‘..“-‘i.'$l'5‘:. f}?'~'.'}?."!?n‘.,'f‘r.l. ', ‘ DEVELOPMENT OF'AN INSTRUMENT} .. .. --“;- ....... T0 MEASURE EARLY ELEMENTARY-SCHOOL -‘ CHILDREN’S ATTITUDES TO‘WARDS SCHOOL } ‘ Thesis for the Degree of Pb. D. 7 MICHIGAN STATE UNIVERSITY ' CAROLE ANN BEERE 1970 mm W\\\\\ \\\\\Dt\ll\\\§}\;§\\§\\§l 73‘ Mach" 3 1293 10197 Mlchlgan Sta“? 1 Umvcrszty' f is" i This is to certify that the thesis entitled DEE'ELOPYEA'I OF AN INSTRUMEXI TO }£-\SL"RE EARLY ELB’EX‘IARY SCHOOL CEILL‘TEX'S ATIIIYDES IOR‘ARDS SCHOOL presented bg Carole Ann Beere has been accepted towards fulfillment of the requirements for . Ph.D. d Education __ egreein____ -\ Major professor r" ‘ . / fl, {'1 ' C —. C . ' n L314. LA; L m}; ( 0-169 M,‘ j” 7“) r2 0%) sh? 19> It“ J» ”‘1‘ Pa. : due .I “HIM “a I :0“ .' y «Isl-fw- . . ”a an. M ‘ war 3m A. t . . ' £4.23. :3- ‘ .. e . ‘ ‘l ‘ J ‘ ‘ A y: o V , -P‘. use-qt: :33} > Om" .‘au , ._ .u'v ‘fty-“W” W . I'm“; .{flru‘inu :rav ’0‘?- W? ‘9 kw”, 2m! “-“w’ ABSTRACT DEVELOPMENT OF AN INSTRUMENT T0 MEASURE EARLY ELEMENTARY SCHOOL CHILDREN'S ATTITUDES TOWARDS SCHOOL By Carole Ann Beere Objective Procedures have been developed in the areas of intelligence and achievement testing which allow for group administered tests to be used with early elementary children. This, however, has not been the case in the area of attitude measurement. This research project was designed to fill that gap by developing an attitude assessment instrument appro- priate for use with groups of early elementary children. Specifically, the objective of this study was to develop an attitude instrument to measure second grade students' attitudes towards school. Procedures The specific attitude assessment procedure with which this study is concerned involves oral and pictorial presentation of items with pictorial presentation of response choices. The instrument is similar to a Likert scale in that there are five response choices ranging from very positive to very negative. However, this instrument, being especially designed for young children, presents oral items and pictorial response choices. Five faces ranging in facial expression from very happy to very sad serve as the response choices. ”a“ . t9 “52‘; C. Beere In order to develop an instrument, a group of forty items, designed to measure attitude towards school, was administered to the second graders in thirteen schools in three districts in Southern Michigan. One hundred forty-five of the students who completed the instru- ments were randomly selected for the instrument development sample, and two types of additional data were gathered about this subset of children. First, each child was individually interviewed by a member of the research team to ascertain the child's attitude towards school. Second, each child in the instrument development sample was rated by his teacher regarding the child's attitude towards school. The information provided by the interview and teacher rating was gathered to help select the most "valid" subset of items from the forty items initially included in the instrument. Results Item selection procedures were actually conducted twice. The proce- dures followed in the two sets of analyses were identical. However, the results of the first analyses, which were based on seventy-three of the 145 subjects in the instrument development sample, were not stable when cross-validated on data from the remaining seventy-two subjects. There— fore, the analyses were repeated combining the data from all 145 subjects. This reduced the influence of random error in the correlations used for item selection, and thus increased the likelihood that the magnitude of the item selection correlations would be maintained on subsequent cross— validations. The first step in the analyses was to determine whether a single set of items could be selected which would correlate well with the inter— view, the teacher rating, and the total test score. Low correlations between teacher rating score and each of the other two measures suggested C . Beers sets of items should be selected: one set to correlate well teacher rating and a second set to correlate well with interview ' Aftotal test score. After the two sets were selected, Hoyt's analysis of variance W was used to compute the reliability of each. The twenty items \ t“ 7. “(’11 which correlated well with teacher rating had a reliability of 8,0 ' {'- \\ .th'e twenty items which correlated well with interview and total test 3'.\--" score had a reliability of .87. ’ f. u- \ Fran the results of the analyses, the researcher concludes that _ ‘.‘ 3 . A... 'l. 'Mtitude towards school which functions well with early elementary . :‘lu ‘},®l.1dren. Stated another way, drawings of human faces provide a viable this project developed a group administrable instrument for measuring - fiestas for presenting the response choices for an attitude scale. \ 2C- 0‘ t .. vr, 431‘. ' o .2 ,‘~t DEVELOPMENT OF AN INSTRUMENT T0 MEASURE EARLY WARY SCHOOL CHILDREN'S ATTITUDES TOWARDS SCHOOL BY Carole Ann Beers A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOROF PHILOSOPHY ' VIN—"t of Counseling, Personnel Services and Educational Psychology 1970 - " © Copyright 'by _ . Carole Ann Beers 1971 ACKNOWLEDGMENTS The writer is grateful to the many persons who contributed to the successful completion of this study. My chairman, Dr. Robert C. Craig, and my doctoral committee members, Dr. Randall Harrison, Dr. Mary Ellen McSweeney, and Dr. William Mehrens provided welcome guidance and support. Appreciation is extended to Dr. C. Philip Kearney and Dr. Thomas P. Wilbur whose patience and support made this project possible. Special thanks are due to John Schweitzer for his unending patience in providing help in using the computer; to Allen Ahola for creating the sketches for the pictorial items; and to Geraldine Coleman for providing general assistance whenever problems arose. The writer is particularly indebted to two people. First, Vicki Bowzer did a superb job of typing this dissertation, and maintained a cheerful attitude throughout the numerous revisions of the paper. Second, my husband, Donald, contributed much direct assistance with this project, and most of all, provided love and understanding and made many personal sacrifices —- without which this project would never have become a reality. iii ‘ rs..- trsrorTABLEs. . . . . . . . LIST 0? FIGURES . O O O I O ' O I O I C PROBLEM STATEMENT. . . . . . Rationale . . . . . . Hypothesis. . . . . . . RELATED LITERATURE . . . . . Attitude Theory . . 71.17, iv Importance of Attitudes . . Importance of Attitude Assessment . Inadequacy of Present Procedures. TABLE OF CONTENTS Attitude Objects. . . . . Attitudes and Behavior. Attitudes and Experience. Components of Attitudes . . . Characteristics of Attitudes. Definitions of Attitudes. . . Emphasizing Response in General . . Emphasizing Overt Action Tendency . Emphasizing Affective Responses . . Emphasizing Internal Process. . . . Attitudes Compared to Other Summary . . . . . . . . . . . . Constructs. Attitude Measurement Procedures Disguised - Structured Techniques . Disguised - Non-structured Techniques Non-disguised - Non-structured Techniques Non-disguised - Structured Techniques . . “Relacd st“ 188 O a c O o o I o a o s o s o . y -:T‘: PM 0 . I O O O I D O O I O O C I O I O Page ix Page O‘UINH 9 12 .~'l !’ mums (Continued) DESIGN. Overview . . . . . . . . . . . . . Sample . . . . . . . . . . . . . . Instrument . . . . ... . . . . . Instrument Design. . . . . . . Instrument Administration. Instrument Scoring . . . . . . . a o 0 Validation Measures. . . . . . . . Interviews . . . . . . . . . . . Interview Scoring. . . . Teacher Rating . . . . . . . . . Summary. . . . . . . . . . . . . . IV RESULTS . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . Preliminary Analysis . . . . . . . Sample Items . . . . . . . . . Interviewer Reactions. Item Selection . . . . . . . . . . First Analysis . . . . . . . . . First Analysis - Discussion. . . Second Analysis. . . . . . . . . Second Analysis — Discussion . . Reliability. . . . . . . . . . . Supplementary Analysis . . . . . Comparisons Between Students Interviewed During First Week and Students Interviewed During Third Week . . . . . . . . . Discussion of Week 1 — Week 3 Comparisons . . . . r Analysis of Data From Students Not Interviewed. . . . . . . . . Norms. C O l I I I O I I I I ' O O smry o o o o o I o s I s o o o o Page 89 Page SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . 104 Summary. . . . . . . . . . . . . . . . . . . . 104 Purpose. . . . . . . . . . . . . . . . . 104 Review of the Literature . . . . . . . . . . 104 Design . . . . . . . . . . . . . . . . . . . 106 Results I I I I I I I I I I I I I I I I I I I 107 ' Conclusions. . . . . . . . . . . . . . . 109 Advantages and Limitations of . the General Procedure. . . . . . . . . . . . 109 ' Use of the Instruments . . . . . . . . . . 110 Implications for Future Research . . . . . . 111 Overall Conclusion . . . . . . . . . . . . . 112 xiflBLIOGRAPEY. . . .. . . . . . . . . . . . . . . . . . . 114 a List of Oral Items . . . . . . . . . . . . . . . 119 : Instrument . . . . . . . . . . . . . . . . . . . 121 : Children's Study Manual. . . . . . . . . . . . . 138 : Sentence Completion Stems. . . . . . . . . . . . 154 , Direct Questions I I o I o s I o s a o o o o o I 156 I.‘ : Descriptions of Children . . . . . . . . . . . . 158 ‘fif.. _ vi TABLE 10 11 12 LIST OF TABLES Sample Size . . . . . . . . . . . . . . . Correlations Between Interviewer, Tape Scorer 1, Tape Scorer 2, and Tape Scorer 3. . . . . . . . Correlations Between Interview Score, Teacher Rating Score, and Total Test Score. . . . . . . Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score: Item Selection Group. . . . . . . Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score: Cross-Validation Group. . . . . . . . . . Correlations Between Each Item and the Teacher Rating Score: Item Selection Group . . . . . . Correlations Between Each of the Twenty Items Selected to Correlate Highly with Teacher Rating Score: Cross—Validation Group . . . . . . Correlations Between Interviewer, Tape Scorer 1, Tape Scorer 2, and Tape Scorer 3. . . . . . . . Correlations Between Interview Score, Teacher Rating Score, and Total Test Score. . . . . . . . Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score: All Subjects. . . . . . . . . . . . . . Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score Using Twenty Items Which Correlated Highly with Interview Score and Total Test Score . . . . . . . . . . . . . . . . . . . . . . Correlations Between Each Item and the Teacher Rating Score: All Subjects . . . . . . . . . . . Page 49 66 67 69 72 74 75 80 82 84 86 LIST OF TABLES (Continued) TABLE Page 13 Correlations for Those Interviewed During . Week One and Those Interviewed During ? Week Three. . . . . . . . . . . . . . . . . . . . 90 5 14 Total Score Means and Standard Deviations for Week One Interviewees and Week Three Interviewees: Forty Items. . . . . . . . . . . . . 92 15 Interview Score Means and Standard Deviations for Week One Interviewees and Week Three Interviewees. . . . . . . . . . . . . . . . . . . . 92 . 16 Means and Standard Deviations For Each of Four Groups: Based on Twenty Items Selected for High Correlation With Interview Score and Total Test Score. . . . . . . . . . . . . . . 95 17 2 x 2 Analysis of Variance Comparing ' Means From Table 16 . . . . . . . . . . . . . . . . 95 I 18 Means and Standard Deviations For Each of l Four Groups: Based on Twenty Items Selected ‘ For High Correlation With Teacher Rating Score. . . 96 i 19 2 x 2 Analysis of Variance Comparing 'MeansFromTable18....... 97 20 Norms for Twenty Items Selected For Their High Correlation With Interview Score and Total Test Score. . . . . . . . . . . . . 100 21 Norms for Twenty Items Selected For Their High Correlation With Teacher Rating Score. . . . . . . . . . . . . . . . . . . . 101 E viii LIST OF FIGURES Classifications of Attitude Assessment Procedures . . . . . . . . . . . . . . . Graph of Correlations Between Each Item and Interview Score and Each Item and Total Test Score (First Analysis). . . . . . . . . . . . . . . . . . Graph of Correlations Between Each Item and Interview Score and Each Item and Total Test Score (Second Analysis) . . . . . . . . . . . . . . . . . Page 28 70 83 rF—_____ CHAPTER I PROBLEM STATEMENT Procedures have been developed in the areas of intelligence and achievement testing which allow for group administered tests to be used with early elementary children. This, however, has not been the case in the area of attitude measurement. The procedures used in this study were designed to fill this gap by developing an attitude assessment instrument appropriate for use with groups of early elementary school children. Rationale The rationale for this project is based on three assumptions: (1) attitudes are important; hence (2) attitude assessment is important; but (3) present methods are inadequate for attitude assessment, espe— cially assessment of groups of students in early elementary grades. The hypothesis is that the technique used in this study can yield a valid and reliable group administrable instrument for assessing the attitudes of students in early elementary grades. The three assumptions will be discussed first; following this, the reasons for expecting the hypothesis to be supported will be given. Importance of Attitudes The following diSCussion is intended to show that attitudes are important as: (l) a central concept in social psychology; (2) a major influence on behavior; and (3) an effect of school as well as a partial cause of students' performance level in school. The notion that attitudes are a central concept in social psychology is certainly not new. In 1935, Allport wrote of their importance, saying: The concept of attitude is probably the most distinc- tive and indispensable concept in contemporary American social psychology. No other term appears more frequently in experimental and theoretical literature . . . . This useful . . . concept has been so widely adopted that it has virtually estab— lished itself as the keystone in the edifice of American social psychology [p. 798]. Apparently, the importance of attitudes in social psychology did not abate during the two decades after Allport wrote those words; he repeated them in his 1954 article in Lindzey's Handbook of Social Psychology. Campbell, and Diab shared Allport's belief that the concept, "attitude," is central to social psychology. In 1963, Campbell wrote: Social attitude provides a traditional theme in social psychology at the levels of both theory and empirical research . . . . In several of the decades of quantitative social psychology's brief history, research on social attitudes has been the dominating concern [p. 95]. In 1967, Diab wrote: "The concept of attitude, more than any other concept in the field of social psychology, has been widely used in attempts to explain and predict behavior [p. 140]." The magnitude of some of the bibliographies accompanying articles on attitudes also reveals the place of attitudes in the field of social PSYChology. In 1940, Day published an article in which he reviewed the research on attitudes conducted between 1937 and 1939. The bibliography 3 for the two year period included 200 entries. The 1954 edition of the Handbook of Social Psychology included 136 references in the bibliography accompanying the article discussing attitudes (Green, 1954); and its 1968 counterpart included 153 references (Scott, 1968). The emphasis which social psychology places on attitudes reflects the view that much of human behavior is greatly affected by attitudes. Remmers (1954) expressed this quite succinctly by saying that "attitudes are theoretically a component of all behavior, overt or covert [p. 3]." Evans (1965) elaborated on this by giving examples of how attitudes influence behavior: It would be difficult to overstress the influence of attitudes and interests in the lives of individ— ual people. They determine what a man will do or say in particular situations, what he will enjoy or dislike, his approach to other people, and his reac- tions to events in his own life and in the world around. The expression of attitudes, either in actions or in words, provides clues to personality and to needs, and makes possible the kind of under- standing which is necessary for the formation of stable relationships [p. 1]. Since this research is particularly concerned with the attitudes of school age children, consideration should be given here to the importance of the relationships between school, children, and attitudes. School influences children's attitudes, and likewise, children's attitudes influence children's performance in school. Cohen (1965) was quite explicit in discussing the important contribution of the school in shap- ing children's attitudes: There is increasing recognition among profes— sional educators and psychologists that children learn much more in school than is included in the formal curriculum. In addition to skills, facts, and modes of social behavior, they learn attitudes and values. The child who is learning to multiply may simultaneously be learning that work can be exciting and that it is fun to learn to do new 4 things. Alternatively, he may be discovering that attempts at learning in school usually result in failure and that it is probably better not to make any effort at all than to risk almost certain failure by trying. . . . The attitudes that the child develops can affect his ability and desire to learn and his willingness to expose himself to new experiences. Moreover, the atti- tudes that he develops in the school setting may well determine his learning behaviors long after he has completed his formal schooling [p. 1]. This view is shared by educators, as exemplified by Evans (1965) who said: Modes of thinking and acting, attitudes and interests are . . . acquired and developed during school days, and these may become permanent, remaining effective and observ- able long after the greater part of the subject matter learnt [sic] has been over— lain or foregotten [p. 1]. As stated earlier, in addition to school affecting children's atti- tudes, children's attitudes also affect children's performance in school. According to Mehrens and Lehmann (1969): Research has demonstrated that, at best, only about 50 percent of the variation in grades is accounted for by cognitive or intellectual factors. A large part of the unaccounted variance is, no doubt, due to affective factors, some of which are attitudes [p. 256]. Carter (1969) agreed that attitudes greatly affect school performance. He cited specific research studies before concluding: The practical importance of measurement of attitudes towards school is obvious, since such attitudes have been shown repeatedly to be significant determiners of school achievement [p. 186]. Carter (1969) went somewhat further than Mehrens and Lehmann when he added that "in the determination of achievement, attitudes are no less important than abilities . . . [p. 186]." Given this reciprocal relationship in which School affects attitudes, and attitudes affect school, it is quite understandable why Mehrens and 5 Lehmann (1969) stated that "education must be as concerned with attitudes, values, and interests . . . as it is concerned with the development of cognitive skills and knowledge [p. 236]." It was stated earlier that the rationale for this study is based on three assumptions. Based on the discussion presented above, it is hoped that the reader will now accept the first aSSumption that "attitudes are important." Importance of Attitude Assessment The second assumption upon which the rationale for this study rests is that "attitude assessment is important." To some extent, this assump— tion follows naturally from acceptance of the first assumption: if atti— tudes are as important as social psychologists indicate, then attitude assessment must also be important. For example, given that attitudes strongly influence behavior, then studies of behavioral change are likely to include an examination of attitude change. Attitude measurement procedures are necessary if one wants to answer questions such as the following: (1) What is the role of experience in the formation of atti- tudes? (2) How does one proceed to deliberately alter another's attitude? (3) What is the relationship between personality structure and attitude change? In other words, any study of attitude, attitude formation or attitude change requires a procedure for the meaSurement of attitudes. Krech, Crutchfield and Ballachey (1962) called attention to this need for attitude meaSurement techniques saying: "To use the concept of attitude in understanding and predicting action, we need reliable and valid measures [p. 147]." Earlier it was shown that attitudes are important in young children 88 well as in adults; therefore, it logically follows that attitude 6 assessment is also important for young children as well as for adults. Cohen (1965) implied this when she wrote: In attempting to measure students' attitudes, it is obviously important to study children in their earliest grades in school. Being first, these grades set the tone for much that is to come and may well foster attitudes in the child that persist throughout the balance of his education [p. 3]. In addition to the reasons for early measurement set forth by Cohen, consideration should also be given to the need for obtaining baseline data which will allow for studies of attitude change during the school years. If one is interested in measuring attitude changes as a result of school in general, or as a result of special programs within the school, it would be necessary to obtain baseline measures on young chil— dren. For example, one might be interested in evaluating a curriculum revision in a school in terms of attitude outcomes. Pretest attitude measures, as well as post test measures, would be helpful. It is quite likely that one or both of these measures would be taken during the early elementary grades. It is hoped that the reader is now convinced of the validity of the assumption that "attitude assessment is important." This has been shown to be especially true at the early elementary grades. Inadequacy of Present Procedures Since "attitude assessment is important," one would expect that extensive research has been done to develop reliable and valid proce- dures to implement assessment. However, most research involving attitude menaurement does not use well developed scales. As Shaw and Wright (1967) Stated: "All too frequently the investigator simply asks a few unstan- dardized (unevaluated) questions and assumes that the attitude has been I 7 reliably and validly measured [p. 559]." Shaw and Wright went on to point out that there are several negative consequences of inadequate instrumentation. These include: (1) research conducted by different investigators is not comparable; (2) the quality of the instruments used remains quite poor; and (3) researchers must spend much time and energy developing a new instrument for each project undertaken. The severity of the consequences of inadequate instrumentation was stressed by Shaw and Wright (1967) when they said: "attitude research has been hindered by the inaccessability of existing attitude scales, resulting in less— than-optimum advances in the scientific analysis of attitudes [p. x]." The methods now used for measuring young children's attitudes are even more inadequate than the methods used for the assessment of adults' attitudes. Professional test developers have been quite successful in developing reliable and valid instruments for measuring the cognitive outcomes of school. However, as Mehrens and Lehmann (1969) pointed out, there has not been "as much Success in measuring noncognitive characteris— tics as . . . in meaSuring achievement or aptitude [p. 204]." Cohen (1965) in an earlier quotation, stressed the importance of studying atti— tudes of young children. However, she went on to deplore the lack of research in the area: Despite the importance of such early measurement, however, relevant research on attitudes toward school among young children is extremely limited. While there has been some research that has related attitudes to achievement at higher grade levels, this work was generally concerned with attitudes towards specific courses as they were related to achievement and did not deal with generalized, broader attitudes abOut the school experience [p. 3]. Cohen (1965) attributed the lack of attitudinal research with children to a lack of instrumentation; she pointed out, however, that the lack of 8 instrumentation is understandable on the basis of the technical problems which exist when traditional methods of attitude assessment are used with young children: Obvious problems arise in attempting attitude measurement in the primary grades; it is hardly surprising, therefore, that little research on young children's attitudes has been conducted. On the technical level, reading skills are extremely limited, and phrasing meaningful questions which the children can understand and to which they can respond is difficult [p. 3]. In addition to the technical problems mentioned by Cohen, there are addi— tional problems connected with any attempt to obtain verbal measures of young children's attitudes. As Yarrow (1960) pointed out: The verbal expression of one's attitudes (and values) calls for introspection on the part of the subject, requiring him to call up within himself a cognitive organization of a given object or class of objects or set of events and to express some preference, inclination, or interpretation with regard to it [p. 656]. It is questionable whether children in the primary grades can carry out the mental steps necessary to give a verbal response. It will depend, to some extent, on the child's intelligence and previous experiences, as well as the nature of the specific task required and the nature of the testing situation. The lack of adequate methods and instruments is even more apparent when one searches for an instrument that can be group administered to early elementary children. VA search of the published literature up to 1970 failed to turn up a single instrument that would satisfy these two criteria: group administrable and appropriate for pre-literates. Two non-published instruments were discovered and will be discussed later. The third assumption upon which the rationale for this study rests 13 that "present methods are inadequate for attitude assessment, especially 9 for use with groups of students in early elementary grades." Based on the above discussion, it is anticipated that the reader will accept the reasonableness of this assumption. Hypothesis Given that the reader accepts the rationale for undertaking this project, attention can now be directed to a discussion of the hypothesis. The hypothesis states that "the technique used in this study can yield a valid and reliable group administrable instrument for assessing the attitudes of students in early elementary grades." Before providing the justification for expecting this hypothesis to be supported, it is neces- sary to provide a brief description of the procedure used in this study. The measurement technique was patterned after those already being used for group administered achievement tests in the early elementary grades. The procedure includes oral presentation of an item and picto- rial presentation of response choices. For the orally presented items, the teacher (or other reader) reads the item to the entire class (or other group); each individual child responds by placing a mark across the picture in his own test booklet which best represents his response. In this case, response choices are drawings of human faces ranging in facial expression from very positive through neutral to very negative. The items are designed to reflect attitudes toward school. In addition to the orally presented items, the instrument also includes pictorially Presented items. With this brief description in mind, how reasonable was the hypothesis before the data collection even began? In order to believe that any tech- n1Que could measure the attitudes of early elementary students, one had to 10 accept that young children have measurable attitudes toward the specific attitude object being studied. Yarrow (1960) indicated that children do have measurable attitudes, when she said: Although language skills of young children are severely limiting for research, the child's ability to communicate feelings through the limited language he possesses is also notable. Without logical and coherent replies to research stimuli, the pre-school child is often well able, with a few choice words, to convey the emotional significance of the stimulus situation. Also, when the content or context of the research is in the familiar world of the child, he may be remarkably perceptive in the variables being studied [p. 658]. Obviously, the school is part of the familiar world of the child. The viewpoint of Yarrow is substantiated by the research efforts of Cohen (1967). In reporting the results of her research study, she con— cluded: "The results reported here suggest that in terms of children's responsiveness . . . it appears feasible to use a projective technique to assess young children's attitudes towards school [p. 306]." Obviously, Cohen could not correctly conclude that any technique was adequate for assessing young children's attitudes towards school unless young children do, in fact, have attitudes towards school. There were at least two reasons to believe that children would be better subjects for the development of this measurement technique than adults would be. First, school age children are accustomed to taking tests. Therefore, the requirement of responding to a battery of questions Was not an unusual experience for them. The second advantage to using School age subjects was their availability. As Krech, Crutchfield, and Ballachey (1962) point cut: "The use of attitude scales is, for rather Obvious reasons, restricted to situations where the individuals being meaBured are readily available to the experimenter . . . [p. 169]." ‘1IIFII-IIIIIIIIIIIIIIIIIIII----------------------------------- T‘ 11 ‘ As was pointed out earlier, dependency on techniques which require reading and writing, limits the researcher to studying only literate subjects. Techniques which substitute oral responses for written responses are impractical on the grounds that they can not be group administered. To avoid these limitations, the technique in this study uses oral and pictorial presentation of the stimuli and pictorial presen— tation of the response choices. As Jahoda, Deutsch, and Cook (1951) stated: "Pictorial methods are particularly useful in the study of children or of those with limited literacy [p. 184]." Horowitz and Murphy (1938), who have conducted attitude research using pictorial techniques, agreed that one large advantage to a pictorial technique is "the possibility it opens of tapping emerging conceptions of the ' world at an age when verbalization is not yet satisfactory [p. 136]." One of the most potent reasons for believing that the technique described in this study would yield a good procedure for attitude measurement is that the technique has already been used in an instru- ment to measure "readiness." For example, the Metropolitan Readiness 22555 follow the format described earlier in which items are presented orally and response choices are presented pictorially. The tests can be group administered at the end of kindergarten or the beginning of first grade. The test manual reports reliabilities for the total test ranging from .90 to .95, depending on the method of estimation and the Particular sample tested; the manual reports predictive validities, Obtained by correlating the score on the Metropolitan Readiness Tests with a sub—test score on the Metropolitan Achievement Tests, ranging from .43 to .68 (Hildreth, Griffiths, & McGauvran, 1969). These corre- 1ations, for validity, and especially for reliability, were quite h... A _ 12 encouraging; they indicated that the test format would have potential for use with groups of early elementary children. It is anticipated that the evidence marshalled here has shown why the researcher originally believed that the hypothesis on which this project is based is a reasonable hypothesis: "the technique used in this study can yield a valid and reliable group administrable instrument for assessing the attitudes of students in early elementary grades." Summary This project tests a procedure and develops instruments for use with groups of early elementary students to assess their attitudes towards school. The rationale for the project is based on three assump— tions: (1) attitudes are important; hence (2) attitude assessment is important; but (3) present methods are inadequate for attitude assess— ment, especially for use with groups of students in early elementary grades. The hypothesis is that the technique used in this study can yield a valid and reliable group administrable instrument for assessing the attitudes of students in early elementary grades. Arguments have been presented above to demonstrate the validity of each assumption and the reasonableness of the hypothesis. CHAPTER II RELATED LITERATURE There is an enormous volume of literature relating to the study of attitudes. A review of all the related literature is both unnecessary and irrelevant in this paper. However, it does seem appropriate to provide an overview of the relevant areas, in the hope that the reader will see how this study relates to the field of attitude measurement as it is today. For organizational convenience, the related areas will be presented in four sections: (1) attitude definition; (2) attitude theory; (3) attitude measurement; and (4) related studies. Definitions of Attitude Specification of the definition of attitude, based on a review of the literature, is not the simple task that one might expect. There is a profusion of published definitions -— some are quite similar, others are distinctly different —- as well as numerous published reviews of different definitions (See Allport, 1935; Campbell, 1950; Green, 1954; and Scott, 1969). Twenty-five differently worded definitions were encountered in preparing this review of the literature. An examination of the assembled definitions indicated that they could be generally divided into four categories; but the categories are pg; considered mutually exclusive. The common element of the vast major- ity of the definitions of attitude is tendency or readiness to respond. 13 ‘F— 14 The major element in the first category of definitions is response, and there is no specification of the type of response. The second and third categories are subsets of this first category: category two emphasizes overt action tendency as the type of response and omits mention of affec- tive responses; inversely, category three emphasizes affective responses and omits mention of overt action tendencies. The fourth category of definitions stresses the presence of an internal process, but makes no mention of response. Although the various definitions of attitude will be discussed in this four category framework, no claim is made that this is the only, nor even the best way, to organize the various definitions. The definitions could easily be separated on a different element of the definition, such as specificity versus generality; i.e., definitions which consider attitudes as a general disposition of the individual could be ' isolated from those definitions which consider attitudes in regard to specific referents. However, the four category system relating to response element serves as the organizational framework here. Emphasizing Response in General In the largest of the four categories, the response element is present, but no particular emphasis is given to any one aspect of response -- neither affective response nor action tendency is stressed over the other. 0f the definitions in this category, Allport's defini— tion (1935) is by far the most frequently quoted: "An attitude is a mental and neural state of readiness, organized through experience, exerting a directive or dynamic influence upon the individual's response to all objects and situations with which it is related [p. 810]." Nelson's definition (1939), though phrased differently than Allport's, says essen— tially the same thing. Nelson defines attitude as: "a felt disposition m A r__'————_——————‘ 15 arising from the integration of experience and innate tendencies which disposition modifies in a general way the responses to psychological . objects [p. 381]." Similarly, Campbell's definition (1950) stresses response to an object but unlike Allport and Nelson, Campbell omits the role of experience: "An individual's social attitude is a syndrome of response consistency with regard to social objects [p. 31]." Rather than just use the term "response," Hartley and Hartley (1952), Cook and Selltiz (1964), Krech, Crutchfield, and Ballachey (1962), and Rokeach (1968) specifically include both affect and action in their definitions of attitude. Hartley and Hartley define attitude as: an abstraction used to refer to an individual's inferred characteristics that account for such consistency in his behavior and expressions as he may manifest, characteristics that are not prescribed by situational imperatives [p. 683]. Cook and Selltiz define attitude as: An underlying disposition which enters, along with other influences, into the determination of a variety of behaviors toward an object or class of objects, including statements of beliefs and feelings about the object and approach— avoidance actions with respect to it [p. 36]. Krech, Crutchfield, and Ballachey define attitude as: "an enduring system of positive or negative evaluations, emotional feelings, and pro or con action tendencies with reSpect to a social object [p. 177]." Their definition is quite similar to the definition quoted above from Campbell. Although Rokeach defines attitude in terms of a set of beliefs, his definition is still similar to those of Hartley and Hartley, Cook and Selltiz, and Krech, Crutchfield, and Ballachey. Rokeach defines attitude as: "a relatively enduring organization of interrelated beliefs that describe, evaluate, and advocate action with respect to an object or _ 16 situation, with each belief having cognitive, affective and behavioral components [p. 132]." It is interesting to note that of the seven defi— nitions cited thus far, only those of Krech, Crutchfield, and Ballachey, and Rokeach include the term "enduring" in the definition. The definitions offered by Remmers (1954), by Cardno (1967), and by Mehrens and Lehmann (1969), though still in this category, place some additional emphasis on the overt behavioral component of response. Remmers defines attitude as "an affectively toned idea or group of ideas predisposing the organism to action with reference to specific attitude objects [p. 3]." Shaw and Wright (1967) phrase Cardno's definition of attitude as "an existing predisposition to respond to social objects which, in interaction with situational and other dispositional variables, guides and directs the overt behavior of the individual [p. 2]." Mehrens and Lehmann provide a similar definition: "Attitudes are predispositions to respond overtly to social objects [p. 258]." Before leaving the first category of definitions, it should be noted that the dictionary definition , provided in English and English' A Comprehensive Dictionary of Psychological and Psychoanalytical Terms (1958), also falls into this category. Their dictionary definition states that attitude is: an enduring learned predisposition to behave in a consistent way toward a given class of objects; a persistent mental and/or neural state of readiness to react to a certain object or class of objects, not as they are but as they are conceived to be [p. 50]. Emphasizing Overt Action Tendency ‘ The second category of definitions emphasizes action tendency while emitting mention of affective responses. Examples of this type of defini- tion are provided by Hovland, Janis, and Kelley (1953), and by Likert A 17 (1932). Hovland, Janis, and Kelley define attitudes as: "implicit responses which are oriented towards approaching or avoiding a given object, person, group, or symbol [p. 7]." Likert considers attitudes to be "dispositions toward overt action [p. 7]." It could be argued that Likert's definition is essentially the same as that given by Mehrens and Lehmann, and that the two definitions should be placed in the same category. Emphasizing Affective Responses The third category is the inverse of the second; in the third cate- gory, the definitions emphasize affective responses and omit mention of action tendencies. Definitions of this type are offered by Rosenberg (1956), and Jahoda, Deutsch, and Cook (1951). Rosenberg defines attitude as simply: "a relatively stable affective response to an object [p. 367]." The definition given by Jahoda, Deutsch, and Cook is quite similar. They define attitude as: "a more or less enduring predisposition to respond affectively toward a specified entity [p. 112]." Emphasizing Internal Process As mentioned earlier, one category of definitions omits altogether the element of response, emphasizing an internal process instead. Exam- ples of these are the definitions given by Ferguson (1939), Fishbein and Raven (1967), Katz and Stotland (1959), Sherif, Sherif, and Nebergall (1965) and Thurstone and Chave (1929). Ferguson defines attitude as: "the acceptance value of a belief [p. 665]." Ferguson actually rejects a behavioral component of attitude. He says: "Whether or not a person acts in accordance with his attitudes is a question quite apart from a definition of measurement of it [p. 665]." Fishbein and Raven's 18 definition is conceptually related to Ferguson's. They define attitude as "the evaluative dimension of a concept —- 'Is it good or bad?‘ [p. 188]." Katz and Stotland's definition, which also falls into this category, is: "tendency or disposition to evaluate an object or the symbol of that object in a certain way [p. 428]." The definition offered by Sherif, Sherif, and Nebergall also ignores a response element of attitude. They write: "Attitudes refer to the stands the individual upholds and cher- ishes about objects, issues, persons, groups or institutions [p. 4]." The definition given by Thurstone and Chave belongs in this category, too. They state that attitudes are: "the sum-total of a man's inclinations and feelings, prejudice or bias, preconceived notions, ideas, fears, threats, and convictions about any specific topic [p. 7]." The definition which was selected as the definition for this project was presented by Shaw and Wright in their 1967 book, Scales For The Measure— ment of Attitudes. There they state that: attitude is best viewed as a set of affective reactions toward the attitude object, derived from the concepts or beliefs that the individual has concerning the object, and predisposing the individual to behave in a certain manner toward the attitude object. Although intimately related to attitude, neither the propositions that the individual accepts about the object (beliefs) nor the action tendencies are a part of the attitude itself [p. 13]. In the classification system presented above, this definition clearly belongs in category three which emphasizes affective responses. How— ever, Shaw and Wright do not omit mention of elements of categories two and four; rather they are explicit that the "attitude itself" does not include the beliefs (category four) nor the action tendencies (category two). A r.“ 19 The selection of Shaw and Wright's definition as Egg definition for this paper was a somewhat arbitrary decision: it is the definition which this writer finds most acceptable. However, no attempt is being made to convince the reader that Shaw and Wright's definition is the only correct definition. To the contrary, the other definitions have been presented to emphasize that there is no single, generally accepted definition of attitude. As Scott (1968) writes: "These determinants [of definition] can be expected to develop and change over the years; hence it is unrealistic to expect a single, final definition of 'attitude' to emerge within the foreseeable future [p. 205]." Attitudes Compared to Other Cppgtructs As well as stating what an attitude is, the literature regarding . attitudes devotes extensive space to stating what an attitude is not, i.e., distinguishing it from other psychological constructs. The most frequent distinctions are made between attitude and each of the follow- ing five terms: opinion, belief, habit, value, and interest. Not only is there considerable diversity of opinion regarding the definition of attitude, there is also disagreement regarding the relationship between attitudes and each of these other concepts. For the purpOSes of this paper, it seems sufficient to present the most popular viewpoint regard— ing each of these terms. The term most frequently associated with attitude is opinion. There is considerable agreement that opinions are verbalized expressions of attitudes (McNemar, 1946; Remmers, 1954; Rokeach, 1968; Scott, 1968; Thurstone, 1928). English and English (1958) go one step further and point out that: "since verbal statements of opinion are often sought as a revelation of attitude, opinion and attitude are often used A fi ' 20 interchangeably [p. 359]." Hartley and Hartley (1952) oppose this use , of the term, "opinion." They claim: It scarcely seems helpful to consider opinion as ‘ merely the 'verbalization of an attitude.‘ Opinion is a fact of a different psychological order; it differs in its functional relation to behavior. It comes into being just when, and to the extent that, attitudes are not adequate to enable the individual or the group to cope with the situation [p. 657]. However, Hartley and Hartley (1952) do concede that: "every opinion is, ' in large degree, an expression of one or several attitudes . . . [p. 658]." English and English (1958) provide a definition of opinion which is inde— pendent of attitude: a belief that one holds to be without emotional commitment or desire, and to be open to revalua- tion since the evidence is not affirmed to be convincing. It is capable of verbal expression under appropriate circumstances, at least to ] oneself [p. 358]. In general, belief is defined to be: "the emotional acceptance of a proposition or doctrine on what one considers to be adequate grounds ' [Shaw & Wright, 1967, p. 4]." Whereas, attitude is usually considered to include an affective reaction, belief deals more with the mere accep— tance of the existence of an idea or relation. As Fishbein and Raven (1967) state, belief is: "the probability dimension of a concept -— '13 its existence probable or improbable?‘ [p. 188]." Before leaving the concept, "belief," one should note that the definition of belief implied by Rokeach (1968) when he refers to attitude as a set of beliefs, is not the same as the definitions of belief offered here. Rokeach defines belief as "any simple proposition, conscious or unconscious, inferred from what a person says or does, capable of being preceded by the phrase 'I believe that . . . ' [p. 112]." fl 21 E22i£ differs from attitude in several ways: (1) habit places considerable emphasis on an action component (English & English, 1958); (2) habits tend to be rather automatic (Scott, 1968); and (3) habits lack a cognitive or conative aspect (Remmers, 1954). English and English (1958) define habit as "an acquired act, usually a relatively simple one, that is regularly or customarily manifested [p. 235]." Definitions of the term value emphasize an evaluative component. English and English (1958) define value as: ”the worth or excellence, or the degree of worth, ascribed to an object or activity or a class thereof [p. 576]." In comparing value with attitude, Hollander (1967) claims that attitudes are more susceptible to change than are values. Yarrow (1960) reports several similarities between the two concepts: "The core characteristics of attitudes and values are . . . an involve- ment in the object (or principle or event), a consistency in orienta- tion toward the object, and a potential effect on behavior relating to that object [p. 647]." Just as the term attitude has numerous different definitions, there is little consensus on the definition of interest. As English and English (1958) point out, interest is "a term of elusive meanings. It is not clear how far the several meanings . . . are distinct [p. 271]." Remmers (1954) compares attitude and interest and concludes that interest indi- cates a desire to hold the object before consciousness; whereas attitude indicates reactions in terms of direction. Mehrens and Lehmann (1969), however, compare the two terms on the basis of their referents. They consider attitude to be feeling toward an object, social institution or group; and they consider interest to be feeling toward an activity. 22 Attitude Theory There are several theoretical issues in the attitude literature which were considered in the planning of this project. Some of the issues are directly relevant; others serve to provide a general aware— ness of the current thinking in the field of attitude study. It should be noted at the outset that there is a reciprocal relationship between this research and existing theory; the theory contributed to the design of the procedure; the procedure can provide a means for furthering theory. Attitude Objects Although there seem to be very few theoretical points on which there is general agreement, social scientists do agree that any attitude which exists, exists toward a specific referent. The definitions discussed earlier varied in the label they attached to the referent (e.g., psycho- logical object, attitude object, social object), but most definitions, either implicitly or explicitly, included the attitude referent. Although there is virtually no limit to the number of attitudes and attitude objects for a given person, an attitude can exist only when the attitude object is in the individual's psychological world (Krech, Crutchfield, & Ballachey, 1962). Attitudes and Behavior There are two theoretical questions which are at least somewhat relevant to this study: (1) What is the relationship between attitude' and behavior? and (2) What is the relationship between attitude and experience? 23 The first question is extremely important in that attitude measure— ment is actually measurement of behavior. Attitudes can not be directly measured, and so the social scientist is forced to make inferences about attitudes from data gathered about behavior. The behavioral measures can vary greatly. As Cook and Selltiz (1964) point out, the behavioral measure may be: "Language in which the individual reports his feelings about the attitude-object, performance of a task involving material related to the object, or actions toward a representative of the object— class [p. 37]." However, despite the complete dependence on behavioral measures of attitude, there is not a one-to—one correspondence between attitude and behavior. Shaw and Wright (1967) point out that two people with the same attitudes toward a particular referent may exhibit very different behavior when confronted with the object. The attitudes make it more likely that the behavior will take a particular form, but having a particular attitude does not guarantee a particular behavior. Rokeach's theory offers a possible explanation for the lack of a one-to-one corre— spondence between attitude and behavior. Rokeach (1968) claims that: Since an attitude object must always be encoun— tered within some situation about which we also have an attitude, a minimum condition for social behavior is the activation of at least two inter— acting attitudes, one concerning the attitude object and the other concerning the situation [p. 132]. Thus, even if the attitude towards a specific object is held constant, the behavior depends on the situation in which the object is encountered. Returning now to answer the original question, behavior reflects atti- tudes, but there is by no means a one-to—one correspondence between behavior and attitude. 24 Attitudes and Experience The question of the relationship between attitude and experience is, in essence, a question regarding the formation and changeability of atti- tudes. The relationship is best summed up by Hartley and Hartley (1952) who say: "attitude is a product of experience, but it enters into subse— quent experience as a directive factor [p. 654]." Saying that "attitude is a product of experience," is actually saying that attitudes are learned. This is another one of those rare statements in attitude theory on which there is consensus among the social scientists. As Rokeach (1968) points out: "all writers agree that attitudes are learned through the principles of learning . . . [p. 112]." Although this assumption does not affect the design of the technique in this study, it does have implications for the use of the procedure. For example, since experience affects attitude, one could alter experience in an attempt to alter responses to the instrument. Components of Attitudes From a theoretical standpoint, there are several ways to dissect attitudes. Two examples worthy of mention are: (l) examine the compo— nents of an attitude; and (2) examine the characteristics of an attitude. Whether or not one accepts the component approach to attitudes depends to a large degree on the definition of attitude which is accepted. Iirech, Crutchfield, and Ballachey (1962), who define attitude as "an Genduring system of positive or negative evaluations, emotional feelings, sand pro or con action tendencies with respect to a social object [p. 177], aspecify three components of attitude: cognitive component, feeling Ciomponent, and action tendency component. They consider the cognitive Clamponent of an attitude to consist of "the beliefs of the individual 25 about the object Ip. 140];" and the action tendency component “includes all the behavioral readinesses associated with the attitude Ip. 140]." Social scientists who speak of the components of an attitude are refer— ring to these three a8pects, though the three components may be labelled differently. Characteristics of Attitudes Examining the characteristics of an attitude is a second way to dissect an attitude. A review of the literature revealed various characteristics including direction, consciousness, saliency, overtness, magnitude, consistency, intensity, stability and isolation from other attitudes. Of these, the most frequently mentioned are direction, Inagnitude, intensity and saliency; the most frequently measured are direction and magnitude (Scott, 1968). Direction of an attitude is the favorableness or unfavorableness component. Is the affect asso— ciated with the attitude object positive or negative? Scott (1968) points out that direction can be considered as the "increase or decrease of psychological distance from the object Ip. 205]." In order to use a multiple-choice format to measure the direction of an attitude, one nmast have three response choices. If the number of response choices Vwere increased to five, one could also estimate the magnitude of the Eittitude. Magpitude deals with the degree of favorableness or unfavor— iil3leness associated with the attitude object. Does an individual feel ‘Viary favorably toward the attitude object, or does the individual feel 1:airly favorable? In addition to direction and magnitude, frequent mention is made of intensity and saliency. Intensity of an attitude refers to the igflsilrength of the feeling -— i.e., the strength of conviction —- associated in... A 26 with the attitude. Intensity is concerned with how certain the individual feels about his opinion. Saliency of an attitude refers to the centrality of the attitude with reference to the individual's constellation of attitudes. Saliency is concerned with whether the attitude is central or peripheral. The measurement of saliency is based on the subject's tendency to spontaneously introduce the atti- tude variable. It is more realistic to consider attitudes in terms of their characteristics. As Thurstone (1928) points out: "attitude is a complex affair which can not be wholly described by any single numerical index [p. 530]." Thurstone applies an analogy of measur— ing a table. One does not merely measure a table; rather, one measures a particular characteristic of a table, such as height. Likewise, one cannot measure an attitude; instead, one must measure a characteristic of an attitude, such as direction (Thurstone, 1928). Attitude Measurement Procedures Since the goal of this project is to develop an instrument for attitude measurement, it seems appropriate to review many of the \rarious measurement procedures which have already been used. Because tihere is such a wide variety of techniques, it is helpful to discuss tihem within a classificatory framework. In 1940, Day proposed a classification system including: "four major categories, namely, case methods, opinion polls, arbitrary scales, and experimental scales 1:13 - 396]." Since that time, numerous additional techniques have I:"Esen used to assess attitudes; hence, Day's categories are inadequate. JE‘1 1950, Campbell suggested a four-category system which will serve as 27 the basis of this review. Campbell‘s categories were: (1) disguised — structured techniques; (2) disguised - non—structured techniques; (3) non—disguised — non—structured techniques; and (4) non-disguised — structured techniques (p. 15). (See Figure l for the classification of the various attitude assessment procedures.) Disguised — Structured Techniques The disguised - structured techniques include assessment procedures based on the performance of objective tasks, and also many of the proce— dures which use physiological measures or measures of unconscious behavior. The attitude assessment procedures based on the performance of objective tasks assume that: "performance may be influenced by atti— tude, and that a systematic bias in performance reflects the influence of attitude [Cook & Selltiz, 1964, p. 50]." As Cook and Selltiz (1964) point out, the objective tasks "are presented as tests of information or ability, or simply as jobs that need to be done [p. 50]." In each case the respondent is unaware of the actual purpose of the task. One objective task for attitude measurement is the error—choice technique, in which the respondent is presented with a series of ques- tions each having two response choices. Both of the responses are incorrect, but in opposite directions. For example, one response might grossly overestimate a correct numerical answer while the other response might grossly underestimate it. Inferences are made about attitudes based on the direction of the incorrect answers selected. A second example of the use of objective tasks is based on the respondent's memorizing material. The respondent is given material to memorize which reflects positive affect towards the attitude object a:tl-cl material to memorize which reflects negative affect towards the 28 .mmmzomuomn. kzmzama.‘ month: ".0 mzo_.....omh....zo< m Hominy—5.255 .m : .xcosta 695283.39 , .. its... .33....65... .953. uewN_n=.¢wmmo : E ”c. ecznzuai 9.39.... Bosuconom £92923. .5: J<=m_> in £25353 .2 69:5? 39:32:... 5.2.9.53 foam EctoEEou 09:92.99 52.3.93: v.33 u4 .0 6:25.32... £2.22... awesomuomm omozuimao $222193 3.532... "2 n .26. bison .29....” 6:22.... W m Jew—=96... .25.. £5596... 9...... J30 £233.: .0 Eases h x .mwzamaflz mo.><:mm maoazouz: oz< ..:a. 3 ... 42.5.2.5 93.559“ 9.3030 9.39 u 99:33.. .293 .265qu .mo_n9u 93.22.59 9.5.59.6 62.650qu p 69.3“ visas—=3 «3.9... 6.2.3... 9:333 .269 .o 5.99.2.9 £an 3339.2. £0.53 aafiu .0 9935.39 ¢ .659... 2.822.... “hum... wo:.:..:..< oz< mmJf salivation, inferences could be made regarding the subject's percep— tzion of the "goodness" of the attitude object. When GSR is the depen— dent measure, the conditioning aspect of the above description is Omitted; the GSR measures may be taken either while the subject reviews statements related to the attitude object or when the subject is in the pT—‘esence of the actual attitude object. Although they are not actually physiological measures, other nn‘iiisures of unconscious behavior are also used to draw inferences 30 about an individual's attitudes. These "other measures" include eye movement, hand movement, blushing, and general activity level. Disguised - Non-structured Techniques The second category of attitude assessment procedures includes the disguised - non-structured techniques. This category can be further divided into two sub—categories: projective techniques and observation techniques. Projective techniques may involve either a verbal or a visual presentation of the stimulus. Word association, sentence completion, and story completion tasks each involve the verbal presentation of a stimulus. Frequently used visual projective techniques include the Thematic Apperception Test (TAT), the Rorschach Test, and the Rosenzweig Picture Frustration Test. These attitude assessment techniques are taken directly from clinical procedures. "The respondents are shown a picture and asked to tell a story about it, to indicate what they think it represents, or to choose among alternative interpretations [Shaw & wright, 1967, p. 32]." As in their clinical application, scoring is primarily subjective and depends on the interpretive skill of the scorer . Closely related to these projective techniques are the procedures 1>ased on partially structured stimuli: The characteristic common to techniques in this category is that, while there may be no attempt to disguise the reference to the attitudinal object, the subject is not asked to state his own reactions directly; he is ostensibly describ— ing a scene, a character, or the behavior of a third person. He may be presented with a photo- .graph of a member of the object-class and asked to describe his characteristics; or he may be presented with a scene in which.members of the 31 object—class are present and asked to describe it, to tell a story about it, to predict the behavior of one of the characters, etc. [Cook & Selltiz, 1964, p. 47]. In evaluating the use of projective methods of attitude assessment for children, Horowitz and Murphy (1938) state: The advantage of these free methods is that current problems, anxieties, or preoccupations of the child are likely to be most readily exposed and that the child's way of perceiving or organizing experience is more spontaneously revealed; the disadvantage is that without more Specific narrowing of the field, the precise attitudes sought by the experimenter may not be revealed [p. 135]. As mentioned earlier, the disguised — structured techniques could be further divided into two sub-categories, one of which was the observa— tion techniques. Observation techniques are applied under standardized and unstandardized conditions. Observation under unstandardized conditions may take the form of tailing an individual in order to record his behavior. If the experi— menter knows when relevant situations are likely to occur, he can sample from the times, thereby reducing the time spent observing each subject. Obviously, these methods of attitude measurement are inefficient; the experimenter might observe the subject for extended time periods with— out the subject displaying any relevant behavior. A second form of unstandardized observation relies on the products of an individual as the dependent measure. The experimenter can learn about an individual's attitudes by studying the individual's artwork, correspondence, diary, or handwriting. Because observation given unstandardized conditions is "quite undependable under normal circumstances, so it is almost universal prac— tice to present some standardized stimulus . . . [Scott, 1968, p. 210]." 32 The observation techniques in which the experimenter standardizes the conditions may appear to the subject as life—like or may admittedly be staged. When the subject is asked to role play, it is obviously a staged situation. However, there are standardized procedures which are life— like. The subject might be asked to sign a petition relevant to the attitude object. DeFleur and Westie (1958) conducted a study in which the dependent measure was the subject's signature on a photographic release. College students were shown pictures, each of which included one male and one female of different races. The subjects were then asked if they would pose for similar pictures. They were to indicate their response by completing a photographic release form on which they indicated the use they would allow the pictures to be put. Thus, this procedure involved standardized conditions, but was still life—like. It was not even necessary for the subjects to be aware that they were participating in an experiment. Non:gisguised - Non—structured Techniques The third category of attitude assessment procedures are the non— disguised - non-structured techniques. This category includes the free- response interview and questionnaire approaches as well as the self— report studies. In the self-report methods, the subject is asked to explain his beliefs about the attitude object, relate how he feels towards it, or recall how he behaved relevant to it. The biographical studies are also classified as non-disguised - non-structured techniques. In discussing these procedures which rely on open—ended questions, Scott (1968) states the following disadvantages: I . . . (1) it is time-consuming for the interviewer, who must record the subject's answer verbatim, (2) it is time—consuming for the coder, who must 33 content-analyze the response according to mean— ingful dimensions, (3) respondents tend to ramble, so that interviewing time is not used efficiently, (4) in their replies many respondents fail to touch the dimension that the researcher is interested in, so their responses are uncodeable, and (5) any particular attitude is assessed from just a small number of questions (due to time limitations in the interview), and hence cannot be so reliably measured as from the multiple— question approach possible with closed questions [p. 211]. Campbell (1950), however, points out an advantage to this type of procedure: " . their primary asset is that of securing an expres— sion of attitudes in a more natural and spontaneous form . . . Ip. 18]." However, if valid data can be obtained using a procedure other than interviewing, then Campbell's praise of Open—ended questions loses its impact and Scott's criticisms become correspondingly more important. Non-disgpised - Structured Techniques The fourth category of procedures are the non-disguised — structured techniques. This category differs from the third category in that the questions are no longer open-ended. Closed questions, as in category four, have the advantage of forcing the respondent to reply along the dimension of interest to the experimenter. The techniques in this cate— gory are the classic attitude tests and attitude scales. The task for the subject may range from providing a yes or no response to a single item to selecting one out of several responses for each of a series of items. Scott (1968) offers several reasons for preferring the latter procedure: Multiple item assessment may be preferred for a variety of reasons: that single items are subject to too much random response error, that any single item necessarily reflects attributes other than the one in which the investigator is interested, that a composite score constructed from multiple items yields a better representation 34 of the intended attitude, that any inference from a measure to a construct requires multiple measures [p. 211]. Attitude scales comprise the major portion of this category. The various scaling procedures include Thurstone's method of equal appear- ing intervals, Likert's summative scales, Guttman's scalogram analysis, and Guttman's facet analysis, Coomb's unfolding technique and Lazarfeld's latent structure analysis. The first three, Thurstone, Likert, and Guttman, are the best known; hence a somewhat detailed description will be given of each. The goal of each of these procedures "is to yield a set of items or item clusters that can be scored identically for all persons in a particular population of interest [Scott, 1968, p. 232]." In the development of a Thurstone scale, the researcher begins with a large pool of statements relevant to the attitude object. Initially the researcher screens the items to omit any irrelevant, factual, or double—barreled items. A panel of judges is then asked to sort the set of items into eleven piles based on the degree of favorableness or unfavorableness towards the attitude object expressed in the statement. Piles numbered one, six, and eleven are defined for the judges by the researcher. The remaining piles are not defined, but the judges are informed that the psychological distance between any two adjacent piles should equal the psychological distance between any other two adjacent piles. After the sorting is completed, two statistics are computed for each statement: (1) the median, and (2) the inter-quartile range. The items selected for the final attitude scale should have differing median values so as to cover the psychological continuum, and each should have a small inter—quartile range, indicating that the judges were pretty much in agreement with each other. When the final instrument is administered to subjects, each subject is asked to indicate all those statements which 35 he endorses. The subject's score ie either (a) the average of the items endorsed, with each item contributing its scale value to the average, or (b) the median scale value of the items endorsed. Thurstone's method of equal appearing intervals is subject to criti— cism because its construction is so laborious. Remmers attempted to deal with this problem by modifying the procedure slightly. Remmers' master scales are constructed in the same manner as Thurstone‘s technique except reference to a Specific attitude object is omitted from the statements during the stage in which the judges sort the statements. In this way, the same set of scaled items can be used with many different attitude objects simply by inserting reference to the appropriate attitude object at the time the instrument is being readied for administration. Likert's summgtive scales differ from Thurstone‘s scales in that the former is not dependent on a panel of judges to sort the statements. In constructing a scale according to the Likert technique, the researcher still begins with a pool of items relevant to the specific attitude object. As in the Thurstone technique, the items are screened to eliminate factual statements, double-barreled statements and irrelevant items. However, whereas Thurstone-type items allow the respondent only two choices, endorse or not endorse, Likert-type items can provide the respondent with three or five choices ranging from very positive through neutral to very negative. The entire set of items which pass the initial screening are pretested in one of two ways. In some cases, the items may be given to two groups which are known to have opposite attitudes relevant to the attitude object. When this procedure is used, items are selected for the final instrument if they succeed in discriminating between the two groups. A second tech— nique allows for the items to be pretested on an undifferentiated sample. 36 Then items are selected for the final instrument on the basis of an item analysis or a high correlation between the item score and the total score. The Likert technique is referred to as "summative scaling" because the total score is the sum of each item score; the item score for a five choice Likert item can range from one to five with one indicating the very negative response and five indicating the very positive response. It is best if half of the items are favorable to the attitude object and the other half are unfavorable. Regardless of which technique is used for item selection, the final instrument should be pilot tested on a new population in order to determine its reliability and validity. Unlike the Thurstone technique where one individual's score is interpretable, " . . . the interpretation of Likert scores is based upon the distribu— tion of sample scores; i.e., a score has meaning only in relation to scores earned by others in the sample [Shaw & Wright, 1967, p. 25]." There are several problems in using Likert scales. In writing about the Likert technique, Shaw and Wright (1967) point out that: "Likert-type scales are often reliable and valid, but they probably should be treated as ordinal scales [p. 24]." In practice, however, the data are usually treated as interval data. Shaw and Wright also point out the difficulty of interpreting a total score which is equal to the "neutral" category multipled by the number of items. He cau- tions against interpreting that as a neutral attitude since it could be obtained if the respondent has selected the very positive choice for half of the items and the very negative choice for the other half of the items, as well as if the respondent had always chosen the "neutral" category. 37 There has been extensive research comparing the Thurstone and Likert procedures. Edwards and Kenney (1967) reviewed the literature and con- cluded that: According to the evidence at hand, there is no longer any reason to doubt that scales constructed by the method of summated ratings and containing fewer items will yield reliability coefficients as high or higher than those obtained with scales constructed by the Thurstone method [p. 252]. Combining this statement with the fact that Likert scales are easier to construct leads one to conclude that the Likert technique is preferable to the Thurstone technique. Guttman's contributions to the development of attitude assessment procedures include both scalogram analysis and facet analysis. Theoret- ically, in a scale constructed by Guttman's scalogram analysis, all persons who respond positively to a given item have a higher attitude "score" than all persons who respond negatively to that same item. Items are selected in such a way that anyone responding favorably to a given item will respond favorably to all of the items which are earlier in the scale; likewise anyone responding unfavorably to an item will respond unfavorably to all items later in the scale. Still speaking theoret— ically, this means that an individual's total score on a Guttman scale indicates his response on every item in the scale. In other words, any given score can be obtained in only one way and knowledge of a person's score allows reproduction of the person's complete pattern of responses. Of course, Guttman concedes that one would not actually obtain a perfect scale using attitude items. Rather, he establishes a criterion for acceptable scales that are not perfectly scalable (Stouffer, 1950). Whereas the procedures for scale construction discussed thus far 1131y on the statistical analysis of empirical data to select items for 38 the scale, Guttman's facet analysis provides a logical, semantic frame— work for the a priori construction and selection of items. Before any actual item writing is begun, the experimenter decides on the facets which he wishes his items to cover. Furthermore, the researcher decides on the levels which he wishes the items to cover. For example, there might be six levels including societal stereotype, societal norm, personal moral evaluation, personal hypothetical action, personal feeling, and personal action (Jordan, 1969). The facets selected depend on the researcher's objectives and the attitude object being studied. The actual task for the subject when asked to respond to a scale constructed by facet analysis is to select one response for each item in the scale. There are usually three possible responses including one favorable response, one unfavorable response, and one neutral or uncer— tain response. As Diab (1967) points out, the scaling procedures discussed above have two characteristics in common: (1) in the Thurstone technique, the Likert technique, and in Guttman's scalogram analysis, the individual‘s attitude is represented by a single preference score on a positive— negative continuum; and (2) in all four techniques, including facet analysis, the respondents are aware that their attitudes are being assessed (p. 141). Two other procedures in the category of non—disguised — structured techniques are worthy of mention. These are Bogardus' Social—Distance Scale and Osgood‘s Semantic Differential. ngardus' Social Distance Scale is used to measure attitudes towards racial or ethnic groups. The respondent is presented with a series of items descriptive of social relationship. These items are placed in 39 order from the most distant relationship to the closest relationship and might range, for example, from fellow employee to marital partner. The respondent is asked to indicate which items express relationships which the respondent would be willing to have with a member of the racial or ethnic group being studied. Bogardus' Social Distance Scale, though it has applicability to a narrow range of attitude objects, is simple to administer and score; however, scores are difficult to interpret. Osgood's Semantic Differential consists of a series of concepts with pairs of adjectives beneath each concept. The adjective pairs are placed at the ends of a seven point scale. The task for the subject is to place a check mark on the scale for each adjective pair. The check.mark indi- cates how the respondent rates the concept on the given adjective continuum. Unlike most of the procedures discussed thus far, the Semantic Differential yields more than one score. The Semantic Differential can be factor analyzed and the respondent assigned a score for each factor. Commonly, three factors are used: evaluative, potency, and activity. Regardless of which non-disguised - structured technique is employed, there are certain characteristics which a good attitude instrument possesses. These guidelines for an adequate attitude instrument include: (1) it reflects the intended preperty veridically; (2) responses are unaffected by irrelevant characteristics in either the subject or the situation; (3) the process of measuring the attitude does not alter it; (4) the instrument has adequate concurrent validity; (5) the instrument is sufficiently reliable; and (6) the instrument is easy to construct, administer, score, and interpret (Scott, 1968, p. 251). 40 Related Studies The research studies which are most relevant to this project use children as the subjects and include pictures in some part of the proce- dure. Nine such studies will be described here. Three studies were conducted in which pictures were used as the stimulus for a projective interview. Two of the studies were investi— gating attitudes related to school whereas the third study was concerned with racial attitudes. In 1952, Trager and Yarrow used pictures to study the racial atti— tudes of children between the ages of five and eight. In an interview setting, the child was shown a picture of several children playing while one child was off to the side. The picture was drawn so as to suggest racial overtones. Although the interviewer began by asking the subject some general questions about the picture, the questions gradually were made more specific to focus on the racial aspect of the picture (Yarrow, 1960). Biber and Lewis conducted a study in 1949 in which they used a projec— tive test in an attempt to measure first and second graders‘ expectations and feelings regarding school. Biber and Lewis described their procedure as a structured projective technique, because the subject was not only handed the picture, but also was given an oral description of what was happening in the picture. Each subject was asked to explain why the events in the picture were taking place as they were, or the subject was asked to predict what would happen next (Biber & Lewis, 1949). In 1965, Cohen conducted a similar study to measure children's atti— tudes towards school. First and second grade children were shown a series 41 of twenty pictures, each of which was school related. As in the Biber- Lewis study, each child was individually asked to tell what had led up to the situation or what was likely to happen next. Cohen (1967) reported that: "Most of the children appeared to enjoy the interview [p. 306]." Cohen (1965) concluded that: " . . . while a technique of this type probably cannot presently be considered suitable for mass administration as part of large-scale evaluation programs, its further development as a research tool may be warranted [p. 1]." Cohen's above conclusion is a major criticism of the projective techniques described above: they are unsuitable for group administra- tion. Because the instrument must be individually administered, the cost of administering it to large groups is usually prohibitive. As a result, projective techniques for measuring attitudes are used only as research tools. Horowitz (1936) tried several approaches using pictures to measure children's racial attitudes. In one technique used with children as young as kindergarteners, the subject was given a page of twelve photo— graphs of children's faces -— four Caucasian, four light Negro, and four dark Negro. The subject's task was to rank order the faces on the basis of whether the subject liked the picture. Caution needed to be exercised in drawing inferences from this technique, as a preference for one's own racial group was not necessarily a rejection of another racial group. Horowitz' second technique used the same photographs as above. However, now the subject's task was to select companions for a variety of imagined situations. For example, the subject might be asked to: "Show me all the boys you would want to eat lunch with." No limit was placed on the number of answers the subject could provide. 42 In the third technique, the subject was shown a series of photos of social situations. Half of the pictures included only Caucasian boys; the matched half included one Negro boy in each picture. For example, one picture showed Caucasian boys playing checkers and the matched picture showed a Negro boy along with the Caucasian boys play— ing checkers. The subject's task was to tell, for each picture, whether he would or would not like to participate in the pictured activity. Although Horowitz fails to provide any systematic validity data, he does report some of the boys' spontaneous verbalizations as evidence of validity. For example, Horowitz (1936) reported that one boy who consistently omitted Negroes from his choices during the first two tech— niques, offered a spontaneous comment, saying: "I don't like colored boys." Horowitz' techniques suffer from the same major drawback as the projective measures discussed above: they were individually administered. However, with some modifications, they could be group administered. A second criticism is that the techniques are appropriate for measuring attitudes to a limited number of attitude objects —— those that can be pictorially communicated. Radke used a similar technique to measure racial attitudes of chil— dren. The children were shown thirty-four sets of four pictures of faces. In each set, two faces were Caucasian and two were Negro. Each set of pictures was presented with a question, such as: "Which one of the girls in this row is lazy and never bothers to do anything?" The subjects' task.was to indicate a pictorial answer for each question. Although the test produced significant individual and group differences, it was not considered satisfactory for general use (Campbell, 1950, p. 27). 43 Dorkey and Amen (1947) experimented with a procedure closely related to the one described in this proposal. However, they were interested in measuring anxiety rather than attitude. Their subjects ranged in age from three years, ten months to five years, eight months. In an individual setting, each subject was shown a series of pictures, each of which included a child doing something. The subject was given a verbal description of the pictured scene. For example, one picture showed a child walking alongside his mother while the mother pushed a baby carriage with the baby in it. In this and every other picture, the face of the child lacked facial features. The task for the subject was to select for each picture, either a happy or a sad set of facial features to place on the pictured child's face. According to the researchers, the procedure was quite successful. In the literature search conducted for this research, three refer— ences were found which refer to techniques identical to the techniques described for this project. In 1966, Fox, Luszki, and Schmuck published Diagnosing Classroom Learning Environments, a book designed for classroom teachers. In the book, they refer to an attitude measurement procedure like the one to be used in this research. They wrote: One technique that has proved useful for finding out the feelings of these pupils is a scale represented by faces; this can serve where the response to an item is in terms of degree of emotional response, from very positive to very negative. The teacher asks the pupils to put an X under the face . . . that shows how he feels [p. 19]. This is all the book included about the procedure. Although they stated that the "technique proved useful," there is no way to know what they meant by "proved." Obviously, no conclusions can be drawn about the 44 technique from what Fox, Luszki, and Schmuck wrote. Furthermore, the book does not refer the reader to any additional sources regarding the technique. Dysinger (1970) developed an instrument called "When Do I Smile," which is quite similar to the instrument described in this proposal. Dysinger's instrument includes twenty—one, school—related statements which the child reads to himself; for each statement, the subject responds by marking one of five drawings of boys' faces ranging from very happy to very sad. There is no published material related to the instrument. A telephone call to Dysinger's associate revealed that the instrument had been used in one school district in Florida, and the results indicated that discrimination could be obtained. No attempt was made to validate the procedure; however, the Spearman-Brown reliability was computed at .82 based on 200 cases. Test-retest reliability with a four month interval was computed to be .42 based on 97 cases. Dysinger's associate indicated that the instrument was considered experimental and little research had been conducted regarding it. Frymier has developed a similar instrument called the "Faces Scale." There are two forms of the instrument with eighteen items in each.form; the items' content is related to self—concept. The teacher reads the items and the subject responds to each item by marking one of two possible response choices -— a happy face or a sad face. Again, there is no pub— lished material regarding the instrument. Personal correspondence from Frymier (1970) stated: It is still very much a research tool in a develop— ment phase, and though we have had some success with it, we have also had a number of problems. It should only be used as a research instrument and results generated looked at very carefully by those who are conducting the research. 45 Unfortunately, no additional information was given. As a result, there is no information regarding the reliability or validity of the instrument. Furthermore, there was no indication of what the specific problems were that Frymier referred to in his correspondence. Summary This review of the literature has attempted to provide the reader with sufficient information for him to see how the proposed project relates to the field of attitude research: numerous definitions of "attitude" were presented and the definition by Shaw and Wright (1967) was identified as Egg definition for the proposed project; a few aspects of attitude theory were discussed briefly; a relatively comprehensive overview of attitude measurement procedures was presented; and some studies closely related to the proposed research.were discussed. CHAPTER III DESIGN Overview As stated in Chapter I, this project was designed to test a proce— dure and develop an instrument which can be used with groups of early elementary children to assess their attitudes towards school. Specifi— cally, the measurement technique involves oral and pictorial presentation of items with pictorial response choices. The instrument is very similar to a Likert scale in that there are five response choices ranging from very positive through neutral to very negative. However, whereas the typical Likert scale presents the response choices verbally, this instru— ment presents the response choices pictorially. A forty item instrument was administered to all second graders in thirty classrooms. In addition, two validation measures were obtained for a subset of the second graders. The validation measures included: (1) an interview in which the child was asked questions specifically directed towards ascertaining his attitude towards school; and (2) a rating by the teacher of the child's attitude towards school. The data gathering procedures spanned a three week period with the inter— views divided between the first and third weeks, and all of the instru— ments administered during the second week. Following is a more detailed description of the sample, the instrument, and the validation measures. 46 47 Sample The sample for this project was drawn from three school districts in Southern Michigan. The elementary schools in the three districts were divided into two categories: (1) schools which in 1969-70 received State aid for culturally and economically deprived students (funded schools); and (2) schools which in 1969—70 did pp£_receive State aid for culturally and economically deprived students (non-funded schools). The major differences between the two groups are: (l) the funded schools have a much larger percentage of Negro children than do the non—funded schools; and (2) the funded schools have a lower average achievement level than do the non-funded schools. Schools were randomly selected from each of the two groups and all of the second grade classrooms from a selected school were included in the sample. Six funded schools were selected for a total of fifteen classes; seven non-funded schools were selected for a total of fifteen classes. The total enrollment of the fifteen funded classes was 270 students; 241 (89.26%) of these students actually completed the instru- ment. The major cause of failure to complete an instrument was absen- teeism. However, several children, though present, scribbled on the instrument rather than marking responses; these instruments were invalidated by the researcher. The total enrollment of the fifteen non— funded classes was 293; 279 (95.22%) of these students actually completed the instrument. Again, the major reason for failure to complete the instrument was absenteeism, with the instruments of a few students being invalidated by the researcher. The greater percentage of absenteeism 48 in the funded schools is consistent with the usual attendance patterns in the two sets of schools. The sample on which the instrument was actually to be developed, that is, the validation sample, was a subset of the sample described above. Eighty-four students from the fifteen funded classes were randomly identified as part of the validation sample. By random selec— tion, the funded schools were divided into two groups so that students in eight classes were part of the "Week One" group, and students from seven classes were part of the "Week Three" group. A complete set of data, including the instrument and both.validation measures, was prereq— uisite if the student were to be included in the analysis. A complete set of data was obtained from seventy (83.33%) of the students selected from funded classes. Eighty—one students from the fifteen non—funded classes were randomly identified as the other portion of the validation sample. Again using random selection, the non—funded schools were divided into two groups so that students in seven classes were part of the "Week One" group, and students from eight classes were part of the "Week Three" group. A complete set of data was obtained from seventy— five (92.59%) of these students. Again it was consistent that a larger percentage of complete sets of data was obtained from the non—funded schools, as the funded schools typically have a higher rate of absenteeism. Combining the samples for the funded and non-funded schools, one finds that there were thirteen schools selected having a total of thirty second grade classes. Of the 563 students originally scheduled to complete the instrument, 520 (92.36%) students actually did so. From the subset of 165 students randomly selected for the validation sample, complete sets of data were obtained from 145 (87.88%). Table 1 summarizes the information regarding the sample size. Sample Size 49 TABLE 1 Funded Non-Funded Week 3 Total Week 1 Week 3 Total Schools 3 3 6 4 3 7 Number of Classes 8 7 15 7 8 15 Total Enrollment 147 123 270 152 141 293 Completed Instruments N 133 108 241 148 131 279 % 90.48 83.80 89.26 97.37 92.91 95.22 Scheduled Interviews 40 44 84 35 46 81 Completed Interviews N 36 34 70 32 43 75 % 90.00 77.27 83.33 91.43 93.48 92.59 50 Instrument Instrument Design Items The initial item pool for the instrument consisted of sixty—one items of which fifty—one were in the oral presentation style and ten were pictorial items. It was felt that an instrument including all sixty—one items would overtax the attention span of the children. Therefore, a decision was made to limit the instrument to forty items. The twenty-one items which.were omitted from the final instrument were those judged to be most ambiguous or least likely to be in the experi— ential world of all of the children who would be completing the instru- ment. Of the forty items included in the final form of the instrument, thirty-two are in the oral presentation style and eight are pictorial items. (A list of the thirty—two oral items is included as Appendix A. A copy of the eight pictorial items can be found in the instrument included as Appendix B.) For twenty—two of the oral items, a favorable attitude is expressed by marking one of the happy faces; for ten of the oral items, a favorable attitude is expressed by marking one of the sad faces. For all eight of the pictorial items, a favorable attitude is expressed by marking one of the happy faces. Response Choices As mentioned earlier, the instrument is like a Likert scale in that each item has five possible response choices ranging from very happy through neutral to very sad. Each response choice is a drawing of a 51 face, 1 1/4 inches in diameter enclosed in a 1 9/16 inch square. Each face includes the head outline, brows, eyes and mouth. The head outline and the eyes are identical for all of the response choices; the brows and mouth vary on each of the five faces. These characteristics were considered necessary and sufficient based on Harrison's statement (1964) that: "The brow—eye-mouth combination . . . seems to account for a large part of the expression within a given face Ip. 132]." The same set of five response choices is used with each item. Two possible sources of error variance were considered when the faces were designed. One possible source of variance involves the sex of the pictured faces. What happens when it is the same as the sex of the subject responding to the instrument? What happens when the picture is of a different sex from the subject responding to the instrument? In order to hold constant the effect of this variable, the faces lack any sexual characteristics. Similarly, the racial characteristics of the faces were considered as a possible source of variance. What are the consequences when the race of the pictured face and the race of the responding individual are the same? What happens if they differ racially? In order to hold constant the effects of this variable, the faces lack any racial charac— teristics. Furthermore, since traditionally, black lines on white paper are considered symbolic of the Caucasian race, the response choices are printed on light green paper. (Copies of the faces appear in the instru— ment which is included as Appendix B.) The faces were pilot tested on twenty—two children ranging in age from four years through nine years, with an average age of seven years. Each child individually was shown the ten possible pairs of faces 52 [(5 x 4) e 2 = 10]. For each pair, the child was asked to identify which face was happier (or sadder). Nineteen of the children responded to the ten pairs of faces without error. The other three of the children made a total of six errors, of which two were obviously the result of careless— ness. On the basis of the results from the pilot test, it was concluded that the relationship between the affective qualities of the various faces was sufficiently clear for most students. Format Each instrument consists of seventeen pages. The first page includes three rows of five faces. In each row, the order of the faces is identiw cal with the very happy face at the left end and the very sad face at the right end. To the left of each row of faces is a small picture of an object likely to be familiar to young children. The first page provides the response choices for the sample items. Pages two through nine each display four rows of five response choices. Again, to the left of each row is a small picture of an item familiar to young children. The pictures of familiar items on pages two through nine, as well as the picture on the sample page, serve the same function that item numbers serve for adults: to aid the child in finding the correct row in which to mark his answer for each orally presented item. Pages ten through seventeen of the instrument each display one pictorial item and a set of five response choices. Since each of these pages displays a different pictorial item, it was unnecessary to provide an additional picture to serve the function of a page number. The pictorial item itself was used to indicate to the child whether he was on the correct page. 53 All seventeen pages of the instrument were printed in black ink on light green paper. As mentioned earlier, colored paper was used to create racial ambiguity. Inserted, but not stapled, in each instrument was a piece of bright yellow construction paper which the children could remove and use to help them stay in the correct row when answering each oral item. Instrument Administration The instrument was administered to each of the thirty selected classes during the second week of the three week data gathering schedule. All of the second graders present in a given classroom completed the instrument at the same time; however, the second grade enrollment in the participating classes ranged from seven to twenty—five. Since it was felt that the teachers might influence their students' responses, the administration of the instrument was conducted by college coeds hired and trained by the researcher. (See the section titled "Interviewers" on page 55 for details regarding the hiring and training of the research team.) The instrument administrator was required to distribute the materials, explain the directions, and read the oral items. A sixteen page manual was provided to the administrator from which she was to read verbatim the directions and the items. The manual also includes directions to the administrator which were not to be read to the children. These appear on the left hand side of each page and are printed in italics. (See Appendix C for a copy of the manual.) The first five pages of the manual concentrate on explaining the task to the children. First the children are introduced to the "item number" pictures and the faces. The emotional quality of each of the five faces is explained to the children; several different adjectives 54 are used to describe each.face. Next the children are shown how to use the accompanying piece of construction paper as a line guide to help them stay in the correct row when answering an oral item. The children are then asked to complete each of the three sample items. The first sample item asks: "How do most people feel when they eat ice cream or candy?" The extreme happy face is pre—marked in all of the instruments with a large "X." The second sample item asks: "How would you feel if you fell and broke your leg?" In this case, the children are asked to place an "X" over the response they would choose. After allowing suffi— cient time for the children to mark a response, the administrator suggests that most people probably marked the very sad face, but it is stressed that there are no right or wrong answers in this task. In the third sample item, the children are directed to mark the face which "is not happy and not sad." Again, after allowing time for the children to mark a response, the administrator explains their probable answer. Pages six through sixteen of the manual include the thirtyetwo verbal items and the verbal directions to accompany the eight pictorial items. The total time for test administration including distribution and collection of materials varied from approximately 30 to 45 minutes. Instrument Scoring In order to facilitate scoring, data processing numbers were printed on the instruments. The column number for each item was printed in the "item number" picture box for the first thirty—two items and in the picto— rial item itself for the last eight items. The score to be assigned to each response was printed in the box with each face. The numbers were printed in very small type so they would be unlikely to influence the 55 child's responses. In some cases the numbers reading from left to right ranged from one to five; in other cases, they ranged from five to one. In all cases a high score on an item was associated with positive feelings about school and a low score was associated with.negative feelings. Validation Measures Two validation measures were obtained for each of the students in the validation sample. One measure was based on the results of an indi— vidual interview of the child conducted by a member of the research team. The second measure was a rating of the child supplied by the teacher. Interviews Interviewers Seven college coeds were hired for the purpose of conducting the interviews. (NOTE: These same seven women served as the test adminis- trators.) Six of the women were undergraduates at Michigan State University; one was a Michigan State University graduate student and experienced teacher. Two sessions of three hours each were conducted for the purpose of training the women to conduct interviews (and administer the instruments). During the training sessions, the interviewers were taught how to use the random numbers provided by the researcher to identify the students for interviewing. Furthermore, they were taught how to conduct the interview. Particular emphasis was placed on the interviewer avoiding any inadver- tent reinforcement of the child's responses. During the training sessions, the interviewers were also taught how to score their own interviews. 56 In assigning interviewers to schools, attention was given to match— ing the race of the interviewer with the dominant race of the school. The three Negro interviewers were assigned to the six predominantly Negro schools which included a total of sixteen second grade classes. The four Caucasian interviewers were assigned to the seven predominantly Caucasian schools which included a total of fourteen classes. In addition, care was taken to see that the same member of the research team did not conduct the interviews and administer the instruments in any given school. Two people carried out these two different tasks during two separate weeks. The Interview Each child was individually interviewed for 8—12 minutes by a trained member of the research team. Of the usable interviews, 47% were conducted during the first week of the three week data gathering period; the remain- ing 53% of the usable interviews were conducted during the third week. (An interview was "usable" only if it was part of a complete set of data for that child.) The purpose of the interview was to ascertain the child's attitude towards school, without differentiating between the social and academic aspects of school. In order to accomplish this, each child was asked to respond to thirty sentence completion stems and answer twenty—five direct questions. (See Appendix D for a copy of the sentence completion stems and the direct questions.) Each of the thirty sentence completion stems and twenty-five direct questions included a direct reference to school, the class, or the teacher. This increased the probability that the child's response would provide information regarding the attitude object of interest —— school. The one exception to this was the first direct question, which asked: "What is the best time of the day?" This 57 question, asked before the child's attention had been focused on school and school—related topics, sought to ascertain whether the child believed a time during the school day or after the school day was the best time of day. Interview Scoring Each interview was first scored by the individual who conducted the interview. The interviewer was directed to record the score he would assign immediately upon completing the interview. In addition, all of the interviews were independently scored by three women, two of which were graduate students and experienced teachers. To make this additional scoring possible, each interview was tape recorded. Thus, the three scorers were able to independently listen to an audio tape of each inter— view and assign a score. In scoring an interview, each scorer and the interviewer indepen- dently assigned an integer value ranging from one through five, with one indicating that the child had expressed extremely negative feelings towards school and five indicating that the child had expressed extremely positive feelings towards school. In order to minimize the error variance in the scores independently assigned, the three scorers and the interviewer were each provided with a written description of what each numerical score should represent. The written description described the child in terms of his general attitude towards school and the teacher, his motivation level, his performance level, his self-concept regarding school, and his interest in school. (See Appendix E for a copy of the guidelines for scoring an interview.) 58 Teacher Rating In addition to being interviewed, each child in the validation sample was rated by his teacher. The method for teacher ratings paralleled the method for interview scoring. Each teacher was provided with the same written description which.was given to the scorers and interviewers. The teacher was directed to use what she knew about her students, and the guidelines established in the written descriptions to assign a rating of l, 2, 3, 4 or 5 to each of the specified children. The rating assigned reflected the teacher's perception of the child's feelings about school. The score assigned to the child by his teacher was treated as a second validation measure. Summary This project was designed to develop an attitude assessment instru— ment appropriate for use with groups of early elementary children. The instrument development sample included a total of 145 students selected from thirteen schools in three districts in Southern Michigan. In addi— tion, 375 of their classmates completed the instrument. Each of the students completed a forty item instrument designed to assess his atti— tude towards school. Administration of the instrument involved oral presentation of thirty-two items. The students recorded their responses by marking a picture of one of five faces which ranged from very happy through neutral to very sad. For the remaining eight items, the item, as well as the response choices, was presented pictorially. In addition to completing the instrument, each child in the instru— ment development sample was individually interviewed by a member of the 59 research team. The purpose of the interview was to ascertain the child's attitude towards school and thereby validate the instrument responses. Approximately half of the instruments were administered during the week preceding the instrument administration while the remainder of the inter— views were conducted during the week following the instrument administra— tion. For additional validation information, each child in the instru— ment development sample was rated by his teacher to indicate the teacher's perceptions of the child's attitude towards school. CHAPTER IV RESULTS Overview Since the major focus of this project was to develop an instrument appropriate for use with groups of early elementary children, the major thrust of the analysis focussed on instrument develOpment. However, some preliminary analyses were conducted to determine whether the major analyses were likely to be productive. The outcomes of the preliminary analyses were positive. The main analyses were conducted twice. The second set of proce— dures, which duplicated the first set, were based on a larger N in an attempt to obtain more stable results. The purpose of the main analyses was to identify the best subsets of items from the forty item instrument administered to the subjects. Supplementary analyses were also conducted. These analyses included comparisons of subsets of students from the instrument development sample. Furthermore, analyses were conducted to compare subsets of the students who completed the instrument but were not in the instrument development sample. The results obtained from these students were used to develop some tentative norms. Following is a more detailed discussion of the preliminary analyses, item selection procedures, and supplementary analyses. 60 61 Preliminary Analysis Sample Items Before investigating the reliability and validity of the children's responses, an examination was made to determine whether it was likely that the children were actually able to perform the required task. In order to ascertain this, two of the three sample items were scored from each of the 520 students who completed the instrument. The first item, which asked, "How do most people feel when they eat ice cream or candy?" could not be scored since the correct response was printed on the instru— ment. This sample item provided the children with an example of how to record their responses. The second sample item asked, "How would you feel if you fell and broke your leg?" The child's response was considered correct if he marked either of the two sad faces. Of the 520 students who completed the instrument, only fourteen (3%) did not respond correctly to the second sample item. The third sample item, which was apparently more difficult, directed the children to: "Make an X on the face that shows best how you would feel if you didn't feel happy and you didn't feel sad?" Only the neutral (middle) face was considered as a correct response to this item. There were fifty—seven children (11%) who responded incorrectly to the third sample item. Counting up the number of students who responded correctly to the two sample items, it was found that 455 (87%) of the 520 students who completed the instrument were able to correctly respond to both sample items. It was decided that this percentage figure was sufficiently high to suggest that the task itself was not overly difficult. 62 Interviewer Reactions As a second step in the preliminary analysis, a conference was held between the researcher and each of the seven college coeds who conducted the interviews and administered the instruments. The purpose of the conference was to learn whether any of the women had encountered anything which would shed light on the data analysis. A brief summary of their comments is presented here with the understanding that these are the women's purely subjective reactions. During the conference, the women were asked to relate their impres- sions of the interviews. Only one of the seven interviewers reported that any of her interviewees seemed disturbed by the presence of the tape recorder. The other six reported that their interviewees either ignored the tape recorder or made a special effort to speak directly into the microphone. Five of the interviewers stated that they felt that the students were not honest during the interview. One of the women reported that the children kept looking to her for an indication of the appropriate- ness of their responses; three of the women reported that the children's responses often sounded as though they were designed to please the interviewer. Of the seven interviewers, only two did not question the honesty of the children's responses. During their conferences, each interviewer was asked whether the children appeared to have any particular difficulty during the interview. Only one child encountered so much difficulty that he could not be interviewed. This child was Mexican—American and did not have sufficient knowledge of English to comprehend the questions. 63 The interviewers reported that the sentence completion stems were more difficult for the children than were the direct questions. One item, "At school the grown—ups think I . . . ," seemed to be especially difficult for the children as many of them did not understand the term "grown-ups." None of the interviewers reported that they had any difficulty scoring the interviews. During the conference, the women were asked for their impressions regarding the instrument administration. Five of the women, in their role as instrument administrators, reported that the children appeared to very much enjoy completing the instrument. Three women indicated that the students had requested to keep the instrument; two women, without being asked, stated that the children obviously preferred the instrument to the interview. Apparently not all children enjoyed the instrument, however; two women reported that they had administered the instrument to a class that had become bored. The women were asked whether the children had appeared to under~ stand the directions. Most women reported that, if anything, the directions were too repetitive. Only one woman reported that she had a class which had difficulty understanding the directions. It is known that that particular class had all "slow learners" who, though technically second graders, were in a combined class of first and second graders. Unfortunately, there was some "cheating" observed during the test administration. Four women reported that they had observed children display some of the following behaviors: yelling responses aloud, whispering responses to children nearby, or copying responses off 64 another child's instrument. Fortunately, the proportion of children who displayed these behaviors was quite small. Another undesirable behavior which was noted by some of the instru— ment administrators was that some children would mark their responses before an item was read. Again, this behavior was displayed by only a very few children. When the women were asked to relate the teachers' reactions to the teacher rating form, five women indicated that the teachers had not commented about rating their students. Two women, on the other hand, reported that teachers had complained about the task. One teacher was quoted as having said: "All kids like school and love the teacher." During the conference, each woman was asked if she noticed anything unusual during the interviews or instrument administration which might have implications for the research project. Most of the women had no observations to offer. One woman, however, noted that the children seemed to distinguish between "social school" and "academic school." She reported that the vast majority of participating students seemed to like "social school," but their attitude towards "academic school" might be quite negative. Although this observation did not affect this research, it does have implications for future research. An attitude scale towards "social school" and an attitude scale towards "academic school" might well be developed. After considering the information acquired during the conferences held between the interviewers and the researcher, it was concluded that the instrument development could proceed. It was somewhat discouraging to learn that some students had "cheated"; however, the number'bheating" was considered sufficiently small so as not to greatly affect the item 6S selection procedures. It was encouraging that the vast majority of the students appeared to understand the directions. It was further encourag— ing to learn that most of the students appeared to enjoy completing the instrument. Item Selection First Analysis The purpose of the data analysis was to identify a subset of the "best" twenty items from the forty item instrument. In order to accom— plish this, it was decided to divide the students into two groups -— one to be used for item selection and one for cross-validation. There were a total of 145 students with complete sets of data including completed instrument, interview, and teacher rating. To divide the data from these 145 students into two groups, the data were first assigned to one of four groups: (1) funded school, interviewed during first week of project; (2) funded school, interviewed during third week of project; (3) non-funded school, interviewed during first week of project; and (4) non-funded school, interviewed during third week of project. Each of these four groups was then randomly divided into two groups: (1) item selection group; and (2) cross—validation group. This two step procedure for dividing the data into two groups insured that in each group there was equal representation of students from funded and non—funded schools, and students interviewed during weeks one and three. The item selection group included a total of seventy—three students; the cross-validation group included a total of seventy-two students. The early portions of the analysis focused on the item selection group. 66 The first step in the analysis was to ascertain whether it was reasonable to combine the four scores assigned to each interview. As one may recall, each interview was scored by the interviewer and addi— tionally, by three scorers working independently. The correlations were computed between the scores assigned by each of the four. As one can see from Table 2, these correlations, all positive, ranged from .66 to TMREZ Correlations Between Interviewer, Tape Scorer 1, Tape Scorer 2, and Tape Scorer 3 Tape Scorer l Interviewer Tape Scorer 2 Tape Scorer l Tape Scorer 2 Tape Scorer 3 .82. Based on these correlations, it was concluded that the scores from the four sources could reasonably be added together and expressed as a single "interview" score. The second step in the analysis was to learn whether it would be reasonable to select a single set of twenty items which.would correlate well with the interview score, the teacher rating score, and the total 67 test score Cmeasure of internal consistency). Therefore, the correla— tions were computed between these three measures. As one can see from Table 3, the correlations between teacher rating score and each of the TABLE 3 Correlations Between Interview Score, Teacher Rating Score, and Total Test Score Total Test Score Teacher Rating Score Teacher Rating Score .07 Interview Score .42 .23 other two measures, though positive, are quite low. Based on these correlations, it was concluded that two subsets of twenty items would have to be selected -— one set to correlate with both the interview score and the total test score, and a second set to correlate with the teacher rating score. In order to select the twenty items which correlated best with the interview score and the total test score, two sets of correlations were computed: (1) each of the forty test items was correlated with the interview score; and (2) each of the forty items was correlated with the 68 total test score. These correlations, which are diaplayed in Table 4, were used to select the best twenty items. A graph, displayed in Figure 2, was constructed with the item—total test score correlations along the abscissa and the item-interview test score correlations along the ordinate. A line was superimposed at the median on each axis. The items in quadrant 1 had the highest correlations with both the interview score and the total test score. Thirteen of the fourteen items in quadrant 1 were therefore selected for the "new" instrument. Since item 45 was the only pictorial item which would have been selected, it was decided to omit pictorial items from consideration. (NOTE: This decision was reversed when the reanalysis was conducted.) The seven additional items were selected by simultaneously moving leftward and downward from quadrant 1. It was decided to include the twenty oral presentation style items which simultaneously satisfied two conditions: (1) correlated at least +.l7 with the interview score; and (2) correlated at least +.32 with the total test score. The items which satisfied these conditions included items numbered 11, 12, l3, 14, 15, l6, 19, 24, 25, 28, 30, 31, 32, 33, 36, 37, 38, 39, 40, and 41. (NOTE: The forty items were numbered ll—SO throughout the entire research project. These numbers corresponded to the columns on the data card in which the responses were punched.) Given that the twenty selected items were the "best" items for the item selection group in terms of their correlations with the interview score and the total test score, attention could now be directed towards examining how well these items functioned in the cross—validation group. Using the seventy—two subjects in the cross—validation group, the ~ correlation was computed between the interview score and the total test score which was now the sum of the twenty selected items. The 69 TABLE 4 Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score: Item Selection Group Correlation With Correlation With Item No. Interview Score Total Test Score 11* .29 .53 12 .39 .65 13 .38 .56 14 .17 .32 15 .39 .69 16 .22 .48 17 .14 .45 18 -.O9 .24 19 .22 .54 20 .23 .11 21 .10 .31 22 -.O4 .50 23 .12 .35 24 .48 .69 25 .34 .63 26 .02 .24 27 .16 .29 28 .37 .67 29 .07 .37 3O .22 .49 31 .41 .72 32 .34 .67 33 .17 .54 34 .09 .56 35 .08 .49 36 .38 .66 37 .19 .80 38 .25 .38 39 .17 .34 40 .33 .72 41 .22 .62 42 .07 .51 43 .06 .27 44 .22 .19 45 .22 .59 46 .14 .44 47 .04 .56 48 .21 .53 49 .23 .41 50 .17 .43 * The forty items were numbered 11-50 throughout this entire research project. These numbers corresponded to the columns on the data card in which the responses were punched. Item-Interview Correlations 7O .461 .421 .38. .341 .30" .26" .22 ‘9 £2 .18‘ .141 010‘ £24 {D 502-1 .Im ‘ .10 .16 .22 .28 Item—Total Correlations FIGURE 2 Total Test Score Graph of Correlations Between Each Item and Interview Score and Each Item and 71 correlation between these two measures, which.had been .42 based on data from the item selection group, drOpped to .23 when data from the cross— validation group were used to compute the correlation. (NOTE: The corre— lation increased from .42 to .48 for the item selection group when the total test score was computed as the sum of the twenty selected items.) In addition, using the data from the cross—validation group, corre— lations were computed between each item and the interview score. These correlations are presented in Table 5. Whereas these correlations, when based on the item selection group, ranged from +.l7 to +.48, the correla- tions using the cross—validation group ranged from —.O7 to +.32. As one can see from Table 5, for eighteen of the twenty items, the correlation between item and interview score was lower for the cross—validation group. In other words, the items which seemed to correlate well with the inter— view score when the data from the item selection group were used, did not correlate very highly when the data from the cross-validation group were used. The outcome was somewhat better when one considered the correlation between the item and the total test score. (See Table 5.) Since the total test score was now the sum of the scores on only twenty items, the correlations for the item selection group were recomputed. For seventeen of the twenty items, the correlations increased. The range of the item— total test correlations for the item selection group was now +.36 to +.78. Looking at the correlations between item and total test score for the cross-validation group, one finds that for sixteen items, the correlations were lower than those obtained using the item selection group. The range of correlations based on the cross—validation group was +.34 to +.68, which was only slightly lower than the range for the item selection group. 72 TABLE 5 Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score: Cross-Validation Group Correlation With Correlation With Item No. Interview Score Total Test Score* 11 (.29)** .08 (.29)*** (.53)** .51 (.57)*** 12 (.38) .13 (.38) (.65) .53 (.74) 13 (.38) .12 (.38) (.56) .60 (.56) 14 (.17) .08 (.17) (.32) .62 (.40) 15 (.39) .03 (.39) (.69) .66 (.72) 16 (.22) .29 (.22) (.48) .48 (.59) 19 (.22) .06 (.22) (.54) .54 (.55) 24 (.48) .13 (.48) (.69) .60 (.77) 25 (.34) .06 (.34) (.63) .40 (.65) 28 (.37) .21 (.37) (.67) .63 (.73) 3O (.22) .08 (.22) (.49) .49 (.44) 31 (.41) .20 (.41) (.72) .58 (.76) 32 (.34) .10 (.34) (.67) .57 (.71) 33 (.17) .08 (.17) (.54) .47 (.58) 36 (.38) .15 (.38) (.66) .68 (.70) 37 (.19) —.O7 (.19) (.80) .56 (.78) 38 (.25) .12 (.25) (.38) .34 (.41) 39 (.17) .32 (.17) (.34) .36 (.36) 4O (.33) .09 (.33) (.72) .61 (.76) 41 (.22) .17 (.22) (.62) .46 (.55) * Total test score is defined as the sum of the item scores on the twenty selected items. ** Correlations computed from item selection group data using all forty items. These were originally presented in Table 3. *** Correlations computed from item selection group data using only twenty items selected for their high correlation with interview score and total test score. As is shown, the correlation with the interview score remains unchanged when only twenty items are used. However, the correlation with.the total test score changes as the total test score is now defined as the sum of twenty items rather than the sum of forty items. 73 As was mentioned earlier, it was decided that two subsets of twenty items were to be selected —— one set to correlate with both the inter— view score and the total test score and a second set to correlate with the teacher rating score. Using the data from the same item selection group as above, correla— tions were computed between each of the forty items and the teacher rating score. As one can see from Table 6, there were a considerable number of low positive and negative correlations. As a result, in order to select twenty items, items with correlations as low as +.03 were chosen. To maintain consistency with the above procedures, all pictorial items were excluded. The twenty items which were selected included items 13, 15, l6, 17, 18, 19, 21, 22, 23, 25, 27, 33, 34, 36, 37, 38, 39, 40, 41, and 42. Using the seventy—two subjects in the cross-validation group, the correlation was computed between teacher rating score and total test score. This correlation, which had been +.O7 for the item selection group (or +.18 for the item selection group when the total test score was the sum of the twenty items only) increased to +.24 when the data from the cross-validation group were used. In addition, the correlations between each of the twenty selected items and the teacher rating score were computed for the cross—validation group. These are displayed in Table 7. For twelve of the twenty items, the item-teacher rating score correlation increased when the data from the cross-validation group were used. The range of correlations was greater for the cross—validation group: whereas the range of item-teacher rating score correlations on the twenty selected items for the item selection group was +.O3 to +.33, the range for the cross—validation group was —.08 to +.38. 74 TABLE 6 Correlations Between Each Item and the Teacher Rating Score: Item Selection Group Correlation With Teacher Item No. Rating Score 11 * -.01 12 -.O7 13 .13 14 -.O7 15 .03 16 .04 17 .11 18 .33 19 .20 20 -.02 21 .15 22 .20 23 .07 24 .01 25 .11 26 -.13 27 .07 28 -.04 29 -.25 3O .03 31 —.09 32 —.07 33 .04 34 .10 35 -.10 36 .04 37 .04 38 .19 39 .05 40 .05 41 .10 42 .08 43 .09 44 .15 45 -.11 46 .03 47 —.03 48 .15 49 .06 50 -.01 * The forty items were numbered 11-50 throughout this entire research project. These numbers corresponded to the columns on the data cards in which the responses were punched. 75 TABLE 7 Correlations Between Each of the Twenty Items Selected to Correlate Highly with Teacher Rating Score: Cross-Validation Group Correlation With Item No. Teacher Rating Score 13 (.13)* .22 15 (.03) .27 16 (.04) -.O8 17 (.ll) .16 18 (.33) .03 19 (.20) .24 21 (.15) .03 22 (.20) .03 23 (.07) .26 25 (.11) -.O6 27 (.07) .38 33 (.04) .06 34 (.10) .09 36 (.04) .18 37 (.04) —.08 38 (.19) .02 39 (.05) .10 4O (.05) .16 41 (.10) .23 42 (.08) .15 * Correlations computed on item selection group using all forty items. These were originally presented in Table 5. 76 First Analysis — Discussion There were some outcomes noted during the item selection analysis described above which warrant some discussion. The first such outcome was the low correlations between the teacher rating score and each of the other two measures —— interview score (.23) and total test score (.07). There are at least three possible explanations for these low correlations. One explanation suggests that different teachers used different anchor points for their rating. In other words, some teachers had a tendency to rate high whereas other teachers had a tendency to rate low. This could result in greater variance between teachers than within teachers. The outcome of this would likely be low correlation with total test score and/or interview score. There was no way of actually testing this explanation. Since no more than one teacher ever rated a given student, it was impossible to test whether the between teacher variance exceeded the within teacher variance. A second procedure which might shed light on this explana— tion was also impossible. Had each teacher rated a large number of students, one might expect that if the teachers were using the same anchor points, the sets of ratings assigned by different teachers would have equal means. However, since each teacher rated, at most, seven students, this does not seem a reasonable assumption. Since there was no way to actually test this first explanation, it remains a plausible hypothesis to be tested in future research efforts. A second explanation for the low correlations between teacher rating score and each of the other two measures —— total test score and interview score -— is that the teacher rating score actually measures something 77 different than do either of the other two measures. Teachers obviously based their scores on different data than the interview or the instrument. Perhaps the data which teachers have available for judging the attitudes of their students lead to different conclusions than the interview or the instrument. If this explanation is, in fact, accurate, then future research projects may want to investigate the relationships between the attitudes towards school which students express to their teachers and the attitudes towards school which students express to an outside researcher. The third explanation, closely related to the second, suggests that students "faked" their interview and instrument responses to the same extent. However, teachers based their ratings on behavioral observations which were not subject to the same "fakability." If this explanation is accurate, then the items selected for their high correlation with the teacher rating score are likely to be better indicators of a student's attitude, whereas items selected for their high correlation with the interview score are better indicators of what a student wants to communi— cate is his attitude. A second outcome noted during the data analysis and warranting some discussion was the change in correlational values obtained from the item selection group and from the cross—validation group. When twenty items were selected for their high correlation with the interview score and total test score, the correlation between interview score and total test score which had been +.42 for the item selection group (or +.48 if the total test score was the sum of the twenty items rather than all forty items) dropped to +.23 for the cross-validation group. Furthermore, for eighteen of the twenty items, the correlation between item score and interview score was lower in the cross—validation group than in the item 78 selection group. However, when twenty items were selected for their high correlation with the teacher rating score, the results were quite the opposite. The correlation between teacher rating score and total test score, which had been +.O7 for the item selection group (or +.18 if the total test score was the sum of the twenty selected items rather than the sum of all forty items), increased to +.24 for the cross—validation group. Furthermore, for twelve of the twenty items, the correlation between the item score and the teacher rating score was higher in the cross—validation group than in the item selection group. These results led the researcher to conclude that the number of cases in the item selection group was too small to be used for item selection. Obviously, the results were not stable when one went from the item selection group data to cross-validation group data, although they might be if the results were based on an increased number of cases; i.e., the influence of random error was quite significant when the data from only seventy-three cases were used for item selection, and hence, the initial values of the correlations were not maintained on cross— validation. Thus, the decision was made to increase the size of the item selection group by combining the data from the item selection and cross— validation groups. This would not guarantee better item selection, but it would reduce the influence of random error. Thus with the larger N, the item selection correlations are more likely to be maintained on subsequent cross-validations. Second Analysis The second analysis began by combining the seventy—three students in the item selection group with the seventy—two students in the cross— validation group to form a single group of 145 subjects. Repeating the 79 analysis reported above, the first step was to ascertain the reasonable— ness of combining into a single score, the four scores independently assigned to each interview. Correlations were computed between the score assigned by the interviewer and each of the scores assigned by the three tape scorers. These correlations, displayed in Table 8, were judged sufficiently high to warrant combining the four independently assigned scores into a single score to be labelled interview score. TABLE 8 Correlations Between Interviewer, Tape Scorer 1, Tape Scorer 2, and Tape Scorer 3 Tape Scorer 1 Tape Scorer 2 Interviewer O‘ \1 Tape Scorer l Tape Scorer 2 .73 Tape Scorer 3 .67 Still repeating the procedures conducted in the first analysis, the second step in the analysis was to determine the reasonableness of select— ing a single set of twenty items which would correlate highly with the interview score, the teacher rating score, and the total test Score. Therefore, the correlations were computed between these three.measures. 80 As one can see from Table 9, the results paralleled the results obtained in the first analysis: the correlation between teacher rating score and total test score and the correlation between teacher rating score and interview score are both quite low. On the other hand, the positive correlation between total test score and interview score is significantly different from zero. TABLE 9 Correlations Between Interview Score, Teacher Rating Score, and Total Test Score Total Test Score Teacher Rating Score Teacher Rating Score .14 Interview Score .34 .09 As a result of these correlations, the decision was again reached to select two sets of twenty items: one set to correlate well with both the interview score and the total test score, and a second set to corre— late well with the teacher rating score. In order to select the twenty items which.correlated best with the interview score and the total test score, the procedures used in the 81 first analysis were followed. Two sets of correlations were computed: (1) each of the forty items was correlated with the interview score; and (2) each of the forty items was correlated with the total test score. From these correlations, which are displayed in Table 10, it was possible to again construct a graph with the item-total test correlations along the abscissa and the item-interview correlations along the ordinate. Again lines were superimposed at the median on each axis. All of the items falling in the first quadrant of the graph, displayed as Figure 3, were included in the "new" instrument. Since only fourteen items were in the first quadrant, it was necessary to select an additional six items. This was accomplished by simultaneously moving leftward and downward from quadrant one. It was decided to select the twenty items which simulta— neously satisfied two conditions: (1) correlated at least +.l4 with the interview score; and (2) correlated at least +.45 with the total test score. The items which satisfied these conditions included items 11, 12, 13, 15, 16, 19, 24, 25, 28, 30, 31, 32, 33, 36, 40, 41, 45, 48, 49, and 50. The initial instrument of forty items included thirty—two oral presentation items and eight pictorial items; the "new" instrument of twenty items included sixteen oral presentation items and four pictorial items. The item—total test correlations which were presented in Table 10 were computed with the total test score defined as the sum of all forty items. With the "new" instrument consisting of only twenty items, it was possible to recompute the item-total test correlations defining the total test score as the sum of the twenty selected items. These correla— tions are presented in Table 11. For thirteen of the twenty items, the correlation between item and total test score increased when the total 82 TABLE 10 Correlation Between Each Item and Interview Score and Between Each Item and Total Test Score: All Subjects Correlation With Correlation With Item No. Interview Score Total Test Score 11 .20 .53 12 .27 .60 13 .27 .56 14 .13 .44 15 .24 .68 16 .26 .45 17 .10 .45 18 -.O6 .22 19 .15 .51 20 .10 .13 21 .10 .15 22 -.Ol .50 23 .09 .35 24 .33 .65 25 .24 .53 26 .19 .32 27 .12 .38 28 .30 .61 29 .05 .45 30 .17 .50 31 .31 .63 32 .22 .59 33 .14 .51 34 -.03 .43 35 .01 .41 36 .27 .65 37 .09 .71 38 .19 .38 39 .24 .34 4O .21 .66 41 .22 .58 42 .07 .49 43 .13 .28 44 .14 .25 45 .14 .59 46 .07 .38 47 .06 .56 48 .20 .50 49 .19 .45 50 .24 .47 ItemrInterview Correlations 83 .33-1 .30-4 .27-4 .24-1 .21-1 .18-1 GS (3 .15-1 96 ED .12-1 £94 .06 «- .03—1 .00" .031 -.06"' @@ Item-Total Correlations FIGURE 3 Graph of Correlations Between Each Item and Interview Score and Each Item and Total Test Score 84 TABLE 11 Correlations Between Each Item and Interview Score and Between Each Item and Total Test Score Using Twenty Items Which Correlated Highly With Interview Score and Total Test Score Correlation With Item No. Total Test Score 11 .54 (.53)* 12 .65 (.60) 13 .60 (.56) 15 .68 (.68) 16 .52 (.45) 19 .55 (.51) 24 .68 (.65) 25 .56 (.53) 28 .66 (.61) 3o .47 (.50) 31 .68 (.63) 32 .60 (.59) 33 .54 (.51) 36 .65 (.65) 40 .70 (.66) 41 .56 (.58) 45 .58 (.59) 48 .51 (.50) 49 .43 (.45) 50 .45 (.47) * Item-total test score correlations computed when the total test score was defined as the sum of all forty items. These values were originally presented in Table 9. 85 test was defined as the sum of the twenty selected items. Also computed was the correlation between interview score and total test score. This correlation, which had been +.34 when the total test score was defined as the sum of forty items, increased to +.39 when the total test score was defined as the sum of the twenty selected items. Obviously, there could be no change in the item-interview correlation as a result of limiting the number of items. Since the above set of twenty items was selected to correlate highly with the interview score and total test score only, it was necessary to select an additional set of twenty items which would correlate highly with the teacher rating score. Again using the data from all 145 subjects, correlations were computed between each item and the teacher rating score. These corre- lations are displayed in Table 12. As one can see, there were a considerable number of low positive and negative correlations. As a result, in order to select twenty items, items with correlations as low as +.O7 were selected. The items which were selected included items 13, 15, 17, 18, 19, 21, 22, 23, 27, 34, 36, 38, 39, 40, 41, 42, 44, 47, 48, and 49. Like the instrument above, this "new” instrument included sixteen of the thirty-two oral presentation items, and four of the eight pictorial items. Second Analysis — Discussion As was the case with the first analysis, there were several outcomes from the second analysis which warrant discussion. First, it was found that, as in the first analysis, the correlations between teacher rating score and total test score, and teacher rating score and interview score were quite low. The three possible explanations which were suggested in 86 TABLE 12 Correlations Between Each Item and the Teacher Rating Score: All Subjects Correlation With Teacher Item No. Rating Score 11 -.03 12 .03 13 .17 14 .02 15 .13 16 —.01 17 .13 18 .20 19 .21 20 -.05 21 .10 22 .13 23 .14 24 .04 25 .05 26 .05 27 .21 28 .06 29 -.05 3O .05 31 .04 32 .03 33 .05 34 .09 35 -.02 36 .10 37 -.01 38 .12 39 .07 40 .10 41 .14 42 .10 43 .04 44 .12 45 -.01 46 .00 47 .07 48 .14 49 .07 50 .06 87 the discussion following the first analysis are offered again as explana- tions for these relationships. The three possible explanations are: (1) different teachers used different anchor points for their ratings so that some teachers tended to rate high.and others to rate low, regard— less of the characteristics of the students rated; (2) the "attitude toward school" on which teachers based their ratings, was actually a different variable than that measured by the interview or the instrument; or (3) students' "faking" affected the interview and instrument response, but not the teacher rating. A second point which warrants discussion is the low correlations between each item and the teacher rating score. Despite these low correlations, it was decided to select the twenty "best" items to form an instrument. After selecting the items, the correlation between total score based on the twenty items and teacher rating score was computed. The correlation computed was +.24 which was considerably lower than +.39, the correlation between interview score and total test score. (NOTE: The total test score in each of these two correlations is a different value based on a different set of twenty items.) Obviously then, a more valid set of items could be selected when attitude towards school was defined by the interview and total test score than when attitude was defined by the teacher rating. A third outcome of the second analysis which warrants some discussion is the overlap between those items selected for their high correlation with the interview score and total test score, and those items selected for their high correlation with.the teacher rating score. There were eight items which.were selected to be in both sets of twenty items. They were items 13, 15, 19, 36, 40, 41, 48, and 49. (See Appendix A for a 88 copy of the oral presentation items, and see Appendix B for a copy of the pictorial items.) An examination of the content of these eight items offers no clue as to why they were selected for both instruments. A fourth point worthy of discussion is the relationship between the items selected from the first analysis and the items selected from the second analysis. Despite the fact that the first analysis was based on only seventy—three subjects and the second analysis was based on 145 subjects, there was actually very little difference between the sets of items which were selected. For both sets of items -- those selected for their high correlation with the interview score and total test score, and those selected for their high correlation with the teacher rating score —— the differences between items selected from the first analysis and those selected from the second analysis were a result of a change in a decision rule. When the first analysis was conducted, the decision was made to exclude all items of the pictorial form. As a result, for both sets of items, twenty oral presentation items were selected. Since no decision was made to exclude pictorial items when the second analysis was conducted, the sets of items were no longer exclusively oral presenta- tion items: four pictorial items were included in each set selected from the second analysis. However, the remaining sixteen oral presentation items selected for each set based on the results of the second analysis were also selected for each set based on the results of the first analysis. In other words, the only difference between the sets selected from the first analysis and the sets selected from the second analysis was the inclusion of four pictorial items in each of the two sets selected from the second analysis and the consequent drOp of four oral presentation items from each of the two sets selected from the first analysis. 89 Reliability Along with focusing on item selection, this research also measured the reliability of the various instruments. Using Hoyt's (1941) analysis of variance method for computing reliability, it was found that the reliability of the original forty item instrument was .93. As one would expect, this reliability decreased when the number of items was reduced from forty to twenty. Again using Hoyt's method of computing reliability, the reliability for the twenty items selected for their high correlation with the interview score and total test score was .87; the reliability for the twenty items selected for their high correlation with the teacher rating score was .80. Both of these reliabilities are considered quite adequate for an attitude assessment instrument. Supplementary Analysis Comparison Between Students Interviewed DuringgFirst Week and Students Interviewed During Third Week As described earlier, all students completed the instruments during the second week of the three week data gathering period. However, sixty- eight students were interviewed during the first week, i.e., before they completed the instruments, and seventy-seven students were interviewed during the third week, i.e., after they completed the instruments. Were there any differences between the results obtained from each of these groups? Separate correlations were computed for each of the two groups: first week interviewees and third week interviewees. The results are displayed in Table 13. When comparing the item—interview score 90 TABLE 13 Correlations for Those Interviewed During Week One and Those Interviewed During Week Three Item-Interview Score Item-Teacher Rating Score Item-Total Score Week 1 Week 3 Week 1 Week 3 Week 1 Week 3 ll .17 .26 .03 -.08 .61 .47 12 .32 .29 .06 -.Ol .63 .55 13 .25 .35 .16 .16 .58 .53 14 .02 .30 .09 -.O6 .48 .37 15 .18 .34 .13 .10 .69 .67 16 .08 .51 .04 -.09 .41 .42 17 .10 .13 .01 .22 .49 .42 18 —.16 .02 -.01 .38 .22 .25 19 .07 .27 .22 .19 .34 .63 20 .10 .07 .14 .07 .16 .16 21 —.01 .19 .15 .06 .41 .36 22 -.O7 .11 .15 .09 .44 .50 23 .22 -.02 .14 .15 .55 .18 24 .36 .42 .17 —.09 .72 .56 25 .23 .29 .03 .04 .57 .49 26 .26 .12 -.01 .11 .43 .27 27 .23 .07 .05 .32 .37 .38 28 .19 .45 .18 -.08 .63 .58 29 .04 .13 .01 .15 .45 .40 3O .19 .20 -.O9 .14 .43 .53 31 .34 .38 -.01 .05 .54 .67 32 .16 .30 .11 -.06 .60 .58 33 .03 .30 -.O6 .09 .40 .54 34 -014 .08 ‘006 019 048 037 35 .08 .02 -.01 -.O6 .45 .33 36 .18 .39 .05 .13 .66 .64 37 .15 .14 .02 -.O7 .74 .66 38 .11 .25 .10 .14 .43 .38 39 .28 .26 .22 -.09 .23 .42 40 .24 .32 .01 .15 .69 .59 41 .23 .23 .01 .25 .48 .67 42 .01 .13 -.10 .28 .34 .63 43 .13 .12 .01 .09 .35 .26 44 .35 -.02 -.O8 .30 .28 .21 as 007 .25 -003 -e01 058 .58 46 005 015 "e08 004 013 .51 47 .07 .11 -.13 .24 .53 .57 48 .19 .25 .15 .11 .36 .61 49 .40 .01 -.02 .15 .60 .34 50 .31 .16 .08 .05 .40 .58 Total Test .32 .45 .10 .16 91 correlations computed for those interviewed during week one with those interviewed during week three, one finds that for twenty-seven of the forty items, the correlations increased for those interviewed during the third week. The probability of this occurring by chance alone is less than .05. Thus, twenty—seven increased correlations out of a possible forty is statistically significant. Furthermore, the interview—total test score correlation increased from +.32 for those interviewed during the first week to +.45 for those interviewed during the third week. When one compares the item-teacher rating score correlations (Table 13) computed for those interviewed during the first week with the corre— 1ations computed for those interviewed during the third week, one finds that twenty—one correlations increased. Since one would expect, by chance alone, to find twenty increases, this is not statistically significant. Comparing the item-total test score correlations (Table 13) for those interviewed during week one with the correlations computed for those interviewed during week three, one finds that the correlations decrease for twenty-three items. This is not a statistically signifi— cant number of decreases. Two additional comparisons were.made between those interviewed during week one and those interviewed during week three. First, the total test score means (based on all forty items) were computed for both groups. The data, which are presented in Table 14, showed that the total test score mean was significantly higher for those interviewed during week one compared to those interviewed during week three. Second, the interview score means were computed for both groups. The data, which are presented in Table 15, showed that those interviewed during 92 TABLE 14 Total Score Means and Standard Deviations for Week One Interviewees and Week Three Interviewees: Forty Items Week Week One Three 68 77 Mean 148.28 135.86 Standard 27.51 28.13 Deviation t=2.69 d.f.=l43 dC<.Ol TABLE 15 Interview Score Means and Standard Deviations for Week One Interviewees and Week Three Interviewees Week Week One Three 68 77 Mean 13.71 14.82 Standard 3.06 3.11 Deviation t=2.15 d.f.=l43 £<.05 93 week three had a significantly higher mean score than those interviewed during week one. Thus, the students who were interviewed during week one scored significantly higher on the instruments but were assigned significantly lower scores from the interview. Discussion of Week 1 — Week 3 Comparisons There are at least two possible explanations for the increases in item—interview score correlations. One explanation suggests that either: (1) being interviewed before completing the instrument somehow results in lower item-interview score correlations; or (2) completing the instru- ment before being interviewed somehow results in higher item-interview score correlations. In other words, the process of measuring the atti— tude actually affects later measurements of the same attitude. The second explanation suggests that the increased correlations actually result from the improved scoring skills of the interviewers and tape scorers. The interviews were scored in approximately the same order in which they were administered; i.e., all of the interviews conducted during week one were completely scored before any of the interviews conducted during week three. It is conceivable that the interviewers and tape scorers were more skillful at scoring the third week's interviews after they had the experience of scoring the first week's interviews. A replication of this study with the interviews scored in random order should provide the data necessary to support or reject this hypothesis. As mentioned earlier, the students who were interviewed during week one scored significantly higher on the instruments and the stu- dents who were interviewed during week three were assigned significantly higher scores from the interview. This finding suggests that students 94 score higher on the second attitude measure -— regardless of which measure is used second. Analysis of Data From Students Not Interviewed As was described in Chapter III, many more students completed the instruments than were interviewed; every second grade student in a selected class completed the instrument, but only five or six students from each class were interviewed. There were 376 students who were not interviewed, but who had completed the instrument. Two analyses were conducted using the data from 334 of these studentsl: one analysis was based on the twenty items selected for their high correlation with the interview score and the total test score; the second analysis was based on the twenty items selected for their high correlation with the teacher rating score. In the first analysis, the dependent variable was the score on the twenty items selected for their high correlation with the interview score and the total test score. A two~way analysis of variance was computed: sex was one independent variable with male and female being the two levels; funding was the second independent variable with students in those schools which received State aid for culturally and economically deprived students (funded) and students in those schools which did not receive State aid for culturally and economically deprived students (non—funded) as the two levels. The means and standard deviations for the four cells are presented in Table 16. The results of the analysis presented in Table 17, indicate no significant interaction between sex l The sex of the remaining forty-two cases was unknown, and thus, they could not be classified for the analyses. 95 TABLE 16 Means and Standard Deviations For Each of Four Groups: Based on Twenty Items Selected for High Correlation With Interview Score and Total Test Score Funded Non—Funded N = 93 N = 71 ' Male i = 66.62 31' = 67.35 S.D. = 16.27 S.D. = 15.59 N = 85 N = 85 Female 3? = 71.56 31" = 71.51 S.D. = 15.84 S.D. = 15.55 TABLE 17 2 x 2 Analysis of Variance Comparing Means From Table 16 Sum of Mean Source Squares df Square F P Sex 1763.43 1 1763.43 7.0317 .008 Interaction 12.81 1 12.81 .0511 N.S. Error 82757.40 330 250°78 Total 84542.34 333 96 and funding. Likewise, main effects for funding are non-significant. However, a statistically significant difference (aC’ = .008) was discovered between the mean score for girls and the mean score for boys. The mean score for girls is significantly higher. In the second analysis, the dependent variable was the score on the twenty items selected for their high correlation with the teacher rating score. The means and standard deviations for the four cells are presented in Table 18. TABLE 18 Means and Standard Deviations For Each of Four Groups: Based on Twenty Items Selected For High Correlation With Teacher Rating Score Funded Non—Funded N -- 93 N = 71 Male '56 = 72.23 ‘1? = 73.52 S.D. . 14.48 S.D. . 13.25 N = 85 N = 85 Female 31' = 79.58 32’ = 77.82 S.D. = 11.36 S.D. = 11.34 A two—way analysis of variance was again computed with sex and funding as the two independent variables. The results, presented in Table 19, are consistent with the first analysis. No significant interaction 97 TABLE 19 2 x 2 Analysis of Variance Comparing Means From Table 18 Sum of Mean Source Squares df Square F P Funding 6.03 l 6.03 .0374 N.S. Sex 2918.93 1 2918.93 18.0983 .000 Interaction 192.11 1 192.11 1.1912 N.S Error 53222.40 §_Q. 161.28 Total 56339.47 333 between sex and funding is found as well as no significant main effects of funding. However, again, a statistically significant difference ( aCL-c .0001) was discovered between mean score for girls and mean score for boys. The mean score for girls is significantly higher. Thus, one can conclude from these analyses that, at the second grade, girls have a more positive attitude towards school than do boys. Furthermore, one can conclude that, at the second grade, students attending schools where there is a high concentration of culturally and economically deprived students do not differ in their attitudes toward school when compared with students who attend schools, in the same districts, where there is not a large percentage of deprived students. 98 Elm As was mentioned in the preceding section, data were available from 376 students who were not included in the instrument development sample. These data were used to develop norms. The characteristics of the norm group were presented in Chapter III under the heading "Sample." There, it was stated that the sample included thirty classes of second graders selected from thirteen schools in three districts in Southern Michigan. Since the students who were interviewed were randomly selected from this group, then it is reasonable to assume that those not interviewed, i.e., the students in the norm group, were also a random subset. The students in half of the classes attend schools which, in 1969-70, received State aid money for having a large proportion of culturally and economically disadvantaged students. These fifteen classes have a very large percentage of Negro children. Since the results reported in the preceding section indicated no main effects for funding but did show.main effects for sex, it was decided to develOp separate norm tables for boys and separate norm tables for girls, but to group the students in funded and non—funded classes together. The norms which are presented below are considered only tentative as they are based on a relatively small number of cases, and hence, the values are unlikely to be very stable. The norm tables for girls are based on data from 170 students and the norm tables for boys are based on data from 164 students; however, the norm tables for boys and girls combined are based on data from all 376 students, as the fortyétwo students whose sex was unknown to the researcher could be included only 99 with this group. Because of the small number of cases and hence the inferred instability of the norms, only selected normative points are given in the norm tables. Further evidence of the tentativeness of the norms is provided by the fact that these norms were developed for the sets of twenty items when the items were administered in the context of a total of forty items. How the extra twenty items affect responses on the set of items normed is not known. Table Despite these cautions, two sets of norms were developed. 20 displays the norms for the scores from the set of twenty items which 'were selected for their high correlation with the interview score and total test score. Table 21 displays the norms for the scores from the set of twenty items which were selected for their high correlation with the teacher rating score. Summary The analysis of the data was organized in three stages: preliminary analysis, item selection, and supplementary analysis. The preliminary eartalysis, which involved studying students' responses to the sample items Eirui interviewing members of the research team, indicated no significant Problems: it was reasonable to proceed with the analysis. The first item selection analysis was based on data from approxi— lanaitZely half the subjects (73 out of 145). The remaining data were set aside to be used in cross-validation. Early analysis indicated that ‘:"C> subsets of items should be identified: one subset to correlate ' ‘V‘ildl with the interview score and total test score and a second subset t:<> «correlate well with the teacher rating score. This conclusion 100 TABLE 20 Norms for Twenty Items Selected for Their High Correlation With Interview Score and Total Test Score R A W S C O R E Percentile Rank Girls Boys Boys & Girls 1 37 29 3O 5 45 41 43 10 49 47 48 20 57 53 54 25 61 56 57 30 - 62 57 6O 40 67 64 64 50 72 67 69 60 77 7O 74 7O 81 75 78 75 84 78 81 80 86 81 84 90 9O 90 9O 95 96 94 95 99 100 98 99 101 TABLE 21 Norms for Twenty Items Selected for Their High Correlation With Teacher Rating Score R A W S C O R E Percentile Rank Girls Boys Boys & Girls 1 41 33 36 5 6O 46 51 10 64 55 58 20 7O 62 65 25 72 65 67 30 73 67 7O 40 77 71 73 50 80 74 77 6O 82 76 8O 7O 85 8O 83 75 87 82 85 80 88 84 87 90 92 91 92 95 95 94 95 99 98 97 98 102 resulted from the low correlation obtained between the interview score and teacher rating score (+.23). After the two subsets were selected, an attempt was made to cross-validate the selections using the data from the remaining subjects. The cross-validation was judged unsuccessful. A reanalysis of the data was conducted using the entire sample of 145 subjects. By using the larger sample size, the influence of random error would be reduced and thus, the results would be more stable. Again two sets of items were selected: one set to correlate highly with inter— view score and total test score and one set to correlate highly with teacher rating score. The only difference between the sets of items selected from the first selection procedures and those selected from the asecond selection procedures was the presence of pictorial style items in tihe latter sets. This, however, was the result of a change in the Inesearcher's decision rule and not a result of the analyses outcomes. Using Hoyt's (1941) analysis of variance method, the reliability for: the twenty items selected for their high correlation with interview SCtxre and total test score was computed to be .87; the reliability for the: twenty items selected for their high correlation with teacher rating was computed to be .80. The supplementary analysis included a comparison of the data from those: interviewed during the first week of the study and those inter— viewed during the third week of the study. It was found that twenty— seven.c&'the item-interview score correlations were higher for those interviewed during the third week. It was also found that students interviewed during the first week scored significantly higher on the instruments but were assigned significantly lower scores from the interview. 103 Additional supplementary analyses were conducted comparing those students who were not interviewed. A 2 x 2 analysis of variance was computed with male-female as one independent variable and funded — non— funded as the second independent variable. Regardless of which of the two sets of twenty items were used as the dependent variable, the follow— ing conclusions were reached: (1) no significant interaction; (2) no main effects from funding; and (3) significant main effects from sex, with girls scoring higher than boys. The data used for the supplementary analyses were also used to develop tentative norms for the instrument. Separate norms were computed for boys and for girls. CHAPTER V SUMMARY AND CONCLUSIONS Summary Purpose This study was designed to examine the feasibility of an attitude assessment procedure appropriate for measuring young children's attitudes towards school. The rationale for the project was based on three assump— tions: (1) attitudes are important; hence (2) attitude assessment is important; but (3) present methods are inadequate for attitude assess- ment, especially assessment of groups of students in early elementary grades. The basic objective of this study was stated in terms of an hypothesis: the technique used in this study can yield a valid and reliable group administrable instrument for assessing the attitudes of students in early elementary grades. Evidence was presented in Chapter I to demonstrate the validity of each assumption and the reasonableness of the hypothesis. IReview of the Literature There is an enormous volume of literature relating to the study of attitudes. Certain aspects of the literature, particularly relevant to this project, were selected for review. It was felt that an awareness of the complexities involved in defining the term "attitude" contributes to an appreciation of the problems involved in conducting attitudinal 104 105 research. Therefore, considerable attention was given to reviewing and classifying the published definitions. The definition given by Shaw and Wright in their 1967 book, Scales For The Measurement Of Attitudes, was identified as the definition for this research. Their definition states: attitude is best viewed as a set of affective reactions toward the attitude object, derived from the concepts or beliefs that the individual has concerning the object, and predisposing the individual to behave in a certain manner toward the attitude object. Although intimately related to attitude, neither the propositions that the individual accepts about the object (beliefs) nor the action tendencies are a part of the attitude itself [p. 13]. In addition to reviewing the definitions of attitude, certain aspects of attitude theory were reviewed. However, no attempt was made to even mention all of the various aspects of attitude theory nor to review any one aspect in great depth. Those aspects which were discussed included: (1) the relationship between attitudes and behavior; (2) the relationship between attitudes and experience; (3) the components of attitudes; and (4) the characteristics of attitudes. Because the focus of this study was attitude assessment, a compre- hensive review of existing measurement procedures was presented. The review was organized around a four-category system for classifying atti— tude assessment techniques. The categories include: (1) disguised — sstructured techniques; (2) disguised — non-structured techniques; (3) non—disguised - non-structured techniques; and (4) non—disguised — structured techniques . The final portion of the literature review discussed specific sstudies which are related to this project. Nine such projects were. Ciescribed. 106 Design The specific attitude assessment procedure with which this study has been concerned involves oral and pictorial presentation of items with pictorial presentation of response choices. The instrument is very similar to a Likert scale in that there are five response choices rang— ing from very positive through neutral to very negative. However, whereas the typical Likert scale presents written response choices, this instru— ment, being especially designed for young children, presents the response choices pictorially. Five faces ranging in facial expression from very happy through neutral to very sad serve as the response choices. In order to test the procedure and develop an instrument, a group of forty items, designed to measure attitude towards school, was admin— istered to all second graders in thirteen schools in three districts in Southern Michigan. Since the classroom teachers might have inadvertently biased their students' responses, college coeds were hired and trained to conduct the instrument administration. One hundred forty-five of the students who completed the instru- ments were randomly selected for the instrument development sample, and two types of additional data were gathered about this subset of children. First, each child was individually interviewed by a member of the research team. The purpose of the interview was to ascertain the child's attitude towards school and thereby provide construct validation for the instrument responses. Second, each child in the instrument develop— ment sample was rated by his teacher to indicate the teacher's percep— tions of the child's attitude towards school. The information provided by the interview and teacher rating was gathered to help select the most "valid" subset of items from the forty items initially included in the instrument. 107 Three hundred seventy—five students completed the instrument but were not part of the instrument development sample. The data from these students were gathered to enable the researcher to compare specific groups (e.g., male—female) in terms of their responses to the instrument. Results The data analysis consisted of three phases: (1) preliminary analysis; (2) item selection; and (3) supplementary analysis. The preliminary analysis included an examination of the students' responses to two of the sample items included at the beginning of the instrument, and an individual conference with each member of the research team. From studying the responses to the sample items, it was concluded that the students were able to comprehend the directions which they were to follow; i.e., they appeared to be able to communicate by marking pictures. The conferences with the members of the research team failed to indicate any unexpected problems which would distort the data. The analyses related to item selection were actually conducted twice. The procedures followed in the two sets of analyses were iden~ tical. However, the results of the first analyses, which were based on seventy-three of the 145 subjects in the instrument development sample, were not stable when cross~validated on data from the remaining seventy“ two subjects; i.e., the correlations on which the item selection was based were not maintained on crOSdealidation. Therefore, the analyses were repeated combining the data from all 145 subjects. This was done to reduce the influence of random error in the correlations used for item selection, and thus increase the likelihood that the magnitude of the item selection correlations would be maintained on subsequent cross— validations. The first step in the analyses was to determine whether a 108 single set of items could be selected which would correlate well with the interview score, the teacher rating score, and the total test score. Low correlations between teacher rating score and each of the other two measures suggested that two sets of items should be selected: one set to correlate well with teacher rating score and a second set to correlate well with the interview score and total test score. After the two sets were selected, Hoyt's (1941) analysis of variance method was used to compute the reliability of each. The twenty items which correlated well with the teacher rating score had a reliability of .80; the twenty items which correlated well with the interview score and tOtal test score had a reliability of .87. The supplementary analysis included a comparison of the data from those interviewed during the first week of the study and those inter- viewed during the third week of the study. It was found that twenty- seven of the item-interview score correlations were higher for those interviewed during the third week. It was also found that students interviewed during the first week scored significantly higher on the instruments but were assigned significantly lower scores from the interview. Additional supplementary analyses were conducted comparing those students who were not interviewed. A 2 x 2 analysis of variance was computed with male-female as one indpendent variable and funded - non— funded as the second independent variable. Regardless of which of the two sets of twenty items were used as the dependent variable, the follow— ing conclusions were reached: (1) no significant interaction; (2) no main effects from funding; and (3) significant main effects from sex, with girls scoring higher than boys. 109 Using the data from the students who were not interviewed, two sets of norm tables were constructed -— one set for the twenty items which correlated highly with interview score and total test score and another set for the twenty items which correlated highly with teacher rating score. Conclusions Advantages and Limitations of the General Procedure Since the instrument developed in this project is quite similar to a Likert scale, it possesses many of the advantages and limitations of a Likert scale. The advantages of these scales include their ease of construction, administration, scoring, and interpretation. Unfortunately, these advantages become liabilities when they are abused. Because a Likert scale is easy to construct, some research is conducted with inadequately developed and inadequately tested Likert scales. The same danger of misuse exists with the type of instrument developed in this project. Like the traditional Likert scale, there are two problems in inter— preting scores for the school attitude instrument. First, one individual's score alone is meaningless. Scores are interpretable only if normative data are provided. Of course, this does not preclude using the scores to compare groups of respondents. A second problem in interpreting scores arises when a respondent's total score is equal to the neutral category multiplied by the number of items in the scale. Caution must be exercised in interpreting this score as representing a neutral atti— tude since it could be obtained if the respondent selected the very 110 positive response for half of the items and the very negative response for the other half of the items, as well as if the respondent had always chosen the "neutral" category. There is an additional limitation to the instrument developed here which does not apply to traditional Likert scales. The generalizability of the procedure is limited to those situations in which the response choices can be expressed by varying facial expressions. For example, whenever the item asks, "How would you feel if . . . ?" or "How do you feel when . . . ?" faces can probably be used as the response choices. However, if the item were to ask, for example, "How often do you . . . ?" or "Do you agree or disagree with the statement that . . . ?" there would be no way to represent the response choices using varying facial expressions. Use of the Instruments Although these instruments are still in the early stages of develOp— ment, there are likely to be individuals who wish to use them. However, several cautions are in order for those who do intend to use the instru— ments. The first investigation of the procedure could not possibly study all aspects of the instruments. As a result, some relevant data are still unavailable. For example, the only available reliability figures for the instruments were based on internal consistency; no test—retest reliability figures are available. Only construct validity was inves— tigated; predictive and concurrent validity were not considered. No investigations were made to relate responses on the instruments to behavior. Although norms were develOped for each of the instruments, they are, at best, tentative norms since they are based on a relatively small 111 number of cases. Furthermore, these norms were developed for the sets of twenty items when the items were administered in the context of a total of forty items. How the extra twenty items affect responses on the set of items normed is not known. Until better norms are developed, it would be better to restrict the use of the instrument to comparing groups of students rather than concluding anything about individual students. Since development of the instrument was based on college coeds administering the instruments, the findings in this report do not apply if the classroom teacher administers the instrument. There is reason to suspect that if teachers were to administer the instruments, they might inadvertently influence their students' responses. However, until this area is studied, one cannot predict, and hence cannot compen— sate for, the influence the teacher would have. Implications for Future Research It was stated in Chapter II that there was a reciprocal relation— ship between this project and existing attitude theory: the theory contributed to the design of the procedure and the procedure can provide a means for furthering theory. Since it has been demonstrated that the procedure is a viable means for assessing the attitudes of young children, and since this procedure is less time-consuming and less expensive than the techniques of individual measurement used previously, it is hoped that this project will stimulate new research in the area of children's attitudes. Furthermore, it is h0ped that this project stimulates educators to investigate.measuring affective outcomes of education. Previously, the excuse of no adequate instrumentation was given when 112 educators chose to ignore the non-cognitive outcomes of education. However, this excuse is no longer valid. In addition to using this procedure to conduct additional research relating to the field of attitudes in general, attention should be given to conducting additional research to improve the quality of the instru— ment. For example, a member of the research team suggested that children differentiate between the social and academic aspects of school. Perhaps, two different instruments should be develOped -- one to measure attitude towards the academic aspects of school and a second to measure attitude towards the social aspects of school. Furthermore, additional research is needed to ascertain the effects of the faces' sex and race on the respondent's answers. This project held constant the effects of sex and race by using drawings of faces having no sexual nor racial characteris— tics. However, it is possible that a better instrument could be developed if the sex and race of the faces matched the sex and race of the respondent. Overall Conclusion Several aspects of this study provide support for an optimistic overall conclusion. The positive correlations between the selected items and the interview score (+.39) and between the selected items and the teacher rating score (+.24) provide construct validation for the items. The high reliabilities which were obtained (.80 and .87) suggest that an underlying construct is being reliably measured. Therefore, what is the general conclusion which can be drawn from this study? Although it is tempting to state that a group administrable instrument has been developed which measures early elementary children's attitudes towards school, caution must be exercised in forming such a sweeping conclusion. After all, it was shown in Chapter II that social 113 psychologists have yet to agree on even the definition of attitude. Thus, this researcher would modify the conclusion to state: this project developed a group administrable instrument for measuring atti— tude towards school which functions well with early elementary children. Stated another way, drawings of human faces provide a viable means for presenting the response choices for an attitude scale. B IBLIOGRAPHY BIBLIOGRAPHY Allport, G. W. Attitudes. In C. Murchison (Ed.), Handbook of social ,psychology. Vol. II. New York: Russell, 1935. Pp. 798—844. Allport, G. W. Historical background of modern social psychology. In G. Lindzey (Ed.), Handbook of social psychology. Vol. 1. Theory and method. Cambridge, Mass.: Addison-Wesley, 1954. Pp. 3-56. Allport, G. W. Attitudes in the history of social psychology. In M. Jahoda & N. Warren (Eds.), Attitudes: Selected readings. Middlesex, England: Penguin Books, 1966. Pp. 15—21. Ashman, G. R. Attitudes toward Negroes as reflected in the picture preferences of institutionalized white morons. Unpublished master's thesis, Michigan State College, 1953. Bain, R. Theory and measurement of attitudes and Opinions. Psychological Bulletin, 1930, 27, 357-379. Bartlett, C. J., Quay, L. C.,'& Wrightsman, L. S. A comparison of two methods of attitude measurement: Likert—type and forced choice. Education and Psycholpgical Measurement, 1960, 20, 699-704. Biber, B., & Lewis, C. An experimental study of what young children expect from their teachers. Genetic Psycholpgy Monographs, 1949, 40, 7—97. Brown, J. F. A modification of the Rosenzweig picture—frustration test to study hostile interracial attitudes. Journal of Psychology, 1947, 24, Campbell, D. T. The indirect assessment of social attitudes. Psychological Bulletin, 1950, 47, 15-38. Campbell, D. T., & Fiske, D. W. Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bplletin, 1959, 56, 81-105. Campbell, D. T. Recommendations for APA test standards regarding construct, trait, or discriminant validity. American Psyghologist, 1960, 15, Campbell, D. T. Social attitudes and other acquired behavioral dispositions. In S. Koch (Ed.), Psychology: A study of a science. Vol. 6. Investi— gations of man as socius. New York: McGraw—Hill, 1963. Pp. 94-172. 114 115 Cantril, H. (Ed.) Gauging public opinion. Princeton, New Jersey: Princeton University Press, 1944. Cantril, H. The intensity of an attitude. Journal of Abnormal Social Psychology, 1946, 41, 129-135. Carter, H. C. Measurement of attitudes toward school. California Journal of Research, 1969, 20, 186-192. Cohen, S. R. An exploratory study of student attitudes in the primary grades. Unpublished manuscript, Educational Testing Service, 1965. Cohen, S. R. An exploratory study of young children's attitudes toward school. Proceedings of the Seventyrfifth Annual Convention of the American Psychological Association. Lancaster, Pa.: American Psychological Association, 1967. Conrad, H. 8. Some principles of attitude measurement: A reply to "Opinion-attitude methodology." Psychological Bulletin, 1946, 43, 570-589. Cook, S. W., & Selltiz, C. A. A multiple-indicator approach to attitude measurement. Psychological Bulletin, 1964, 62, 36-55. Cooper, J. B., & Pollock, D. The identification of prejudicial attitudes by the Galvanic skin response. Journal of Social Psychology, 1959, Day, D. Methods in attitude research. American Sociological Review, 1940, 5, 395-410. DeFleur, M. L., & Westie, F. R. Verbal attitudes and overt acts: An experiment on the salience of attitudes. American Sociological Review, 1958, 23, 667-673. Diab, L. N. Some limitations of existing scales in measurement of social attitudes. Psychological Reports, 1965, 17, 427-430. Diab, L. N. Measurement of social attitudes: Problems and prospects. In C. W. Sherif & M. Sherif (Eds.), Attitude;ego—involvementL and change. New York: John Wiley & Sons, Inc., 1967. Pp. 140—158. Doob, L. W. The behavior of attitudes. Psychological Review, 1947, 54, Dorkey, M., & Amen, E. W. A continuation study of anxiety reactions in young children by means of a projective technique. Genetic Psychology Monographs, 1947, 35, 139-183. Dysinger, D. W. When do I smile? Unpublished, 1970. Edwards, A. L. Techniques of attitude scale construction. New York: Appleton-Century-Crofts, Inc., 1957. 116 Edwards, A. L., & Kenney, K. C. A comparison of the Thurstone and Likert techniques of attitude scale construction. In M. Fishbein (Ed.), Attitude theoryyand measurement. New York: John Wiley & Sons, Inc., 1967. Pp. 249-256. English, H. B., & English, A. C. A comprehensive dictionarygof psychologr ical and psychoanalytical terms. New York: David McKay Company, Inc., 1958. Evans, K. M. Attitudes and interests in education. London: Routledge & Kegan Paul, 1965. Ferguson, L. W. The requirements of an adequate attitude scale. Psycho— logical Bulletin, 1939, 36, 665-673. Ferguson, L. W. A study of the Likert technique of attitude scale construc— tion. Journal of Social Psychology, 1941, 13, 51-57. Fishbein, M., & Raven, B. H. The AB scales: An operational definition of belief and attitude. In M. Fishbein (Ed.), Attitude theory and measurement. New York: John Wiley & Sons, Inc., 1967. Pp. 183-189. Fox, R., Luszki, M., & Schmuck, R. Diagnosingiclassroom learning_environ— ments. Chicago: Science Research Associates, Inc., 1966. Fromme, A. On the use of certain qualitative methods of attitude research: A study of Opinions on the methods of preventing war. Journal of Social Psychology, 1941, 13, 429-459. Frymier, J. R. Personal Correspondence. January 30, 1970. Green, B. F. Attitude measurement. In G. Lindzey (Ed.), Handbook of social psychology, Vol. 1. Theory and method. Cambridge, Mass.: Addison—Wesley, 1954. Pp. 335-369. Greenberg, J. W., Gerver, J. M., Chall, J., & Davidson, H. H. Attitudes of children from a deprived environment towards achievement—related concepts. The Journal of Educational Research, 1965, 59 (2), 57—62. Greenwald, A. G. On defining attitude and attitude theory. In A. G. Greenwald & T. C. Brock (Eds.), Psychological foundations of atti— tudes. New York: Academic Press, 1968. Pp. 361-390. Harrison, R. P. Pictic analysis: Toward a vocabulary and syntax for the pictorial code, with research on facial communication. Unpub— lished doctoral dissertation, Michigan State University, 1964. Hartley, E. L., & Hartley, R. E. Fundamentals of social psychology. New York: Knopf, 1952. ‘ Hess, E. Attitude and pupil size. Scientific American, 1965, 212, 46—54. Hildreth, H. G., Griffiths, N. L., & McGauvran, M. E. Manual of direc— tions: Metropolitan readiness tests. New York: Harcourt, Brace & World, Inc., 1969. 117 Hollander, E. P. Principles and methods of social psychology. New York: Oxford University Press, 1967. Horowitz, E. L. The development of attitude toward the Negro. Archives of Psychology, 1936, No. 194. Horowitz, E. L., & Horowitz, R. E. Development of social attitudes in children. Sociometry, 1938, 1, 301—338. Horowitz, R., & Murphy, L. B. Projective methods in the psychological study of children. Journal of Experimental Education, 1938, 7, Hovland, C. 1., Janis, I. L., & Kelley, H. H. Communication and persua- sion. New Haven: Yale University Press, 1953. Hoyt, C. Test reliability estimated by analysis of variance. Psychometrika, 1941, 6, 153-160. Insko, C. A. Theories of attitude change. New York: Appleton-Century- Crofts, 1967. Jahoda, M., Deutsch, M., & Cook, S. Research methods in social relations. Part 1: Basicgprocesses. New York: Dryden Press, 1951. Jordan, J. Guttman facet design and development of a cross cultural atti— tudes towards mentally retarded persons scale. Unpublished, Sept., 1969. Katz, D. Do interviewers bias poll results? Public Opinion Quarterly, 1942, 6, 248-268. Katz, D., & Stotland, E. A preliminary statement to a theory of attitude structure and change. In S. Koch (Ed.), Psychology: A study of a science. Vol. 3. New York: McGraw—Hill, 1959. Pp. 423—475. Krech, D., Crutchfield, R. 8., & Ballachey, E. L. Individual in society. New York: ‘McGraw—Hill, 1962. Kuder, G. F., & Paulson, B. B. Exploring_ghi1dren's interests. Chicago: Science Research Associates, 1951. Leeds, C. H. A scale for measuring teacher—pupil attitudes and teacher— pupil rapport. Psychological Monographs, 1950, No. 312. Likert, R. A technique for the measurement of attitudes. Archives of Psychology, 1932, No. 140. Link, H. C., & Freiberg, A. D. The problem of validity versus reliability in public opinion polls. Public Opinion Quarterly, 1942, 6, 87-98. Malpass, L. F. Some relationships between students' perceptions of school and their achievement. Journal of Educational Psychology, 1953, 44, 118 McNemar, Q. Opinion-attitude methodology. Psychological Bulletin, 1946, 43, 289-374. Mehrens, W. A., & Lehmann, I. J. Standardized tests in education. New York: Holt, Rinehart and Winston, Inc., 1969. Nelson, E. Attitudes. Journal of General Psychology, 1939, 21, 367—436. Proshansky, H. M. A projective method for the study of attitudes. Journal of Abnormal Social Psychology, 1943, 38, 393-395. Remmers, H. H. Introduction to opinion and attitude measurement. New York: Harper and Brothers, 1954. Rice, S. A. Stereotypes: A source of error in judging human character. The Journal of Personnel Research, 1926-7, 5, 267-276. Rokeach, M. Beliefs, attitudes and values. San Francisco: Jossey-Bass Inc., 1968. Rosenberg, M. Cognitive structure and attitudinal affect. Journal of Abnormal and Social Psychology, 1956, 53, 367-372. Scott, W. A. Attitude measurement. In G. Lindzey (Ed.), Handbook of social psychology. Vol. II. Research methods. Cambridge, Mass.: Addison-Wesley, 1968. Pp. 204-274. Shaw, M. E., & Wright, J. M. Scales for the measurement of attitudes. New York: McGraw-Hill Book Co., 1967. Sherif, C. W., Sherif, M., & Nebergall, R. E. Attitude and attitude change. Philadelphia: Saunders, 1965. Sherman, M. Theories and measurement of attitudes. Child Development, 1932, 3, 15-28. Stouffer, S. A. (Ed.), Measurement andgprediction. Princeton: Princeton University Press, 1950. Thurstone, L. L. Attitudes can be measured. American Journal of Sociology, 1928, 33, 529-554. Thurstone, L. L., & Chave, E. J. The measurement of attitude. Chicago: University of Chicago Press, 1929. ‘Yarrow, M. The measurement of children's attitudes and values. In P. H. Mussen (Ed.), Handbook of research methods in child development. New York: Wiley & Sons, 1960. Pp. 645-687. APPENDICES ll. 12. l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 119 Appendix A ORAL ITEMS How do you feel when you are sitting in your seat in school? How would you feel if someone told you that school would end forever tomorrow, and you would never go to school again? How would you feel if you were at home talking to your family and someone asked you to tell them what you did at school today? How would you feel if the school rules were changed, so that you went to school everyday, including Saturday and Sunday? How do you feel when you get up in the morning and get dressed to come to school? How would you feel if your teacher told you there would be no more reading-classes? How do you feel when you are sitting at your seat doing some work and the room is very quiet? How would you feel if your teacher said you could take your books home from school today? How do you feel when your teacher calls your name, asks you a question and waits for you to answer? How would you feel if your teacher gave you directions for some work, you started to work, and then the student next to you began talking to you? How would you feel about being a school teacher when you grow up? How would you feel if your teacher moved away and you got a new teacher? How would you feel if you were in the grocery store, and your teacher walked in and said "hello" to you? How would you feel if your mother said you looked sick, and you had to stay home from school? How do you feel when you are doing your reading? How would you feel if you were in the classroom with your teacher, none of the other boys and girls were in the room, and she was talking to you? 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 120 Appendix A - Continued How would you feel if your teacher invited you over to her house? How would you feel if the school day were changed, and you only went to school in the morning instead of both the morning and the afternoon? How would you feel if someone told you you had to go to school this summer? How would you feel if your teacher said you could take your school work home? How do you feel when you are doing your math? How would you feel if tomorrow were a holiday and there were no school? How do you feel when you are sitting at your seat, and the teacher is in front of the room talking to all the boys and girls? How would you feel if you had to go to a different school and you could not attend school any more? How do you feel when you are doing your school work at home? How would you feel if there were no schools in ,(city) ? How do you feel when you come to school in the morning? How would you feel if you had a schoolroom at home? How would you feel if you stayed in the room with the teacher when the other boys and girls left? How would you feel if your teacher said there would be no more classes in math? How would you feel if your teacher were coming over to your house tomorrow? How would you feel if your mother were coming to school this afternoon? " E 44 'o I: M a) {-4 <1 H V «O @718 V N ‘ ll ~ II ____—__————— / l“) "1‘ ‘ m II Q N N11 . OJ ('7 'U (D :3 G H u G O U I CO c '0 fl necyI-nnu. .U-IAJNIV it v- ‘u..-.v-:EU.:HAH< — 1.4.5 .. . In N | Q N N Q G) I) D V N 'C ' D '- D I) I) O F O oonoausoo I m xfiocomm< U) N N t M I v ' N N ' owdaflufioo I m 593%? 143 Q 0N ' Q N oosafiuaoo 1 m Nfionmmn< 126 “I Q .— C) V I) N .. N ‘ v N N V oosafiusou 1 m Naofioam< v N a v» u v 0 70 v N _ _Nn N v n en ownsfiucou I m Narcoom< 128 .- 0 V N N I) h» ' M N .— fl an oosafluaoo I m Naraoma¢ 129 n NV *w C N ow fl an omsoausoo 1 m xfioamoo< 130 posoausou I m xaoaoaa< 1.5.5 possauooo 1 m swosmoo< 134 : 71s.: ' s 4 3 2 1 136 a1 h. _Um< cmssfiuaoo 1 m savaaaa< 138 .uosuwwou Bonn ussou Haas ms was ammo Ou usflom .3ou osmm onu so moomm o>Hw mum Hosea sea «0 ousuofio on» 0» uxoz .HoeNo men on wsausfioo on Ham oasonm so» .Amz men on uofloo oaomuo>m .uoaxoon “30% so mono onu oxaa umsfl mum zone sumaxoon >8 so mmomw pom mousuofio onu wow 30% on .xaasmmnmo soumwa 302 0°“ 50% Hamu H Hausa woman was upon no .uoaxoon Mao» so muwu3 no: on .%m3 woos as on Hafi3 mascuoo ow mxmmo usom mmo Mamas ommoam .30» ou uso mmmm on wsfiom an H umsu mum: mousuofia mo umaxoon m o>ma H A .ooHp .Hsmaouuom mxooa pH .moomzs: wmm> pom pom wmm> mxooa momm poo mHnH .oosm xwwww sew 0H Hesse .pofiuuos .uomo: Ho .ooanoouu mauuwa m . mxooa uH .wsfisuoaom odors %aom£so me u“ oxfia mxooa oomm axon one .uodm swedew oxw 0H Nixon .pmm muo> nos homes huo> won me on sons mxooa osomsom oxHH mxooa uH .omm was home: soosuoaIsH ma uH .umsufim wsfiosoum uos ma ua use mawaaam uos ma comm masses may I .uosm mdwdw mew ow Hesse .oomw oasmomuwm .pmauwaumm .pme m oxwa mxooa pH .oafiam oommmoao 6 mm: oomm osooom mafia .uodm orgasm uxw 0H stom .oomm momma .HSmuowso .233 m 3:: 963 S .625 .392 m3 6 was 38 $43 25 88% 43% New 3 “Ewes .onmem uaouomm«w m macaw meow Loom .mmomm o>am mamm osu poem HHH3 50% Bow some sH .momao Hock does so% damn ou “sou Sumo pom uaoummwao mum monouUfin may .onsswdwn uwdefiodwqw ow Howmooc .moomw mom mououoflm mo msou nuaB moaaam ow umaxoon may sowwdewqxo§uo we women smSOesw ANNA .33 .85 3.858 54.3. 3 $3 .8: .8838. 34 94.4 .mimuusomsmmufiuofiumso segue Joew :63 3 H58. {833% 38.5888 N3... $.14 833.3. as 8 H58 . a8 53% :33 sHSQgH omaowucoo I o Narcooa< 140 .Am>omm may no mHDuUHm msu Sufi3 mcflwmn nowas .3ou uxwa mSu Hmwaa meuma_na0h wQHHw .aowunmsv uan msu Hon mvmmu mum m3 302 .mumw Hamuwmno umoa .ummfimama .ummuawuuo man so cmxuma sump mm: :x: cm om .maafim momma .an msu nufis moan msu .mmw .Hmmm vase? oamoom uuoa‘soa y’L’}¥’rrr‘y>y’yr’y’7t 28% $03 3: mfiimo €23 .wmfiu mo 36 ”SH 93. E Ems dam 58mm $0: on son 25 838% «E. .23 clean a 3 SE .mnfiavmxaa umafi t 30H “mafia mag Hmva: uH woman cam umaxoon “50% Eoum meu a Soaamh use» m>oawu 302 .umcumwou macm wafiow %Hu m.umA .Hmmm cases 90% 30: ummn w3osm umnu wowm mac m£u mmouum :x: mm Spas waB_amnu wcm .GOfiummsv mnu usonm xawnu ou mum so» .woamamms mafiauoaom ma Hmow wasoa so» 30: unonm mcowummnd meow 30% xmw Cu wawow an H 302 vmsaauaoo I o Nauammm< 33 32:84 383 SE §m§ 23% <35»: 55m 33‘: 34 :3 38.3 on 3325 :ogqsofifiu Sam m§§ ma2§§ .N 3m=§m .33; .583: RE 33%. 39358 3 §§ 58 3w 3: 3 £85 .33 $123 315% 3% 533 5&3: 55m 33m Jan 53 33.3. .Eeofiofia § :5. ms 35 141 puma uoa can wanna uoa mu moan scans .vmm Hmmm u.:vav 50% van hmama ammm u.awaw :0» ma Hmmm manna so» so: ummn msoam umcu 00mm mSu no :N: am mxmz_ .hMHmo zoz .Hmnumwou whoa mac ow m.umq .xuma.ou mmumm no 309 unwfiu msu vafim no» mam: ou haao mumnu mum hose .90% 3mm ou waHow am H :oaummav wnu saga ow ou waasuoc m>ms mmnsuofia mmmnu .Hmnauaum .50» xmd H coaummav mzu uaonm Hmmm 50% so: maocm umnu comm onu mmouum :N: man a oxma,ou mum :0» 30M comm a« van Maw H acaummav wnu on haaammuuo amumHH ou and new .mamm mnu mud mmumm mnu umaxoon can udoam50Hnu 30H comm aH .Hmwm 6H503.mmw.303 xuma umafi vanosm no» .muozmcw waoua can unwwu on mum mum:H .maous uoa mu uH .vmn umnu Hmmm u.avasos 30% mmswumn .m00Mw nonuo mnu mo mac vmxuma so» ma .Hw>m30m .vmannouu cam knamaus .uusn .cmm hum> mxooa umnu momm cam onu wmxuma hanmnoum :05 mo umoz wufi ca Hm>05m mnu nu“? 3cm mnu aw menu «so no :x: cm pan :0» can .Hmmm vaaoz so» 30: ummn macaw umnu moan mnu :0 :N: man a xuma,mammw cases 50% 3o: no xdany wwma Mbow MMomm 02¢ 44mm :0» MH «Ammm DOM 94:03 30m ma Gowumosa uch mnH vmsafiucoo I o Kawamaa< .3§Howm E E58 :3 8 so; $3.5 .5§ 3 3.931 3R853 33 g :03 3H 34 3 £35 142 .Homm mHsoa 30% so: mzonm umzu oomm ozo xuma mam hHHamoumo amumHH Soz woomHn oawfiu oau opmn maohuo>o moon .mmmy mnu Sofia maHmop umnu mmomm mo 30H onu Roma: Hoxuma.uzo% oomHm fl.HHv .mwmm ono dado 3oz .umvna uoxuma u:o% nun ou manuon Soaps 30% HHou HHHS H .oomHm Mach moox so» mHmn cu mH mHna .uaoumHMHv mum Boa comm mo oumum onu um mounuoaa 05H .mGOHooouHm onHom mam hHHSHoumo amumHH .muosmam mucus no uanu on mum ouona .uonawwoa Mao» mm oomm mama man Jame u.aom 90% «H hypo? o.aov om .aoaumosv huo>o uaonm mms mamm osu Homm HHH3 ocomuo>o ooz .3ou nomm aw oomm mac HHao Mama.ou ohm 50h Honamamm .Hoom no» so: ozonm omSu oomm onu do =N: man m oxmz. .Hoow vHaoB so» 30: m3onm umon umnu.MMMMImmm onu xuma coco mam aOHumosu ono osonm xaHSH .moaomamn waHnooaom «H Hoom anoz 90% 30a uaonm maOHu Imoav meow 50% xmm HHH3 H .ouoa mono macauoouwv onu omomou ma.uoH menu ow HHHB o3 Bo: usonm maOHumoav ham mumnu mu< .mmm no: mam hammn uoa mH oomw mHna .oomw ovaHE may woxumfi thmnoua 50% we owe: moacHuaou I o vaamma< 1VNOIld0 .343 .33 5333 359 §ofi3so 5% 333 .NBB =3§§o§§ 9: 5 8% 333: go 8 R58 .nomoaooa 5&5 mg 3 §§3u “so 33 .oflém 143 .Hzoa uxwz A.qav ~%mvou Hoonom um vwv 00% 00:3 50% wmxmm 080: um maomaom ma Hmww waaos 30% 30: macaw uwau moan mcu 00 :N: cm 030: ~%000u Hoonum um 0H0 00% um:3.awau Hamu 00 90% 00300 0000300 000 %Hfiamm H00% ou mcfixamu 080: um mumB 50% mg Hmmm 30% vflaos 30m .m%m am nuws mafimmn zofis3 .30u 0x00 0:0 00 0300 meuma H=0% m>oanaoz “.mav Newmwa Hoonum 0u anus um>ma 00% 000 30uu080u 00000 Hoofiom «a Hmmm waaoz 50% 30: 03030 umnu 000m 050 00 :N: :0 sumzv .cwmwm Hoonum cu 0w Hm>m= Hafiz 00% 0:0 .30uu080u um>0u0m 000 HHHB Hoosum umnu 00% maamu maoma0m .cowummav mnu ou amumHA .mmmmn 0:0 nufis mcfiwmn sofin3 .30» 0x00 msu “00:: umxuma H=0% m>oz A.NHV .Hoocom ca umwm u:0% cw wawuuam mum 50% 00:3 Hmmm 50% 30: 030:0 umnu 000m 0:» no :N: 00 0x02 ~H00£00 dfi ummw u:0% cw wcwuuam mum 00% 00:3 Homm 30% 00 30m vmdafiuaou I o xavcmam< .0330 .3389 $3 33 $30 8 3.33 3 33E .3280 144 M I .. .423 .30330 03 8003 033 3033 3000 330% 30 w3333o3 H003 30% 303 3003 03030 3033 0003 033 30 :x: 30 030: 330333 %30> 03 8003 033 030 3303 0800 w33o0 3000 330% 30 3333330 030 30% 3033 H003 30% 00 303 .m>0Hm 0 30 0333033 033 3333 033m03 30333 .303 3x03 033 03 303308 330% 0>oz A.mHV .000H0 w330003 3033030 0>03 30>03 0H3o3 30% 3033 30% 33 H003 0H303 30% 303 03030 3033 0003 033 30 :N: 30 0302 300000H0 3330003 0308 03 03 0H303 03033 30% 0Ho3 3030003 330% 33 H003 30% 0H303 303 .3003 0 30 0333033 033 3333 033303 30333 .303 3x03 033 03 3300 303308 330% 030: A.0HV .3333308 033 33 H00300 303 %0003 30m 30% 3033 H003 30% 303 03030 3033 0003 033 30 :N: 30 0302 3H00300 03 0800 03 0000030 303 030 3333308 033 33 33 303 30% 3033 H003 30% 00 303 .330 H00300 03 3303 30% 3033 00 .0033030 0303 00H33 H00300 033 33 H003 30% 0H303 303 003333300 I o 3303033< .0303 30053 033 023 3033 33 33 3% 33: 145 .H08.3oz A.o~v .030H 330 30330033 0 303030 03 30% 0300 3030003 330% 3033 H003 30% 303 03030 3033 0003 033 30 :x: 30 030: 3303030 03 30% 303 03303 030 30330033 0 30% 0300 .0803 330% 0HH00 3030003 330% 3033 H003 30% 00 303 .320333303 0 me 0333030 033 3333 033003 30333 303 30333 033 30033 303308 330% 000H3 A.mHv .33030 0303 033 3333 0.303 .0803 03003 H00300 330% 0303 0H300 30% 33 H003 0H303 30% 303 3003 03030 3033 0003 033 30 :x: 30 0302 3%0003 H00300 8033 0803 03003 330% 0303 0H300 30% 0300 3030003 330% 33 H003 30% 0H303 303 amdw 0 30 0333033 033 3333 033303 30333 .303 300H 033 03 3300 303308 330% 0302 A.mHV 003333300 I 0 33030330 146 .300300 8033 0803 3030 03 003 303 33 3003 03303 303 303 3003 03030 3033 0003 033 30 :3: 30 0302 3300300 8033 0803 3030 03 003 303 030 .3030 003003 303 0300 303308 3303 33 3003 303 03303 303 .303030 033 30 0303033 033 3333 033303 30333 .303 3x03 033 03 303.303 330.3 0303 3:38 .303 03 03030 030 030 03030 3300033 033 33. 3030003 3303 308 303 33 3003 03303 303 303 3003 03030 3033 0003 033 30 :3: 30 0302 3303 03 :03303: 0300 030 33 003303 3030003 3303 030 .03030 3300033 033 33 0303 303 33 3003 303 03303 303 .33033303 30 3303 0 30 0333033 033 3333 033303 30333 .303 30333 033 30033 303308 3303 00033 3.333 .33030 0303 033 3333 03 0833 0.33 .3030 00308 3030003 3303 33 3003 03303 303 303 3003 03030 3033 0003 033 30 :x: 30 0302 33030003 303 0 303 303 030 3030 00308 3030003 3303 33 3003 303 03303 303 .20033 0 30 0333033 033 3333 033303 30333 .303 3003 033 03 303308 3303 0302 A.~Nv .33 3033 303 3033 3030003 300300 0 33303 33030 3003 03303 303 303 3003 03030 3033 0003 033 30 :N: 30 0302 ~33 3033 303 3033 3030003 300300 0 33303 33030 3003 303 03303 303 003333300 1 0 33030333 147 .wcficuoa mnu CH Hoonum ou uam3 haco 30% «H Hmmm vaso3 30% 30; ummn msonw uwnu mumm man so :x: cm mxmz waoocumuwm mnu wcm waficuoa mzu :uon we vmmumaw wcaauoa mnu ca Hoonom ou ucma mace 50% was .vmwsmno mums %mw H00£om man ma Hmmm do» vaso3 30w .sz whmmoz A.wmv .080: um um: unH> ou :om vmufi>afi umnummu Mach ma Hmwm vanes 50% 30: ummp mzonm umnu momm mnu no :x: an mme wmmao: Hm: ou um>o 30% wmufi>afi umnummu Mach ma Hmmm so» vH303 30m ammmmflm m we muauofia mnu nufia mcwwmn sofin3 .3ou umufim mnu “mun: meuma use» woman “.mmv .mwma uxm: mnu ou Gaza .wasoum mmam «no on nufia .aooummmao msu ca so» ou mafixamu muma umnummu Hack ma Hmwm cano3 50% Bo: maonm umsu momm wnu no :N: am mxmz wSOh ou wcaxamu mm3 mnm wam .aoou mnu aH mum3 mauaw van when umnuo man no maoz .umnummu “90% nua3 aooummmau mnu ca muwa do» ma Hmmm no» vaaoa 30m .HHmmoz A.omv .wafivmmu mum 30m cmnz Hmmu 90% 30: umwn mzonm umsu womm man no :N: am wxmz wwaanmmu Hack wcfiow mum so» Gmsz Hmmm 90% cc 30: .Qzoz A.mmv wmaafiucoo I o xavcmna< .0000 000% 00 w000000 000 00% 000 000H0 0:0 00 w00xamu 00 0050000 000 0003 H00w 00% 30: 0000 03000 0000 000“ 000 00 :x: 00 0x0: ~0H00w 000 0%0n 000 HH0 00 w00xH00 0000 000 m0 0000M 00 00 0000000 000 000 .0000 000% 00 w000000 000 00% 00:3 H000 00% 00 3oz .MUDQ 0 m0 0000000 000 0003 000w0n 00003 .300 0x00 000 00 003008 000% 0>0z A.mmv .30000000 000 0003 Hoonum M0 H00m 0H003 00% 30: 0000 03000 0000 000m 000 00 :x: 00 0x02 waoonom 00 0003 00000 000 %000H00 0 0003 30000800 ma H000 00% 0H003 30: ammmmflm 0 m0 0000000 000 0003 000w0n 00003 .300 0000 000 00 00x000 000% 0>0z A.~mV .0000 000% w0000 000 00% 0003 H000 00% 30: 0000 03000 0000 000m 000 00 :x: 00 0&0: NSuma 000% w0000 000 00% 0003 H00m 00% 00 303 ammwmmm.0 no 0000000 000 0003 00Hw0n 00003 300 00000 000 00000 00x000 000% 000Hm A.Hmv .0m00 000 0009 148 .0800 x003 H00000 000% 0x00 0H000 00% «0 H000 0H003 00% 300 03050 0000 000m 000 00 :x: 00 0x0: N080: x003 H00000 000% 0300 0H000 00% .00% 0H00 0000000 000% ma H00m 00% 0H003 30m .mmHm 0 m0 0000000 000 0003 00fiw0n 00003 .300 000a 0:0 00 00x008 000% 030: A.omv .Hoosom 00 000800 0000 00000 0H003 00% 300& 00% m0 H00m 0H003 00% 300 0000 03000 0000 000m 000 00 :x: 00 0x02 ~00aa00 00:0 Hoonom 00 cm 00 000 00% 00% waou 0000500 ma H00w 00% 0H003 3oz .momm 0 00 0000000 000 £003 00Hw0n 00003 .300 0000 0:0 00 00x0ma 000% 0>0z A.m~v 000000000 I u x000000< .00000 000% 00 8000000000 0 000 00% 00 0000 00003 00% 300 0000 03000 0000 0000 000 00 :x: 00 0002 00800 00 8000 I000000 0 000 00% 00 0000 00% 00003 300 admmmmmmm 00 00 0000000 000 0003 000000 00003 .300 0000 000 00 000008 000% 0>0z 0.00v .0000008 000 00 000000 00 0800 00% 0003 0000 00% 300 0000 03000 0000 0000 000 00 :x: 00 0002 00000008 000 00 000000 00 0800 00% 0003 0000 00% 00 300 .00H0 0 00 0000000 000 0003 000000 00003 .300 0x00 000 00 000008 000% 0>02 0.00v . 0%000v 00 0000000 00 0003 00000 00 0000 00003 00% 300 0000 03000 0000 0000 000 00 :N: 00 000: w 00000v 00 0000000 00 0003 00000 00 0000 000 00003 300 .0003000 00 0000000 0 0003 000000 00003 .300 0x00 000 00 000008 000% 0>0z 0.00v .0800 00 0003 000000 00000 000 00% 0003 0000 00% 300 0000 03000 0000 0000 000 00 :x: 00 0002 00800 00 0003 000000 00000 000 00% 0003 0000 00% 00 300 .00943 00 00000 0 00 0000000 000 0003 000000 00003 300 000 00000 000008 000% 00000 0.00v 149 .0000 0x00 000 00 0000 000000 .30z .000000 300 0 00 00 00 000 000 0008 %00 000000 00 00 000 00000 00% 00 0000 00003 00% 300 0000 03000 0000 0000 000 00 :x: 00 000: 00008 %00 000000 000000 000 00000 000 000000 000000000 0 00 00 00 000 00% 00 0000 00% 00003 300 .000 00 00 0000000 000 0003 000000 00003 300 0000 000 00 000008 000% 0>0z A.¢mv 000000000 I 0 00000000 150 .0000 0000 000 00 0000 302 .000000000 0000 800000000 000% 0000> 00 000800 0003 000008 000% 00 0000 00003 00% 300 0000 03000 0000 0000 000 00 :0: 00 0002 .800000000 000% 00 0000 000 00 000 00003 000 0000000000 0000 000000 00 000800 0003 000008 000% 00 0000 00% 00003 300 .0000000 000 00 0000 0 00 0000000 000 0003 000000 00003 .300 0000 000 00 000008 000% 0>0z 0.000 .30000800 0800 000% 0000> 00 00000 0003 0000000 000% 00 0000 00003 00% 300 0000 03000 0000 0000 000 00 :0: 00 0002 030000800 00000 000% 00 00>0 000800 0003 0000000 000% 00 0000 00% 00003 300 ammmfim 0 00 0000000 000 0003 000000 00003 .300 0000 000 00 000008 000% 0>02 0.000 .0000000 0008 0008 00 0003 00000 00 0000 00003 00% 300 0000 03000 0000 0000 000 00 :x: 00 0002 00008 00 0000000 0008 00 00 00003 00000 0000 0000000 000% 00 0000 00% 00003 300 .900 20000>m0MH 0 00 0000000 000 0003 000000 00003 .300 0000 000 00 000008 000% 0>0z 0.000 .0000000 000 0003 000000 00%000 00% 00 0000 00003 00% 300 03000 0000 0000 000 00 :x: 00 000: 00000 00000 000 0%00 00000 000 0003 0000000 000 0003 8000 000 00 00%000 00% 00 0000 00% 00003 30: .0000 0 00 0000000 000 0003 000000 00003 300 00000 000 00000 000008 000% 00000 0.000 .0000 0000 000 00 0000 302 000000000 I 0 00000000 151 .0000 0000000 0000 00 00000000 000 00000 000 300 0000 03000 0000 0000 000 00 :0: 00 0002 00000 00000000 000 00000 000 00 300 .0000000 00000000 00 00000 0 00 0000000 0000000 000 00 0000000 000 00 0000 .0000 0x00 000 00 000B A.m00 .8000 00 0000 0000000 000 000>00 0000 00000000 000 00000 000 300 03000 0000 0000 000 00 :0: 00 0000 00000 00000000 000 00000 000 00 300 .00000000 00 00000 0 00 0000000 0000000 000 00 0000000 000 00 0000 .0000 0x00 000 00 0009 0.000 .0000 000000 000 0300 0000003 0000 0000 000 000 000 00000 000 300 03000 0000 0000 000 :x: 000 0 0003 000: .0000 0000 000 000 000 300 00000 00009 .000000000 0000000 000 00 0000 .0030000 000000 0 00 0000003 0000 0 000 000 0 00 0000000 000 00 0000 .000 00000 000 00 0.000 0.00V .00 300 8000 0000 0000 00 0000 000 0000 0008 00 000 00» .0000 000 00 000 00000 0>00 0800 000 .000 000 0< .0000 000 0000008 00 000000000 000 003000 00000 0003 03 .0000 0000 00 0000000 000 00000 00 0003 000000000 009 .00000000 000 00 000000000 0000000 000000 00 0003 03 00000 000 00 0000 000 00 .0000000 0000 00000 00 00000 00 .000008 300000 0000 0000 000000 00 00» 000000000 I 0 00000004 .0000 000.30 000 :0 3 00000 <08 005 000 3 00000 0030000 00500 00 0.58. 0:0 080 00§§o§0 0: 000: 152 .CBOU mawocmn “50% usm .wmnmwcww Ham mum m3 302 .Hmmw musuowa mnu cw cmuwaanu mnu xcwnu 50% 30: ummn mSOnm umSu womm man so :x: am mxmz mammm %mnu xaHSu 90% ow 30m .mHAMu w um wafivmmu convafizu msu mo musuowa mnu um x004 .mwma uxmc mnu ou cusa A.omv .mammm vHHsu mnu xcfinu no» 30: ummn m3osw umnu momm mnu no :N: am mxmz. wmammm tawnu mau xaanu ao% ow 30$ .vumonxamno mSu mo uaoum ca wafiwawum wHHnu mSu mo muSuUfla mnu um xooq .mwma uxwa m:u ou cusa A.mcv .mammm uawnu mnu xawnu so» so: ummn macaw umnu momm mnu no :x: cm mxmz wmammm vaficu mnu xcanu 30% ow 30m .ummm may um wcfi Ixuoz wafinu m «o wuqufia mnu um x004 .mwmm uxmc mnu cu nude A.qu .Hmmw musuuaa mnu ca cmuvawno mSu xawnu so» 30: ummn mSOnm umnu momm mnu :0 :x: cm mxmz wammm cmuvafisu mSu xcwcu 30% ow 30m .mon wam Hufim mnu ou wawxamu umcommu mnu mo manuUfim mnu um xooq .mwma uxma mcu ou dune A.n¢v .Hmmm wusuofia mnu a“ cmuvafino mcu 3o: umwn mzozm umsu mumm mnu no :x: cm mxmz wammm %mnu xcasu 30% ow 30$ .mmpamnmxoon mnu ou uxmc wafivCMum Huaw cam won mnu mo musuofia m£u um x004 .mwma uxmc mnu ou cusa A.oqv wmscfiucoo I U xavcmam< .gusus 33m 303% 38 .3989. .§§ooa mgsfifi §§a§s 33 H338 153 Umsawuaou I o xwvamaa< 3% gm ESQSEV 2: mggsa 5% 5%:on mi V23: 9. 10. ll. 12. 13. 14. 15. l6. 17. 18. 19. 20. 21. 22. 23. 154 Appendix D SENTENCE COMPLETION STEMS I am happiest in school when . The boys and girls in my class . In school I try very hard to . . . When it's time to go to school, I feel . In school I really like . . At school the grownups think I . In school I am . . In school I am afraid of . I think school is . . The children in my class . . . In school the most important thing is . . . In school I want to . . . I intend to stay in school until . . In school I hate . When I'm in school my parents expect me . In school I think I can . I think my teacher is . The only trouble with school is I wish school were . . . When it's time to leave school, I feel . What makes me mad in school is . . . In school I hardly ever . . I get very upset in school when . . 24. 25. 26. 27. 28. 29. 30. 155 Appendix D - Continued In school I love . The peOple in school think I . My teacher thinks I t School is . Our class . When I'm in school my parents get upset when I . I go to school because . 10. ll. 12. 13. 14. 15. l6. 17. 18. 19. 20. 21. 156 Appendix D - Continued DIRECT QUESTIONS What is the best time of the day? Do you talk about the things that happened at school when you go home? If you were at home now, what would you tell about your day at school today (yesterday)? Would you rather go to school or stay in your house and watch tele— vision? Does your teacher like you? What is the best thing about your teacher? What is the worst thing about your teacher? How would you feel if your mother came to school to visit your class? Who is your best friend? Does he (she) like school? Why do you think they have school? If you could change school, what would you change? What parts of school are fun? What parts of school are not fun? Why do you go to school? Think about this. If tomorrow morning your mother says that you may go to school or you may stay home if you want to, which would you do? Would you go to school or would you stay home? Do you ever play school when you are at home? If yes, what do you do when you play school? If you did not go to school during the day, what would you do? Would you rather go to school or (#17) ? Is school a good place or a bad place? Why? Do you think boys and girls should have to go to school? Tell me three good things about school. 22. 23. 24. 25. 157 Appendix D - Continued Tell me three bad things about school. What things What things Do you like Do you hate about school make you happy? about school make you sad? school? Do you like it a little or do you like it a lot? school? 158 Appendix E DESCRIPTIONS OF CHILDREN* Children who fit the description in this column receive a score of 1 This child . . . Hates school. Views school as the worst part of each day. Eager to quit school. Hates the teacher. Will exert much effort to avoid talking to the teacher. Goes to great length to avoid being around the teacher. Hates to even see the teacher. Never seems motivated to perform in school. He rejects teacher's attempts to motivate him. Regardless of who attempts to motivate the child, they are rarely successful. Barely participates in school: e.g. Does not cooperate with activities of the class; Completes few assignments; Skips school. Believes his school experiences are utter failures -- both in the social and academic spheres. Believes the teacher hates him. * The original form of this document displayed the five larger boxes (descriptions of child 1, 2, 3, 4, and 5) next to each other on a single page. Because of space considerations, it was impossible to reproduce it in that format here. 159 Appendix E - Continued Children who fit the description in this column receive a score of 1.(con't) This Child O O O Hates to participate in school activities (academic & non-academic); Shows no interest in school; Is very inattentive; Is extremely unenthusiastic; Has very negative attitudes towards school. 160 Appendix E - Continued Children who fit the description in this column receive a score of .2. This child . . . Dislikes school. Views school as a necessary but negative part of the day. Dislikes the teacher. Prefers not to talk to the teacher. Prefers not to be around the teacher. Has no internal motivation towards school. Resists teacher's attempts to motivate him. Attempts to do as little as possible in school. Inclined to avoid doing tasks unless pressured by the teacher: e.g. Spends as little time as possible in class; Occasionally fails to turn in assignments; Stays home for any slight excuse. Believes most of his school experiences are unsuccessful —— both social and academic experiences. Believes the teacher dislikes him. Does not like to participate in school activities (academic & non- academic); Is disinterested in school; Is inattentive; Is unenthusiastic; Has negative attitudes towards school. 161 Appendix E — Continued Children who fit the description in this column receive a score of 2 This child . Is indifferent towards school. Goes to school because he is supposed to. Is indifferent to the teacher. Speaks to teacher when spoken to. Seems indifferent to being around the teacher. Is not self—motivated in school. However, he does respond to external motivation. Performs according to the minimum requirements of school. Carries out what is necessary — no more, no less. Believes most of his school experiences are neither successes nor failures. Or Believes he has approximately equal number of successful and unsuccessful experiences in school. Believes the teacher is indifferent to him. 162 Appendix E - Continued Children who fit the description in this column receive a score of i This child . Likes Views school. school as a positive experience most days. Likes Likes Likes the teacher. to talk to the teacher. to be around the teacher. Much of the time is not dependent on others for motivation in school. Occasionally does more than the minimum requirements of school: e.g. 'Might offer or indicate a willingness to stay after school or help the teacher; Occasionally does extra academic work. Believes most of his school experiences are successful in both social and academic spheres. Believes he is liked by his teacher. Participates in school activities (academic & non-academic); Interested in school; Is attentive; Is enthusiastic; Has positive attitudes towards school. 163 Appendix E - Continued Children who fit the description in this column receive a score of §_ This child . . Loves school. Views school as the high point of the day. Loves the teacher. Loves to talk to the teacher. Loves to be around the teacher. Loves to see the teacher. Is very highly motivated in school without any external pressure. Motivation is internalized. Applies to both the academic and non— academic aspects of school. Frequently, actively seeks to do more than the minimum require- ments of school: e.g. Voluntarily stays after school to help or comes into the classroom early; Does extra academic work; Stays home only when forced to by parents' insistence; Volunteers to help the teacher. Believes he is very successful in school -- socially and academically. Believes he is really liked by his teacher. 164 Appendix E - Continued Children who fit the description in this column receive a score of _5_ (con' t) This child . . . Is eager to participate in school activities (academic & non- academic); Is very interested in school; Is very attentive; Is extremely enthusiastic; Has very positive attitudes towards school. "I11111111111111.1311?