage is irrelevant to African schools, the first three show how the technique can simplify the assessment of attitudes in African schools and definitely all the four advantages are relevant to American schools. As a matter of fact, this study emanates from a committment to helping teachers understand their students 11 better through use of simplified and appropriate tools. The utility of the Semantic Differential Technique has been well documented and demonstrated with respect to how it compares with other instruments. For instance, the tech- nique has been proven to be comparable to Thurstone Scales and Guttman's Scallogram (Osgood, 1961, p. 194). However, the simplicity that accompanies the construction of the technique and the rather straightforward 'D' statistics used for analysis of response data make the Semantic Differential a convenient tool for teachers. Most teachers can use the technique even with no training in statistics and psycho- metrics. This issue increases the importance of the deter- mination of reliable Semantic Differential formats for different grades. The determination of reliable formats has the potential of providing teachers with the simplest yet the most reliable means of measuring attitudes, etc. of the students in order to facilitate better instruction. Generalizing Results to Other Instruments This study has potential for generality to other rating instruments that use formats similar to Semantic Differential Technique. One such instrument is the Likert Scale which is administered using formats identical to the ones used in the study although a different design is used. Generalization to the Likert Scale can be done by examining the reliability coefficients associated with various formats at different grade levels in the Evaluative factor. This is 12 because the Likert scale is a one dimension scale that is based primarily on the evaluative factor. Reading Attitudes and Literacy The choice of the concept for the study was deliber- ate. READING was chosen because in most of the African countries the main educational concern is the elimination of illiteracy. In Kenya, the Government has even legislated universal primary school education with the aim of elimin- ating illiteracy. However, it would be dangerous for the various governments to think that achievement tests provide all the information necessary to monitor these literacy programs. The ultimate success of a reading program's effect- iveness should be judged not solely on the basis of how well students learn but also in terms of their attitudes to read- ing. This applies to both the African countries and America. However, even in America, while schools have begun to recog- nize the importance of developing in students positive attitudes toward reading, "it is safe to say that basic skills of instruction receive the greatest emphasis"(Epstein 1980, p.8). Measuring a program's effectiveness based on ability scores alone is limited. A more balanced goal would be achieved by investigating skills improvement together with attitude improvement. Rowell (1967, p.3) insists that "If improvement in reading skills takes place without a concurrent improvement in attitude toward reading, the progress is only partial and at best, may be of short 13 duration." Therefore, monitoring the attitudes of the various participants in the literacy programs is important. This study will provide an example of the factors that are important in measuring various attitudes toward reading. Since attitudes are learned (Lemon,l973) the literacy pro- grams should endeavor to evaluate their programs to deter- mine if the participants are changing their attitudes in a way corresponding to the improvement in their reading skills. This study provides tools on how to monitor the changes in reading attitudes at different grades and even ages. CHAPTER II REVIEW OF LITERATURE Introduction The literature review in this chapter covers various issues that are relevant to this study. First, relevant literature which focuses attention on the Semantic Differ- ential as a scaling technique is examined. Then the liter- ature on the dimensionality (underlying factors) of the technique and subsequently the dimensionality of reading attitudes is reviewed. Finally, the literature on relia- bility is reviewed. Various terms are defined during the process of this literature review. Semantic Differential - A Scaling Method Semantic Differential is a scaling technique. Therefore one way of understanding the technique is to give a little relevant background of the general scaling theory. According to Allen, and Yen (1979, p. 179), scal— ing is a branch of measurement theory that focuses on rationales and mathematical techniques for determining what numbers should be used to represent amounts of property being measured. With respect totthe Semantic Differential, one would therefore say that this is an attempt to assign 14 15 numbers to the properties being measured. Warren S. Torgerson (1958) hasproposed three approaches to scaling methods. There is the subject-centered approach (method). In this kind of scaling method, "the systematic variations in reactions of the subjects to the stimuli are attributed to individual differences in the subjects" (Torgerson, 1958.,P- 46). The subjects are the ones that are scaled. Scaling methods that fall in this category include those that diff— erentiate between individuals like aptitude, achievement, and intelligence tests. Obviously the Semantic Differential does not fall in this category. There is the Stimulus-centered (or judgement) Ap- proach. In this kind of scaling method, "the systematic variations in the reactions of the subjects to the stimuli are attributed to differences in the stimuli with respect to a designated attribute" (Torgerson, 1958, p.46). The purpose of any experiment that utilizes this kind of scaling has the major purpose of scaling the stimuli which alone are assigned scale values. The Semantic Differential falls in this category and so does the Likert technique. Hence the purpose of the Semantic Differential is to scale the various bipolar adjectives used as stimuli during the experiment. The third category is a combination of the Judgement approach and stimulus-centered approach. The approach used in this third category is called Response Approach and it involves scaling both the subjects and the stimuli. 16 Accordingly, "the variability of reactions to stimuli is ascribed to both variation in the subjects and in the stim- uli" (Torgerson, 1958, p.46) . The Guttman technique pro- vides a good example for this category. The Semantic Diff- erential does not belong to this category but this category is included in this review in order to put SD in a proper perspective with respect to the scaling theory. However, according to Torgerson, (1958) the various scaling methods classified under the same category as the SD, (which as indicated above constitutes the Judgement Approach) must have certain important metric properties. Some of these metric pr0perties include property of equal intervals within the scale and zero points falling at the same place on each scale (i.e. at the centroid). Samuel Messick (1957) has tried to investigate these properties with respect to the Semantic Differential. According to Messick, these properties have to be established for the various factor analysis studies on the SD response to be meaningful for the studies are based on the assumptions of equal intervals within the scale and the zero points fall- ing at the same place on each scale. In order to summarize Messick's findings, it may be appropriate here to briefly review what is involved in SD ratings. The SD measures people's reaction to things in terms of ratings on scales defined with contrasting ad- jectives at each end. For example, a seven-point SD scale 17 may be as follows: Pleasant - - - — - - - Unpleasant 3 2 l 0 -1 -2 -3 The SD data are coded numerically as if the scales are bipolar, equal interval scales each passing through the origin (zero in the above example). Hence the coding as- sumes equal intervals within the scale and zero points falling at the same place on each scale. This is similar to the metric properties referred to above. Accordingly, Messick, (1957) applied the method of successive intervals to SD data to determine if the assump- tion of equal intervals is warranted. He found that cate- gory boundaries were similarly spaced on all of the nine scales he considered, but not exactly in the proper pos- itions for equal intervals. One side of a scale generally has slightly larger intervals than the other side. He also found that the end intervals tend to be larger than those toward the middle. However, Messick indicates that despite the deviations from equal interval assumption, one does not go far wrong in making this assumption. The correlations between the assumed and scaled boundaries were greater than .98 for every scale considered. Therefore one could make the assumption about this metric property on the Semantic Differential. Hence the SD fits in Torgerson's (1958) class- ification of scaling methods as discussed with respect to the metric property of equal intervals. The assumption of the zero point was not 18 conclusively demonstrated in Messick's (1957) study. The study revealed that the center point of SD scales is not true zero, but rather a point lying about .2 scale units from the true zero. But according to Heise, (1969) this may be due to left-right bias which was also reflected in Messick's study. It is therefore possible that this prop- erty can be assumed to exist in the SD scales. If this is the case, then the SD has metric properties similar to those possessed by other methods in the Judgemental and even Re- sponse Scaling Categories. This means that the SD compares very well to the other scaling methods with respect to the above metric properties. Robert B. Kane (1968) has also attempted to test whether in the reordering of SD Scales one answer on the SD is substantially changed from what is wouléiotherwise be. Kane calls this kind of error the proximity error. The findings of the study indicate that the proximity error is not a problem hence the ordering of the various SD Scales does not cause a problem in SD research. Multi-Dimensional Advantage According to Charles Osgood et al. (1961) the Semantic Differential is rather special and different from some of the scaling methods discussed above in the sense that the technique can scale a stimulus in more than one dimension. For instance, methods like Likert technique and Guttman technique are only unidimensional (Osgood, et a1., (1961), 72; Snider and Osgood, i1969 ) while the SD has the 19 potential of being two-dimensional, three-dimensional, four-dimensional, etc. Factor analyses of Semantic Differential data have consistently shown that there are three major dimensions of rating responses. These three dimensions (or factors) are Evaluation, Activity, and Potency. The Evaluation dimension includes bipolar adjectives like good-bad, nice-awful, beautiful-ugly, pleasant-unpleasant, etc. The potency factor includes bipolars like strong-weak, large—small, heavy-light. The Activity Dimension includes adjective pairs like fast-slow, active-passive, sharp-dull. (Osgood, et al. .1961 , 44). One of the studies available is the one by Osgood, et a1. (1961) in which seventy-six adjective contrasts were chosen from Roget's Thesaurus and the corresponding bipolar scales were used to rate twenty different concepts. Corr- elations between the ratings on different scales were cal- culated and factor analysis was carried on these. The factors that were identified from this study were Evaluation, Potency and Activity, respectively. This sequence of fac- tors, i.e. Evaluation first, then Potency, then Activity is often referred to as EPA structure in SD data. A number of studies exist that support the EPA dimensions of the Semantic Differential. For instance, David R. Heise (1965) had 1,000 concepts rated on eight scales by Navy enlistees. Factor analyses of the data based on mean ratings for the 1,000 different words 20 yielded the EPA structure. Similarly, Bopp, J. (reported in Osgood, C. et a1. 1961) had forty schizophrenics rate thirty-two words on a thirteen—scale form. The EPA struc- ture emerged from the factor analysis of the response data. DiVesta, (1966) had one hundred concepts rated on twenty- seven scales by subjects in Grades two through seven. The EPA structure emerged though there was some tendency for Potency and Activity to merge into a single dimension up until the fifth grade. This dimension was labelled by DiVesta as the Dynamism dimension. The study by Nyberg and Clarke, (1979) on using the Semantic Differential for measuring the attitudes of Elementary grade children toward school subjects basically revealed the same EPA dimensions. Number of Dimensions According to most SD research, there are basically the three EPA dimensions (Osgood, C. et al. 1961 ; Wright, 1958 ; DiVesta, 1966 ; Heise, 1965 ). The stress on these three dimensions almost always suggests that addition- al dimensions do not exist. However, it is apparent from other research sources that the dimensionality issue is not settled. There are some studies that suggest that there may be more than three dimensions of the Semantic Differ- ential rating response. For instance, Green, and Goldfried (1965) in their study of the bipolarity of the Semantic Differential, used unipolar rather than bipolar adjective scales and found more than three factors. In a study by 21 Bruce W. Tuckman (1976) using twenty-eight bipolar adjective scales, four factors were found to characterize the data. On the other hand, the study by Wiggins, and Fishbein (1969) found that there are different dimensions according to types of subjects. Some types of subjects were found to employ a two-dimensional structure (EP), others a three- dimensional factor structure (EPA), and still others a four-dimensional structure (EPA with either the Evaluation or the Activity dimension splintering into two factors). Borgatta (1964) and Norman (1963) have found that when adjective ratings are used to assess persons, five factors emerge from the response data. Yonker, et a1. (1974) using the SD among tenth grade studentsto measure their self-concept found different factor structures for males and females. For males, the factors were evaluation and potency while for females, evaluation, activity and intellectual ability factors were identified. For the combined groups of males and females five interpret- able factors were identified. On the other hand, in the study by Hecht and Bonnie (1976) only two factors emerged from the subsequent factor analysis of SD data. Similarly, there are some studies that suggest that only one dimension emerges when the Semantic Differential is used to rate some concepts. For example, Solly and Stagner (1956), Dyer (1964), Nisbertt and Gordon (1967) have obtained measurements of self-concept using the Semantic Differential with only the Evaluation dimension 22 represented. A study by Ranson, et a1, (1973) on how students evaluate instruction revealed only one dimension. However, studies carried out by Aiken (1965), Griggs (1959), Pervin and Lilly (1967), Kubiniec (1970), Farr and Kubiniec (1972), on the measurement of the same self-concept con- firm the basic EPA dimensions. There is some form of di- mensionality dilemma (Yonker, et a1. 1974) in the Semantic Differential research. Therefore it will not be sur- prising to find different dimensions characterizing the data in this study of the reading attitudes of African and Amer- ican Elementary level students. It may be also possible to get the basic EPA factor structures plus one (or more) additional factor(s) as in the study by Patricia Maslon and Merrifield (1973) in a school situation. Pan-Cultural Factor Structure A number of studies have also been carried out to determine if the EPA (Evaluation, Potency and Activity) structure is idiosyncratic to English or whether it holds within other languages. Accordingly, Suci (1960) had illiterate Navajo, Hopi, and Zuni respondents make ratings by pointing. The data obtained revealed Evaluation and Potency factors. The Activity factor did not appear sepa- rately. This may be due to the fact that only two dimen- sions characterize the adjective pairs that were used in these languages, or it may be that Suci did not use enough Activity scales, or the concepts used did not introduce enough Activity variance. However, other studies on the 23 cross-cultural aspects of the dimensions have confirmed the EPA structure in a number of languages. For instance in the studies by Jakobovits (1966) and Osgood (1964), the EPA structure has been identified in twenty-four different lan- guages. In Jakobovits' study, fifteen languages were used. These languages were American (English), Arabic, Cantonese, Dutch, Finnish, Flemish, French, Greek, Hindi, Italian, Jap- anese, Kannada, Serbo-Croatian, Swedish and Spanish. In each of these languages, a set of fifty bipolar scales were developed and these were used to rate one hundred concepts. The factor analyses revealed the usual EPA structures in all the languages except the Hindi and Arabic. According to Jakobovits, this demonstrates the pan-cultural nature of EPA structure of the Semantic Differential. Further work to study the generality of the EPA factor structure across cultures has been conducted by Kumata and Schramm (1969) using subjects from Korean, Japanese and American cultures. In their study, Kumata and Schramm set out to investigate the following: First, given that the differential has indicated that a common EPA structure in the area of connotative meaning exists among persons in American culture, do the same semantic factors operate in the judgments of Japanese and Korean people? Secondly, do the same set of different factors operate within a bilingual (and bicultural) group? That is if an individual knows two languages, will he make 24 the same meaning judgements when he is operating in one language as when he is using the other? The subjects selected for this study were Japanese, Korean and American students. All three groups were college students at a midwestern university. Length of residence in the United States for the two foreign groups ranged from three months to six years. All groups received two administrations of the test with three weeks between testings. The Korean and Japanese groups received one form in English and the other form in their native language. The American subjects received the identical English form on both administrations. All the instructions were given in the language appropriate to the version of the test. It was assumed that conducting the administration in the manner would encourage "thinking" in the appropriate language. The administrator for each of the three groups was a member of that group. i.e. a Korean tester for the Koreans, a Japanese for the Japanese group, and an American for the American subjects. In total, thirty concepts were rated by the subjects. The scales were picked from the original study by Osgood, et a1. (1952). The results showed that the first and dominant factor (after rotation) for all six analyses was the Eval- uation factor. This first factor is in agreement with the first factor in the EPA factor structure and it accounted for thirty-four percent of the variance. The second factor 25 identified in the study was labeled as dynamism. The high- est loadings and the most restricted scales for this factor were the scales strong-weak, active-passive, fast-slow and sharp-dull. Hence this factor seems to combine both the potency and activity factors identified in the Osgood orig- inal study. Kumata and Schramm (1956) suggest that the same two factors (evaluation and dynamism) are utilized by the students from the three cultures, i.e. American, Japanese and Korean. This is true within the same language (i.e. English) and the same factors are utilized by bilinguals regardless of language used. The study conducted by Tanaka, Oyama, and Osgood (1969) using American andJapanese subjects also verified the EPA factor structure among the subjects from the two cultures. However, this study suggests that although the same factors exist in the two cultures, the scale composi- tion of semantic spaces can be to some extent modified in terms of the objects of judgement. Factors may vary with the concepts being rated. Some scales may change their meaning in the sense of factorial composition in the semantic space. This means that although the EPA factor structure characterizes the data, the same scale may change its meaning according to the concept being rated. This phenomenon was found in both the American and Japanese subjects. Gerald Elsworth et a1. (1977) using the Semantic 26 differential among Australian subjects to measure the self-perception in students for Australian Council for Educational Research identified seven factors. These factors were orderliness, warmth-supportiveness, satisfac— tion, clarity, energy-enthusiasm, non-conformity and creativity. The study seems to suggest that the kind of conceptsbeing rated determine the factors that subsequently emerge in the factor analysis. In a study conducted in South Africa by Hudson (1970), using mine laborers, it was found that the samples that had less than four years of education had not devel- oped a more than two-dimensional way of thinking about things. On the other hand, samples that had more than four years of education had developed a kind of three-dimensional way of thinking about things. This should suggest that the Semantic Differential should identify only two dimensions among the people in this particular sample that had less than four years of education while at least three dimen- sions should be identified among people with more than four years of schooling. In a separate study conducted in Taiwan and America, Lao (1978) identified the same factor structures among American and Chinese students. The basic three factors identified by Osgood characterized the responses of the subjects from the two cultures. Maclay and Ware (1970) found Evaluation, Potency and Activity factors among the Zuni and Navajo. Their sample consisted of pe0p1e ranging 27 from the teens to over sixty. Osgood (1970) found a nine- ty percent agreement between the factor structures of Navajo, Mexican-Spanish American, Anglos and Japanese. From this he concluded that "there is a world view that is relatively stable despite differences in both language and culture. . . . certain aspects of cognitive behaviour may remain stable despite cultural and linguistic differences (Issa and Dennis, 1970, p. 412)." From this brief literature review, the EPA factor structure has been confirmed in cultures like the American culture, Cantonese, Dutch, Finnish, Flemish, Australians, French, Greek, Italian, Japanese, Mexicans, Kannada, Serbo- Croatian, Swedish, Spanish, etc. This suggests that the same dimensions (evaluation, potency, and activity) should characterize the respones of both African and American stu- dents who constitute the subjects of the current study. Dimensionality Issue Among Children Applications of the semantic differential with adults are numerous. However, there are relatively fewer applications with children as subjects. Some of these applications of the technique also confirmed the basic eval- uation, potency and activity factors among children. In one experiment by DiVesta (1962) the Evaluation factor characterized the responses of children when rating color. Kagan, Hosken, and Watson (1961) found that in six- and seven-year old children semantic differential ratings of mother, father, and self, the Evaluative dimension was the primary structuring principles used in classification. 28 In another study conducted by DiVesta and Stover (1962) it was demonstrated that the Evaluative dimension was dominant even in the rating of nonsense figures by children. These three studies seem to suggest that the Evaluative dim- ension predominantly characterizes the responsesof children. In a study conducted by Ervin and Foster (1960) in which the first- and sixth-grade children rated pictures of faces, two factors, i.e. Evaluation and Potency, were identified. However the study cautions against using scales that have metaphorical meanings because these kind of scales may confuse children. Maltz (1969) in a study of SD ratings by second- fourth- and sixth-grade children identified the presence of the Evaluation, Potency and Activity factors hence con- firming the EPA factor structure among the children. This study also showed that changes in ratings became more appar- ent with increasing age differences. The meanings of concepts were less consistent in the younger children than in the older ones. In two studies conducted by Francis J. DiVesta (1969) using subjects from grades two through seven, there was evidence for the stability of Evaluation, Potency, and Activity factors down to the second-grade level. Progress- ive refinement and differentiation were evident in the unrotated principle-axes factors as reflected in the shift from a predominantly two-factor system (composed of the 29 Evaluation and Dynamism factors) among the lower grades to the three-factor EPA factor structure among higher grades. In the first study (by DiVesta, 1969) subjects were one hundred children from each of the grade two through six inclusive. One half of the sample was from a city school in a lower—middle socioeconomic area. The other half of the sample was from a campus school of a state college and was representative of a middle socio-economic area. Five fac— tors emerged from the subsequent factor analysis of the response data. However, most of the variance was accounted for by first three factors, namely, Evaluation, (40.62 per- cent), Potency (15.60 percent) and Activity (11.54 percent). The fourth and fifth factors defined dimensions involving sensory discriminations. In the second study (DiVesta, 1969), the subjects again consisted of children from grades two through seven. Six factors emerged from the subsequent factor analysis. These factors were labeled as Evaluation, Potency, Activ- ity, warmth, Tautness (represented by scales such as loose- tight, soft-hard) and Novelty-Reality (represented by scales like same-different, real-make believe, new-old, etc.) Again, the first three factors, i.e. Evaluation, Potency, and Activity, accounted for most of the variance. However, one difference between the first and the second study was that in the second study there was a gradual increase in loadings on the Evaluation factor from around .13 in the second grade to around .63 in the seventh grade. 30 Additionally, in the second study, the Potency and Activity factors were combined into one factor (labeled Dynamism) in the lower grades (two through four). However, in this second study, like the first one, from the fifth grade through seventh grades the breakdown of the Dynamism factor clearly emerged in the form of its two components, the Potency and the Activity factors. Other studies exist in which the issue of the dimen- sionality of the SD ratings among elementary grade children has been treated. For instance, Sunder (1962) using sub- jects from grades two, four and six found ten factors char- acterized the reponse data of students from the three grades. The first factor was Abstract Evaluation, the second, Con- crete Evaluation, the third, Softness, the fourth, Noise, the fifth, Quantity-size, the sixth, Primacy, the seventh, Quantity-weight, the eighth, Activity, the ninth, warmth and the tenth factor was Dryness. However, according to DiVesta, (1969) these factors represent the four factors. For instance, the first, second, third and sixth factors comprise the Evaluation dimension; the fifth and seventh factors comprise the Potency dimension; the fourth and eighth factors are related to Activity dimension; and the ninth and tenth factors make up the Warmth dimension. The Semantic Differential has also been utilized in schools and amongcfifildren for purposes other than identify- ing dimensionality of attitudes held. But from some of these applications, the dimensionality question can be 31 addressed. For instance, Richard Jantz et al.(l978) used the Semantic Differential to study the attitudes of the three to eleven year old children toward the elderly. The results show that the technique is reliable among children of these ages but more significantly, the Evaluation factor was identified. Anttonen et a1. (1977) used the Semantic Differential to monitor the attitudes of third and fifth graders and they found that evaluation factor was the most important factor and clearly distinct by as early as third grade. In summary, the above literature review suggests that the semantic space of elementary grade children can be defined by one, two, three, or more than three factors. Generally, ratings by children from lower grades (like one through four) are characterized by one or two factors. These factors almost always are evaluation and dynamism, in that order of importance. Ratings by children in higher grade (five through seven) are characterized by all the three factors in the EPA factor structure and it is possible even to identify additional factors. However, there is some evidence that the general EPA structure exists among elemen- tary level grade children irrespective of their grades (Nyberg, and Clark 1979). Nyberg and Clark used the Semantic Differential to measure the attitudes of elementary children from Alberta, Canada, toward school subjects. They identified the 32 Evaluation, Potency (which was called Difficulty) and Activity (named Usefulness) factors. Dimensionality of Reading Attitudes Since this study attempts to investigate the dimen- sionality of reading attitudes, it is only appropriate that at this point the relevant research with respect to the issue should be discussed. According to Ira Epstein (1980) investigation of the dimensionality of reading attitudes should facilitate knowledge of various factors that con- tribute to the development of these attitudes. Attitudes The review on the dimensionality of the Semantic Differential has revealed the dilemma that characterizes the research on this matter. This dilemma haunts one even in the research on the use of the SD to determine the dimen- sionality of Reading attitudes. The nature of what is involved when pe0ple talk about attitudes may be partly responsible for this dimensionality dilemma (Epstein 1980). For instance, although the concept of attitude is frequently cited. there are multiple and often confusing definitions of the term. Scott (1968, pp. 205) after reviewing various definitions attached to the term attitude concludes that “it is unreasonable to expect a single final definition of 'attitude' to emerge within the forseeable future." Similarly, Green (1977, p. 111) states that "no single definition can be found that will satisfy all 33 those who study the topic. This fact is largely a conse- quence of the broadness of the concept, which permits var- ious definitions reflecting the theoretical point of view of the individual student of attitudes." According to Donlon (1974) this broadness of the concept (attitude) may partly contribute to the dimensionality dilemma. However, in spite of the multiplicity of definitions of the concept of attitudes, certain dimensions emerge from the literature as being common to most attitudes. The dimensions seem to suggest that the EPA structure is rele- vant for characterizing data from most attitudes, although often there are additional dimensions. For instance, Epstein (1980) has identified three basic dimensions of attitudes. The first dimension is that of attitudes as mental constructs. This is inferred from the definition of attitudes by Green (1977) which views attitudes as constructs similar to intelligence. This is similar to the Potency dimension in the EPA structure. Epstein also points out an effective dimension in attitudes. This dimension is evaluative in nature and is comparable to the Evaluation factor in the EPA structure. Then there is the dimension which is supposed to predispose an individual to respond. This is equivalent to the Activity dimension in the EPA structure discussed above. Therefore according to Epstein (1980), the three dimensions from the EPA factor structure are recognizable in attitudes. Epstein points out that reading attitudes are not different. 34 Scott (1968) has provided eleven dimensions of attitudes. These dimensions involve: l. Direction: the position along the attitude continuum ranging from positive feelings at one extreme to negative feelings at the other. 2. Magnitude: the degree to which the attitude is favor- able or unfavorable 3. Intensity: the degree of commitment with which an indi- vidual maintains a particular attitude position. 4. Ambivalence: the extent to which an individual main- tains both positive and negative feelings toward the attitude object. 5. Salience: the importance or prominence of the attitude to an individual 6. Affective Salience: the degree to which the evaluative or feeling component of an attitude exerts a greater influence over an individuals' views than does the cognitive or behavioral component. 7. Cognitive Complexity: the number or "richness" or ideas an individual has about the attitude object 8. Overtness: the extent to which an attitude will be ex- pressed in action tendencies on the part of an individual. 9. Embeddedness: the degree to which an attitude is asso- ciated with or related to other concepts. 10. Flexibility: the extent to which an attitude can be modified. 11. Consciousness: the degree to which an individual is aware of a given attitude. According to Lemon (1973) these eleven dimensions can be represented by three major components: cognitive, affective and behavioral. The cognitive component refers to the knowledge, information and perception held by an individual and this may be equivalent to the Potency 35 (e.g. difficult-easy) dimension of the EPA structure of the Semantic Differential. The affective component concerns the individual's evaluation of the attitude object. This is definitely equivalent to the Evaluation dimension of the EPA structure. The behavioral component is supposed to refer to the individual's action tendency toward the attitudinal object. This may be considered to be equivalent to the Activity dimension. Therefore it looks like even in the above long list of eleven dimensions of the attitudes, three basic EPA dimensions can be identified. Then one could reasonably hold that although attitudes may be char- acterized by many factors, three major underlying factors are Evaluation, Potency and Activity. Reading Attitudes The same issues that relate to attitudes in general have been raised about the area of reading attitudes. For example, Mikulecky (1976, p. 10) reviews literature on reading attitudes and concludes that no uniform definition exists. This compares with the views of Scott (1968) and Green (1977) on the existing definitions of attitudes in general. On the matter of the dimensionality of reading attitudes, various dimensions are also identified but the basic EPA dimensions seem to emerge (Reed, 1978; Summers, 1970: Epstein, 1980). Epstein (1980) confirmed that all the eleven general dimensions identified by Scott (1968), and dis- cussed above, as characterizing general attudes, i.e. 36 direction, magnitude, intensity, ambivalence, salience, affective salience, cognitive complexity, overtness, em— beddedness, flexibility and consciousness also characterize attitudes toward reading. However, Epstein shows that these eleven dimensions can be adequately represented by the three dimensions, Evaluation, Potency and Activity. Therefore, although in a research similar to the current study it would not be surprising to obtain numerous factors, one can rea— sonably expect at least the main EPA structure factors (i.e. Evaluation, Potency, and Activity) to emerge in the subsequent factor analysis of the response data. But there are possibilities for just two factors to characterize reading attitudes. For instance the study carried by Frances Bennies(l973) on the reading attitudes of elementary school children revealed only two factors, Evaluation and Potency. This finding has been supported by another study conducted by Rausbury (1973) Which was aimed at determining the cognitive components of attitudes toward reading. The basic factors identified in the study were Evaluation and Potency factors. Similarly, Lowery and Grafft (1968) have found that the reading attitudes towards paperback books are characterized by two basic dimensions, i.e. Evaluation and Potency. However, there are instances of only one factor characterizing attitudes toward reading. Factor analysis of the Likert-type instrument developed by Estes (1972) for measuring reading attitudes shows that the Evaluative 37 factor seems to be the dominant factor in the assessment of reading attitudes. Estes uses different scales like, enjoyment, boring, waste of time, interesting, rewarding, not good, dull, pleasant, but the variance in these scales is shown to be accounted for mainly by the Evaluation factor. Kamper (1973) in the investigation of seventh grade students' attitudes toward reading similarly shows that the Evaluation factor seems to be the dominant factor. However, in this particular study girls generally were more positive than boys. Subjects that were classified in the study as reading above their grade placement were found to have significantly more positive attitudes toward reading than subjects classified below grade placement. This sugg- ests that the Evaluation factor may be the most important factor in attitudes toward reading but at the same time within this one factor, there may be differences according to the sex and the ability of the students. The above literature review on reading attitudes suggests that one can expect the SD ratings of the concept of READING to be characterized by one dimension, two dimen- sions or three dimensions or even more than three dimensions. Evaluation of the Semantic Differential Osgood, et a1. (1961) have suggested that the SD should be evaluated as a measurement instrument using the criteria of objectivity, validity, sensitivity, comparabil- ity, and reliability (pp.125-177). 38 An instrument is valid when it measures what it is supposed to measure. The validity of SD has been estab- lished in terms of face validity and correlation with external criteria (Osgood, 1961; Marks, 1965). The Semantic Differential is sensitive because it yields distinctions as fine, or even finer, than those made on common sense grounds (Osgood, et a1. 1961, pp. 166-169). The criterion of comparability is met by SD when it is applied across the range of situations relevant to what is being measured and the results are interpreted in a constant fashion (Osgood, et a1. 1961, p. 169). The different situations include different subjects, different cultures (Kumata, and Schramm, 1956). and different concepts (Osgood et al. 1961, p. 177). However, the focus of this study is on the evalu- ation of SD with respect to the reliability of various SD formats. According to Osgood, et a1. (1961, 126-140), three kinds of reliabilities can be computed from any of the three formats used in this study. These are item reliability,(p. 126), factor-score reliability (p. 138) and concept-reliability (p. 140). These reliabilities are explained in the following way: The basic score obtained from the semantic differen- tial is a digit value (e.g. one through seven for a seven point format SD) corresponding to a subject's check-mark with which he indicates his judgment of a particular concept against a particular scale. We shall use the term item reliability to the re- producibility of these’Basic scores. These item scores are typically averaged within factors . . . we use the term factor-score reliability to refer to the reproducibility 6f these values under retest conditions. The several factor-scores for a single 39 concept serve to allocate this concept to a point in the semantic space which defines the meaning of this concept; we shall refer to concept-meaning reliability when dealing with the reproducibility of pointsvin the semantic space with repetition of the measurement Operation (Osgood, C.E. et a1. 1961, p. 126). However since there is more involved in the esti- mation than the simple test-retest reliability which Osgood et a1. (1961, see above) use for their definition of relia- bility, an attempt will be made first to review literature that provides a complete conceptual framework of "reliabil- ity" then some additional literature dealing specifically with the reliability of SD will be examined. Reliability Reliability refers to consistency of scores. It is concerned with the degreecuiagreement or consistency between two independently discerned assessment scores. A correla- tion coefficient is used to express the degree of relation- ship between the two sets of scores. In educational and psychological assessment, it is unreasonable to expect scores that are completely consistent. For instance, in the case of measuring attitudes toward reading, some vari- ations in measurement are inevitable. This may be due to the fact that the attitude held by subjects may be unstable or the measuring procedure or instrument may have changed from one assessment to another. Therefore, the results of assessment may reflect upon the "true" amount of attitude held by the subjects as well as how accurate a measuring device is. 40 Many factors influence scores and affect reliability. Emotional strain, fatigue, guessing, poor conditions during the administration of the test, scoring error, memory effects, etc. (Magnusson, 1966; Mehrens, and Lehmann, 1978 ). However, according to Gronlund (1976, p. 106), "the more consistent the test results are from one measure- ment to another, the less error present and, consequently, the greater the reliability." The goal of reliability estimates is to determine the consistency of our results. Consistency, however, can be thought of in several ways. For instance, consistency can be over a period of time, over different samples of questions, within the measure itself, or between different raters. This means that there are different ways of esti- mating reliability. This study focuses on one kind of reliability, namely, the reliability within the measure itself. This way of estimating reliability is called the internal consistency method. Coefficient Alpha, which is an estimate of internal consistency will be used. Nunnally and Durham (1975) support the use of Coefficient Alpha because, in most situations, it provides a good estimate of reliability. This method will be used for estimation of reliabilities in this study. In the study conducted by Askov (1971) the SD was demonstrated to have an interval consistency (reliability) which was as high as .925. This information should provide a guide on what to 41 expect in this study. Nunally (1967) (in Elsworth and Coulter, 1977, p. 35) has suggested that for scales, re- liabilities of 0.50 to 0.60 may be adequate. 'Reliability and the SD Technique Reliability of the Semantic Differential can be approached in terms of conceptualizing the observed rating in terms of true variance and error variance. According to the classical theory of measurement, reliability of an instrument is due to the true variance (Magnusson, 1967, p. 64: Mehrens and Lehmann, 1978, p. 90; Allen and Yen 1979, pp. 61-65 ). Reliability indicates the extent to which the observed variance (from the ratings) is due to true score variance. If error variance accounts for a big portion of the observed variance, then the reliability will be low because the large error variance will reduce the true variance. In the Semantic Differential, true variance is that derived from actual variations in effective responses and error variance is that due to other factors. Error var- iance has two sources, random errors and bias (non-random errors). However, for reliability of an instrument, random errors contribute to unreliability but the non- random or systematic errors become part of true variance. Both kinds of errors will be reviewed below for various reasons. First, the literature on random errors in SD ratings will be reviewed in order to show potential sources of unreliability. The literature on the bias in SD ratings 42 will subsequently be reviewed to show that SD data needs to be interpreted cautiously especially where instances of bias discussed in this literature are possible. Sources of Random Errors One example of the source of random error has been associated with the grade level of subjects. DiVesta and Dixon, (1966) found in their study of the test-retest reliability of children's ratings on the Semantic Differen- tial that reliabilities were higher in the higher grades than lower grades. The same study found that the Evaluation scales were more reliable at all grade levels than the Potency and Activity factors. Research also seems to indicate that ratings are more stable for some concepts than for others, and this seems to be related to the number of meanings for a con- cept. For example, Norman (1959) found that the concepts leper and tornado were more stable than the concept star. Studies by Lauria (1959) and Peabody (1962) suggest that concepts whose true values are neutral are rated with less reliability. The individual differences between subjects seem to introduce something to consider when reliability is analyzed. For instance, Norman (1959) has found that some subjects are more stable than others in making their rat- ings. According to Norman's research, there is a tendency for those subjects who use the end points of the scales 43 more often to have lower test-retest stability. Addition- ally, Norman's study indicates that certain scales gener— ally have greater stability. These are scales with bigger factor loadings on the Evaluation factor. The other aspect of reliability discussed in the literature on the reliability of Semantic Differential relates to comparisons of the reliabilities of individual scales and those of factors which consist of various scales (bipolars). According to Mehrens and Lehmann (1978) and Magnusson (1967), increasing the length of a test will increase the reliability of the test. Accordingly, Norman, (1959); Nyberg and Clarke (1979) found that factor scores are more reliable than single ratings and most of the gain in test-retest stability can be attained by averaging three or four scales. However, going to eight scale factor score seems to add very little additional reliability (Norman, 1959). This is rather surprising since one would expect continual gains in reliability as more relevant items are added into the total score. The other reliability related issue discussed and researched is the reliability of group means. Many SD studies do not focus on an individual's rating of a concept but on a group mean. In such a case, there is averaging both across scales (factor scores) and across persons. According to DiVesta and Dixon (1966); and Miron (1961), the averaging of group means across scales and persons make the 44 group means highly stable (reliable) even when the samples of subjects involved in calculating the means are as small as thirty. Bias. Examples of bias in SD ratings include social desirability, and scale checking styles. Social desirability. Social desirability is an example of biased error. This can be illustrated by a study by Nickols and Shaw (1964). Nickols and Shaw hypoth- esize that subjects are more sensitive to the social repercussions of their ratings when dealing with salient objects. Then the SD is more transparent as a measure of attitude. According to them, then, social desirability may enter as a factor in SD ratings of salient objects. This interpretation receives support from a study by Ford and Meisels (1965) which showed that the social desirability of SD scales corresponded directly to their loading on the Evaluation dimension. (The Potency and Activity di- mensions are unrelated to social desirability.) Therefore it may be true that direct SD rating of objects may not be an efficient approach to measurement when salient or delicate topics are involved, because subjects can distort their responses in the direction of social desirability. However, Heise (1969, p. 408) suggests that before making a firm conclusion on the impact of social desirability on SD, "one would like to see replications which involve more 45 than two attitude objects, in which subject's need for approval is an actual control variable, in which a criter- ion is used to show that the SD measurements are made on all three dimensions." (p. 408). Yonker, et a1. (1974) have further shown that although social desirability is real, its influence does not render the measures of SD invalid. Error due to scale-checking styles. Important differences appear between persons in scale-checking styles. Some subjects appear to use the end points of scales more often and to avoid the intermediate discriminatory positions (Peabody, 1962). According to Arthur (1962), this is a stable trait of individuals over time and over different sets of concepts. When measurements of scale-checking style are made, the test-retest and split-half correlations of measurements are above .70. Scale-checking styles introduce biased error by moving measurements toward the end points or midpoint of a scale. Additionally, it seems likely that the consistency and extent of end-point checking is related to the true scores for concepts, (Heise, 1969, p. 409). If a subject with such a bias rates a "quite good" concept on five eval- uation scales, he is likely to check the positive end on all five scales; if he rates a "slightly good" concept, he is more likely to alternate between the end points and mid- points of the scales. The result would be that when aver- aging his ratings to get factor scores, we would find 46 systematic deviations from true scores, and the amount of deviation would be larger for the more polarized concepts. Kahneman (1963) calls this exaggeration error. Exag— geration error is a non-random deviation of a subjects' rating from the true score which varies with the true score for the concept and the individual's propensity to exagger- ate when using rating scales. This error introduces prob- lems in SD methodological research and complicates cross- person comparisons of attitudes by raising a question like, Does the obtained difference in scores reflect true differ- ences in affective responses or merely differences in propensity to exaggerate? A number of studies on scale-checking styles exist and these suggest that certain classes of people are more likely to emphasize scale end-points. Age: Children tend to use the end points and center more than adults (Osgood, 1961, p. 85). The study by Scares & Scares, (1981) also indicates that age is a factor. In terms of IQ, low IQ is associated with greater use of end point but mainly among young children, (Stricker and Zax, 1966). According to the study by Mogar (1960), high scores on the F scale tend to use the extremes more often. Sex: There are some studies to suggest that females use the extremes more than males (Dixon, and Dixon, 1964; Goldfried, Kissell, 1963). These studies suggest sex differences with respect to checking styles. But at the 47 same time there are other study reports that show no sex difference (Stricker, and Zax, 1966). Neurosis: Studies by Arthur, 1965; Zax, Gardiner, Lowy, 1964, indicate that neurotics use the extremes more. However, the study by Luria shows no differences between neurotics and controls in the use of scales. Psychosis:_ Studies by Neuringer, 1963; Marks, 1965, indicate that psychotics use the extremes more than normals. Grade: According to Leo M. Harrill (1971) students in lower grades (eg. second grade) are more likely to in- dicate extreme positions (p. 11). Summary 1. The Semantic Differential Technique is shown to be a scaling method with demonstrated properties like equal intervals within the scale and zero points falling at the same place on each scale. Where these properties do not hold the effect is negligible (Messick, 1957; Heise, 1969). 2. There are basically three dimensions emergingfrom factor analysis of the SD response data. These dimensions are Evaluation, Potency and Activity (i.e. EPA Dimensions). However, there is evidence of additional factors and in some cases even less than three factors have been identified. This has been particularly true among younger children and people from non-Western cultures where often one or two dimensions have characterized the response data. 3. Reading attitudes tend to display the general EPA Dimensions although additional' dimensions have been identified in some studies. 4. The reliability of SD is influenced by factors like the social desirability of concepts being rated, checking styles of respondents, age, IQ, etc. However, SD has beenshown to be reliable for students even as low as 48 second grade (Harrill, 1971). According to Elsworth, and Coulter— (1977), for scales of SD, reliabilities of 0.50 to 0.60 may be adequate although the ideal would be to get reliabilities as high as 0.80 or greater. CHAPTER III METHODOLOGY AND PROCEDURES Construction of Scales The scales used in this study are based on the work done by Nybergy, et a1. (1979) on the attitudes of children to school subjects and the work that has been done by Osgood, Suci, and Tannenbaum (1957) on the measurement of meaning. Eighteen scales (or bipolar adjectives) are used to represent three factors, namely, Evaluative, Difficulty (Potency) and Usefulness (Activity). The names of two factors were changed to make the interpretation of the information from school subjects more meaningful. The names were changed by Nybergy and his colleagues (1979). Below is a breakdown of the scales under each of the fac- tors represented. Evaluative Factor 1. Nice - Awful 2. Boring - Interesting 3. Dislike - Like 4. Dead — Alive 5. Lively - Inactive 6. Good - Bad 49 READING. 50 Usefulness Useful - Useless Important - Unimportant Worthless - Valuable Helpful — Unhelpful Necessary - Unnecessary Meaningful — Meaningless Difficulty Hard - Easy Light - Heavy Clear - Confusing Difficult - Simple Strange - Familiar Understandable - Confusing The scales will be used to rate the concept Construction of Formats Three formats of the Semantic Differential tech- nique were constructed for this study. These were three- point, five-point, and seven-point format. Each of these formats was used to rate the concept (READING). The same kind and number of bipolar adjectives was used in each of the formats. However, the order of these adjective pairs was different on each format. This provided three formats with the same bipolars ordered differently. First the order of the bipolars on the seven-point format are 51 determined. The order of the bipolars in the three-point and five-point formats is done randomly based on the ordering used in the seven-point format. This differential ordering of bipolar adjectives meant, for instance, that the first bipolar on the seven- point format became the ninth on the five-point format, and the fourteenth on the three—point format, etc. This was deliberately done to minimize the chances of subjects in the study going back to syncronize their responses on one format with those on other formats within the time con- straints provided by the testing. According to research by Kane, Robert B. (1968) the order in which SD adjective scales are presented does not lower the reliability of the instrument. Hence the differential ordering of bipolars used in the construction of formats can achieve their objective without lowering the reliability of the instruments. Construction of Directions To ensure uniformity of the directions for rating the formats, some directions were prepared in advance. An example of these directions is contained in Appendix A. Construction-Orientation Material Since the concept of reading can take on different meanings, a questionnaire was constructed to create some commonality in the basic concept of reference among the various subjects. The subjects completed the questionnaire 52 prior to completing the SD forms. The purpose was not to measure any reading but rather to enable the respondents to focus their attention on one similar kind of reading. An example of the questionnaire is contained in Appendix A. Translation The Semantic Differential adjective pairs were translated into Swahili so that the African subjects could use the instrument in a language most familiar to them. This was also done for the materials on the Direction and Orientation. The Standard English-Swahili Dictionary was used for the translation of the scales. A sample of the Swahili version of the scales, directions and orientation questionnaire is contained in Appendix B. Administration The administration was done in two steps. The first step involved the administration of the Orientation questionnaire to each of the respondents in the study. Immediately after the first step, the three formats of the Semantic Differential were administered to all the sub- jects. Although all the subjects rated the concept (READING) using all the formats, the order of the formats was varied in such a way that there were three different orders. One order (called Sequence A) had the formats in a seven, five, three sequence, then the other (Sequence B) had the formats in three, seven, five sequence, and 53 finally the sequence of five, three, seven was used to create a third ordering (Sequence C) of the formats. The various formats were presorted in these three sequences. Each sequence was administered to one third of respondents in each grade selected for the study. For instance, in fifth grade, one-third of the students took the Sequence A, one-third the Sequence B, and one-third the Sequence C. The pile of sequences of formats was randomly ordered again so that there was no predictive way of knowing the sequence to be administered to any subject. Population The sample used for the study consisted of students from grades four, five, six, and seven. The ages of these children ranged from nine to thirteen years. The sample consisted of students from Western Kenya (i.e. African students) and some American students selected from Lansing, Michigan. The African students were randomly selected from two schools, i.e. Momboha Primary (Elementary) School and Itumbu Primary School. The American students were selected from Lansing Christian Elementary Schools on Logan and South Pennsylvania. Below is a table that shows the number of students for every category used in the study. The total number of subjects in the study was 448. Convenience sampling was used to select schools were the study was but the selection of the children within each school was done randomly. 54 The American subjects were selected from private schools, i.e. the two Christian schools, and may therefore be more representative of the American middle class than of children in public schools. However, the African subjects were from families that rely on subsistence agriculture and their parents are basically self-employed peasants living in rural Kenya. While the American subjects come from homes where parents can read and write, most of the African sub- jects come from homes where parents do not know how to read nor write. The African subjects View education as a means of getting salaried jobs in government or modern private sector. Compulsory schooling characterizes the educational systems of both America and Kenya. However, in Kenya, the compulsory schooling applies to the elementary grades while in America, compulsory schooling extends to secondary grades. Additionally, compulsory education in Kenya has only been operative for the last seven years. The American subjects come from a society where compulsory education, at least for elementary grades, has been a prominent feature for a relatively longer period than in Kenya. Therefore, the American subjects were from a society that is charac- terized by a higher literacy rate than the Kenya subjects. All the American subjects spoke English as their first language and the same language was used as a medium 55 of instruction. The African subjects spoke more than one language. The language used for instruction in classes was English but Swahili is used mainly for informal interaction outside the classroom. However, the children in these grades know better Swahili than English because while they begin learning English when they go to schools, Swahili is learned almost from the cradle. The American subjects represent the urban American population while the African subjects were mainly drawn from rural setting and are more likely to be representative of children in rural elementary schools. 56 TABLE 1 THE SAMPLE American Subjects African Subjects Age Grade N Age Grade N 9—10 4 55 9-10 4 57 10-11 5 55 10-11 5 55 11-12 6 58 11-12 6 56 12-13 7 55 12-13 7 57 Scoring In a typical seven-point scale, the scores are numbers one to seven. The low end on the scale, repre- senting adjectives like distasteful, weak, passive, bad, mild, unpleasant, etc., are assigned the value of one while seven is assigned to the high end of the scale, representing adjectives like tasty, strong, clean, good, etc. The numbers two to six are assigned to the points between the low and high ends of the scale. The responses are scored according to the numerical value corresponding to the points checked by the subjects. This scoring procedure was used for all the three formats used in the study. However, for the five-point scales, the high end was assigned the value of five and the values of two-four were assigned to the points between the low and high ends of the scales. Similarly, 57 the high end of the three point scale was assigned the vlaue of three and the value of two was assigned to the point between the high and low end. Analysis Part of the focus of the analysis was on comparing the formats within each sub-population with respect to concept reliability. Additionally, an attempt was made to examine the factors underlying the subjects' reading attitudes, and the reliabilities of the various factors. Factor Analysis The SPSS factor analysis sub-program, i.e. princi— pal factoring with iterations (PA2) was used for the analysis of the data from all the formats. This method utilizes the correlation matrix with communality estimates in the main diagonals to extract factors. The first pro- cedure involved extracting factors that had eigenvalues greater than or equal to 1.00. The second procedure involved specifying the extraction of only three factors. The information on the percentage of variance accounted for by each of the first three factors was based on the output from the second procedure. Varimax rotation procedure was then used to sim- plify the factor structures and hence facilitate easier interpretation of the information on the factors. This rotation resulted in some factors loading highly on some I variable but almost zero on other variables. The result 58 led to the identification of clusters of variables on which different factors loaded significantly. To facili- tate interpretation of factors, it was decided a prior that only factor loadings in excess of .30 would be con- sidered significant. The kind of scales loading signifi- cantly on a factor were used to give the factor a label. From the SPSS computer factor analysis output, the following information was obtained: 1. The composition of the first three factors from the ratings by American subjects. From this information, the factors were given their labels (names). 2. The composition of the first three factors from the ratings by African subjects. Again factor labels (names) were obtained. 3. The order of the factors from the American data was observed. This was basically to determine the factor that was extracted first, the factor extracted second, and the third factor. The order in which the factors were extracted and the percentage of variance they accounted for was used to make some inference on the importance of these factors. 4. The order of the factors from the African data was observed and inference was made about the importance of these factors on the basis of their variance and their order of extraction. 5. Some inference was then made on the difference or similarity of the order different factors followed in both the African and American factor structures. This information was used to determine whether the factor structures of the data from the American and African subjects were similar or different. Factor score reliability. This is basically the estimate of the internal consistency on the ratings of the combined scales representing each of the three factors (i.e. Evaluation, Difficulty, and Usefulness). For 59 instance to obtain the factor score reliability for Eval- uation, the SPSS subprogram was used to calculate the internal consistency on the ratings of the six scales, i.e. good-bad, nice-awful, interesting-boring, lively-inactive, like—dislike, alive—dead. This provided the factor score reliability for Evaluation. Similarly, the six scales representing the Difficulty factor (i.e. simple-difficult, easy-hard, light-heavy, understandable-confusing, clear- confusing, familiar-strange) were used to obtain reliabil- ity estimates for the Difficulty. The factor score of the Usefulness factor was obtained from the internal consis- tency of the combined scales representing this factor. These scales were helpful-unhelpful, meaningful-meaning— less, necessary-unnecessary, useful-useless, important- unimportant, valuable-worthless. Concept Reliability This consisted of computing the internal consis- tency coefficients of the scores obtained from all ratings on each format for the eight grades. This gave 8 x 3 coefficients of internal consistency. This also provided an idea of which format provides the most reliable measures of the reading attitudes at different grades and in different cultural groups. The SPSS computer program was used for obtaining estimates of Reliability. Criteria for most reliable format. A comparison of the reliabilities from various formats was made within 60 each grade (among African and American subjects, sepa- rately) using the concept of reliabilities. The format with the biggest magnitude of reliability was considered the most reliable. However, some tests of statistical significance using F tests were computed to determine if formats with the biggest reliabilities in every sub-popula- tion are significantly different from formats with the second biggest reliabilities at p=.05. The use of an F test is based on the sampling distribution of l-rl/l-ro (where re and r1 = coefficient alpha) which is distributed as an F. (Feldt, 1969) However, based on the review of literature, it was also decided that any reliability coefficients below .50 would be rejected as being too low. CHAPTER IV PRESENTATION AND ANALYSIS OF RESULTS Introduction In this chapter the results from the ratings of the concept of READING using the three formats of the Semantic Differential technique by American and African subjects from fourth to seventh grades are presented and analyzed in three sections. The first section presents and analyzes the results on the reliability of the formats. First, reasons are provided for using the total concept reliability for comparing the reliabilities of the formats. The subse- quent analysis of format reliabilities shows that among the American subjects, five-point and seven-point formats are more reliable than the three-point format. The results from the African subjects show that the five- point format is more reliable than other formats in the fourth and fifth grades but in the sixth and seventh grades, both the five-point and seven-point are more reliable than the three-point format. The second section presents and analyzes the results on the factors (dimensions) and the factor struc— tures. The criterion for accepting a factor loading as 61 62 being significant is .30 so that any factor loading above this is considered to be significant. Accordingly, among the American subject, the first factor loads significantly on scales representing the Evaluation dimension, the second factor loads significantly mainly on scales repre- senting the Difficulty dimension while the third factor has significant factor loadings on scales representing the Usefulness dimension. This suggests an Evaluation-Diffi- culty-Usefulness factor structure. The results from the African subjects show that Factor I loads significantly on scales representing the Usefulness dimension, Factor II has significant factor loading on scales representing the Evaluation dimension while Factor III loads significantly on scales representing the Difficulty dimension. Hence it is concluded from observing the results generated by various formats that Factor I is Usefulness, Factor II is Evaluation, and Factor III is Difficulty. The third section presents some results on the reliabilities of the factors represented by the eighteen scales in this study. The purpose is to determine the most reliable factor in American and African subjects of different grades. The analysis of the results shows that among the American subjects the most reliable factor is represented by the six Evaluation scales. This factor is followed by the factor represented by the Difficulty scales and the least reliable factor is the one represented 63 by Usefulness scales. This is true for all grades from which the subjects were taken. Hence there is no factor- grade interaction. Among the African subjects, the most reliable factor is Usefulness followed by Evaluation and then Difficulty. This is also true from fourth grade to seventh grade. Comparison of Reliabilities of Formats For the purpose of determining the most reliable formats, reliability coefficients generated on all eighteen scales for each format (i.e. Concept reliabilities) were used. Osgood, et. al.(l961, p. 126) provide additional ways of comparing the reliabilities of the formats. For instance, there is factor-score reliability which refers to the reliability of factors represented by various scales. There is also item reliability which is associated with the reliability on the ratings from each individual adjective pair (scale). The factor-score reliability could be used for comparing the formats. However, concept reliabilities were used in this study for comparing the formats because the main focus of the study was to compare the formats in terms of ratings on the total concepts rather than individ- ual items or even factor-scores. Reason for Using Three Dimensions in Factor Analysis At this point it may also be necessary to explain why the subsequent results from factor analysis only utilize 64 information from the first three factors. First, this was determined by the nature of this study which was confirma- tory rather than exploratory. The scales used in the study were deliberately selected to represent three factors and the factor analysis was aimed at confirming these three dimensions. Additionally, there was some information from the initial factor analysis that reinforced the decision to use the first three factors. It was observed that in almost all the cases, the first three (out of eighteen) factors accounted for sub- stantial percentages of variance. Tables 2 and 3 provide the percentages of variance accounted for by the first three factors from the analysis of the ratings by American and African subjects, respectively. The magnitude of these percentages suggest that the attitudes toward reading are mainly characterized by three factors. Reliable Formats for American Subjects Generally, the five-point and seven—point formats yield more reliable ratings than the three-point format among the American subjects. This is true for both unstandardized and standardized reliability coefficients. This conclusion is based on the observation of the differ- ences between the magnitudes of these reliability coeffi- cients but in some cases this is also true when tests of statistical significance are used to determine whether the differences are statistically different at .05 level of 65 TABLE 2 PERCENTAGE OF VARIANCE ACCOUNTED FOR BY THE FIRST THREE FACTORS (AMERICANS) Grade Format PCT of Variance 4 3 pt 49.6 4 5 pt 52.0 4 7 pt 65.2 5 3 pt 52.9 5 5 pt 60.5 5 7 pt 67.5 6 3 pt 48.0 6 5 pt 63.4 6 7 pt 66.3 7 3 pt 57.7 7 5 pt 63.2 7 7 pt 63.3 significance. A comparison of the reliability coefficients from the ratings on different formats in different American grades confirms the superiority of five-point and seven- point formats over three-point format in terms of the magnitudes of the reliability coefficients. Table 4, which is based on unstandardized reliability coefficients, attempts to portray these differences. The same superiority of the five-point and seven- point formats emerges from the analysis of standardized 66 TABLE 3 PERCENTAGE OF VARIANCE ACCOUNTED FOR BY THE FIRST THREE FACTORS (AFRICANS) Grade Format PCT of Variance 4 3 pt 59.8 4 5 pt 65.5 4 7 pt 65.5 5 3 pt 50.1 5 5 pt 52.7 5 7 pt 56.3 6 3 pt 51.1 6 5 pt 64.4 6 7 pt 54.2 7 3 pt 49.4 7 5 pt 62.1 7 7 pt 61.1 TABLE 4 UNSTANDARDIZED RELIABILITIES FOR AMERICAN SUBJECTS Grade 3 Pt Format 5 Pt Format 7 Pt Format 4 .77 .86 .89 5 .80 .85 .90 6 .74 .89 .89 7 .75 .87 .88 67 reliability coefficients for the same group of subjects, as shown in Table 5. TABLE 5 STANDARDIZED RELIABILITIES FOR AMERICAN SUBJECTS Grade 3 Pt Format 5 Pt Format 7 Pt Format 4 .76 .87 .89 5 .78 .85 .89 6 .74 .90 .89 7 .74 .87 .88 Tests of Statistical Significance Tests of statistical significance using an F test were performed to determine whether the differences between reliability coefficients from various formats were statis- tically significant. The standardized reliabilities were used for this purpose and the 0.05 level of significance was adopted for obtaining critical values. The F ratios were computed using l-rz/l-rl. The degrees of freedom used for obtaining the critical values were computed in the following way. This method has been suggested by Feldt (1969). 2 _ 2A _ 2A . v2 ' A-I V1 ‘ 2B-AB-A‘ where A - Eff—2'2“ ° arri’ 4 2 2 2 (df1+2)(df4) (df3+2)(df2) B = o (df4-2)(df4-4)(dfi) (df2-2)(df2-4)(df3) 68 and df1 (Nl-l) df4 = (NZ-1) df3 = (NZ-1)(k2-1) df2 = (Nl-l)(k1-1) and N1 and N2 are the number of subjects that used formats I and II respectively, while k and k2 refer to the number 1 of scales (which was equal to eighteen in both cases). The results for the degrees of freedom, the F ratios obtained and the conclusions concerning the signif- icance of the differences are provided in Appendix C for both the American and African subjects. However, provided below is a summary of the results on the significance of the differences in the reliability coefficients from the ratings by American subjects using the various formats. 1. In fourth grade, the five-point and seven-point formats were significantly more reliable than the three-point formats. The difference between the five- point and seven-point formats was not significant. 2. In the fifth grade, only the seven-point format was significantly more reliable than the three-point format. Other differences were non-significant. 3. In the sixth grade, the five-point format was significantly more reliable than the three-point format. Similarly, the seven-point format was more reliable than the three-point format. The difference between the five-point and the seven-point formats was non-significant. 4. In the seventh grade, the results were very similar to what was obtained in sixth grade. Both the five- point and the seven-point formats (separately) were significantly more reliable than the three-point format. 69 Summagy Among the American subjects, the five-point and seven-point formats proved to be more reliable than the three-point format. In fourth grade, both the five-point and the seven-point formats have relatively bigger relia- bility coefficients than the reliability coefficient of the three-point format but the differences between these coefficients are not statistically signicant. In the fifth grade, the seven-point format has a reliability coefficient which is significantly greater than that of the three-point format but there is no statistical differ- ence between the three-point and five-point formats and between five-point and seven-point formats. However, in fifth grade, like fourth grade, the reliability coeffi- cients of the five-point and seven-point formats are bigger in magnitude than that of the three-point format. In sixth and seventh American grades, the five-point and seven-point formats have reliability coefficients which are bigger in magnitude and are significantly different from the three- point formats. This seems to suggest that in sixth and seventh grades, the five-point and seven-point formats may be more reliable than the three-point format. 70 Reliable Formats for African Subjects Among the African subjects, the five-point format seems to be the most reliable one in the lower grades (i.e. fourth and fifth grades) while both the five-point and seven-point formats are relatively more reliable in sixth and seventh grades. The three-point format is generally less reliable than the other formats in all grades. In fact lowest reliability coefficient is provided by ratings on the three-point format in the African fourth grade subjects. The two tables below provide summaries of the unstandardized and standardized (respectively) reliability coefficients generated by ratings on the various formats used in different grades. TABLE 6 UNSTANDARDIZED RELIABILITIES FOR AFRICAN SUBJECTS Grade 3 Pt Format 5 Pt Format 7 Pt Format 4 .33 .86 .54 5 .59 .77 .71 6 .77 .78 .80 7 .79 .85 .88 The reliability coefficient of .33 provided by the ratings on the three-point format in fourth grade is the lowest in the whole study and is unacceptable because the standard acceptable reliability is .50 and above (see 71 Methodology). Therefore, the three-point format may not be appr0priate for African fourth graders. TABLE 7 STANDARDIZED RELIABILITIES FOR AFRICAN SUBJECTS Grade 3 Pt Format 5 Pt Format 7 Pt Format 4 .41 .88 .59 5 .57 .80 .71 6 .77 .81 .82 7 .79 .88 .88 Similarly, the standardized reliability coefficient of .41 (from the table above) is unacceptable for the same reason that acceptable reliability should only be above .50. However, it can be noted that the five-point format provides some very high reliability coefficients among fourth graders and fifth graders relative to other formats. This is more pronounced in fourth grade than in fifth grade. In the sixth grade the reliability coefficients of the five-point and seven-point format are almost the same and in the seventh grade both formats have the same relia- bility coefficients. The reliability coefficients gener- ated from the ratings on the three-point formats are smaller than those from the five-point and seven-point formats in all the African grades. 72 Tests of Statistical Significance Tests of statistical significance to determine whether differences between the more reliable and less reliable formats were significant were also performed. Conditions similar to the tests of significance on the data from American subjects were maintained. Appendix C has a table that gives the results. An outline of these results is also presented below. 1. In fourth grade, the five-point format was signifi- cantly more reliable than the three-point and (sepa- rately) the seven-point formats. The difference between the three-point and seven-point formats was not significant. 2. In the fifth grade, both the five-point and the seven- point formats were significantly more reliable than the three-point format while the difference between the five-point and the seven-point formats were non- significant. 3. In the sixth grade, no significant differences were obtained in the reliabilities of the formats. 4. In the seventh grade, both the five-point and seven- point formats were significantly more reliable than the three-point format. The five-point and seven-point formats were not significantly different in their relationships. Summary Among the African subjects, the five-point format appears to be more reliable than the three-point and seven- point in fourth and fifth grade but in the subsequent grades, both the five-point and seven-point formats are more reliable than the three-point formats. The summary of the results on the reliable formats 73 for both the American and African subjects is given in Table 8 below. TABLE 8 SUMMARY ON RELIABLE FORMATS Subjects' Nationality Grade Most Reliable Formats American 4 5 pt and 7 pt American 5 5 pt and 7 pt American 6 5 pt and 7 pt American 7 5 pt and 7 pt African 4 5 pt African 5 5 pt African 6 5 pt and 7 pt African 7 5 pt and 7 pt Information on Factors Factor analysis of the ratings revealed the exis- tence of the three factors identified by various studies conducted by Osgood and others as discussed in the review of literature. These factors are evaluation, potency and activity which are given the names of evaluation, diffi- culty and usefulness, respectively, in this study to make them relevant to the school situation. However, although the factor analysis of the American subjects ratings show that the most important factor is the Evaluation, followed by the Difficulty (Potency) factor, and the third factor 74 is the Usefulness (Activity) factor, the factor analysis of the ratings of African subjects showed that the most important factor, with minor exceptions, was the Useful- ness factor, followed by the Evaluation factor and the third factor was the Difficulty (Potency) factor. Factors from American Subjects' Ratings In the factor analysis of the ratings from almost all the formats in all the American subjects and grades used for the study, the first factor was identified as the Evaluation, the second was the Difficulty (Potency) and the third factor was the Usefulness (Activity). The only exception was the factor structure obtained in fourth grade on the ratings from the three-point format. An attempt is made below to present the results on the factor structures of the data. These factor structures are pre- sented according to the grades and the format that was used to get the ratings. The results that seem to suggest the same factor structures are presented first and then the factor structure from the ratings on the three-point format by American fourth graders is presented last because it differs from all the other factor structures. All the first eleven factor structures that are presented below suggest an Evaluation, Difficult, Usefulness factor structure. 75 Factors from the Five-Point and Seven-Point Formats Used in American Fourth Grade For the sake of interpreting factor loadings, .30 was made (in the section on methodology) to be the cutoff for significant factor loadings. Any factor loading above this cut-off is considered significant and any loading below that is rejected. On this basis, Factor I on Table 9 is Evaluative because most of the scales that load highly on the factor are evaluative. These scales are alive-dead, like-dislike, good—bad, nice-awful, and interesting-boring. Factor II is clearly a Difficulty factor because the scales that consistently load significantly on it are simple- difficulty, easy-hard, understandable—confusing, clear— confusing, and familiar-strange. These scales represent Difficulty. Factor III loads significantly on scales like helpful-unhelpful, useful-useless, valuable-worthless, important-unimportant, and necessary-unnecessary. These scales represent a Usefulness dimension and therefore it is reasonable to conclude that Factor III is basically a Usefulness factor. In this case, since the table gives results of the factor analysis of the Semantic Differen— tial ratings using five-point format among American fourth graders, the table therefore provides the factors and the factor structure provided by the data obtained from this particular format in fourth grade. The factor analysis of the ratings by fourth grade American subjects using a seven-point format is provided 76 TABLE 9 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AMERICAN FOURTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .31 .13 .12 2. Like .61 .28 .22 3. Good .58 —.09 .18 4. Lively .09 .41 -.17 5. Nice .84 .31 -.08 6. Interesting .69 .02 .19 Difficulty 7. Simple .13 .20 -.01 3, Easy .45 .64 .22 9. Light .16 .07 .30 10. Understandable .25 .51 .24 11. Clear .02 .57 .15 12. Familiar .07 .30 .49 Usefulness 13. Helpful .13 .09 .73 14. Meaningful .25 .09 .40 15. Necessary .05 .30 .74 16. Useful -.01 —.16 ’.40 17. Important .11 .04 .40 18. Valuable .28 .26 .78 77 in Table 10. The Factor I loads highly on scales like like-dislike, good-bad, lively-inactive, nice-awful, interesting-boring. This is an Evaluation factor. Factor II loads significantly on scales like simple-difficult, easy-hard, light—heavy, understandable-confusing, clear- confusing, and familiar-strange. This is therefore a Difficulty factor. Factor III loads significantly of scales like valuable-worthless, useful-useless, important- unimportant, necessary-unnecessary. This is clearly a Usefulness factor. Therefore Factor I is Evaluation, Factor II is Difficulty and Factor III is Usefulness. Factors from Ratings on Different Formats Used by American Fifth Grade Subjects The next three tables give the various results of factor analysis of the Semantic Differential ratings by American fifth grade subjects using three-point, five- point and seven-point formats. Table 11 gives the factor structure of the ratings on the three-point format. Factor I loads significantly on alive-dead, like—dislike, good—bad, lively-inactive, nice-awful, which are evaluation scales. This is clearly an Evaluation factor. Factor II loads highly on light- heavy, clear-confusing, familiar-strange. This may be a Difficulty factor. Factor III loads significantly on valuable-worthless, important-unimportant, useful-useless, necessary-unnecessary, meaningful-meaningless. These 78 TABLE 10 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AMERICAN FOURTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation CI 1. Alive .29 .75 .16 2. Like .73 .07 .36 3. Good .57 .19 .10 4. Lively .41 -.16 .16 5. Nice .65 .24 -.40 6. Interesting .78 .09 .20 Difficulty 7. Simple .62 .62 -.25 8. Easy -.01 .87 .21 9. Light .15 .71 .08 10. Understandable .18 .50 -.05 11. Clear .65 .42 .11 12. Familiar .20 .71 .06 Usefulness 13. Helpful .09 .93 .17 14. Meaningful .38 .29 .69 15. Necessary -.02 .09 .34 16. Useful -.05 .05 .49 17. Important .84 .02 .34 18. Valuable .06 .24 .89 79 TABLE 11 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT FORMAT BY AMERICAN FIFTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .70 .12 -.19 2. Like .30 .57 .31 3. Good .81 .13 .16 4. Lively .56 .21 .05 5. Nice .78 -.03 .00 6. Interesting .04 .59 -.06 Difficulty 7. Simple .29 .08 -.51 8. Easy .04 .10 -.47 9. Light .23 .86 .05 10. Understandable .10 .06 -.01 11. Clear .11 .35 .04 12. Familiar .04 .34 -.19 Usefulness 13. Helpful -.03 .14 .02 14. Meaningful .76 .12 .36 15. Necessary -.14 .02 .30 16. Useful .25 .43 .69 17. Important .22 .09 .73 18. Valuable .21 .01 .35 80 adjective pairs represent the Usefulness factor and in fact Factor III is clearly Usefulness. Table 12 gives the factor analysis results of the ratings by fifth graders using five-point formats. Factor I loads highly on alive-dead, like-dislike, good-bad, nice-awful, interesting-boring. This is clearly an Evalu— ation factor. Factor II loads significantly on scales like simple-difficult, easy-hard, light-heavy, understandable- confusing, clear-confusing. This is a Difficulty factor. Factor III loads significantly on useful-useless, neces- sary-unnecessary, important-unimportant, which are Useful- ness scales. But this factor is not as interpretable as Factor I and Factor II but the message concerning the Cluster of scales is still clear enough. Table 13 presents the factor structure of the ratings on the seven-point format by fifth grade American subjects. Factor I loads significantly on alive-dead, like-dislike, good-bad, lively-inactive, nice-awful, interesting-boring. This is clearly an Evaluation factor. Factor II loads significantly mainly on scales repre- senting the difficulty factor like simple-difficult, easy- difficult, light-heavy, understandably-confusing, clear— confusing, and familiar—strange. this is a Difficulty factor. Factor III loads significantly on all the Useful- ness scales except one. This does not happen with scales representing other factors and hence indicates that Factor III is Usefulness. 81 TABLE 12 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AMERICAN FIFTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .77 .12 .12 2. Like .40 .09 .54 3. Good .85 -.12 .22 4. Lively .22 .01 .13 5. Nice .61 .22 —.01 6. Interesting ~50 .30 .49 Difficulty 7- Simple .08 .67 -.11 8. Easy .08 .72 -.01 9. Light -.08 .63 -.41 10. Understandable .00 .82 .25 11. Clear .28 .54 .27 12. Familiar .10 .02 .80 Usefulness 13. Helpful .17 .21 .37 14. Meaningful .23 -.05 .15 15. Necessary .49 -.08 .47 16. Useful .29 .00 .68 17. Important .75 -.01 .30 18. Valuable .86 -.09 .28 82 TABLE 13 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AMERICAN FIFTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .62 .14 .22 2. Like .85 —.02 .26 3. Good .67 .34 .05 4. Lively .66 .43 -.03 5. Nice .82 -.10 -.04 6. Interesting .70 .28 .19 Difficulty 7. Simple .11 .96 .11 8. Easy .11 .87 .20 9. Light .20 .74 -.11 10. Understandable .12 .36 -.02 11. Clear -.05 .92 .07 12. Familiar .55 .41 .51 Usefulness 13. Helpful -.03 .02 .28 14. Meaningful -.19 .58 .33 15. Necessary .19 .18 .42 16. Useful .45 .49 .57 17. Important .09 -.04 .70 18. Valuable .42 -.03 .43 83 In summary, all the formats used among American fifth grade suggests that the first factor or dimension characterizing the attitudes of the subjects toward reading is an Evaluation, the second one is the Diffi- culty dimension and the third is the Usefulness dimension. Factors from American Sixth Grade Subjects' Ratings Tables 14, 15, and 16 present the factor struc- tures obtained from the factor analysis of the ratings by American sixth grade subjects using three-point, five- point, and seven-point formats to express their attitudes toward reading. Table 14 was obtained from the factor analysis of the ratings on the three-point format. Factor I has consistent significant factor loadings on evaluation scales. These scales are alive—dead, like-dislike, good- bad, lively-inactive, nice-awful, interesting-boring. Factor II loads significantly on adjective pairs like simple-difficult, easy-hard, light-heavy, Clear-confusing. This is a Difficulty factor. Factor III loads signifi- cantly mainly on Usefulness scales and hence is a Useful- ness factor. Table 15 presents the factor structure from the ratings on the five-point formats by American sixth grade subjects. Factor I loads significantly on adjective pairs like alive-dead, like-dislike, good-bad, lively-inactive, nice-awful, interesting-boring. Factor II loads 84 TABLE 14 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT FORMAT BY AMERICAN SIXTH GRADE CHILDREN Scales Factor I Factor II‘ Factor III Evaluation 1. Alive .64 .29 -.32 2. Like .48 .19 .17 3. Good .50 .03 .52 4. Lively .56 -.01 -.06 5. Nice .16 .48 .46 6. Interesting .69 .11 .03 Difficulty 7. Simple .05 .53 -.04 8. Easy -.06 .69 .10 9. Light .09 .85 .08 10. Understandable .39 -.16 .00 11. Clear .37 .56 .16 12. Familiar .18 .01 .83 Usefulness 13. Helpful .67* .06 .18 14. Meaningful .01 .06 .21 15. Necessary -.12 .01 .32 16. Useful .09 .23 -.49 17. Important -39 -.45 -33 l8. Valuable ~59 -.03 ~14 85 TABLE 15 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AMERICAN SIXTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .73 .ll .02 2. Like .61 .10 .21 3. Good .61 .04 .51 4. Lively .74 .00 .27 5. Nice .49 .03 .62 6. Interesting .66 .18 .45 Difficulty 7. Simple .52 .39 .08 8. Easy .71 .16 -.03 9. Light .18 .61 .53 10. Understandable .06 .84 .01 11. Clear .16 .90 .08 12. Familiar .16 .71 .13 Usefulness 13. Helpful .32 .15 .31 14. Meaningful .33 .39 .02 15. Necessary -.01 -.03 .81 16. Useful .10 .25 .81 17. Important -.05 .15 .30 18. Valuable .11 .07 .64 86 significantly on simple-difficult, light-heavy, under- standable—confusing, clear-confusing, and familiar- strange. Factor III loads significantly on helpful- unhelpful, necessary—unnecessary, useful-useless, impor- tant, unimportant, valuable-worthless. This seems to suggest that the first, second, and third factors are Evaluation, Difficulty, and Usefulness, respectively. Table 16 presents the factor structure obtained from the ratings by sixth grade subjects using seven-point formats. Factor I loads highly mainly on evaluation scales like alive-dead, like-dislike, good-bad, lively- inactive, nice-awful, interesting—boring. Factor II has significant factor loadings on simple-difficult, easy- hard, light-heavy, understandable-confusing, clear-con- fusing, and familiar-strange. Factor III loads signifi- cantly mainly on usefulness scales like useful-useless, valuable-worthless, helpful-unhelpful, etc. It is reasonable to conclude that the first, second and third factors are Evaluation, Difficulty, and Usefulness, respectively. In brief, the attitudes of American sixth grade subjects toward reading are characterized by the three dimensions of Evaluation, Difficulty, and Usefulness, in that order of importance. 87 TABLE 16 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AMERICAN SIXTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .78 .01 .42 2. Like .76 .02 .34 3. Good .82 .35 -.16 4. Lively .40 .89 -.02 5. Nice .74 .07 -.06 6. Interesting .67 .39 .03 Difficulty 7. Simple .06 .40 .29 8. Easy .02 .65 .44 9- Light .08 .47 .27 10. Understandable .03 .61 .19 11. Clear .10 .22 -.03 12. Familiar .91 .ll .01 Usefulness 13. Helpful -.04 .39 .41 14. Meaningful .68 .16 .35 15. Necessary .09 .22 .87 16. Useful .04 .03 .74 17. Important -.01 .05 .90 18. Valuable .12 -.02 .60 88 Factors from American Seventh Grade Subjects' Ratings The factor analysis results of the attitudes of the American seventh grade subjects toward reading are given in Tables 17, 18, and 19. The three dimensions identified among the fourth, fifth and sixth grade sub- jects emerge from the factor analysis of the data from seventh grade subjects. Factor I loads significantly on scales representing the Evaluation dimension, Factor II loads significantly on scales representing the Difficulty dimension and Factor III has significant loading on scales representing the Usefulness dimension. This is true of all the formats used. For instance, in Table 17, which presents factor analysis of ratings on three-point format, Factor I loads significantly on alive-dead, like-dislike, good-bad, lively-inactive, nice-awful, interesting-boring. These scales represent the Evaluation dimension. Factor II loads significantly on simple-difficult, easy-hard, clear- confusing which represent the Difficulty dimension. Factor III loads significantly on helpful-unhelpful, useful-use- less, important-unimportant, valuable-worthless. These scales represent the Usefulness dimension. Table 18 represents the factor analysis results from ratings using five-point formats. Again Factor I loads significantly mainly on evaluative scales like alive, like, good, lively, nice and interesting. Factor II 89 TABLE 17 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT FORMAT BY AMERICAN SEVENTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .84 .02 -.02 2. Like .55 .04 -.41 3. Good .37 .06 -.ll 4. Lively .53 .14 -.32 5. Nice .62 .21 .21 6. Interesting .75 -.05 .03 Difficulty 7. Simple .20 .85 .19 8. Easy .14 .71 .28 9. Light .26 .25 .16 10. Understandable .06 .10 .11 11. Clear .17 .40 .08 12. Familiar -.23 .10 .04 Usefulness 13. Helpful .15 .14 .80 14. Meaningful -.02 .02 .25 15. Necessary .10 .71 .08 16. Useful -.03 .21 .74 17. Important -.03 .21 .74 18. Valuable .05 .24 .54 90 TABLE 18 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AMERICAN SEVENTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .83 .17 .03 2. Like .82 .02 .25 3. Good .85 .09 -.09 4. Lively .88 .01 .23 5. Nice .58 .33 -.28 6. Interesting .44 .25 —.16 Difficulty I 7. Simple .16 .73 .17 8. Easy -.01 .70 .28 9. Light .14 .91 .18 10. Understandable .06 .93 .16 11. Clear .25 .74 .13 12. Familiar .76 .22 .23 Usefulness 13. Helpful .24 .15 .58 14. Meaningful .07 —.05 .33 15. Necessary .26 .01 .59 16. Useful -.01 -.10 .69 17. Important .04 -.ll .17 18. Valuable -.02 .27 .67 91 loads significantly on simple, easy, light, understandable, and clear which represent the Difficulty dimension. Fac- tor III also loads significantly mainly on scales repre- senting the Usefulness dimension. In Table 19, which presents the factor structures from ratings using seven-point formats by American seventh grade subjects, Factor I loads significantly mainly on scales representing the Evaluation dimension. Factor II loads significantly mainly on scales representing the Dif- ficulty dimension. Finally, Factor III loads significantly on scales like helpful-unhelpful, useful-useless, valuable- worthless, necessary-unnecessary, which represent the Usefulness dimension. Therefore in all the three formats, the first, second and third factors that characterize the attitudes of the American seventh grade subjects toward reading are Evaluation, Difficulty and Usefulness, respectively. Summary The overwhelming evidence from eleven sources, consisting of three three-point formats, four five-point formats, and four seven-point formats show that the atti- tudes of the American subjects from grades four to seven toward reading can be defined adequately by three factors, Evaluation, Difficulty, and Usefulness. The only format that provides contradictory evidence is the three-point format factor structure which is given in Table 20. 92 TABLE 19 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AMERICAN SEVENTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .24 .77 -.01 2. Like .39 .58 .07 3. Good .52 .31 .22 4. Lively .74 .26 .06 5. Nice .64 .48 .14 6. Interesting .77 .05 .09 Difficulty 7. Simple .09 .87 -.03 8. Easy .53 .47 .09 9. Light .27 .ll -.03 10. UnderStandable -02 .70 .16 11. Clear .40 .53 .12 12. Familiar .07 .61 .09 Usefulness 13. Helpful .20 -.07 .63 14. Meaningful .12 .04 .70 15. Necessary .14 -.11 .62 16. Useful .03 .13 .68 17. Important -.19 .42 .69 18. Valuable -.04 .27 .60 93 TABLE 20 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT FORMAT BY AMERICAN FOURTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .05 .50 .40 2. Like .30 .58 .32 3. Good .01 .73 .61 4. Lively .65 .17 .46 5. Nice .67 .22 .20 6. Interesting .24 .28 .31 Difficulty 7. Simple .51 .49 .23 8. Easy .37 .70 .18 9. Light .62 .26 .04 10. Understandable .13 .59 —.04 11. Clear .50 .75 -.17 12. Familiar .78 .33 .09 Usefulness 13. Helpful .71 -.10 .34 14. Meaningful .22 .22 .60 15. Necessary .19 .07 .69 16. Useful .01 .13 .58 17. Important .61 .26 .11 18. Valuable .32 -.22 .63 94 Factors from African Subjects' Ratings The three factors identified in the factor analy- sis of the response data of American subjects are also identified among African subjects but there is a slight difference in the order of the factors in terms of their importance. While in the American results the first fac- tor is Evaluation, followed by Difficulty and then Useful- ness, among the African subjects, the first factor is Usefulness, followed by Evaluation, then Difficulty. This means that the most important factor in the attitudes of African subjects toward reading is Usefulness, then, Eval— uation, and third, Difficulty. A grade by grade presenta- tion and discussion of the factor analysis results of African subjects will be discussed below. Factors from African Fourth Grade Subjects' Ratings For the purpose of making inference about the dimensions that characterize reading attitudes held by African subjects, only the information provided by the factor analysis results of the ratings from the five-point format by African fourth grade subjects was used. This is because the reliabilities of the three-point and seven- point were so low that they were unacceptable. Therefore the factor structure provided by the five—point format was the only one that was used for analyzing the relevant factors and their importance. The reliability of the five-point format was quite high, .86, while the 95 reliabilities of the three-point and seven-point formats were .33 and .54, respectively. Hence, a decision was made to use the five-point factor structure for this analy- sis. The factors extracted from the ratings from the three- point and seven-point formats were not used to make any de- cision on the factor structures of the African subjects. This is because of the unreliability of the formats. The information concerning the extracted factors from the ratings using the five-point formats by African subjects is presented in Table 21. Factor I is dominated by significant factor loadings on scales which represent the Usefulness factor. These scales are helpful-unhelpful, useful-useless, meaningful-meaningless, necessary-unneces- sary, important-unimportant. In fact most Usefulness scales load significantly on Factor I. Factor II loads sig- nificantly on four of the Evaluation scales and Factor III loads significantly on a similar number of scales repre- senting the Difficulty. Clearly, the first, second and third factors are Usefulness, Evaluation, and Difficulty, respectively. Tables 22 and 23 are presented to show the uninter- pretable results from the factor analysis of the results from the three-point and seven-point formats, respectively. Factors from African Fifth Grade Subjects' Ratings Tables 24, 25, and 26 present information of the factors that characterize the attitudes of African fifth 96 TABLE 2 1 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AFRICAN FOURTH GRADE CHILDREN Scales‘ Factor I Factor II Factor III Evaluation 1. Alive .11 .77 .43 2. Like .17 .12 .19 3. Good .14 .43 .06 4. Lively -.03 .48 -.27 5. Nice .61 -.18 .05 6. Interesting —.09 .42 .01 Difficulty 7. Simple .13 .45 .24 8. Easy -.07 -.05 .10 9. Light .06 -.Ol .57 10. Understandable .01 .02 .32 11. Clear .32 -.35 .41 12. Familiar .19 .13 .49 Usefulness 13. Helpful .30 .13 .ll 14. Meaningful .69 .01 .07 15. Necessary .52 .15 -.14 16. Useful .52 -.16 .31 17. Important .58 -.16 .31 18. Valuable .07 -.01 -.28 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT 97 TABLE 2 2 FORMAT BY AFRICAN FOURTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .16 .37 —.12 2. Like .41 .75 -.03 3. Good .28 .87 .05 4. Lively .28 .68 .14 5. Nice -.08 .63 .18 6. Interesting -.22 .83 .04 Difficulty 7. Simple .14 .13 .67 8. Easy -.01 .30 .75 9. Light .35 .06 .33 10. Understandable .13 .64 .25 11. Clear .47 .12 .14 12. Familiar .43 .01 .59 Usefulness 13. Helpful .38 .14 .22 14. Meaningful .97 .12 .14 15. Necessary .61 .27 .01 16. Useful .58 .06 .39 17. Important .82 .26 .02 18. Valuable .51 -.35 -.32 98 TABLE 2 3 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AFRICAN FOURTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .16 .37 -.12 2. Like .41 .75 -.03 3. Good .28 .87 .05 4. Lively .28 .68 .14 5. Nice -.08 .63 .18 6. Interesting -.22 .83 .04 Difficulty 7. Simple .96 .13 .14 8. Easy -.01 .30 .75 9. Light .35 -.06 .33 10. Understandable .13 .25 .64 11. Clear .14 .12 .77 12. Familiar .01 .43 .59 Usefulness 13. Helpful .38 .14 .22 14. Meaningful .97 .12 .14 15. Necessary .61 .27‘ .01 16. Useful .58 .06 .39 17. Important .82 .25 .02 18. Valuable .51 -.03 —.32 99 grade subjects toward reading. The information from all the three formats suggests that the factors seem to follow the order of Usefulness, Evaluation and Difficulty, in terms of their importance. For instance, in Table 24, which presents the results from the factor analysis of the ratings using three-point format, all the six scales representing the Usefulness factor load significantly on Factor I. Factor II has significant loadings on five of the scales repre- senting the Evaluation factor while Factor III has signif- icant factor loadings on four of the scales representing the Difficulty dimension. The structure, whereby the first factor is Usefulness, the second, Evaluation, and the third, Difficulty, emerges from these factor analysis results. The results from the five-point ratings, presented in Table 25, show that Factor I loads significantly on the six scales representing the Usefulness dimension. Factor II loads significantly on five of the six scales repre- senting the Evaluation dimension. These scales are alive—dead, like-dislike, good-bad, nice-awful, interesting- boring. Factor III loads significantly on scales which mainly represent the Difficulty dimensions. The Usefulness, Evaluation, and Difficulty factor structure emerges here again. In Table 26, presenting the results from the factor analysis of the ratings using seven-point formats, Factor I 100 TABLE 24 EXTRACTED FACTORS FROM RATINGS USING"THREE-POINT FORMAT BY AFRICAN FIFTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .02 .06 -.13 2. Like .51 .76 .01 3. Good .03 .84 -.13 4. Lively .68 .37 .01 5. Nice -.50 .31 -.05 6. Interesting .06 .70 -.08 Difficulty 7. Simple .01 .06 .64 8. Easy .15 .09 .40 9. Light .38 .15 .06 10. Understandable .01 -.16 .07 11. Clear .04 -.03 .73 12. Familiar -.08 .73 .40 Usefulness 13. Helpful .41 -.12 -.08 14. Meaningful .88 -.10 -.03 15. Necessary .75 .01 .04 16. Useful .85 —.13 -.02 17. Important .61 .13 .05 18. Valuable .74 -.02 .33 101 TABLE 25 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AFRICAN FIFTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .16 .85 .18 2. Like .31 .51 —.01 3. Good .09 .58 .40 4. Lively -.07 .02 -.31 5. Nice .68 .23 .22 6. Interesting .07 .38 —.15 Difficulty 7. Simple .44 -.05 .54 8. Easy -.30 .26 .02 9. Light -.07 -.23 .45 10. Understandable -43 .41 .50 11. Clear .04 -.02 .47 12. Familiar .02 .38 .33 Usefulness 13. Helpful .78 -.07 .04 14. Meaningful .46 .26 -.24 15. Necessary .64 .27 .23 16. Useful .60 .33 .40 17. Important .61 .08 .17 18. Valuable .87 .02 .30 102 TABLE 26 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AFRICAN FIFTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .25 .67 -.14 2. Like -.14 .75 .20 3. Good .58 -.02 -.08 4. Lively -.30 .75 .24 5. Nice .23 .47 -.24 6. Interesting .37 .57 -.21 Difficulty 7. Simple -.01 .03 .23 8. Easy —.18 ‘ .06 .37 9. Light .11 .07 .57 10. Understandable .28 .27 .61 11. Clear .14 .06 .62 12. Familiar .10 .40 .34 Usefulness ' 13. Helpful .79 .05 .10' 14. Meaningful .75 -.01 .13 15. Necessary .43 .25 .05 16. Useful .39 .18 .04 17. Important .38 .22 .04 18. Valuable .10 -.06 .19 103 loads significantly on helpful-unhelpful, useful-useless, necessary-unnecessary, important-unimportant. The factor loads significantly mainly on the kind of scales which in fact represent the Usefulness dimension. Factor II loads significantly on five of the scales representing the Evaluation dimension and on only one from another dimen- sion. This is clearly an Evaluation factor. Factor III loads significantly on scales which are mainly drawn from the Difficulty dimension. In fact five out of the six scales of the scales representing the Difficulty dimension load significantly on Factor III. The Usefulness, Evalua- tion and Difficutly factor structure seems to be suggested by these results as characterizing the attitudes of African fifth grade subjects toward reading. Factors from African Sixth Grade Subjects' Ratings The Usefulness, Evaluation, and Difficulty dimen- sions clearly emerge in the attitudes of the African sixth grade subjects towards reading. Tables 27, 28, and 29 provide this information. For instance, in Table 27, six 0f the scales representing the Usefulness dimension load significantly on Factor I. The scales representing the Usefulness factor dominate the Factor I in terms of scales with significant factor loadings. Factor II loads significantly on scales like alive-dead, good-bad, lively-inactive, nice-awful, which represent the Evaluation factor. Factor III loads 104 TABLE 27 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT FORMAT BY AFRICAN SIXTH GRADE CHILDREN Factor II Factor III Scales Factor I Evaluation 1. Alive .27 .56 .08 2. Like .31 .48 -.30 3. Good .37 .74 .19 4. Lively .02 .41 -.11 5. Nice .19 .54 .14 6. Interesting .07 .60 -.15 Difficulty 7. Simple .39 .19 -.02 8. Easy .55 .18 -.35 9. Light -.22 .04 .49 10. Understandable -.04 -.07 .35 11. Clear -.08 .04 .31 12. Familiar .21 -.04 .68 Usefulness 13. Helpful .67 .38 -.16 14. Meaningful -55 .48 .05 15. Necessary .55 .48 -.08 16. Useful .70 .28 .01 17. Important .60 .19 -.19 18. Valuable .54 -.23 .16 105 significantly on scales like light-heavy, understandable- confusing, clear-confusing, familiar-strange, which repre- sent the Difficulty dimension. Therefore, in this table, which gives results from the data on three-point format, the first, second and third factors are Usefulness, Evaluation, and Difficulty, respectively, in the same order of importance. Results from the ratings using five-point formats given in Table 28 show that Factor I loads significantly all the six scales representing the Usefulness dimension, similarly, Factor II loads significantly on all the scales representing the Evaluation dimension. Factor III loads significantly on four of the six scales representing the Difficulty factor but only loads on one other scale repre- senting a different scale. This is therefore clearly a factor representing Difficulty, while the first and the second factors represent Usefulness and Difficulty, respectively. The table presenting the results from the factor analysis of the ratings on the seven-point formats (Table 29) by African sixth grade subjects also shows the same dimensions. Factor I seems to load significantly on all the scales representing the Usefulness dimension, Factor II has significant loading on all the six scales repre- senting the Evaluation dimension. Factor III loads sig- nificantly on five of the scales representing the Difficulty dimension out of the six observed significant 106 TABLE 28 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AFRICAN SIXTH GRADE CHILDREN Factor II Factor III Scales Factor I Evaluation 1. Alive .15 .72 .06 2. Like .38 .74 -.21 3. Good .75 .08 -.04 4. Lively -.01 .44 -.48 5. Nice -.10 .03 .44 6. Interesting .32 .76 .03 Difficulty 7. Simple .09 .70 .19 8. Easy .28 .24 .46 9. Light -.51 .01 .31 10. Understandable -26 .05 .79 11. Clear -.44 .16 -.21 12. Familiar -.13 -.03 .77 Usefulness 13. Helpful .61 .06 .18 14. Meaningful -94 ~28 .06 15. Necessary .76 .25 .02 16. Useful .55 -.02 .30 17. Important .89 .28 .07 18. Valuable .74 .25 .07 107 TABLE 29 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AFRICAN SIXTH GRADE CHILDREN Factor III Scales Factor I Factor II Evaluation 1. Alive .15 .50 -.01 2. Like .20 .54 .29 3. Good .16 .05 .07 4. Lively .01 .91 -.04 5. Nice .04 .38 .09 6. Interesting .19 .30 -.01 Difficulty 7. Simple .33 .27 .43 8. Easy .48 .23 .07 9. Light .03 .05 .74 10. Understandable .20 .02 .52 11. Clear -.11 .04 .39 12. .Familiar .20 .19 .01 Usefulness 13. Helpful .90 .09 -.21 14. Meaningful .83 .03 .01 15. Necessary .76 .05 .13 16. Useful .85 .01 -.09 17. Important .87 .07 -.10 16. Valuable .59 .05 .02 108 factor loadings. This again suggests that the attitudes of the African sixth grade subjects are characterized by Usefulness, Evaluation and Difficulty dimensions in that order of importance. Factors from African Seventh Grade Subjects' Ratings The Usefulness, Evaluation, and Difficulty factors emerge in the factor analysis of the seventh grade ratings on all the three formats. For instance, the data from the factor analysis of the ratings on the three—point format by African seventh grade subjects (presented in Table 30) show that Factor I is a Usefulness factor because it mainly loads signifi— cantly on scales representing the Usefulness dimension like, helpful-unhelpful, necessary-unnecessary, useful- useless, valuable-worthless. Factor II is basically an Evaluation factor because it also loads significantly mainly on scales representing the Evaluation dimension. Factor III loads significantly on simple-difficult, easy- hard, understandable-confusing, clear-confusing and familiar-strange which represent the Difficulty dimension. Hence, a Usefulness, Evaluation and Difficulty is also suggested. Table 31 presents the results of the factor analy- sis of the ratings from the five-point format. Factor I loads significantly on all the scales representing the Usefulness factor. The significant loadings on Factor II 109 TABLE 30 EXTRACTED FACTORS FROM RATINGS USING THREE-POINT FORMAT BY AFRICAN SEVENTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive -.11 .90 .17 2. Like .30 .59 .04 3. Good -.05 .83 .05 4. Lively .02 .48 .17 5. Nice .24 .05' .04 6. Interesting .14 .14 .57 Difficulty \ 7. Simple .40 .24 .56 8. Easy .01 .06 .59 9. Light .34 -.16 .20 10. Understandable .05 .47 .49 11. Clear -.27 .06 .67 12. Familiar .36 .08 .57 Usefulness 13. Helpful .45 .08 .06 14. Meaningful .10 .02 .65 15. Necessary .77 .11 .08 16. Useful .39 -.04 .08 17. Important .57 .01 .12 18. Valuable .48 .23 -.17 110 TABLE 31 EXTRACTED FACTORS FROM RATINGS USING FIVE-POINT FORMAT BY AFRICAN SEVENTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .42 .56 .53 2. Like .25 .50 .33 3. Good .05 .66 .38 4. Lively -.ll .21 .02 5. Nice .26 .30 -.04 6. Interesting .26 .48 .09 Difficulty 7. Simple .09 .18 .38 8. Easy .68 -.02 .37 9. Light .73 .07 .02 10. Understandable .05 .78 -.08 11. Clear -.14 -.04 .70 12. Familiar .08 .15 .87 Usefulness 13. Helpful .70 .09 .10 14. Meaningful .71 .61 .01 15. Necessary .90 .ll -.16 16. Useful .92 .03 -.14 17. Important .59 .24 .12 18. Valuable .95 -.01 .08 111 are dominated by scales representing the Evaluation factor while the significant loadings on Factor III are dominated by scales representing the Difficulty factor. Therefore, the first, second and third factors in this case are Usefulness, Evaluation and Difficulty, respectively. In Table 32, which presents the results from the factor analysis of the ratings using seven-point formats, the significant loadings on Factor I are dominated by scales representing the Usefulness dimension. In fact all the six scales representing the Usefulness dimension load significantly on Factor I. The scales representing the Evaluation dimension dominate the significant loadings on Factor II and the significant factor loadings on Factor III are dominated by scales representing the Difficulty dimen- sion. The results here again show that the attitudes of African subjects toward reading are characterized by, first, the Usefulness dimension, then, the Evaluation dimension, and third, by the Difficulty dimension. Summary The table below (Table 33) gives a summary of the results of this study regarding the factor structure (dimensionality) of the attitudes of the African subjects toward reading. With the exception of results from the ratings using three-point and seven-point formats among fourth grade subjects, which are omitted, the other results 112 TABLE 32 EXTRACTED FACTORS FROM RATINGS USING SEVEN-POINT FORMAT BY AFRICAN SEVENTH GRADE CHILDREN Scales Factor I Factor II Factor III Evaluation 1. Alive .12 .81 -.09 2. Like -.06 .60 .29 3. Good .21 .77 -.01 4. Lively .54 .09 .14 5. Nice .07 .63 -.05 6. Interesting .46 .56 .16 Difficulty 7. Simple .23 -.10 .92 8. Easy I .23 .44 .35 9. Light .13 -.12 .86 10. Understandable .31 .38 .20 11. Clear .05 -.01 .36 12. Familiar .29 .36 .43 Usefulness 13. Helpful .66 .22 .37 14. Meaningful .74 .04 .24 15. Necessary .81 .18 .16 16. Useful .77 .23 .02 17. Important -36 .07 .54 18. Valuable .71 .17 —.04 113 suggest that the first factor is Usefulness, then Evalua- tion, and the third is Difficulty. TABLE 33 SUMMARY OF FACTORS AMONG AFRICAN SUBJECTS Grade Format Factor I Factor II Factor III 4 5 pt Usefulness Evaluation Difficulty 5 3 pt Usefulness Evaluation Difficulty 5 5 pt Usefulness Evaluation Difficulty 5 7 pt Usefulness Evaluation Difficulty 6 3 pt Usefulness Evaluation Difficulty 6 5 pt Usefulness Evaluation Difficulty 6 7 pt Usefulness Evaluation Difficulty 7 3 pt Usefulness Evaluation Difficulty 7 5 pt Usefulness Evaluation Difficulty 7 7 pt Usefulness Evaluation Difficulty Factor Score Reliabilities Factor Score reliability refers to the reliability of each of the three factors represented by the eighteen scales in this study. Difficulty and Usefulness. sented by six scales. Each of these factors is repre- These factors are Evaluation, The purpose of this section is to present information on the reliabilities of these factors 114 and determine which factor is measured more reliably in each grade using the various formats. This will be done for both American and African subjects. Factor Score Reliabilities for American Subjects Below, an attempt has been made to present the factor score reliabilities provided by different formats in all the four American grades used in the study. Factor Score Reliabilities-~American Fourth Grade The results from the ratings by American fourth grade subjects on three-point, five-point and seven-point scales show that the Evaluation factor is more reliable than the other two within fourth graders. For instance, the factor-score reliabilities from the ratings using the three-point formats are as follows: 1. Six Evaluation Scales = .78 2. Six Difficulty Scales = .75 3. Six Usefulness Scales = .38 There is not any significant difference between the Evaluation and the Difficulty factors but the differ- ence between these two factors and the Usefulness factor is significant at .05 level of significance. However, the Evaluative factor has a slightly bigger size of reliabil- ity coefficient than the other two factors. The factor reliabilities from the five-point and seven-point formats are summarized in the table below 115 (Table 34). It appears that on the five-point, the Difficulty factor is more reliable than the Evaluation factor but this difference is not significant at .05 level of significance. The Evaluation factor is more reliable than the two other factors on the information from the seven-point format. TABLE 34 FACTOR SCORE RELIABILITIES FROM FIVE-POINT AND SEVEN-POINT FORMATS USED BY AMERICAN FOURTH GRADE SUBJECTS Factor Format Reliability Coefficient Evaluation 5 pt .78 Difficulty 5 pt .81 Usefulness 5 pt .70 Evaluation 7 pt .77 Difficulty 7 pt .64 Usefulness 7 pt .57 Factor Score Reliabilities--American Fifth Grade The table below (Table 35) provides a summary of the different factor-score reliabilities generated by ratings on different formats by American fifth grade subjects. The Evaluation factor is more reliable than the Difficulty and Usefulness factors on all the three formats. 116 TABLE 35 FACTOR SCORE RELIABILITIES FROM THREE-POINT, FIVE-POINT, AND SEVEN-POINT FORMATS USED BY AMERICAN FIFTH GRADE SUBJECTS Factor Format Reliability Coefficient Evaluation 3 pt .77 Difficulty 3 pt .64 Usefulness 3 pt .57 Evaluation 5 pt .84 Difficulty 5 pt .67 Usefulness 5 pt .76 Evaluation 7 pt .88 Difficulty 7 pt .79 Usefulness 7 pt .71 Factor Score Reliabilities--American Sixth Grade With the exception of the factor score reliabil- ities provided by the ratings on the three-point format, the Evaluation factor is more reliable than the other factors. However, on the three-point format, the Useful- ness factor is more reliable than the Evaluation factor and this is statistically significant at .05 level of signifi- cance. The table below (Table 36) gives a summary of the various factor score reliabilities from the ratings by American sixth grade subjects. 117 TABLE 36 AMERICAN SIXTH GRADE FACTOR SCORE RELIABILITIES Factor Format Reliability Coefficient Evaluation 3 pt .44 Difficulty 3 pt .35 Usefulness 3 pt .71 Evaluation 5 pt .88 Difficulty 5 pt .84 Usefulness 5 pt .75 Evaluation 7 pt .87 Difficulty 7 pt .81 Usefulness 7 pt .72 The Evaluation factor is significantly (at .05 level of significance) more reliable than the Usefulness factor on both the five-point and seven-point formats. Factor Score Reliability--American Seventh Grade The table below (Table 37) provides the factor score reliabilities for the three factors from the ratings by American subjects using the three Semantic Differential formats. The Evaluation factor emerges as being more reliable than the Difficulty and the Usefulness factors. The second most reliable factor is Difficulty and on all 118 three formats, the Usefulness factor is the least reliable. On the three-point format, the Evaluation factor is sig- nificantly (at .05 level of significance) more reliable than the Usefulness factor. On other formats, there is no statistically significant difference between the reliabil- ity of the Evaluation factor and the other factors. This is also true between the factor score reliabilities of Evaluation and Difficulty on the three-point format. TABLE 37 AMERICAN SEVENTH GRADE FACTOR SCORE RELIABILITIES Factor Format Reliability Coefficient Evaluation 3 pt .72 Difficulty 3 pt .52 Usefulness 3 pt .43 Evaluation 5 pt .87 Difficulty 5 pt .86 Usefulness 5 pt .75 Evaluation 7 pt .89 Difficulty 7 pt .83 Usefulness 7 pt .82 Summa y Among the American subjects, the most reliable factor is the Evaluation. The least reliable factor seems 119 to be the Usefulness factor. In most cases, the Evaluation is quantitatively more reliable than the other two factors in terms of the relatively bigger size of its reliability coefficients but this is not always statistically signifi— cant at the .05 level of significance. Factor Score Reliabilities for African Subjects Among the African subjects, the Usefulness factor seems to be more reliable than the Evaluation and the Difficulty factors. There are few cases where this is not observed. An attempt has been made below to present the results concerning the factor score reliabilities using information from different formats in different grades. For instance, the table below (Table 38) gives the factor score reliabilities from the ratings by African fourth grade subjects using the three formats. From the ratings using both the three-point and the seven-point formats, the Usefulness factor is more reliable than the Difficulty and Evaluation factor. This is not, however, statistically significant at .05 level of significance. But the Evaluation factor is significantly more reliable than the Usefulness factor on the five-point format. The general picture seems to be that the Usefulness factor is the most reliable, followed by Evaluation and then the Difficulty factor. 120 TABLE 38 AFRICAN FOURTH GRADE FACTOR SCORE RELIABILITIES Factor Format Reliability Coefficient Evaluation 3 pt .32 Difficulty 3 pt .28 Usefulness 3 pt .45 Evaluation 5 pt .84 Difficulty 5 pt .70 Usefulness 5 pt .77 Evaluation 7 pt .45 Difficulty 7 pt .30 Usefulness 7 pt .52 Factor Score Reliabilities--African Fifth Grade Among the African fifth grade subjects, the Useful- ness factor is more reliable than the Evaluation and Difficulty factors on all the formats. This is true in terms of comparing the magnitudes of the reliability coefficients and also in computing the statistical tests of significance at .05 level of significance. Both approaches yield the same results showing that the Useful- ness factor is more reliable. The Evaluation factor is more reliable than the Difficulty factor on the five-point 121 and seven-point formats while the Difficulty factor is more reliable than the three-point format. However, these differences in the reliabilities of the Evaluation and Difficulty factors are not statistically significant (at .05 level significance). Therefore, the most reliable factor seems to be the Usefulness factor, then the Evalua- tion and then the Difficulty factor. TABLE 39 AFRICAN FIFTH GRADE FACTOR SCORE RELIABILITIES Factor Format Reliability Coefficient Evaluation 3 pt .37 Difficulty 3 pt .48 Usefulness 3 pt .74 Evaluation 5 pt .52 Difficulty 5 pt .45 Usefulness 5 pt .83 Evaluation 7 pt .57 Difficulty 7 pt .45 Usefulness 7 pt .83 Factor Score Reliabilities--African Sixth Grade The size of the reliability coefficients of the Usefulness factor is bigger than the reliability 122 coefficients of the Difficulty and Usefulness factors on all the three formats among African sixth grade subjects. It is only on the three-point and five-point formats that the Usefulness factor is significantly more reliable than the other two factors. This is at .05 level of significance. The Evaluation factor is more reliable than the Difficulty factor on the three-point and seven-point for- mats while the Difficulty factor is more reliable on the five-point format than Evaluation. None of these differ— ences in the reliabilities is significant at .05 level of TABLE 40 AFRICAN SIXTH GRADE FACTOR SCORE RELIABILITIES Factor Format Reliability Coefficient Evaluation 3 pt .35 Difficulty 3 pt .34 Usefulness 3 pt .71 Evaluation 5 pt .30 Difficulty 5 pt .33 Usefulness 5 pt .76 Evaluation 7 pt .50 Difficulty 7 pt .49 Usefulness 7 pt .58 123 significance. However, the most reliable factor seems again to be Usefulness, followed by Evaluation, and then the Difficulty factor. Factor Score Reliabilities--African Seventh Grade From the data generated from African seventh grade subjects, the Usefulness factor has bigger reliability coefficients than the Evaluation and Difficulty factors on all the formats used in this research. On the five-point and seven-point formats, the Usefulness factor is signifi- cantly more reliable than the other two factors (p = .05). TABLE 41 AFRICAN SEVENTH GRADE FACTOR SCORE RELIABILITIES Factor Format Reliability Coefficient Evaluation 3 pt .64 Difficulty 3 pt .71 Usefulness 3 pt .72 Evaluation 5 pt .70 Difficulty 5 pt .53 Usefulness 5 pt .91 Evaluation 7 pt .73 Difficulty 7 pt .72 Usefulness 7 pt .89 124 This shows that the Usefulness factor is clearly the most reliable factor in this case again. The Evaluation factor is more reliable than the Difficulty factor on the five-point and seven-point for- mats while the Difficulty factor is more reliable than Evaluation on the three-point format. These differences in the reliability coefficients are not significantly differ- ent at .05 level of significance. However, just by exami- nation of these reliability coefficients, the Evaluation factor seems to be the second most reliable factor while the Difficulty factor is the third most reliable factor. Summary The most reliable factor when the Semantic Differ— ential technique is used to rate reading by African fourth to seventh grade subjects is the Usefulness factor. The superiority of this factor in its reliability relative to the reliabilities of the Difficulty and Evaluation factors is confirmed on ten out of the twelve observations made using different formats in the four grades. The second most reliable factor is the Evaluation factor and then the Difficulty factor is the third factor in terms of reliability. CHAPTER V CONCLUSIONS, RECOMMENDATIONS AND DISCUSSION Conclusions A number of conclusions are available from this study concerning reliable formats of the Semantic Differ- ential for subjects from different grades, the factor structures or dimensions identified and the factors which the technique is likely to measure more reliably in Ameri- can and African subjects. These conclusions are listed below. 1. Among the American subjects, the five-point and the seven-point formats are generally more reliable than the three-point format. This is true even at the lower grades like fourth grade and fifth grade. Among the African subjects, the five-point format is more reliable than the three-point and the seven-point format in fourth grade. In fifth grade, the five- point format is slightly more reliable than the seven- point and the three-point formats. In the sixth and seventh grades, the five-point and the seven-point are equally reliable but more reliable than the three- point format. Generally, the five-point format is more reliable in the lower grades (i.e. fourth and fifth grades) while in the sixth and seventh grades five-point and seven-point are more reliable than three-point. The reading attitudes of American subjects are charac- terized, first, by the Evaluation dimension, then, the Difficulty dimension and finally, the Usefulness dimension. Since the Evaluation, Difficulty, and Usefulness scales correspond, respectively, to the Evaluation, Potency, and Activity scales in studies carried by Osgood, it seems reasonable to conclude that the EPA (Evaluation, Potency, and Activity) 125 126 factor structure is confirmed among the American subjects. Among the African subjects, the first most important factor characterizing the attitudes toward reading is Usefulness. The second factor is Evaluation, and the third factor is Difficulty. Given that the Usefulness, Evaluation, and Difficulty scales were selected to represent Osgood's Activity, Evaluation, and Potency factors, it appears that among the African subjects, the EPA (Evaluation, Potency, and Activity) factor structure was not confirmed. Instead, the dimensions were in the order of Activity, Evaluation, and Potency. The factor that was measured most reliably by almost all the formats used for rating reading by American subjects was the Evaluation. The second most reliable factor was the Difficulty factor which was followed by the Usefulness factor. Among the African subjects, the factor that was mea- sured most reliably was the Usefulness factor. This means that the ratings on the scales representing the Usefulness factor were more reliable than ratings on scales representing the other factors. The second most reliable factor was the Evaluation factor and the least reliable factor was the one represented by the six Difficulty scales. In both the American and the African subjects, the factors that emerged as being the most important, almost always, were the ones that were measured most reliably and the factors that were second in order of importance were the second most reliably measured. Similarly, the factors that emerged as being third in terms of their importance in characterizing the semantic space of the subjects toward reading were also third in terms of their reliability coefficients. For instance, among the American subjects, the most reliably factor was Evaluation which also was the first factor in importance. In African subjects, the most reliably measured factor was Usefulness and it was also the most important factor. Recommendations Among the American subjects, it is better to use either the five-point or the seven-point formats because they are more reliable than the three-point format. 127 2. Among the African subjects, it is better to use the five-point format of the Semantic Differential in the lower grades (fourth and fifth grades). In sixth and seventh grades, both five-point and seven-point are better for use because they are relatively more reliable. 3. Among the American subjects, information can be obtained more reliably if the Evaluation dimension is made the focus of the measurement using the Semantic Differential technique. 4. Among African subjects, information can be obtained more reliably if the Usefulness (Utility) dimension is made the focus of the measurement using the Semantic Differential technique. Discussion The factor structure of the African subjects differs from that of the Americans in terms of the order of factors. For instance, while the factors among the American subjects are Evaluation, Difficulty (Potency), and Usefulness (Activity), in the same order of impor- tance, the order of these factors is different among the African subjects. Among the African subjects the order of the factors is Usefulness (Activity scales), Evaluation, and Difficulty (Potency). There may be different reasons for the difference in the factor structure. One possible explanation may just be that these differences in factor structures reflect the natural differences between the American and African subjects. This would mean that while the American children would naturally first think of reading in terms of how it is interesting (Evaluation) the African subjects would first think of reading in terms of its utility (Usefulness). 128 So one possible postulate is that these differences reflect the natural differences imposed on the two groups by their cultures. The other possible explanation may be in the problems of translating the scales used in the research from English to Swahili, which is the language used by the African subjects. It is possible that when the English Usefulness scales were translated into Swahili they gained an evaluative connotative meaning and hence although they appear to be representing the Usefulness dimension they are actually the Evaluation scales. It is possible to accept the results on the dimen- sionality of the American subjects toward reading because there are other studies that confirm the similar factor structures among American subjects elsewhere (see the chapter on literature review). However, before the results on the dimensionality of the African subjects are final- ized, some replication studies, using more than the eigh- teen scales may be necessary. According to Osgood, et al (1957), to settle the dimensionality issue within a cul- ture, one needs to use numerous scales. For example, Osgood (1957) had to use about one thousand adjective pairs from the Thesaurus in order to settle the dimension- ality issue among the English speaking Americans. A similar number of scales may be necessary to settle the dimensionality issue among the Swahili speaking Africans. 129 However, it is worthwhile to note that the eigh- teen scales used in this study yield information which confirm the findings from numerous research projects, as discussed in the chapter on literature review. In this case, it may just be true that with the same number of scales, the dimensionality issue among the African subjects has been adequately solved. One final point to note concerns the generalizabil- ity of results of this study to other concepts and groups. Concerning the concept, the information obtained in this study can only be restricted to making inference con- cerning the attitudes toward reading. It may not be appropriate to generalize to other concepts without some additional research. Concerning generalizing to different groups, the study uses a fixed design model which rules out attempting to generalize to other groups. However, the research provides some useful information that may be relevant to identical African and American children in fourth to seventh grades. APPENDICES APPENDIX A APPENDIX A EXAMPLE OF SCALES AND DIRECTIONS IN ENGLISH SCh001 O 0 O O O O O O O O O O O O O O O C Grade 0 O I O C O O O O O O O O O O O O O O NMber O O O O O O O O O O O O O O O O O O ReadingiQuestionnaire Instructions Please answer each of the following statements with either "YES" or "NO." After you have completed responding to all the statements, wait for further instructions before you continue. Here are the statements: OmxlmUI-bWNH O .5 (A) N I'-‘ o o o o o o 15. 16. 17. I love reading. I feel good after I've read a book. I would be happy to get a book for my birthday. Reading school books is a waste of time. Reading stories is fun. It is fun to read books. like to find books to read. would rather play than read. can learn things from reading books. think reading books is silly. want to be absent from reading class. want more time in school to read. H +4 H +4 H +4 H get tired reading stories. It is good to know how to read. I hate reading books in school. I would rather reat than work on other things. Reading is the worst part of my day. 130 131 Instructions Now, how students feel about reading is very important, therefore, you are asked to fill forms given on the next three pages. This is not a test. There are no answers that are 'right' and 'wrong.‘ Each of the two words is a pair of opposites. Mark with (x) the space between each pair of words that tell how you feel about READING. Here is an example of how to do this. If you feel that READING is closely related to one end of the pair, you should place your mark as follows: Slow x Fast Slow __ __ __ __ _x Fast If you feel that READING is neutral with respect to both words in a pair, then place your mark in the middle position. The direction toward which you mark should depend on which of the ends seem most like what you feel about READING. IMPORTANT : 1. You will notice that on one of the forms you are provided with five spaces like in the example above. But you will also notice that there are two other different forms. In one of these, you are provided with three spaces, and in another, seven spaces. Fill the forms depending on how you feel about READING. 2. Place your marks in the middle of the spaces, not on boundaries. THIS NOT THIS 3. Place only one mark between each pair of words. 4. Do not leave out any of the pair. 5. Work quickly. It is better to give your first feelings rather than think hard about each pair of words. 132 READING _ Nice Useful Easy Interesting Important Light Like Clear Valuable Simple Alive Helpful Lively Necessary Familiar Understandable Good Meaningful School........ .......... .Grade.. ..... ...... Name. 0 O ......... O ....... Awful Useless Hard Boring Unimportant Heavy Dislike Confusing Worthless Difficult Dead Unhelpful Inactive Unnecessary Strange Confusing Bad Meaningless 10. ll. 12. 13. 14. 15. 16. 17. 18. 133 READING Interesting School ............... ... Grade.. ................. NameOOOOOOOOOOOOOO Boring Lively Inactive Good Bad Light Heavy Simple Difficult Familiar Strange Nice Awful Like Dislike Alive Dead Useful Useless Helpful Unhelpful Meaningful Necessary Valuable Meaningless Unnecessary Worthless Important Easy Clear Unimportant Hard Confusing Understandable Confusing 134 School........... ....... Grade ........ ... ....... . Name.. .................. READING 1. Simple ___ Difficult 2. Easy __ Hard 3. Light __ Heavy 4. Alive __ Dead 5. Like __ Dislike 6. Good __ Bad 7. Helpful __ Unhelpful 8. Understandable __ Confusing 9. Lively ___ Inactive 10. Meaningful ___ Meaningless. 11. Nice __ Awful 12. Clear __ Confusing l3. Necessary __ Unnecessary 14. Useful ___ Useless 15. Important __ Unimportant 16. Familiar ___ Strange l7. Valuable ___ Worthless 18. Interesting Boring APPENDIX B APPENDIX B EXAMPLE OF SCALES AND DIRECTIONS IN SWAHILI Shule................... DarasaOOCOOOOOOOOOOOOOOO Jina LakOOOOOOOOOOOOOOOO Orodha Ya'Maswali Juu Ya Kusoma Maelezo Tafadhali jibu maswali yafuatayo kwa kuandika 'NDIYO' au 'LA.’ Baada ya kumaliza maswali yote, ngojea hadi upate maelezo mengine kabla ya kuendelea. Haya ndiyo Maswali: \.OCD\1C!\U'IobLAJ(VI--l O +a P'PJ H +4 H +-I H ~q o‘cn b-L» Ri+4 o O O O O O O O O Napenda kusoma. Ninasikia vizuri sana baada ya kusoma kitabu fulani. Ningependa kupata zawadi ya kitabu. Kusoma vitabu vya shule ni kuharibu wakati tu. Kusoma hadithi kunanipendeza. Kusona vitabu kunanipendeza mno. Ninapenda kutafuta vitabu vya kusoma. Ni heri kucheza badala ys kusoma. Ninaweza kujifunza mengi kutokana na kusoma vitabu. Ninafikiri kwamba kusoma vitabu in jambo la upuzi. Sipendi kuwa katika darasa wakati wa kusoma. Ningependa kuongezwa wakati wa kusoma shuleni. Kusoma hadithi kunanichokesha. Ni vyema kujua kusoma. Ninachukia kusoma vitabu shuleni. Ni heri kusoma kitabu badala ya kufanya mambo mengine. Kusoma kunaniharibia siku nzima. 135 136 Maelezo Sasa, wanafunzi waonavyo juu ya jambo la kusoma ni muhimu sana, kwa hivyo, unaulizwa kujaza kurasa tatu zifuatazo. Huu si mtihani, la! Hakuna jawabu yoyote ilio sawa au mbaya. Kila neno 1a maneno mawili uliyopewa ni kinyume cha lingine. Onyesa na (x) katika nafasi ilio katikati ya haya maneno mawili kuonyesha uonavyo moyoni mwako juu ya jambo la KUSOMA. Hebu angalia huu mfano. Kama unaona kwamba KUSOMA kunalingana sana na neno lililo upande mmoja wa haya maneno mawili, basi onyesha hivi: Polepole x Haraka Polepole __ __ __ __ _x Haraka Kama unaona kwamba kusoma hakulingani na upande mmoja wala mwingine, basi weka alama katikati ya maneno hayo. Weka alama hiyo karibu na upande ulioko karibu na lile neno lielezalo maoni yako juu ya KUSOMA. MUHIMU: l. Utaona kwamba karatasi moja ulilopewa lina nafasi taso (5) kama mfano ulioko hapo juu. Lakini utaona kwamba kuna makaratasi mengine mawili yaliyo tofauti. Karatasi moja lina nafasi tatu katikati ya maneno, na lingine lina nafasi saba (7). Tafadhali jaza karatasi katika hizo nafasi kulingana na maoni yako juu ya KUSOMA. 2. Weka alama yako katikati ya kila nafasi wala si upande: HIVI SI HIVI 3. Weka alama moja tu katikati ya maneno hayo mawili. 4. Jibu maswali yote. _ 5. Jibu haraka. Ni bora kujibu vile uonavyo mara tu. 137 Nikwema Kunafaa Nirahisi Kwavutia Nimuhimu Nikwepesi Kwapendesa Niwazi Nibora Nirahisi Nikwepesi Kunasaidia Nikwepesi Nilazima Nikuzoevu Kunaeleweka Nikuzuri Nikusudi Shule ........... ........ DarasaOOOOOOOOOO ........ Jina Lako...... ........ . Nikubaya Hakufaidiki Nikali Kwachokesha Simuhimu Nikulivu Sikupendi Huangaisha Sibora Nikugumu Nidhaifu Hakusaidii Nikuzito Silazima Nikugeni Hubumbuaza Nikubaya Nihafifu 138 Shule....... ............ Darasa.... .............. Jina Lako..... .......... KUSOMA Kwavutia _____ Kwachokesha Nikwepesi _____ Nikuzito Nikuzuri _____ Nikubaya Nikwepesi _____ Nikulivu Nirahisi _____ Nikugumu Nikuzoevu _____ Nikugeni Nikwema _____ Nikubaya Kwapendesa _____ Sikupendi Nikwepesi _____ Nidhaifu Kunafaa _____ Hakufaidiki Kunasaidia _____ Hakusaidii Nikusudi _____ Nihafifu Nilazima _____ Silazima Nibora _____ Sibora Nimuhimu _____ Simuhimu Nirahisi _____ Nikali Niwazi _____ Huangaisha Kunaeleweka Hubumbuaza 139 Shule..... ....... . ...... Darasa.... ..... . ........ Jina Lako ............... KUSOMA Nirahisi __ __ Nikugumu Nirahisi __ __ Nikali Nikwepesi __ __ Nikuvuli Nikwepesi __ __ Nidhaifu Kwapendesa __ __ Sikupendi Nikuzuri __ __ Nikubaya Kunasaidia __ __ Hakusaidii Kunaeleweka __ __ Hubumbuaza Nikwepesi __ __ Nikuzito Nikusudi __ __ Nihafifu Nikwema __ __ Nikubaya Niwazi __ __ Huangaisha Nilazima __ __ Silazima Kunafaa __ __ Hakufaidiki Nimuhimu __ __ Simuhimu Nikuzoevu __ __ Nikugeni Nibora __ __ Sibora Kwavutia Kwachokesha APPENDIX C APPENDIX C 140 RESULTS ON TESTS OF STATISTICAL SIGNIFICANCE ON DATA FROM AMERICAN SUBJECTS (p=.05) Grade Formats V and V F Ratios Decision 4 3 and 5 51 and 49 1.846 significant 4 3 and 7 51 and 49 2.18 significant 4 5 and 7 51 and 49 1.181 non-significant 5 3 and 5 51 and 49 1.466 non-significant 5 3 and 7 51 and 49 2.00 significant 5 5 and 7 51 and 49 1.363 non-significant 6 3 and 5 54 and 54 2.6 significant 6 3 and 7 54 and 54 2.36 significant 6 5 and 7 54 and 54 1.1 non-significant 7 3 and 5 51 and 49 2.00 significant 7 3 and 7 51 and 49 2.166 significant 7 5 and 7 51 and 49 1.083 non-significant 141 RESULTS ON TESTS OF STATISTICAL SIGNIFICANCE ON DATA FROM AFRICAN SUBJECTS (p=.05) Grade Formats V and V F Ratios Decision 4 3 and 5 53 and 53 4.91 significant 4 3 and 7 53 and 53 1.43 non-significant 4 5 and 7 53 and 53 3.41 significant 5 3 and 5 51 and 49 2.15 significant 5 3 and 7 51 and 49 1.48 non-significant 5 5 and 7 51 and 49 1.45 non-significant 6 3 and 5 52 and 52 1.21 non-significant 6 3 and 7 52 and 52 1.27 non-significant 6 5 and 7 52 and 52 1.05 non-significant 7 3 and 5 53 and 53 1.75 significant 7 3 and 7 53 and 53 1.75 significant 7 5 and 7 53 and 53 1.00 non-significant 142 APPENDIX D DESCRIPTION OF THE CORRELATION COEFFICIENTS The correlation coefficients provided in the matrices that appear in the next twenty-four pages omit the decimal points but the presence of these decimal points are assumed to exist in front of each of the coefficients. Additionally, unities which should otherwise appear in the diagonal have also been omitted but these should also be assumed to exist. Rex The following key provides some idea of the scales that are represented by various numbers in the tables. 1 2 OtU'Iub \l 8 9 The first = Alive 10 = Like 11 = Good 12 = Lively 13 = Nice 14 = Interesting 15 = Simple 16 = Easy 17 = Light 18 Understandable Clear Familiar Helpful Meaningful Necessary Useful Important Valuable six scales represent the Evaluation factor, the next six represent the Difficulty factor and the last six represent the Usefulness factor. APPENDIX D moa 143 mom boo mmm moo omo NHH NNM omo mmN mmo Hao boo ohm «om Hoot mHoI mNH moot omo mod mmal moo mNH moot moat mmo mmq mbol Nom mvot oHo ohm omH «mot NNHI what mam mom who «Ho moo mom omv mmol mHHI moat omN HNH HhNI mmol vHv mom «vol mom mhol woo mom moot oHHI mmo mNm woo woo mmo Hmm mmo ohol mom ova NoHI «mm Hod moN who Nmo moo mNN omal moo Hum hva mva Hmv mom mae NNo mmo ham 5mm mmm omHI ovN mho omo mad mmm mmm Ham omol wan mmH mmm «mm how «sol hMHI mmm mHo ohml whol Nom mbv NNHI 5mm hom wmv on ohm mnm omHI boo how boa omol Nmm mod mNN omo mom mmm mmm Hmv hem mom mow omol Noon moo mmo woo Hmm mho mmm oHoI ANN mod mma Hmm mom Hmm Hmm mom em. mm. mm. om. om. mm. on. mm. mo. mm. mm. mm. Nb. om. mm. mv. mh. Nb. Om O‘OHNMQ‘IDSDFM r-lu-lc-Iv-lt-iI-lv-It-lr-l HNMV‘MWFQ m8<2m0h BZHOmImmmmB OZHmD mUZHfiflm madmw mBmDOh ZQUHMWZ< 20mm mBZmHUHmmmOU ZOHBdAmmmOU oam 144 ovN aom Nov mom mmv moo ohm va amm Noa oma mmm mmm moo moa mma oom vma amN mho woo woo oNa amN oNa omo ooN mmm mmm mmal maa mom ooal mav Now Nba ooo moo mam mom moa mum omw mam hmm oma aoa mmv aNa oom vmm amo amq Nov omo mma mmN aoal oam awn oom ooN mom who vam Nmm Noa mmm oma aoa mmm oNN mma who mmo ova aaN omNI oaal mNo oaa mam ooal maol va arm voa ooNI new mmm mmm mma oma Naa amo ohm oool amm «ma oha hum moo obv ovm amm mom Nvm aha mmN oNN Nma mvo vow moa ooN mma amm one mom moo mom mmo mNm ovm omo moo mho who own mbN owm mov mmm moo mum amN on ooN aaN oma mmN mmN aNa mmv NVN mom ooo mow ooN mma mom Nvm a.a m.m N.a o.m v.a m.m m.a m.m mm. ¢.¢ N.a o.m o.a N.v o.a N.¢ o.a o.v a.a o.¢ a.a o.¢ mm. v.v o.a m.m mm. N.v m.a h.m a.a o.m a.a N.v o.a o.v Om x O‘OI—INMV‘IDWI‘Q r-lt-ir-II-lI-lr-iI-lr-iv-l HNMV‘ID‘DI‘Q mfidzmom BZHOmIm>Hm OZHmD mUZHB¢m mnflmw mBmDOm ZdUHmm2¢ 20mm mBZMHUHhmMOU ZOHB¢ammmOU mmo 145 mow ooo aow mwo amo wmm ooo moo omw mmo mwa oma woa oow oma Now mwal mmm moa awa ooo moo owol ooa mow ooN oam ama on omo aoo oao mmw aom mmm mmm amo mmol oam omo oam wmm moo Now amo maol wwa ooo oam mNN owo mww oow owo mool mom mmol oool ooo oom Noa mwm oaw mmm Now ooN mow Nmol oma maw mmm mow ooo wma oma omm omw oNNI omm ooNI oool oNoI mma ooa mom mwN aoN mmo ooo wow mma mmo amN oow mma maal ooN oao mmo moo wool mwo Now aoN moa oNo wmo Naol moo awn mmm mow moa mNo ooa moo mww on aom waw oom Nma moa aoN mma moa omm omo oNo ama mmw aoo mow ooo oow wmo moN ooN wmo omm mmm aoa .oom Now oom wNm oaa wa ooo wwm oww wom o.a o.~ mo. m.a o.a m.a m.a o.N N.N N.N o.a m.a m.a a.~ m.a m.a m.a o.a Om moamowmomo HHHI-lr-lu-lo-II—iv-i HNMQ'LDWI‘CD mUZHeflm mafimw mBmDOm ZflUHmMZd 20mm mBZmHUHmmMOU ZOHB¢ammMOU mfiflzmom BZHOmIZm>mm OZHmD omo 146 omw owo ooo omo moo moo mao moo oma mma omo mmo mmo mmo ooo ooo omo aoal oma wool ooo ama ooo oool mow mmol owm moa moa mmo ooal oow omal moa aow omo ooo woa oool ooo amw owo mmo aoo mool mowl aowl mool ooo omo ooo moo omo awa moat wool maol mmal moo mool moo aaol mma ooo mma oma woa oow owa ooo moo mao owo owo wam omol mmo ooo mma mma owo maa moo moo mma woo wao maa oaa mow oaol ooo moo woa moo owa ooo ooo aoo owo ooa maol ooo moo oow omo ooo owal mmo ooal wma oma owo moo omol ooo mmw omm ooo moo oow mmw ooo oow oool woo oom amw moa ooo omw aoo moo omw oow woa oo. w.o owa ow. m.o maa mo. w.o omal om. m.o omo ow. m.o mool mw. o.o mwo om. m.o ooa om. w.o mwo om. m.o ooo oo. o.o moo ow. m.o oool wo. o.o amw oo. o.o amw ow. o.o oaw ao. o.o aom oo. o.o wow om. m.o I mw. m.o no N O‘OHNMQ‘IDOI‘” HHHHHHHHI—l r-INMV‘IDOI‘Q mBgOh BZHOmImmmmB UZHmD mUZHfig mago m.a:mHna Z<0Hmm2< 20mm m.a.ZmHUHmmMOU ZOHBsnmmmOU omm 147 moo ooo amo woa wwa oao woo mmaI oom wmol oow omo owo ooo woo aoa am. o.w oa maw oom moo aom oaal owo oooI moo ooaI ooo ooo aoo omo ooaI omo mwa om. o.w ma I oow omo oow aoaI ooo moo mool moa amoI oom oaa oow mmw oao mmo o.a o.o oa I omo mmoI aao aoo wmo wmaI waaI aool oam omoI mma wmo mow mow mo. o.w ma I ooo waal mooI amaI omoI ooa mao waw wwal aaa owm mow ooa om. w.w wa I wool ooa omo omaI ooo moo omw omo mma moo omo oaa mo. o.w oa I wwo mmw oow oooI ooaI oaal oooI omo omo oooI ooa mm. m.o oa I oom mmw omw aom moo omo oon moo wmo amo a.a m.o aa I mow omm oww moo ooo oaoI oooI maa moa am. a.w oa I mmo ooo ooa oowI oao omoI oool mon mm. o.o m I owm ooo mmo oma mool maa ooo mm. o.w o I oma wmo ooa mool wmo ooa mo. o.w m I oom omw moo oom ooo o.a o.w o I ooo mow wwm mmw om. o.w m I ooo mmo oom mo. o.w w I ooo omm mm. m.w o I ooo a.a o.w o I om. o.w a mm x oedzmom BZHOmIm>Hm UszD mUzHefim madmw mamam Z¢UHmmz< 20mm mEZmHUHmmmOU onedammmOU moo 148 oow omo mom moo wmo moo woa omw ooo oao maw wmo ooa ooo owm amw wmm omw mmo oaw ooo oma oma mmoI amw aoo amw ooaI moaI oao omoI owo omo omw ooo moa moa mma omo mma woo mao mwo ooo moo moo woo omo oao omo oom waw oow owm mmo moa mmo oao omm mma awm wmw oaw omm omo oow amo mwo woa awo ooo moo oma mmo amo mww waw ool omoI ooo ooo oooI amoI omo oooI mool oaa amoI ooo oow ooa mool moo ama aaa ooaI owo oon aoa oaw mmo mmw amm oom moo moo omo oao woo mmo mao moo mow moo omw mmw aao omm woo mmw omo oom omo aooI aooI wom moa moa oaa mma ooo mmm. wmm ooo oom ooo moo omo wwo moo moo wom omo mwo owo ooo oma amm omw moo moo ooo o.a o.m w.a w.m m.a m.m m.a o.m a.a o.o m.a a.m o.a m.m w.a o.m o.a o.m o.a o.o o.a a.o mo. m.o o.a o.m o.a o.m o.a o.o o.a m.m o.a o.m o.a m.m cm x moaoowmomo I—lI-lHr-ir-lr-ir-II-lu-i HNMV‘MWI‘Q wadzmom BZHOmIZm>mm DZHmD mGZHfiflm madmw mBmHh 240Hmm2d 20mm mBZmHUHmhmOU ZOHfidammmOU ooo 149 mow aoo moo moa I amo moo omo moa ooo wow moo mow ooo owa wao ooo ooo moo mmw mmo omoI mma mma amo awo mao oow aoo mao ooo mmoI ooo moa woa mma waa moa moo ooa mmoI moo moo woo ooo ooa wool mwo omo oma oma ooa mow mow moo mac Hem mmoI ama Abe mean Heo mam Awe Ham mam ooo oaal oooI amo omoI woa woaI ooa mmo oma oon ooa wmo oma omo ooo woa oow ooo owo ooo ooo oma moa moo ooo mwol oaal mool maa owaI moo oaa oooI owa mwaI mma omw omo omo mma ooo wool aoa oao ooa moo oma omaI omoI ooo omo omm mow moo moaI aoo mma ooo ooa aoo ooa omaI ooa ooo mow amo wow ooo waw amo om. m.o oooI mm. m.o aao mm. o.o omoI ow. o.o mmoI ow. m.o ooo mm. m.o ooo wm. m.o amo wm. o.o ooo om. m.o ooo om. m.o mao mo. o.o oaoI wm. m.o mom mw. m.o mom om. m.o aow wm. m.o owo mo. o.o oow om. o.o I mw. o.o 0m x O‘OHNMV‘MWI‘Q t-lI—iHF-lr-lF-lr-lr-II-i HNMQ‘MWI‘Q mfiézmom BZHOmImmmmB UZHmD meHBfim mUZHfiflm m0Hm OZHmD oma 151 moo omm mow wmw omo oow mmw aoo oww omo mam mmo wmw mom oom mao amo wmo ooa ooa owa moa ooo woo maaI amo omoI omo oao ooo aoo ooa mow mow I mow ooo woo ooo omo ooo ooo mow oom omm moa ooo moa omo aoo omo ooa awo awo mom oaa ooo omo moo omo omo oom woo omm wmo moo wow omo oon ooa mon aooI mma ooo ooo moo owo mmoI oam owo omoI oooI omo oooI mmoI mmo mma moo ooo owo I oom mmm aoa wao omo mow moo amo owa moo aom oom owo oom oow oam ooo ooo oaoI moo omo wom mwo omo wao oao wmm woo ooo ooo mmm mwo owo mmo woo mma oom ooo mma amo moa mmo omm mom omo mom awo o.a omo o.a m.m mmo o.a m.m mao o.a a.o omo m.a o.m ooa m.a owm ooo w.a m.m owaI w.a m.m aoo o.a o.m aao m.a a.m mma w.a o.o oooI o.a o.o moo m.a o.m moo w.a m.m ooo w.a o.m wwm m.a o.m moo m.a w.m I m.a m.m 0m x moaoowmomo r-iv-II-lv-ir-lI-lv-lv-iu-i aoowmomo mUZHfidm madmo meHm Z<0Hmm2¢ 20mm mBZmHUHhhMOU ZOHBfiAmmMOU mBfizmom EZHOQIZm>mm OZHmD oww 152 omw mmw oao moa ooo wow owo moa aoo wmo omo owo ooo aoo omo ama aoa aww woo woa ooo mao oma aoo ooo ooo omo oao oao oma omo omo ama mow mma mwo aoa omo ooo oma mmo ooo ooo mmo oao oaa ooo maa mwo omo oow amm woo aow mmo ooo mao mma ooo waa mmo omo woo amo wmo ooa mao omoI moa ooo wma wmol woa oma omo waaI amo ooo mwa moa oao moo omo oooI mwo maoI aao wmo ooa oow mma mwaI moo ooa oma ooaI oma aao moa moo ooa ooo aow omo mwo mmoI ooo moo aoa ooo ooo mmo ooo mmo mwo mmo wwm wmo owm woa mool woa oooI ooo ooo ooo oma aoo moo omo omoI mmo oom mow mmo wmo om. a.a Ame om. e.~ ama mm. m.m emo om. m.~ «no em. w.m mad em. e.~ mmm- am. m.m mma em. m.m ewe em. a.” Sea mm. 4.~ owe Rm. m.m Nee me. e.~ new mm. m.m mom em. a.a Hoe Rm. m.m Ham he. R.~ one am. m.m I mm. m.m an x O‘OHNMQ'MVDI‘Q v-lI-ir-lr-lI-lI-lI-lI-Ir-l aoowmomo maflzmom BZHOmImmmmB UZHmD meHfiflm MQMm ZQUHmmzd 20mm wBZHHUHmmMOU ZOHfimammmOU ooo 153 mww aom wmw moo moo wmo woo omo oaw mmw oom woo oom moo mma woo ooo oma woo mmw ooo mool omaI ooa ooo oao ooo owo ooo omo mool wool mool oma mmo ooo owo wwo mwo ooo moo mam oom I oam woo wmo moo mmo mwo oom oma mow wmm omo mmo omo owo omo moaI ooo oao mmo moo wmo moo mool ooa aoaI mao moo ama omo mmo ooo wwo mmoI moo ooo mmo won omo ooo moo oow wool ooo omo ooo ooo omo omo moo ooo owo woa omo omo mmo oaa oma moo oma mmo oow oooI oao aoo oao oool mmoI ama omo omo moa oma omo maw omo mmo moo omo moa moa ooo ooo mmo aooI ooa wmo ooo aoo oom aom aao I amo ooo am. o.w oaa om. o.o ooa oo. m.o mwo mo. a.w ooo mm. o.w ooo mo. o.w oom om. o.w mao o.a o.w wool o.a o.w omo a.a o.w mmo mo. w.w mmo o.a w.o mmw o.a m.o oom mo. o.w oom a.a o.w moo om. a.w mmo mm. a.w I oo. a.w no N O‘CHNMQ‘IOWI‘Q v-IHI-II-lv-iu—lv-iv-lr—l HNMQ‘IDOI‘CD mUZHBmm ZflUHmmzd 20mm mfizmHUHhhmOU ZOHBflAWmmOU mfidszh BZHOmIm>Hm UZHwD mwo 154 moo omw 3... m2. e3 omw mom wmw mow omw aoo mmw oww oom maa maoI oao oma mma wwo moo ooo wwo ooo oow ooo oow ooa awo owa owo ooo amo omw oow ooaI aaaI woal mao omo mao moo omo aoo moo ooo maa oaa awo moo aoo omo ooo omo mool omoI aoo ooo omo oaa woo oww moo ooo ooo ooo mmo moa aoo moo omo ooa oma maa aao ooo omo mmo ooa ooo ooa oaa ooo owo aow owm mow ooo oom oaw oao oao ooo aao owo wma mwo mma ooo mmo ooo amo mmm omm oom oow omo mmo ooo owo wmo wow wmw moa mom oaw omo mmm omm oow ooo owa oao oaa wmo owo oao aaw aoo omm oww oow moo omm omm ooo w.a w.m ooo o.a m.m mwo o.a o.m ooaI w.a m.m oaa m.a o.m moo a.a m.m ooa w.a a.m oma o.a w.m woa o.a o.m woo o.a m.w moo m.a w.m woo m.a m.m mmm w.a m.m oww o.a m.m mam m.a o.m wom w.a o.o omo w.a m.m I o.a m.m mm x HNMQ‘IDOI‘QO‘O mUZHadm m0¢m0 mBzm>mm Zmm OZHwD moo 155 omo ooo mmm mwm omo aow ooo mwo oao ooo mom moa aoo waw oaa moo mmo oao omw moo aow ooo mwo oao oom waw omw awo moa moa omo moo maw mao moo moo ooo mma oma oow owo oow oow wmo mool awo owo ooa mwa mmo moo mwa wow ooo omw amm owo mmo mow aow oow mom wmo omo aoa amaI omo omoI oool woaI ooa oon woaI owo moaI omo oaal oao awa mmo omo oooI ooaI aao oooI mool moaI ooo wool mom moo woo oao amw oow oaa ooa oow wom ooa omo mow mmo ooo owa mmw omo ooo oow oaa ooo oow owo ooa woo omo owm mom mao moa mao ooo oow wom ooo ooa wom mmw mmo oma amw mmw mam mmo aoo moo om. a.o mma om. o.o maal mw. o.o mao oo. o.o owa om. m.o ooo om. o.o mmoI om. o.o owa om. o.o wwo om. o.o moo oo. m.o oao mm. a.o wma om. o.m oao om. o.o mwa mo. m.o omo oo. m.a moo oo. o.o moo om. m.o I om. m.o Gm x oaoowmomo Hv-il-lI-lt-lv-ir-iv-il-i 14°40‘7meme madszh BZHOmImmmmB UZHmD mUZHedm madmw mBMDom ZdUHmm< 20mm mBZmHUHmmmOU ZOHBdammmOU mmo 156 omo mmm ooo mwm omo aow ooo mwo oao ooo mom mmo aoo waw oaa moo moa oao omw moo aoo ooo mwo aoa ooa waw oow awo moa moa omo moo mao maw moo moo ooo mma oma owo awo oow oow mow moo awo owo ooa mwa mmo moo wwo wow ooo omw amm owo mmo moo aow oow mom wmo omo aoo amal omo omoI oooI woao ooa oon woaI owo moat omo wool oaoI awa mmo omo oooI ooaI aao oooI moo moaI ooo wooI mom moo woo oao amw oow oaa ooa mow wom ooa omo mow mmo ooo owa mmw omo ooo oow oaa ooo ooo owa ooa woo omo owm mom mao moa mao ooo oow wom ooo ooa wom mmw mmo oma amw mmw mam mmo aoo moo o.o o.o mma w.a o.o aaaI m.a w.o ama o.a o.o oow a.a m.o oaa w.o m.m mmoI a.o o.o oom a.a m.o woo m.o a.m ooa w.o m.m omo m.o o.o wma a.a m.o oao w.o m.m mwa o.o m.m omo w.o m.m mao w.o m.m moo a.o a.o I o.o m.o 0m x aoowmomomo mfidzmom BZHOmIm>Hm UZHmD mUZHfiflm mnmm OZHmD mUZHfifim madmw mBmDOh ZdUHmhd 20mm mBZMHUHmmmOU ZOHBdammmOU woo 158 oow aom mmo mmo I omm wow oma mmo aom moo mao ooo maa omm owol ooa oaoI mwaI mmOI mmaI mao oma mooI omo ooo wool oao ooa moo omoI awo moo oooI owo mmoI moa moa mww mmo omo moo ooa ooa moo amo mma ooa ooa wmo ooo omw aoo moo mmo woo moo mooI aoo mwo oow omo mwo wmo aoa I mow awaI ooaI aoaI omoI owaI amoI oom omaI oooI mwaI omoI aoo moo woa ooo ooo oao owo aooI woaI oma ooo omo mmoI mwo oaa oaa ooaI ooo oooI mooI owo ooo moaI wmaI oooI omal moo oon mmo moor moo oon woo aon moaI oaoI oooI aoo oma omm amoI owa wmo owo GNHI mom mean ANN Neon ome Nee ama- mam ooa moo wmm mam Hem eoe oaoI amoI ooaI mool moaI oooI moo mooI woa mmoI aooI oooI aoo wwa oow aoo moo om. ao. wm. mm. om. mm. wm. mm. wo. wo. mo. ao. o.a om. mm. wm. ao. ao. Om moaoowmomo I-lI—lI-lI-lr-II-iI—IHI—I aoowmomo mfiflzmom BZHOmImmmmB UZHmD mUZHfiflm ma¢m0 mBhHm ZdUHmmd 20mm mBZMHUHmhmOU ZOHBfiammmOU moo moo owo 159 mmo moo wom mow mow aoo oww ooo oao aow mom omo mwa moa maa aoo ooal mmo mao moo moa omo oma oooI ooo mom oom ooo mmm wmaI maw aoo mow moa ooa moa mwa aooI ooa moo mmo mmw .man- mooI moaI ooaI owol mooI moa woa oao wwa aom mwa omw woo ooo mao ooa oao oom moo omw omoI wmo oooI woo ooo ooo mool owo moa maa mon woo wom moo moo mom omo mom moaI amo mmo mmaI omaI amo ooa mooI ooo omaI aoaI ooo omoI ooa moo aooI aoo ooo mool aoa ooo moo mwm ooa moo aooI aow maa wao mmw moa oooI wmm oom ooo aow oow mam Hem mam Rem mom nee mma- oam mmo mom- Hoe mom Hue owe ama moo w.a w.w ooo m.a o.o omo o.a o.o aoo w.a w.w moo o.a o.o amoI m.a w.o oma o.a w.w ooa o.a o.o mam o.a w.o moaI m.a o.o mmoI o.a m.w ooa o.a o.w ooo mo. m.w oao o.a m.w aoaI o.a m.w oww m.a w.o oow o.a m.w I mo. o.w no x GOHNMQ‘LDkDI‘m r-lr-iI-II-iI—lr-ir—lI-lu-l aoowmomo mess—mom BZHOmIm>Hm UZHmD mUZHBg man—4mm m.a:mHnH ~440ng 20mm mBZmHUHmmmOU ZOHBfimmmOU oma 160 a: mac oma ooo omo woo omo omw mww omo oao omo omo ooo omaI aooI ooo woa ooa oma oaol moo moo oaol maa aoo aoa oaa wom ooa wma amo ooo owa omw woa maoI omo oma omw wmo wao omo ooo omo ooo oma owo mooI woal aoo wwa oao ooo owaI owa oooI moo moo ooaI moo owo maw woo oaa wao ooo moa oao aaa woo woo aooI moo mwo amoI oool oma aoa mooI owa mooI ama omm wmaI ooaI owa aoaI woo amm oaal mmoI woo ooa mwaI woaI oma woo ooo amo mwa moa ooo moa aon oooI moa ooa omw omw oooI omo woo mow wmaI oooI wma ooo ooo ooaI oooI ooa moa aooI oooI ooa oma ooo woo ooo mmo omo mom oom amo oaal m.a a.o aaoI w.o w.m ooo m.a o.o omo w.a m.o wwa o.a a.o owo a.a o.o moaI o.o o.m owa o.o o.m ooo m.o o.m moo o.a a.o omoI a.a m.o oooI o.a m.o mom m.a o.o mmw o.o o.w amw o.o m.o oao w.a w.o ooo m.o o.m I o.o a.o no x moaoowmomo I-lr-II-IFII—lt-lI-lv-lr-l r-INMQ'IDWI‘CD mfigom BZHOQIZm>mm OZHmD mOZHB/mm w.m—go mEhHm EON—“mg 20mm mBZMHUHhhmOU ZOHBfimmmOU omo 161 omo mow mao ooo wow woo aom mom mmw ama omw mwm mom mom mmo mooI woa mma moa oooI oaoI ooo ooo oooI oooI ooo aoo mmoI owo ooo oooI oooI ooo moo moo oao mooI ooal maoI oaal oon omo omo omo ooo mom omw omw omo amw ooaI moo ooa moo ooa omo oow aow moo omo woa wmo ooo oooI ooa mooI mma moo mow ooo aoo amal aoo owa wmoI woo ama mooI wwo ooo mmo mwo wwo moo mma oooI mooI omo owo owo aooI mma maoI oma omo omo awo omoI oma mmoI maa owo oooI omo wwo mow aow oom omm oom moo aoo owol wmoI omo oaa omw aom mao oooI oaw mow moo omw mwo oooI oooI mooI mooI omw ooo mmo wmo moo mww moo om. o.o moo oo. w.o mmm oo. m.o moo mm. o.o ooo mm. o.o mmw om. o.o aoo mm. o.o ooa o.a a.o woaI wm. o.a oooI om. o.o omo wm. o.o aoo wm. o.o omo om. o.o mmo om. m.o ooo mm. o.a mmo om. o.o wmo wo. m.o I ao. m.o cm x O‘CHNMQ‘MOI‘Q r-lv-le-lI-lI-lI-Ir-lr-i HNMQ‘IG‘DI‘N maflmu EBXHm Z<0Hmm< 20mm mBZmHUHmmmOU ZOHBflammmOU mBflszm BZHOmImmmmB wZHmD mUZHedm oom 162 mow omo mmm mwm I mmw moo mmm oom mmm mow omm mao wmw mmm omaI mon omo aaaI oooI wmo omoI woo amoI wowl omoI omoI ooo oma wwo oom oma aoo moo mmw mmaI mooI momI mwaI moml ome oooI omo aoo aoo mmm maw ama ooo oom oow ooo mmaI mao moo ama wwo ooo wmo ooo omo oma aaoI moa mooI omo moo mma oooI mmo aoo mooI moo ooa mao mon moa woo moo oom moa moo omw mma oaoI owo mmoI ooaI moo omo omw mma mmo won ooa ooa oooI mowI ooo aowI mao owol moo moo amo oom oam moo mwm mam oom wmal mooI ooa oaoI mwo omo mooI moo ooo mmw wom ooo maw owm awo oooI omoI maoI mmaI moa oom mom ooo omo mmo oww ooo moo ooo mmo amo omo aao ooa mon aao mow mom omw wwo moa aoo o.a o.a o.o o.a w.a o.o m.o m.o o.o w.a w.o o.o o.o m.o o.a o.a a.o oo.a Gm aoowmomom medszh BZHOmIm>Hm UZHmD mUZHBmm wZHmD mGZHBAam oow 164 woo oow aoo mmo mow wmo aoo woo oao ooo oow ooo oam mmo woo ooo mma woo owa woo maoI oooI wool moal ooaI woaI oao aoo woa mww owa mmo omo wow wmo ooo moo omo owo aww aoa mow mmo ooo mmoI woo moo woa moo omo ooo omo ooa woo mmo moa ooo omo omo owo wmm ooo mow mma mma aoo moo ooa mmo omo maa omo wwo mwm woo owo mom omo omo oma mow oma wmo wooI oool mao wmo moa oma oao mooI ooo ooal oooI moa mwo ooo mao oaa aoo owo omo oao ooo ooo woo aoaI owa ooa maoI oao omo aoo moal mma moa ooo moo ooo ooo moo ooo owo mwa oma oma amoI woa ooo ooa ooo ooa amo moo omw SH 28 3° 48 one So a: 1.4 m8 SH SH SN 2:. mac awe 2: m: ao. ao. oo. am. om. om. oo. om. oo. oo. oo. mo. om. mm. mo. wo. om. wo. Gm w.o w.o o.o o.o o\~ m.o w.o m.o o.o a.o w.o m.o o.o w.o o.o o.o m.o o.o O‘OHNMQ'IOOI‘QD r-lF-ir-II-lr-Ir-IHr-Ir-l I-INMQ'IDSDFQ maflmw mBZW>Mm ZHm QZHmD mOZHfiflm madmo mBZm>mm Zmm UZHmD mOZHfig whammy mfizmgm ZfiUHmnfim 20mm mBZmHUHmmm—OU ZOHBgmmMOU BIBLIOGRAPHY BIBLIOGRAPHY Aiken, E.G. Alternate Forms of Semantic Differential for Measurement of Changes in Self-Description. Psychological Reports, 1965, 16, 177-178. Allen, M.J. and Yen, W.M. Introduction to Measurement Theory. Brooks/Cole Publishing Co., Monterey, California, 1979. Anatasi, A. Psychological_1esting, 4th edition. New York: MacMillan Publishing C01, 1976. Askov, E.N. Development of an Instrument for Measuring Attitudes Toward Reading in Primary Pupils. 1969, ERIC 027164. . Assessment of Teachers' Attitudes Toward an Individualized Approach to Reading Instruction. Wisconsin University, Madison, 1971 (ERIC 048349). Anttonen, R. et al. A Three Year Study of Children's Achievement and Attitude. Millersville State College, Pennsylvania, Division of Education, 1977, ERIC 154031. Beitner, M.S. Word Meaning and Sexual Identification in Paranoid Schizophrenia and Anxiety Neurotics. Journal of Abnormal and Social Psychology, 1961, 63, 289-293. Bennie, F. Pupil Attitudes Toward Individually Prescribed Lab Programs. Journal of Reading, 1973, 17, 108—112. Borgatta, E.F. The Structure of Personality Characteris- tics. Behavioral Science, 1964, 8-17. Carrol, J.B. Review of the Measurement of Meaning. Language, 1959, 35, 58—77. . Some Cautionary Notes on the Semantic Differen- tial. Paper presented at the meeting of American Psychological Association, Cincinnati, September 1959 (b). 167 168 Cheung, T. et a1. Referent Differences in Pupil Motiva- tions. Educational and Psychological Measurement, Vol. 40, N2, 531-536, Summer 1976. DiVesta, J.F. A Developmental Study of the Semantic Structures of Children. Semantic Differential Technique, Eds. Snider, J.C., and Osgood, C.E., Aldine Publishing Co., 1969. DiVesta, F.J.A. and Dick, W. The Test-Retest Reliability of Children's Ratings on the Semantic Differen- tial. Educational and ngcholoqical Measurement, 1966, 26, 605-616. Dixon, T.R. and Dixon, J.F. The Impression Value of Verbs. Journal of Verbal Learning and Verbal Behavior, Donlon, T.F. Testing in the Affective Domain. Eric Clearinghouse, Princeton, N.J., 1974 (ERIC 103493). Donnahoe, J.W. Changes in Meaning as a Function of Age. Journal of General Psychology, 99:23-28, 1961. Dyer, C.O. Construct Validity of Self-Concept by a Multitrait-Multimethod Analysis. Ann Arbor, Michigan, 1964. Elsworth, G.R. and Coulter, F. Aspiration and Achievement: The Measurement of Self-Perception in Students. Occasional Paper No. 7, Australian Council of Educational Research, Melbourne, Australia, 1977. Epstein, I. Measuring Attitudes Towards Reading. Eric Clearinghouse, N.J., 1980 (ERIC 196938). Estes, T.H. A Scale to Measure Attitudes Toward Reading. Journal of Readin , 1971, 15, 135-138. Estill, J.A. and Filler, R.C. Measurement of Reading Attitudes. Elementary English, 1973, 17, 108-112. Farr, S.D. and Kubiniec, C.M. Stable and Dynamic Compo- nents of Self-Concept. Multivazite Behavioral Research, 1972, 147-163. Glennon, R.C. An Attitude Scale for Reading. The Reading Teacher, 1972, 25, 442-447. 169 Goldfried, M.R. and Kissel, S. Age as a Variable in Connotative Perceptions of Some Animal Symbols. Journal of Projective Techniques and Personality Assessment, 1963, 27, 171-180. Green, D.H. Attitudes: In Motivation in Education, ed. Ball, 3., New York: Academic Press, 1977, 111-129. Griggs, A.E. Validity Study of the Semantic Differential Technique. Journal of Clinical Psychology, 1959, 15, 179-181. Groff, P.J. Children's Attitudes Toward Reading. Journal of Educational Research, 1962, 55, 313-318. Gronlund, N.E. Measurement and Evaluation in Teaching, New York, MacMillan Publishing Co., 1977. Hallworth, H.J. A Factorial Study of Value Judgments Among Adolescent Girls. British Journal of Statistical P§ychology, 16, 1963. Harman, H.H. Modern Factor Analysis, 3rd ed. Chicago, University of Chicago Press, 1976. Harrill, L.M. Evaluation of Several Methods for Measuring Young Children‘s Attitudes. South Dakota UniVer- sity, Vermillion, 1971 (ERIC 056059) Harris, C.W., ed. Problems in Measuring Change. Madison: University of Wisconsin Press, 1963. Hecht, A.R. and Bonnie, H. Development of a Semantic Differential Instrument for Student Evaluation of Counseling Conferences. Paper presented at the annual meeting of the North Central Region EARA, Madison, Wisconsin, July 1976. Heise, D.R. Semantic Differential Profiles for 1,000 Most Frequent English Words. Psychological Monographs, 1965, 79(8) No. 601. . Some Methodological Issues in Semantic Differ- ential Research. Psychological Bulletin, 1969, 72(6), 406-422. Hudson, W. Dimensionality and Levels of Education. Cross- Cultural Study of Behavior, Eds. Al-Issa and Dennis, W., Holt, Rinehart and Winston, N.Y., 1970. 170 Jakobovits, L.A. Comparative Psycholinguistics in the Study of Cultures. International Journal of Psychology, 1966, 1, 15-37. Jantz, R. K. Children' s Attitudes Tow rd t Maryland University College Park, 1976 (ERIC 181081). Kahneman, D. The Semantic Differential and the Structure of Inferences Among Attributes. American Journal of Psychology, 1963, 76, 554-567. Kamper, R.E. An Investigation of 7th Grade Students' Attitudes Toward Reading7as Measured bypthe Semantic Differential. Doctoral Dissertation, University of Pittsburgh, Dissertation Abstracts, 1970. Kane, R.B. Reducing Proximity Error in Administering the Semantic Differential. Purdue University, Lafayette, Indiana, 1968 (ERIC 032761). Kerlinger, F.N. Foundations of Behavioral Research, lst ed. New York, Holt, 1964, Chapter 32. Kim, J.O. and Mueller, C.W. Introduction to Factor Analysis. Sage Publishing, Beverly Hills, 1982. Komorita, 8.8. and Bass, A.R. Attitude Differentiation and Evaluative Scales of Semantic Differential. Journal of Personality and Social Psychology, 1967, 6, 241-244. Kubiniec, C.M. The Relative Efficiency of Various Dimen- sions of Self-Concept in Predicting Academic Achievement. American Educationa; Research Journal, 1970, 7, 321-332. Kumata, H. and Schramm, W. A Pilot Study of Cross- Cultural Meaning. Semantic_Qifferential, Eds. Snider, J.G. and Osgood, C.E., Aldine Publishing Co., Chicago, 1969. Lao, C.R. Levenson's IPC Scale (A Comparison of Chinese and American Students. Journal of Cross-Cultural PS ChOlO ' V01. 9' NO. 1’ MarCh 1978' 113-1240 Lemon, N. Attitudes and Their Measurement. New York: Halstead Press, John Wiley and Sons, Inc. 1973. Lowery, L.F. and Grafft, W. Paperback Books and Reading Attitudes. The Reading Teacher, 1968, 21, 618-623. 171 Luria, Z.A. A Semantic Analysis of a Normal and Neurotic Therapy Group. Journal of Abnermel end Seeiel Psychology, 1959, 58, 216-220. Maclay, H. and Ware, E.F. Cross-Cultural Use of Semantic Differential. Cross-Cultural Study of Behavior, Eds. Al-Issa and Dennis, W., Holt, Rinehart and Winston, N.Y., 1970. Magnusson, D. Test Theory. Addison-Wesley Publishing Co., Reading, Massachusetts, 1966. Maltz, H.E. Ontogenetic Change in the Meaning of Concepts as Measured by the Semantic Differential. Seman- tic Differential Technique. Ed. Snider, J.B., Aldine Publishing Co., Chicago, 1969. Marks, I.M. Patterns of Meaninggin Psychiattic Patients. Oxford University Press, New York, 1965. Maslon, P. and Menifield, P. Semanticsgpace Coordinates in School Situations. Paper presented at annual meeting of Northeastern Educational Research Association, Ellenville, New York, November 1973 (ERIC 092580). Mehrens, W.A. and Lehmann, I.J. Measurement and Evaluation in Educational Psychology. New York: Holt, Rinehart and Winston, 1978. Messick, S.J. Metric Properties of the Semantic Differen- tial. Educational and Psychological Measurement, 1957, 17, 200-206. Miller, G.A. The Magical Number Seven, Plus or Minus Two. Psychological Review, 1956, 63, 81-97. Miron, M.S. A Cross-Linguistic Investigation of Phonetic Symbolism. Journal of Abnggmal and Social Psychology, 1961, 62, 623-630. Mogar, R.E. Three Versions of the F Scale and Performance on the Semantic Differential. Journal of Abnormal and Social Psychology, 1960, 60, 262-265. Neuringer, C. Effect of Intellectual Level and Neuro- Psychiatric Status on the Diversity of Semantic Differential Ratings. Journal of Consulting_ Psychology, 1963, 27, 280. Nickols, S.A. and Shaw, M.E. Saliency and Two Measures of Attitude. Psychological Reports, 1964, 14, 273-274. 172 Nie, N.H. et a1. Statistical Package for the Social Sciences, 2nd ed. McGraw Hill, New York, 1975. Nisbett, R.E. and Gordon, A. Self-Esteem and Susceptibil- ity to Social Influence. Journal of Personality and Social Psychgtggy, 1967, 5(3), 264-276. Norman, W.T. Stability--Characteristics of the Semantic Differential. American Journal of Psyghelegy, 1959, 72, 581-584. . Toward an Adequate Taxonomy of Personality Attributes. Replicated Factor Structure in Peer Nomination Personality Ratings. Journal of Abnormal and Social Psychology, 1963, 66, 574-583. Nunnally, J.C. Psychometric Theopy. New York, McGraw- Hill, 1967. Nunally, J.C. and Durham, R.L. Validity, Reliability and Special Problems in Evaluation Research in Handbook of Evaluation Research. Vol. 2, Eds. Guttentag, M. and Struening, E. Berverly Hills, California: Sage, 1975, 289-352. Nyberg, V.R. and Clarke, S.C.T. Technical Report on School Subjects Attitude Scales (and) Improvingy and Extending the SEhool Subjects Attitude Scales (and) Manual for Administration, Scoring and Interpretation of the School Sub'ects Attitude SEales. Alberta Department of Education, Edmonton, cams, 1979 (ERIC 179575). Osgood, E., Yasumasa, T., and Oyama, T. A Cross-Cultural and Cross-Concept Study of the Generality of Semantic Space. Semantic Differential Technique, Eds. Snider, J.G. and Osgood, C.E., Aldine Publishing Co., Chicago, 1969. Osgood, C. The Cross-Cultural Generality of Visual-Verbal Synesthetic Tendencies. Cross-Cultural Study of Behavior, Eds. Al-Issa and Dennis, W., Holt, Rinehart and Winston, N.Y., 1970. Osgood, C.E., Suci, G.J. and Tannenbaum, P.H. The Mea— surement of Meaning. Urbana: University of Illinois Press, 1961, 3rd edition. Osgood, C.E. Semantic Differential Techniques in the Comparative Study of Cultures. American Anthro- pologist, 1964, 66(3, Pt. 2), 171-200. 173 Peabody, D. Two Components in Bipolar Scales: Direction and Extremeness. Psychological Review, 1962, 69, 65-73. Perrin, L.A. and Lilly, R.S. Social Desirability and Self- Ideal Self Ratings on the Semantic Differential. Educational and Psychological Measurement, 1967, 27, 845-853. Rammers, H.H. Rating Methods in Research on Teaching. In N.L. Gage (ed) Handbook of Research on Teaching, Chicago, Rand McNally, 1963. Ranson, J.T. et al. Changes in Students' Attitudes as Measured by a SD Instrument. National Institute of Education, Washington, D.C., 1973 (ERIC 110645). Ransbury, M.K. Assessment of Reading Attitudes. Journal of Reading, 1973, 17, 25-28. Reed, K. Reading Instruction: The Affective Domain “Measuring Affective Growth in Reading." Paper presented at the annual meeting of the Interna- tional Reading Association World Congress on Reading, Hamburg, West Germany, August 3, 1978 (ERIC ED 160990). Scott, W.A. Attitude Measurement. In The Handbook of Social Psychology, 2nd ed., Vol. 2, Eds. Lindzey, G. and Aronson, E., Reading, Massachusetts, Addison-Wesley, 1968, 204-273. Snider, J.G. and Osgood, C.E., eds. Semantic Differential Technique: A Sourcebook. Chicago, Aldine, 1969. Soares, L.M. and Soares, A.J. Convergent and Discriminant Validation of Self Perception. Paper presented at American Educational Research Association, (L.A. 1981) (ERIC 201646). Solly, C.N. and Stagner, R. Effects of Magnitude of Temporal Barriers, Type of Goal and Perception of Self. Journal of Experimental Psychology, 1956, 51, 62-70. Stagner, R. Impact of War on a Nationalistic Frame of Reference. Journal of Social Psychology, 24: 187-215, 1946. Stricker, G. and Zax, M. Intelligence and Semantic Differ- ential Discriminability. Psychological Reports, 1966, 18, 775-778. 174 Suci, G.J. A Comparison of Semantic Structures in American Southwest Culture Groups. Journal of Abnormal and Social Psychology, 1960, 61, 25-30. Summers, G.F., ed. Attitude Measurement, Chicago: Rand McNally, 1970. Summers, E.G. A Review of Research and Bibliography on Published Attitudinal Instruments in Reading. Paper presented at the annual meeting of Interna- tional Reading Association, Anaheim, California, May 1976, 150 pages (ERIC 123558). Torgerson, W.S. Methods of Scaling. John Wilen and Sons, Inc., New York, 1958. Wells, W.D. and Smith, G. Four Semantic Rating Scales Compared. Journal of A lied Ps cholo , 1960, 44, 393-397. Wiggins, N. and Fishbein, M. Dimensions of Semantic Space: A Problem of Individual Differences. In J.G. Snider and C.E. Osgood (eds.), Semantic Differen- tial Technique: A Souggebook, Chicago, Aldine, 1969. Yonker, Robert J. et al. A Methodological Investigation of the Development of SD to Assess Self-Concept. Paper presented to the National Council of Mea- surement in Education Annual Meeting, Chicago, Illinois, April 1974 (ERIC 090264). Zax, M., Gardiner, D.H. and Lowry, D.G. Extreme Response Tendency as a Function of Emotional Adjustment. Journal of Abnormal and Social Psychology, 1964, 654-657.