AN AEMLYSES OF THE CREEKMORE SCALES OF EIGHT CLOTHING VARIABLES Thais {'01- Hm Degree of M. A. filCHlGAN STATE UNIVERSITY Nelma Irene Fetterman 196-8 ABSTRACT AN ANALYSIS OF THE CREEKMORE SCALES OF EIGHT CLOTHING VARIABLES by Nelma Irene Fetterman The purpose of this study was to estimate the reli- ability and validity of responses made to the Creekmore Scales of Eight Clothing Variables as found in the Importance of Clothing questionnaire and to propose recommendations for the improvement of the scales. The subjects whose responses were analyzed were 269 girls and 236 boys in grades ten, eleven, and twelve of a high school located in a mid~western city of the United States. The Creekmore Scales of Eight Clothing Variables were Likert-type summated rating scales pertaining to aspects of clothing: aesthetics, modesty, interest, social approval, management, comfort, special attention, and psychological dependence. Hoyt‘s analysis of variance method of estimating the reliability of the scores of each of the scales was used. The "interest" scale had the highest reliability coefficient for the girls' responses, the "psychological dependence" Nelma Irene Fetterman scale had the second highest, "special attention," ”modesty," and “social approval" were tied for third highest. The "man- agement," "comfort," and "aesthetic" scales had the lowest reliability coefficients. For the boys' responses to the eight scales all of the reliability coefficients except for the "modesty'I and "management” scales were higher than the coefficients for the girls' responses. The ranking of the scales according to their reliability coefficients for the boys' responses was the same as for the girls' responses except that the ”social approval" scale had a higher relia- bility coefficient than the "modesty" scale. An item analysis consisting of item-total correla- tions and discriminatory power was conducted for each item of each of the scales for the boys' and girlsl responses. On the basis of the results of the item analysis recommenda- tions were suggested for improving some of the scales, es- pecially the "management,“ "aesthetic," and "comfort" scales. An attempt was made to estimate the validity of the scores obtained from the first seven of the eight scales. For each of the 199 girls and l5l boys who completed the scales and the criterion measure, a rank order correlation coefficient was calculated. The rank of each of the seven variables according to their importance was determined by the scores the subject obtained on the first seven scales. Nelma Irene Fetterman This ranking was compared with his ranking of seven words and phrases in the criterion measure which corresponded to the aspects measured in the first seven scales. No conclu- sions could be drawn concerning the validity of the scales because very few rank correlation coefficients were signifi~ cant and the reliability and validity of the criterion meas- ure were unknown. gen/o7 AN ANALYSIS OF THE CREEKMORE SCALES OF EIGHT CLOTHING VARIABLES by Nelma Irene Fetterman A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Textiles, Clothing and Related Arts l968 ACKNOWLEDGMENTS The writer wishes to express her appreciation to her advisor, Dr. Anna Creekmore, for the opportunity of conducting this study and for her guidance and encouragement; to Dr. Mary Gephart for her invaluable suggestions and as- sistance; to Dr. Maryellen McSweeney for her willing advice concerning the analysis; to Dr. Frances Magrabi for assist- ance with the computer programming; and to her friends for their encouragement. ii TABLE OF CONTENTS Page ACKNOWLEDGMENTS. . . . . . . . . .1. . . . . . . . . . ii LIST OF TABLES . . . . . . . . . . . . . . . . . . . . v Chapter I. INTRODUCTION . . . . . . . . . . . . . . . . . 1 II. REVIEW OF LITERATURE . . . . . . . . . . . . . 5 The Development of the Creekmore Scales of Eight Clothing Variables. . . . . . . . 5 Summated Rating Scales . . . . . . . . . . . 20 Reliability. . . . . . . . . . . . . . . . . 23 Validity . . . . . . . . . . . . . . 3 . . . 32 III. STATEMENT OF THE PROBLEM . . . . . . . . . . . 36 Definition of Terms. . . . . . . . . . . . . 37 Objectives . . . . . . . . . . . . . . . . . 38 Assumptions. . . . . . . . . . . . . . . . . 39 IV. PROCEDURE. . . . . . . . . . . . . . . . . . . #0 Description of the Research Site . . . . . . ho Description of the Sample. . . . . . . . . . Al Description of the Questionnaire . . . . . . #2 Selection of Techniques. . . . . . . . . #3 Description of Methods of Analysis . . . . . #8 Limitations. . . . . . . . . . . . . . . . . 55 V. FINDINGS AND DISCUSSION. . . . . . . . . . . . S7 Reliability Coefficients . . . . . . . . . . 58 Aesthetic Scale. . . . . . . . . . . . . . . 59 Modesty Scale. . . . . . . . . . . . . . . . 63 Interest Scale . . . . . . . . . . . . . . . 67 Chapter Page Comfort Scale. . . . . . . . . . . . . . 70 Special Attention Scale. . . . . . . . . . . 7h Management Scale . . . . . . . . . . . . . . 78 Social Approval Scale. . . . . . . . . . . 8l Psychological Dependence Scale . . . . . . . 84 Validity . . . . . . . . . . . . . 87 Recon'mendations for Improvements . . . . . . 90 VI. SUMMARY AND CONCLUSIONS . . . . . . . . . . . . 96 Recommendations. . . . . . . . . . . . . . . lOl BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . l03 APPENDIX . . . . . . . . . . . . . . . . . . . . . . . 108 iv LIST OF TABLES Table Page l. Estimated Reliability Coefficients for 269 Girls' and 236 Boys' Responses to the Creekmore Scales of Eight Clothing Variables. . . . . . . . . . . . . . . . . . 59 2. The Results of the Item Analysis for the 269 Girls' Responses to the "Aesthetic" Scale. . . . . . . . . . . . . . . . . . . . 62 3. The Results of the Item Analysis of the 236 Boys' Responses to the "Aesthetic" Scale. . . . . . . . . . . . . . . . . . . . 6A A. The Results of the Item Analysis for the 269 Girls' Responses to the "Modesty" Scale. . . . . . . . . . . . . . . . . . . . 65 5. The Results of the Item Analysis for the 236 Boys' Responses to the "Modesty" Scale. . . . . . . . . . . . . . . . . . . . 67 6. The Results of the Item Analysis for the 269 Girls' Responses to the "Interest“ Scale. . . . . . . . . . . . . . . . . . . . 69 7. The Results of the Item Analysis for the 236 Boys' Responses to the "Interest" Scale. . . . . . . . . . . . . . . . . . . . 7O 8. The Results of the Item Analysis for the 269 Girls' Responses to the "Comfort" Scale. . . . . . . . . . . . . . . . . . . . 73 9. The Results of the Item Analysis for the 236 Boys' Responses to the "Comfort" Scale. . . . . . . . . . . . . . . . . . . . 7h Table Page l0. The Results of the Item Analysis for the 269 Girls' Responses to the "Special Attention" Scale. . . . . . . . . . . . . . 76 ll. The Results of the Item Analysis for the 236 Boys' Responses to the "Special Attention" Scale. . . . . . . . . . . . . . 77 12. The Results of the Item Analysis for the 269 Girls' Responses to the "Management" Scale . . . . . . . . . . . . . . . . . . . 79 13. The Results of the Item Analysis for the 236 Boys' Responses to the "Management" Scale . . . . . . . . . . . . . . . . . . . 81 lb. The Results of the Item Analysis for the 269 Girls' Responses to the ”Social Approval" Scale . . . . . . . . . . . . . . 83 IS. The Results of the Item Analysis for the 236 Boys' Responses to the "Social Approval" Scale . . . . . . . . . . . . . . 8A l6. The Results of the Item Analysis for the 269 Girls' Responses to the ”Psychologi— cal Dependence" Scale . . . . . . . . . . . 86 17. The Results of the Item Analysis for the 236 Boys' Responses to the "Psychologi- cal Dependence" Scale . . . . . . . . . . . 87 vi CHAPTER I INTRODUCTION Prior to 1947 the major concerns of studies in the area of clothing were garment construction and design. Since that time more and more people have come to recognize the importance of the socio-psychological aspects inherent in the use of clothing. For example, clothing plays a sig- nificant role in the way we perceive peOple; clothing in- fluences the actions of the wearer; clothing partially de- termines the role one plays in society.2 All of these as- pects are indicative of the role clothing plays in social interaction. Because the study of the socio-psychological as- pects of clothing is recent, the tOtal area of attitudes toward and behaviors related to clothing is relatively un- 1Mary Shaw Ryan, Clothin x A Study in Human Behavior (New York: Holt, Rinehart and Winston, Inc., 19667, p. 2. 21bid., p. 3. explored and there is a need for more research.1 At the same time there exists a lack of valid and reliable tech- niques for measuring these attitudes and behaviors. Effec- tive measuring techniques are necessary for "if one does not know the reliability and validity of one's data little faith can be put in the results obtained and the conclu- sions drawn from the results."2 Good measuring devices are not developed in a single effort. It is usually necessary to revise and refine statements in a measuring instrument many times before a useful, valid and reliable technique is developed. Then the refined instrument should be administered to many samples and the reliability and validity of its scores calculated so that its usefulness can be established. One of the objectives of a recent Agricultural Ex- periment Station Project at Michigan State University was the composition of an instrument for measuring various as- pects of attitudes toward and behaviors related to cloth- 1Betty L. Brady, "Clothing Behavior: Refinement of a Measure and Relationships with Social Security and Inse- curity for a Group of College Women" (unpublished Master's thesis, Department of Clothing and Textiles, Pennsylvania State University, 1963), p. 2. 2 Fred N. Kerlinger, Foundations of Behavioral Re- search (New York: Holt, Rinehart and Wihston, Inc., 1965), P. 559. ing.1 It was the desire of the participants of this re- search project that the instrument, Importance of Clothing, hereafter referred to as the Creekmore Scales of Eight Clothing Variables, be used with many samples of populations and that it become the basis for a standardized measure of the attitudes and behaviors of adolescents toward cloth- ing. In order for other investigators to consider the use of the Creekmore Scales of Eight Clothing Variables, it is necessary to determine the degree of success already at- tained for securing reliable and valid responses from the use of the scales. At present only the discriminatory power of the first seven scales and the assurance of the face validity as judged by five authorities are available. More information about the scales is necessary. The purpose of this study is to investigate the reliability, internal consistency and validity of the Creek- more Scales of Eight Clothing Variables. On the basis of 1Anna M. Creekmore, "The Relationship of Clothing to Self-Concept and to Attitudes toward Clothing" (unpublished research in progress for Michigan State University Agricul- tural Experiment Station, Project M-78k). 2Mary Jane Young, "The Relationship of Social Accept- ance to ClOthing and to Personal Appearance of Adolescents" (unpublished Master's thesis, Department of Textiles, Cloth- ing and Related Arts, Michigan State University, 1967), p.77. the results recommendations will be made for improvements in the scales. CHAPTER II REVIEW OF LITERATURE In addition to knowing the reliability and validity of the responses made to the items of a scale, an investiga- tor must know the theoretical basis for the formulation and selection of the statements in the scale if the results are to be properly interpreted. The following discussion will be concerned with the history of the development of the Creekmore Scales of Eight Clothing Variables, a description of sumnated-rating scales, and methods of determining the reliability and validity of Likert-type Summated-rating scales. 6 The Develgpment of the Creekmore Scales of . Eight CTothing Variables In her exploratory study, Creekmore (1963)1 sought to investigate some of the factors which might be responsible for differences in the use of clothing by individuals and to 1Anna Mary Creekmore, "Clothing Behaviors and Their Relationship to General Values and to the Striving for Basic Needs” (unpublished Ph.D. thesis, Department of Clothing and Textiles, Pennsylvania State University, 1963). 5 discover possible relationships among eight basic needs, eight general values, and fourteen clothing behaviors for a group of college women. The investigation was based Upon the theory that needs are most basic to human beings and that in the process of striving to satisfy needs values evolve. These values may be observed in the behavior of the individual and for each value there may be a congruent clOthing behavior. For the study of basic needs, Creekmore added action or activity needs to Maslow's list of needs which consisted of belongingness, safety, physiological, self-actualizing, self-esteem, aesthetic, and cognitive. When investigating the general values of her subjects Creekmore used two values, Exploratory and Sensuous, in addition to Spranger's six value orientations which Vernon, Allport and Lindzey used in their Study of Values. It appeared to Creekmore that there existed two levels of values. The Social, Economic, Exploratory, and Sensuous were the first order while the Religious, Political, Theoretical and Aesthetic types could be refinements from the first order types, and hence more closely related to those parallel to them. To determine which behaviors should be included, Creekmore analyzed the eight specific value orientations and derived eight specific clothing behaviors which corres- ponded to the value orientations. However, literature re- vealed that six more global behaviors were often discussed. Thus, Creekmore proposed fourteen clothing behaviors for her investigation. The eight specific clothing behaviors were altruistic use of clothing, management emphasis, in- terest in the tactual aspects, experimentation, interest in symbolic meaning, emphasis on status symbol use, interest in appearance, and interest in theoretical aspects. The six general clothing behaviors were use of clothing construction, fashion interest, conformity, tool use, emphasis on modesty in dress, and no concern for clOthing. With the eight specific cloching behaviors and six general clothing behaviors as a basis, Creekmore developed 216 statements to which subjects could indicate their agree- ment or disagreement on a five point scale ranging from "al- most always" to "almost never." The 216 statements were pre- tested with 30 undergraduate and graduate students from home economics. Their responses to the items were analyzed and the items which failed to differentiate the high and the low scorers on the measure were deleted. The remaining items were examined and some revisions were made. The revised Clothing Interest Inventory consisted of 130 statements with seven to ten of these statements fitting into each of the fourteen clothing behavior categories. No attempt was made to validate the clothing behav- ior items because no known criterion existed. However, on an odd-even item basis, reliability coefficients were estab- 1ished for the responses of the 30 individuals on the pre- test. All of the scales, except Appearance and Tactual As- pects, had acceptable levels of reliability. The Appearance and Tactual Aspects scales were retained, although they had low reliability, because Creekmore felt that the multidimen- sionality of the variables caused the low reliability. Brady1 analyzed the responses given by Creekmore's 300 subjects to the Clothing Interest Inventory. First fre- quency distributions of the five possible choices for each item were computed. Those statements which had highly skewed distributions of responses were eliminated. The item- total correlations were calculated in each category for the remaining statements. As a result of both procedures 28 items were eliminated. Further revisions were made on the basis of clarity of meaning, agreement with the definition of the behavior category, applicability to peeple or situ- ations, and directness of statement When Brady examined the sections of the inventory and the definitions she found that some overlapping had 1Brady. occurred between sections and as a result additional changes and modifications were made. The following behavior catego- ries were eliminated: altruistic use of clothing; tool use of clothing; no concern for clothing; and theoretical in- terest in clothing. The definition of experimental use of clothing was modified to include "the planning, trying on, and putting together parts of ensembles which the wearer feels are new, different, or unusual.”] The section on tac- tual aspects of clothing was replaced by a section dealing with comfort. The statements concerning judgment of social status were incorporated with the statements about the sym- bolic meaning of clothing. As a result of the revisions made by Brady, the fi- nal form of the refined questionnaire consisted of ten statements in each of the following nine behavior categories: experimental use of clothing; construction; concern for ap- pearance; management; symbolic meaning of clothing; fashion; conformity; modesty; and comfort.2 The 90 statements used in the inventory consisted of 29 of Creekmore's original statements, 6 modifications of the original statements, and 55 new statements composed by Brady. 1Brady, p. 28. 2Ibid., p. 30. 10 When item-total correlations were computed within the nine clothing behavior categories for the responses given by 120 unmarried female Education undergraduate majors, each statement was found to correlate with its respective total above the .01 level of confidence. According to Brady, this indicated that the items in each of the nine behavior categories were internally consistent and discriminating.1 Sharpe2 developed a Likert-type scale to measure the degree of clothing interest and two aspects of clothing importance, importance from a prestige viewpoint and impor- tance in terms of ego satisfaction, for use with college females. Sharpe's initial instrument consisted of 80 state- ments, 7h attitudinal, 1 pertaining to knowledge of fashion designers, and 5 concerning the numbers of specific items in the respondents' wardrobes. The responses obtained in three pretests were examined by the scale value difference method of item analysis and by factor analysis. As a result, statements which gave results that were inconsistent with the total scores or statements which measured some other related 1113101., p. 60. 2Elizabeth Susan Sharpe, "Development of a Clothing Interest-and-Importance Scale" (unpublished Master's thesis, Ohio State University, 1963). 11 variable were eliminated.1 The final form of Sharpe's scale consisted of 1h attitudinal statements of which 10 pertained to clothing interest, 2 pertained to importance from ego satisfaction, and 2 pertained to importance from prestige.2 The split-half reliability of the responses of 54 coeds to the 1h statements was .89 when corrected by the Spearman-Brown pr0phecy formula.3 The 1h attitudinal statements were administered to 2h women, 12 of whom were predicted to have high interest in clothing and 12 of whom would have low interest in cloth- ing, to determine if the scores obtained were valid. Eleven of those predicted to have high interest scores did obtain the highest scores, and 10 of those predicted to have low scores did obtain the lowest scores. When a t-test was ap- plied to the mean scores of the two groups of women, the difference between the scores was found to be highly sig- nificant. Thus, Sharpe assumed that the scale gave valid results.“ A recent Agricultural Experiment Station Project at 1Ibid., p. 38. 2Ibid., p. 39. 3Ibid., p. A0. 1+1pm., pp. no-u1. 12 Michigan State University investigated the relationship of clothing to the self-concept of adolescents.1 One of the features of this project was the composition of an instru- ment to measure various aspects of attitudes and behaviors toward clothing. For the five researchers2 in the Michigan State Uni- versity Project under the direction of Dr. Anna Creekmore the first step in the composition of the instrument was to list ”as many situations or feelings involving clothing as possible"3 and "reasons why clothing was important to in- dividuals”“ so that a basic theoretical structure could be established. "After extensive discussion a compromise list of seven categories"5 involving clothing was formulated and the seven category headings were defined. The categories and 1Anna M. Creekmore, Relationship of Clothing . 2Carolyn Andree Humphrey, Karen Engel, Winifred Sue Hundley, Mary Green Klaasen, Mary Jane Young. 3Carolyn Andree Humphrey, "The Relationship of Sta— bility of Self Concept to the Clothing of Adolescents" (un- published Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 1968), p. 3h. “Mary Green Klaasen, “Self Esteem and Its Relation- ship to Clothing" (unpublished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State Uni- versity, 1967), p. 35. 5Young, p. AD. 13 their definitions were: Aesthetic refers to the use of clothing to achieve a pTeasing or beautiful appearance. Included as a part of appearance is neatness in dress. Comfort means the use of clothing to achieve comfort whether this relates to temperature, physical re- sponse to certain textures, or to acceptance of tightness or looseness in garments. Interest in clothing includes the willingness to give attention, to investigate, manipulate, or experiment with the putting together of the parts of a costume. Management refers to the thoughtful and careful use of time, money, and energy in planning, buying, and using clothing; thus, it can be an economic aspect of clothing usage. Modesty refers to the use of inconspicuous clothing w ich is quite conservative in design, color, fit, and body exposure. Social Approval is the use of clothing to attain a feeling of belongingness or the approval of others in a particular role situation and usually indicates conformity to the grOUp norm. Special Attention is the seeking of prestige and status through the use of clothing. The attention that is sought may be either socially approved or not apprgvgdhdepending upon the reference group. 9 9 : 1Winifred Sue Hundley, "The Relationship of Clothing to Social Class, High School Position and Status Inconsis- tency of Adolescent Boys and Girls" (unpublished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 1967), pp. 25-26. 2Humphrey, pp. 28-29. 3Young, pp. 30-31. 4 Klaasen, pp. 30-31. 1A 1 The seven categories and their definitions repre- sented improvements and refinements of Creekmore's original specific and general behavior categories. The aesthetic cat- gory corresponded to the original specific behavior of "em- phasis on appearance" and included neatness and cleanliness. Comfort was the same as the original specific behavior en- titled ”emphasis on tactual aspects." Interest corresponded to the specific behavior of "experimentation in clothing usage" but deleted the aspect of planning. Management was the same as the original specific behavior of ”management" except for the addition of planning for the acquisition of clothing. Modesty referred to Creekmore‘s original general category of "modesty" but was altered slightly to include use of inconspicuous or conservative clothing whereas pre~ viously it emphasized new unusual designs as well as body expoSure. The category of social approval combined the orig- inal general behavior categories of "conformity" and "tool use." Special attention incorporated the original general behavior categories of "clothing used as a tool" and "fash- ion" and the specific behavior of "emphasis on clothing as a status symbol." This category was altered to include both so- cially approved and disapproved behavior. The specific be- havior, "interest in the theoretical aspects,“ of the original clothing inventory was incorporated into each of the catego- 15 ries and appeared as the last statement in each of the scales. Although there were alterations, deletions, additions and combinations, the seven categories remained very similar to Creekmore's original specific categories with certain as- pects of the general clothing behaviors incorporated where they fit. The second step involved the formulation of state- ments. The researchers Used statements which referred to concrete situations. "I do" statements rather than "I feel" statements were Used because the researchers "felt that the subjects would respond to the 'I do' situations more truths fully than 'I feel' because they would be more aware of their activities than their feelings“1 and there would be less chance of subjects responding in a "stereotype, socia1~ 1y desirable manner.“2 "The original ideas for some of the statements for the scales began with those developed by Creekmore, Brady, and Sharpe because they contained short groups of objective statements on many aspects of clothing."3 Some statements were adopted verbatim and some were modified. In addition 1Humphrey, p. 33. 21bid. 3Young, pp. no-u1. 16 many new statements were developed. Prior to the first pre- test the researchers had collected and developed 170 state- ments which "included all parts of the defined aspects of clothing in an appropriate proportion and at various levels of importance . . . . Negative statements were also intro~ duced to avoid response set."1 Because the scales were de- signed for use with adolescents, both boys and girls, a con- scious effort was made to delete statements which could not be answered by boys or to modify the statements so that they could be answered by the boys. Three pretests of the ”Importance of Clothing” scales were conducted. The first pretest which contained 170 statements was administered to 28 education and retailing majors.2 Following an analysis to determine which statements discriminated adequately between high and low scorers, A7 statements were deleted and other statements were revised. The second pretest of 123 statements was administered to 21 adolescents, A boys and 17 girls,“ and the third pre- test was given to 68 graduate and undergraduate students in 1Ibid., p. 41. 2Hundley, p. 38. 31bid. b’Ibid. 17 an education service course at the Pennsylvania State Uni- versity.1 The responses given during the second and third pretests were analyzed to determine again which statements discriminated best between the high and low scorers. Following the selection of the items for the seven categories, an eighth category, psychological dependence, . was added and defined as being “used to indicate a person's sensitivity to the influence of clothing on his feelings which includes general good feelings, sense of well being and changing of moods."2 The statements of this category were very similar to some statements in Creekmore's original general behavior "clothing used as a tool” but were more limited. The decision to include this category was based upon the need for items to assess the effect of clothing upon moods, emotions and feelings and to measure emphasis placed on end use of clothing. Because the eleven statements for this category were developed after the pretests no in- formation was obtained about the discriminatory power of the items.3 The final instrument entitled “Importance of w 1Young, p. #1. 2Ibid., p. 31. 3Humphrey, p. 36. 18 Clothing" consisted of eight separate scales, one for each of the categories that had previously been defined, and an introductory statement which was not included in the analy- sis of the data. Each of the eight scales contained eleven statements. The respondents were instructed to indicate how they would act in each of the situations described in the statements which generally referred to the school situation. The subjects responded to each statement by indicat- ing the number which corresponded to their behavior in the given situation. The numbers ranged from 1 to 5 and repre» sented the following responses: “Almost Never,“ "Seldom,” “Sometimes," "Usually,“ and "Almost Always,“ respectively. The weights for the responses were identical to the numbers the subject placed beside the statements with the exception of items 2, 6, 10, 60, and 76 for which the weights were reversed because the items were negatively stated. The sub- ject's score for a given scale was obtained by the summation of the weight for the 11 items in that scale. The possible' range of scores in each of the eight scales was 11 to 55. A high score was representative of a favorable at- titude toward the aspect being measured and conversely, a low score represented a less favorable attitude toward the clothing variable. However, the scores could not be used to say how much more favorable one subject was than another nor 19 could these scores be compared with scores obtained from a second administration of the scales to the same group to determine whether there had been changes in attitudes. Up to the present time ideas and statements have been taken from Creekmore's original Clothing Inventory1 and instruments have been developed from her original ideas. Brady2 refined the Creekmore Clothing Inventory and used it to study the relationship between clothing behavior and social security and insecurity. Renn3 used Brady's refine- ment for investigating relationships between clothing behav- ior and desire for social participation. Phillips,“ in her investigation of knowledge and use of brand~named clothing as related to certain personal and social characteristics, used ideas from the original Clothing Inventory. Some of the statements from Creekmore's new scales were used by 1Creekmore, Clothing Behaviors . 2Brady. 3Emma Jane Renn, "Clothing Behavior and Relation— ships to Desire for Social Participation and to Reasons for Desiring Social Participation" (unpublished Master's thesis, Pennsylvania State University, 1965). hEmma Jane Phillips, "The Relationship of the Know- ledge and Use of Brand-Named Clothing with Certain Personal and Social Characteristics of a Selected GroUp of Sorority Women" (unpublished Master' 5 thesis, Department of Textiles and Clothing, University of Tennessee, 1966). 20 Dickey1 in her study. Klaasen,2 Hundley,3 Young,“ Humphrey,5 and Hacklanderé used the "Importance of Clothing" question- naire to obtain data for their studies of clothing as it related to self esteem, social class, peer acceptance, sta- bility of self concept, and concern for the body, respec- tively. Summated Rating Scales Attitudes toward clothing, like other attitudes cannot be measured directly; they can only be measured by the observation of behavioral indicants of the attitudes. Operational definitions specify what behavioral indicants must be observed in order that the attitudes can be 1Lois Edith Dickey, "Projection of the Self through Judgments of Clothed-Figures and Its Relation to Self-Es~ teem, Security-Insecurity and to Selected Clothing Behav- iors" (unpublished Ph.D. dissertatiOn, Deparflment of Cloth- ing and Textiles, Pennsylvania State University, 1967). 2Klaasen. 3Hund1ey. “Young. 5Humphrey. 6Effie Hewitt Hacklander, "The Relationship of Con- cern for the Body and the Clothing of Adolescents" (unpub- lished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 1968). 21 measured. One method of collecting observations of behavioral indicants so that the attitudes can be measured involves the use of self-reporting Summated rating scales. A sumnated rating scale consists of a series of statements to which a subject indicates his reaction.1 These statements should be clear, concise, interesting expressions of opinions and not facts, and expressions of a single atti- tude variable.2’3 It is not necessary that the statements pertain specifically to the aspect being measured. As long as the item is found to be consistent with the total score, it is considered satisfactory.A No attempt is made to have items which will be distributed evenly over a scale of favorableness or unfavorableness (or whatever the scale is measuring).5 Instead only those statements which seem either definitely favorable or definitely unfavorable to the aspect 1Claire Selltiz et al., Research Methodsiln Social Relations (New York: Holt, Rinehart and Winston, Inc., 1965), p. 366. 2Rensis A. Likert, "A Technique for the Measurement of Attitudes,” Archives of Psychology, CXL (June, 1932), p. 12. 3 . . . . Qu1nn McNemar, "Op1n1on-Att1tude Methodology," Psychological Bulletin, XLIII (1946), p. 313. “Selltiz, p. 367. 51b1d., p. 366. 22 in question are used.1 There should be an equal number of favorable and unfavorable statements in the scale.2’3 The subjects indicate their degree of agreement or disagreement with the statement by selecting the apprOpriate response. The responses are usually on a five point scale ranging from a strongly negative attitude through neutral to a strongly positive attitude.“ The responses are arbitrarily weighted with a “1” representing almost total disagreement through "5" representing almost total agreement for a posi- tive statement and the reverse for negative statements. The Likertutype method which is the most frequently used method for constructing sun‘mated rating scales5 does not yield 6,7 equal intervals between the responses. For example, the the distance between 1 and 2 on the continuum does not Ibid. 2Likert, p. 44. 3Allen L. Edwards, Techniques of Attityde Scale Con- struction (New York: Appleton~Century~Crofts, Inc., 1937), 371337— Jum C. Nunnally, Jr., Tests and Measurements (New York: McGraw-Hill Book Company, Inc., 1959), p. 301. 5 6 Selltiz, p. 366. Ibid., p. 369. 7A. N. Oppenheim, Questionnaire Desi n and Attitude Measurement (New York: Basic Boo s, Inc., 19 6), p. 1110. 23 necessarily equal the distance between 2 and 3. The subject's score for a sumated rating scale is the result of the addition of the weights for the responses he made to the individual statements of the scale. This score places the subject on a continuum which has agreement at one end and disagreement at the other end. His position is relative to the positions of the other subjects because the scores from a sumnated rating scale provide only a rank ordering of peOple along a continuum according to their at- titudes as indicated by their scores.1 The score does not provide a basis for saying how much more favorable one sub- ject's attitude is than another's nor for measuring the amount of change after some experience. Instead it is the aim of the Likert-type scaling technique to provide a uni~ dimensional sca1e2 which rank orders subjects along a con- tinuum according to their attitudes relative to the Other- members of the groUp. Reliability After the subjects have been rank ordered along a continuum according to their scores obtained on a Likert-type 1McNemar, p. 298. 2Oppenheim, p. 133. 24 surmated rating scale, it is necessary to determine the ac- 3 of the rank ordered curacy,1 consistency,2 and stability positions. The accuracy, consistency and stability all refer to the reliability of the scale or the ability of the scale to estimate the true scores of the subjects or their true positions on the continuum. To the degree that the obtained scores are free of error in estimating the subject's posi- tion, to that degree the scores are reliable.u There are several methods of determining the reli- ability of scores for scales. One of the methods, the test- retest, involves administering the same test twice to the same grOUp. If a very similar or the same rank ordering of subjects occurs both times the instrument produces stable or reliable results. However, many problems are encountered with this technique of determining reliability because other factors than the stability of responses are involved. The subjects may remember what their responses were to the items v 1McNemar, p. 298. 2. , N. N. Downie and R. W. Heath, Basic Statistical Methods (New York: Harper and Row, 1965), p. 216. BSelltiz, p. 168. “Ibid. p. 166. 25 if the time span is too short;1 intervening experiences may ~' alter the subject's attitudes so that his responses are not the sane32 sometimes the same grOUp is not available for the second administration of the scale;3 if the responses are anonymous so that more "honest“ responses are obtained, there is no method of matching the subject's responses to both scales. AnOther method of assessing the reliability of a scale without having to ask the same questions twice is the split-half technique.4 In this method, the items in the scale are divided in half on an odd-even basis or by random assignment and the product-moment correlation coefficient is calculated between the scores on the two halves. After this process, the Spearman-Brown prOphecy formula is applied to give an approximation of what the reliability would be if the scale is twice as long.5 Again there are drawbacks; if there is an uneven number of items one item will have to be 1G. C. Helmstadter, Principles of Psycholo ical Measurement (New York: Appleton-Century~Crofts, 1965), p. 65. 21bid. BSelltiz, p. 177. “Oppenheim, p. 74. 5Helmstadter, p. 68. 26 mitted from the analysis when the scale is divided in half.1 Another hindrance is the occurrence of an unfortunate split in the items so that too high or too low a correlation results.2 A third method of calculating the reliability of a scale is based on the theory that all measurement contains some error. An individual's obtained score on a scale con- sists of his true score and the error score. The error score is composed of both systematic error and random error. The smaller the random error, the more reliable the measuring instrument.3 The true scores and the error scores are never known but it is possible to estimate the error variance and to know the variance of the obtained scores for the subjects of the sample. From the estimate of the error variance and the known variance of the obtained scores the reliability can be estimated because "reliability is the proportion of error variance to the total obtained variance of the data yielded by the measuring instrument subtracted from 1.00, 1Likert, p. 28. 2Cyril Hoyt, "Test Reliability Estimated by Analysis of Variance," Psychometrika, VI (1941), p. 155. 3Downie and Heath, pp. 215-216. 27 "I The results the index 1.00 indicating perfect reliability. obtained from Hoyt's variance method of estimating the reli- ability of the scores of a scale are equivalent to the re- sults obtained fromlusing the Kuder-Richardson 20 formula2 and similar to the results obtained when the split-half method is used.3 Another interpretation of reliability is that if the items are internally consistent or are homogeneous the scale is reliable.“ To determine if the items are internally con- sistent the subjects' scores for each statement are corre- lated with their total scores for the scale. If the result- ing correlation coefficient for the correlation of the item with the total is zero or very low, the statement is am- biguous5 or measures some extraneous attitude factor.6 If the correlation coefficient is high the statement is consis- tent with the attitude being measured. IKerlinger, p. 434. 2Helmstadter, p. 73. 3Kerlinger, p. 438. “McNemar, p. 297. 5Lee J. Cronbach, Essentials of Psychologjcal Test- 122 (New York: Harper and Bfothers, 1949), p, 371, 6Nunnally, p. 305. 28 The total score for the scale is used as the crite- rion with which to correlate the statements because the total score is the best available criterion.1 It is the most convenient and the most relevant.2 Some authors dispute the use of the total score and have indicated that the total score minus the score for the item should be used as the criterion. However, Oppenheim suggests that this is unnec- essary when the number of statements is large.3 The determination of the discriminatory power of each of the items is another method of determining the in- ternal consistency of the items of a scale. If the scale fulfils its purpose of rank ordering subjects along a con- tinuum and if the content is consistent with what the whole scale measures, the subjects at the more favorable end of the continuum should score higher on the item than the sub~ jects at the less favorable or less agreeable end of the 4 To the extent that the items discriminate the continuum. high and low scorers, to that extent are the items of the scale internally consistent or reliable. 1Oppenheim, p. 138. 2Robert L. Ebel, Measurin Educational Achievement (Englewood Cliffs, N. J.: Prentice-Hall, Inc., 1965), p. 356. 3Oppenheim, pp. 138-139. hSelltiz, p. 184. 29 In order to determine the discriminatory power of the items of a scale, a high grOUp and a low groUp must be selected. Likert suggests that the subjects with the highest 10% of the scores and the subjects with the lowest 10% of the scores be selected.1 Selltiz suggests using the subjects who have the highest and lowest 25% of the scores.2 Kelley has found that using the Upper and lower 27% of the subjects provided the best compromise between making the extreme groups as large as possible and yet making the extreme groups as different as possible.3 Ebel states that in "al- most all situations, the use of 27% for the Upper and lower groups will be convenient as well as most logically defen- sible."h After the sums of the scores for a single item for the high and low grOUps have been calculated and the differ- ence between the sums found and divided by the number of subjects constituting the 27% of the sample, the resulting number is the discrimination index for the item. The larger 1Likert, p. 51. 2Selltiz, p. 368. 3Truman L. Kelley, "The Selection of the Upper and Lower Groups for the Validation of Test Items,“ Journal of Educational Psychology, XXX (1939), p. 24. “Ebel, p. 350. 30 the discrimination index, the better the item differentiates among subjects and the more reliable the scale is. All of the procedures which have been described with the exception of the test-retest method require more than ordinal measurement in order to be Used legitimately. In order to use means, product-moment coefficients of cor- relation and analysis of variance which are necessary for these procedures, the intervals on a scale must be equal but the intervals on an ordinal scale are not equal. If the assumption is made that the intervals on the scale are equal, then the procedures can be used providing the in- vestigator is constantly aware of the possibility of error when interpreting the data and the relations inferred from the data. Many of the psychological measures which are used are basically ordinal but have been treated as though the intervals were equal and the results have been satisfactory. If one fails to make the assumption that the intervals are equal, many powerful methods of analysis are eliminated and only inadequate techniques are left for use in solving problems.1 The recommended procedure seems "to be to treat ordinal measurements as thOUgh they were interval measure- ments, but to be constantly alert to the possibility of 1Kerlinger, p. 427. 31 gross inequality of intervals."I Various statements have been made about what a sat~ isfactory reliability coefficient should be. Oppenheim sug- gests that reliability coefficients of .80 or over are can- men;2 Edwards states that reliability coefficients for scales constructed by the method of summated ratings are usually above .85 even when fewer than 20 items constitute the scale;3 Nunnally claims that although no definite rule states how high a reliability coefficient should be one sus- pects a test that has a reliability coefficient less than .80.h Downie and Heath summarize the situation well when they state: In general, reliability coefficients of well-made standardized tests tend to be high, .90 or above. There is no hard and fast rule that says that any reliability has to be of a certain size before any test or measure- ment instrument can be useful. Today we look upon reli- ability as a relative thing, and there are certain areas and certain techniques where reliability coeffi~ cients fall well below this .90, and the techniques are still used and found to be very useful. Rating scales are an example of this. 'Ibid. 2Oppenheim, p. 122. Edwards, p. 156. 4Tb.) Nunnally, p. 110. SDownie and Heath, p. 220. 32 Validity Validity is generally defined as the ability of a measuring instrument to measure what it was designed to measure.1 When this definition is applied to an attitude scale it is translated to mean the ability of a scale to measure one's true attitude toward an object as it is ex- pressed through behavior as indicated by responses to "I do" statements. AlthOUgh attitudes are not behaviors but pre- dispositions to behavior, the only way that attitudes can be measured is throngh their expression in behavior.2 Herein lies the problem of validation of attitude scales because one must be certain that the behaviors observed are determined by the attitude being measured and not by some Other attitude. What is essential to the approach of validation is a reasonably reliable and valid criterion with which the scores on an attitude scale can be compared.3 One criterion that has been suggested for validating the responses to an attitude scale is observation of the respondent's overt be- havior for if one's attitude is his tendency to react lOppenheim, p. 121. 2Theodore M. Newcomb, Social Psychology (New York: Henry Holt and Company, Inc., 1959), p. 160. 3Sentiz, p. 157. 33 favorably or unfavorably toward an attitude object, a valid test should predict his behavior.1 However, reactions are difficult to observe and they can be influenced by the sub~ ject's desire to create a good impression.2 Therefore, ob- servation of overt behavior may not be a valid and reliable criterion for validating an attitude scale. An0ther suggestion for validating attitude scales is the administration of the scales to criterion grOUps who are known to have favorable and unfavorable attitudes toward the attitude object being measured by the scale. If the scale distinguishes between those who have the particular attitude from those who do n0t, the scale is considered to be valid.3 Oppenheim disputes the use of this procedure when he says: At best there can only be rough correspondence between the two indicators. We cannOt necessarily predict be- havior from attitudes, nor are attitudes readily in- ferred from behavior with any validity, nor is behavior necessarily a more valid expression of an attitude than a verbal statement. The links between attitudes and behaviors are complex, and group membership may or may not be a valid criterion for a particular attitude. 1Nunnally, pp. 312-313. 2 Cronbach, p. 375- 3Sharpe, pp. 40-41. “Oppenheim, p. 75. .' ‘LI: la. 34 A third criterion which can be used for validating an attitude scale is another verbal technique which is known to be reliable and valid and which is known to measure the same attitude.1 Just as Creekmore noted, it is extremely difficult if not impossible to find an adequate criterion to 2 The solution which Selltiz which the scores can be compared. offers is the selection of the most adequate criterion avail- able for the purpose but to be constantly alert to the limi- tations of its use.3 The lack of an adequate criterion for the validation of scales has been the cause of serious criticism for those who use attitude scales which have not been validated.“ Yet no one has devised a more effective method of assessing at- titudes and until such time as a better means is developed attitude scales will have to be used for measuring subjects' attitudes. Oppenheim summarizes the dilemma when he says that "the problem of validity remains one of the most diffi- cult in social research and one to which an adequate solution 1Selltiz, p. 158. 2Creekmore, Clothing Behaviors . . ., p. 144. 3Selltiz, p. 158. l‘Cronbach, p. 375. 35 . . l . is not yet 1n s1ght." The best that one can do 15 to use an available criterion and keep in mind the limitations of the criterion. 1Oppenheim, p. 78. 41 CHAPTER III STATEMENT OF THE PROBLEM As part of a recent projeCt investigating the rela- tionship of clothing to the self~concept of adolescents, the participants of the project composed eight scales for meas- uring attitudes toward and behaviors related to clothing using approximately the same theoretical basis that Creek- ] used for her original Clothing Inventory. The partic- more ipants hoped that the scales would "become the basis for a standardized clothing measure"2 because of the need for tech- niques for measuring attitudes toward clothing. In order to know the value of the scales for measuring attitudes toward and behaviors related to clothing, the results obtained from the administration of the scales must be analyzed to deter- mine whether the responses obtained from the subjects were reliable and valid indications of their attitudes toward and behaviors related to clOthing. lCreekmore, Clothing Behaviors . . . 2Young, p. 77. 36 37 The purpose of this study was to analyze the re- sponses obtained from the administration of a questionnaire, containing the eight scales to determine the reliability and validity of the scores obtained. On the basis of the results of the analysis, recommendations were proposed for improvement of the scales. Definition of Terms In addition to the definitions already given for the clothing variables, the following definitions were used to clarify other terms which occurred in this study: Analysis is the evaluation of the reliability and validity of the scores obtained from the use of the Creek- more Scales of Eight Clothing Variables. Creekmore Scales of Eight Clothing_Variab1es are eight summated rating scales which constitute the Importance of Clothing questionnaire and which were compiled by Creek- more and students for Michigan State University Agricultural Experiment Station, Project 784. S2313 is a device for measuring attitudes and rank orders subjects along a continuum according to their total or summated scores for the scale. Reliability is the accuracy and consistency with 38 which a scale measures a subject's “true" score for a spe- cific attitude. Validity is the ability of a scale to measure the variable it was designed to measure. Attitudes are predispositions to behavior. Discrimination is the ability to distinguish between two groUps according to their scores. Item is a statement used in a scale to obtain a sub- ject's response toward a particular behavioral indicant of an attitude. Objectives The objectives for this study are: 1. To estimate the reliability of the data obtained from the use of the Creekmore Scales of Eight Clothing Variables. 2. To estimate the validity of the data obtained from the use of the Creekmore Scales of Eight Clothing Variables. 3. To propose recommendations for the improvement of the Creekmore Scales of Eight Clothing Variables. 39 Assumptions In order to fulfil the objectives of this study the following assumptions were made: 1. Attitudes toward clothing can be measured by the use of Likert-type summated rating scales. 2. The subjects for this study do possess the atti- tudes toward clothing which are being measured by the scales. 3. The data have been honestly and sincerely SUpplied by the subjects. “” 4. For purposes of estimating the reliability of the responses, the intervals on the scale are equal, i.e., the distance between 2 and 3 on the scales is the same as the distance between 3 and 4. 5. The relationship between the item scores and the total or summated score is linear. 6. The criterion with which the scores from the attitude scales are compared is a reliable, valid and in- dependent criterion. CHAPTER IV PROCEDURE The results of the administration of the Creekmore Scales of Eight Clothing Variables were analyzed according to selected procedures and within certain limitations so that estimates could be made of the reliability and validity of the scores. Description of the Research Site The high school which the subjects for this study attended was located in a mid-western industrial city which had a population of approximately 15,000. The criteria for the selection of the high school were: location in a corrmunity which had a range of family socio-economic posi- tions which approximated that of the United States as a whole; the only high school in the county so that both rural and urban students would be in attendance; an enrollment which was large enOUQh to provide a sample of approximately 500 students. The high school which was selected had an 40 41 enrollment of 1,850 students in grades nine, ten, eleven and twelve.]’2’3’u Description of the Sample . Because the students attended school on staggered schedules and because there were no vacant rooms where ran- domly selected subjects could complete the questionnaire, a random sample was impossible. Consequently, the students in grades ten, eleven and twelve who attended study halls on Monday constituted the non-random sample. The sample for the project was composed of 270 girls and 251 boys but 1 girl and 15 boys did not complete the Creekmore Scales of Eight Clothing Variables. Therefore, all of the discussion of this study referred to the responses of 269 girls and 236 boys except in the investigation of the validity where Only 151 boys and 199 girls completed the criterion measure, the ranking question. 1Klaasen, pp. 41-42. 2Hundley, pp. 29-31. 3Humphrey, pp. 41-42. 1; Young, pp. 34-35. 42 Description of the_Qgestionnaire The first part of a two-part self-administered questionnaire used in the project for the collection of data contained the 89 statements of the Creekmore Scales of Eight Clothing Variables and a list of seven words and phrases which corresponded with the first seven variables of the scales. The statements of the Creekmore Scales were grOUped according to the variables but they were neither separated according to variable nor titled. The subjects indicated their responses to each item by placing in the blank to the left of the statement the number which corresponded to the frequency of their behavior described in the statement. After responding to the statements of the scales, the sub- jects ranked the seven words or phrases according to their degree of importance when they selected clothing for their wardrobes. A phrase referring to psychological dependence was not included in the list. The remainder of the first part of the questionnaire consisted of a Self Rating Inventory. The second part of the questionnaire which was administered one week later asked questions for background information of the sUbjects, for social class, and body characteristics and contained the Self Rating Inventory. The subjects remained anonymous; only their initials, birthdates and sex were requested so that 43 the two parts of the questionnaire could be matched for each subject. Selection of Techniques In compliance with the objectives of this study, the investigator had to select the techniques for assessing the reliability and validity of the Creekmore Scales of Eight Clothing Variables and methods for ascertaining what imprbvements were necessary. The selection of the techniques was based not only on the value of the techniques but also on the practical possibilities open and the resources avail- able for the investigation. In the review of literature three procedures were diSCUSsed for obtaining reliability coefficients for Likert- type summated rating scales: test-retest; split-half; and Hoyt's method of analysis of variance for estimating error variance of measurement. Two of these procedures were not possible. The test-retest method for estimating reliability could not be Used because the same sample was not available for the second administration of the scales. The split-half reliability was not suitable for assessing the reliability because there was an uneven number of items in each scale. If it were used, one item would have to be omitted from the analysis. Secondly, if an unfortunate split occurred when 44 the scale was divided in half, the correlation between the two halves could present a false indication of the reliabil- ity of the scale. The third method, the analysis of variance method for estimating the error variance, could be prohibi- tive if one did not have access to computer facilities. How- ever, with the aid of the computer all of the responses to all of the items of a scale could be analyzed at once and estimates of the error variance and variance for the sub- jects obtained so that the reliability of the scales could be calculated. With these considerations in mind, the anal- ysis of variance method for estimating the reliability of the responses to the Creekmore Scales of Eight Clothing Variables was selected because the investigator had access to a computer system and authorities had shown that the re- sults of this method were comparable to those of the split- half reliability1 and other methods of determining reliabil- ity.2 The third purpose of this study was to propose recommendations for improvement of the scales. The calcula- tion of the reliability coefficient using HOYt's method told if the test was reliable or unreliable but it did not tell 1Kerlinger, p. 436. 2Helmstadter, p. 73. 45 what items should be altered or deleted so that the scales could be improved. Therefore, an item analysis was necessary tO fulfil the second objective. Likert suggested two methods of analyzing items of scales.1 The first method of determining if an item contrib- uted satisfactorily to the total scale was the correlation Of the subjects' scores on an item with their total summated scores for the scale. If the correlation was high the item was considered an asset to the scale and retained; if it was low or zero it contributed nothing and should be altered or deleted. The second method which Likert suggested because he found the first method too laborious was the use of the criterion of internal consistency. In this method the re- actions or responses of the grOUp at one end of the continu- um were compared with the responses of the subjects at the Other end of the continuum. If the item differentiated be- tween the two groups, it was considered to be satisfactory. Both of the methods suggested by Likert were fre- quently used in the field of attitude measurement and were ‘w 1Likert, pp. 48-51. 46 1,2,3,4 Therefore, both recommended for use by many authors. the itemptotal correlation and the criterion of internal consistency were calculated in order that recoumendations could be made for the improvement of the scales. The second objective of this study was to determine the validity of the Creekmore Scales of Eight Clothing Vari- ables. From the review of literature the consensus was that this was one of the most serious problems of attitude meas- urement. Some authors advocated the comparison of the re- sponses to the attitude scales with the overt behavior of the subjects; other authors stated that behavior was not the ultimate criterion for assessing the validity of responses. Those who disputed the use of behavior as a criterion sug- gested that subjects' responses to a reliable and valid scale measuring the same attitudes be used as the criterion. Herein lies the problem because there were no known tech- niques which measured the same attitudes as the Creekmore Scales of Eight Clothing Variables.5 lSelltiz, p. 184. 2Oppenheim, pp. 138-140. 3McNemar, p. 306. “Downie and Heath, pp. 227-229. 5Young, p. 42. 47 The only available criterion which dealt with the same attitudes as the Creekmore Scales was the list of words and phrases which corresponded to seven of the eight atti- tudes measured in the scales1 which the subjects ranked after they responded to the scales. Realizing that single statements were not as reliable as groups of statements when measuring attitudes2 and at the same time desiring a crite- rion measure, the investigator decided that one possible method of assessing the validity of the scales was to com- pare the subject's rank ordering of the words and phrases with his ranking of the clothing variables according to the summated scores obtained on the scales. The method seleCted for comparing the two rank or- derings of seven clothing variables was the Spearman rank order correlation. If the resulting rank order correlation coefficients were significant, the scales were considered to be valid. The methods that were selected to enable the inves- tigator to carry out the objectives of the study were Hoyt's analysis of variance for estimating the reliability coef- ficients of each of the scales; item analysis consisting of 1Psychological dependence was nOt included. 2Oppenheim, p. 73. 48 item-total correlations and discrimination indices so that statements which needed improvement or should be deleted could be identified; the Spearman rank order correlation method for attempting to assess the validity of the Creek- more Scales of Eight Clothing Variables. Description of Methods of Analysis Reliability was defined thrOUQh error; the greater the random error component, the greater the unreliability and the smaller the random error, the greater the reliabil- ity.I By estimating the proportion of error variance to the total obtained variance of the individuals' scores and sub- traCting the result from 1.00, the index of perfect reliability, one could estimate the reliability of responses to scales such as the Creekmore Scales of Eight Clothing Variables. The formula for the reliability coefficient was: rtt = l - ve where Vind V6 is the estimate of the error variance and vind is the variance of the obtained scores 2 of the individuals. 1Kerlinger, p. 434. 21bid., p. 437. 49 By means of one-way analysis of variance with rep- lication, the between items variance, the between individu- als variance, and the error or residual variance could be calculated. From these calculations the between individuals and the error variances were selected for the computation Of the reliability coefficient.1 In order to compute the analysis of variance tables for each of the scales to get the variances due to individuals and error, the data for each subject were read into the CDC 3600 Computer system at Michigan State Univer- sity from punched IBM cards. The one-way analysis of vari- ance with replication routine was employed.2 From the re- sults of the computations by the computer system the error variance and variance due to individuals were extracted and with the aid of a desk calculator the error variance was divided by the variance due to individuals to obtain the proportion of error in the responses. The quotient was sub- tracted from 1.00, the index of perfect reliability, to 1For a complete description of this method consult Hoyt's article and pages 432 - 439 of Kerlinger. 2w111iam L. Ruble et a1, "Analysis of Variance with Equal Frequency in Each Cell," STAT Series Description No. 14 (Agricultural Experiment Station, Michigan State Uni- versity, September, 1967). SO estimate the reliability of each scale or the degree of accuracy with which the scales measured the "true" scores of individuals. This procedure was followed for each of the eight scales and for boys and girls separately. Therefore, sixteen reliability coefficients were calculated for the Creekmore Scales of Eight ClOthing Variables. After the reliability coefficients had been calcu- lated for each of the scales, the individual items were analyzed to ascertain which items contributed significantly to the reliability of the scales and which items did not contribute significantly. The first method was the calcula- tion of the discriminatory power of each of the items or the ability of the items to differentiate between the groups at the two ends of the continuum. In order to calculate the discriminatory power of each item the data were read into the CDC 3600 Computer system and the Computer Institute for Social Science Research Program Analysis of Contingency Tables ACt II1 was used. The results of this program con- sisted of a listing of all the summated scores that the subjects received for the scale, the frequency of each of 1Alan M. Lesgold, "Analysis of Contingency Tables, Act II," Technical Report No. 14 (Computer InstitUte for Social Science Research, Michigan State University, January 12, 1968). 51 the summated scores, the frequency of the choice of the five possible responses for each of the summated scores, and the total frequency of each of the five possible responses for the item. These results were obtained for each item of the scale and were repeated for each of the eight scales for boys and girls separately. After the results were obtained from the computer, the 27% of the subjeCts who had low summated scores and the 27% of the subjects who had high summated scores were se- lected.1 If the cut-off point for the lower 27% occurred at a point where there were more than the necessary number of subjects to complete the lower grOUp, the subjects whose scores were lowest for the item were selected. If the cut- off point for the Upper 27% occurred in a similar situation, the necessary number of subjects whose scores were highest on the item were selected. With the aid of a desk calculator the sums of the weights for the responses for the lower group and for the Upper grOUp were compUted. The discrimination index was then found by calculating the difference between the mean scores of the two groups. The formula which was used for the dis- crimination index was: IKelley. 52 Oz 1U ~ZL where ZLlisithe sum of the weights for the responses for the Upper grOUp 2L.is the sum of the weights for the responses for the lower grOUp n is .27N and N is the total number of boys or girls. The larger the discrimination index for an item, the more differentiating the item was and the more it ful- filled the purpose of distributing the subjects along the continuum. The second method of analyzing the individual items was carried out to determine whether each item measured the same attitude as the total scale measured. By using the same computer program as was used for the discrimination indices, the Pearson product-moment correlation coefficient was cal- culated between the subjects' scores for an item and their summated scores for the scale. The formula which was used for the item-total correlations was: 53 l’ = J1 XL - (Ex) (I!) flNsz- (XXTZ) (le- (2)02) The larger the correlation coefficient was, the greater was the relationship between the subjects' scores on an item and their summated scores for the scale and the greater the relationship between the content of the item and the content of the entire scale. If the correlation coef- ficient was zero.or low, the item was considered inconsist- ent with what the other items of the scale were measuring and should be deleted or altered. The item-total correlations were calculated for each of the items of each of the eight scales and were cal- culated separately for boys and girls. 'In order to attempt to assess the validity of seven Of the eight scales,2 the second objective of this study, the subject's ranking of the seven words and phrases pertaining to the first seven attitudes measured in the scales was cor- related with his rank ordering of the seven attitudes meas- ured on the scales. The rank ordering of the seven attitudes v—v 1William L. Hays, Statistics for Psycholo ists (New York: Holt, Rinehart and Winston, 1965), p. 506. 2Psychological dependence was nOt included in this analysis. 54 on the scales was based upon his summated scores for the scales. The attitude which had the highest score was ranked highest, the attitude with the second highest score was given the second highest rank, and so on. Each subject's summated scores for the first seven scales and his ranking of the items in the criterion measure were read into the CDC 3600 Computer system and the Computer Institute for Social Science Research program for Spearman rank order correlation coefficients was used.1 One rank or- der correlation coefficient was calculated for each subject who completed the scales and the criterion measure which in- volved the ranking of the words and phrases. The formula that was used for the calculation of the correlation coefficients was: 2 2 2 rs : Xx " Z72 ~Zd where ‘dh'z x2 2:71 1x2 was N3-N - t3-t fort=tiesinx 12 12 1John Morris, "Rank Correlation Coefficients," Tech- nical Report 47 (Computer Institute for Social Science Re- search, Michigan State University, January 5, 1967). 2Sidney Siegel, Nonparametric StatiStics for tgg Behavioral Sciences (New Y6? : McGraw-Hill BooR Company, Inc., 1956), p. 207. 55 2y2 was N3-N - t3-t fort'tiesiny 12 12 Zd was the sum of the differences between ranks. After the rank order correlation coefficients had been calculated for the girls, the coefficients were rank ordered from highest to lowest. The number of coefficients that were significant at the .044 level for a one-tailed test was calculated.1 The number of significant correlations which could have occurred by chance alone was subtracted from the total number of significant correlations. Then the percentage of significant correlations, excluding those due to chance, of the total number of calculated correlations was computed to determine the amount of agreement between the results of the scale and the criterion measure. The same procedure was followed for the boys. Limitations Because of the specific circumstances of this study, certain limitations were imposed upon the results Obtained from the procedures which were outlined. The results of this study cannot be generalized to any other subjects than the 1Allen L. Edwards, Statistical Methods for the Be- havioral Sciences (New York: Holt, Rinehart and Winston, 1964), p. 513. 56 ones in this study becaUse the sample was not random. Sec- ondly, the results of the item analysis are dependable to the degree that the scale intervals are equal. The results of the item-tOtal correlations are appropriate to the degree that the relationship between the weights for the responses to the items and the summated scores for the scale contain- ing the items was linear. Finally, the results of the esti- mation of the validity of the Creekmore Scales are signif- icant if the criterion measure was truly reliable and valid. However, single items as in the criterion measure are nOt as reliable for locating subjects along a continuum as grOUps of statements1 and the possibility of drawing un- justified conclUSions should be avoided. With these limita- tions in mind, the investigator analyzed the results. l Oppenheim, p. 73. CHAPTER V FINDINGS AND DISCUSSION The investigation of the reliability and validity of the Creekmore Scales of Eight ClOthing Variables involved four main Operations, three for reliability and one for va- lidity. Hoyt's method of estimating reliability provided a single reliability coefficient for an entire scale and item analysis consisting of discriminatory power and item-total correlations yielded indications of which items were most consistent with the total scale and which items were least consistent. The Spearman rank order correlation method was used to attempt to estimate the validity of the first seven of the Creekmore Scales. At the beginning of the discussion of the results obtained from the analysis of the scales, all Of the esti- mated reliability coefficients for the girls' and boys' re- sponses to the scales will be presented in table form. Fol- lowing the presentation of the reliability coefficients, the scales will be discussed individually. The eleven items which compose a scale will be listed and the over-all 57 58 reliability and the item analysis for the scale will be discu$sed first for the girls' responses and then for the boys' responses. After this procedure has been followed for each of the eight scales, the results of the Spearman rank Order correlations will be described as they relate to the validity of the first seven scales. The final portion of this chapter will deal with suggestions for improvements of the Creekmore Scales and will be based upon the results of the three operations which pertained to the examination of the reliability of the scales. Reliability Coefficients The estimation of the reliability coefficients of each of the scales Used Hoyt's method. The reliability co- efficients which were presented in Table l were based Upon the responses of 269 girls and 236 boys. Four of the scales, ”interest,” "psychological dependence," “special attention," and "social approval," had acceptable reliability coeffi- cients for both boys and girls. The "modesty” scale was sat- isfactorily reliable for the girls but the reliability co- efficient for the boys' responses to the scale was low. The reliability coefficients for the other three scales, "manage- ment," "comfort," and "aesthetic," were below the desired 59 level. For all of the scales except "modesty" and "manage- ment" the reliability coefficients for the boys were greater than or equal to those for the girls (see Table 1). TABLE 1.--Estimated reliability coefficients for 269 girls' and 236 boys' responses to the Creekmore Scales of Eight Clothing Variables. Estimated Reliability Scale Coefficients for Regponses of Girls Boys Interest .77 .81 Psychological Dependence .75 .78 Special Attention .71 .77 Social Approval .71 .71 Modesty .71 .65 Management .67 .65 Comfort .57 .61 Aesthetic .46 .58 Aesthetic Scale (-N 2. When I am shopping I choose clothes that I like even if they do not look the best on me. 3. It bOthers me when my shirt tail keeps coming out. 4. I consider the fabric texture with the line of the garment when choosing my clothes. l A negatively stated item. 60 5. I use clothing as a means of disguising phys- ical problems and imperfections through skillful use of color, line and texture. (-) 6. I wear clothes which have buttons or snaps missing. 7. I pay alot of attention to pleasing color combinations. 8. I keep my shoes clean and neat. 9. I carefully coordinate the accessories that I wear with each outfit. (-) 10. I wear the clothing fads that are pOpular in our school even though they may not be as becoming on me. 11. I spend more time than others coordinating the colors in my clothes. 12. I try to figure out why some people's clothes look better on them than others. The reliability coefficient for the girls' responses to the aesthetic scale, which pertained to the use of cloth- ing to achieve a pleasing or beautiful appearance including order and neatness, was .46. This scale had the lowest reli- ability of the eight scales. When the individual items were analyzed according to their discriminatory power and item-total correlation, some evidence was given for the low reliability. Items 6 and 8 failed to discriminate satisfactorily between the Upper and lower groups and had low item-total correlations. The majority of the girls responded in the same manner to each of the two items. With the clustering together of re- sponses at the more agreeable end of the continuum, the items were not suitable for distributing the subjects along 61 the continuum and the high scores obtained on these items did not correlate well with the total score for the subjects who had low total scores. The clustering of the responses in- dicated a social point of view concerning the behaviors men- tioned in both items. Although the content of the two items was related to the attitude the scale was measuring, the items did nOt fulfil the purpose of distributing the subjects along the continuum for this particular sample. Similar re- marks could be made concerning two negatively stated items, Items 2 and 10, although their discriminatory power and item-total correlations were higher than those for Items 6 and 8. Items 7 and 9 discriminated poorly between the Upper and lower grOUps of girls becaUse the majority of the girls responded in the same way. Items 3 and 12 had lower item- total correlations than the more satisfactory items of the scale, Items 4, 5, and 11 (see Table 2, p. 62). AThe estimated reliability coefficient for the 236 boys' responses to the "aesthetic" scale was .58. Just as for the girls' responses, this scale had the lowest reli- ability of the eight scales for the boys. The results of the item analysis gave some indica- tion of the reasons for the low reliability. The subjects' responses to Item 10 had a very low correlation with their 62 TABLE 2.--The results of the item analysis for the 269 girls' responses to the "aesthetic" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. l 2 3 4 5 Index to Corre- 0115' Col-P lation 2 4 10 so 85 120 .89 8 8 .344 3 24 4 ‘ 46 74 121 1.34 4 7 .380 4 18 37 82 79 53 1.85 1 l .560 5 f 30 35 73 89 42 1.60 2 3 .493 6 7 0 27 56 179 .64 lo 11 .208 7 O l 13 61 194 .74 9 5 .488 8 0 8 42 156 63 .51 1], 10 .260 9 2 4 30 120 113 .96 7 4 .461 10 5 21 78 90 75 .98 6 9 .264 ll 19 49 129 53 19 1.59 3 2 .560 12 30 29 115 69 26 1.23 5 6 .397 aDiscrimination index. bItem-total correlation coefficient. total scores for the scale and the least discriminatory power of all the items of the scale. Although Item 6 had similarly low discriminatory power, the item-total correlation was. higher than for that of Item 10. The lack of discrimina- tion for Item 6 resulted from approximately 71% of the boys 63 receiving the weight of 5 for their responses to the item. Although Items 10 and 6 pertained to the content of the "aesthetic" scale, they were not satisfactory for distribut- ing the subjects along a continuum which is the purpose of a scale. Items 5, 8, and 12 also had low discriminatory power and item-total correlations. Because the responses were not clustered, some other factor must have caused the responses to be inconsistent with the total scores for the scale. Items 3, 4, 7, 9, and 11 were satisfactory items be- cause they discriminated between the Upper and lower groups and had high item-total correlations (see Table 3, p. 64). Modesty Scale 13. Unlined sheer dresses or blouses reveal too much of the body. 14. I select clothes that are conservative in style. 15. I feel uncomfortable when someone has forgotten to close his or her zipper. 16. The first time in the season that I go to a pub- lic beach or pool I feel exposed in my bathing suit. 17. I choose clothing with small prints, even though a larger design looks equally good on me. 18. I feel embarrassed when I see someone in too low cut a dress. 19. I select clothes which do not call attention to myself in any way. ' 20. I feel embarrassed when I see someone in clothes that are too tight. 21. I like dark or muted colors rather than bright ones for my clothes. 64 22. I hesitate to associate with those whose clothes reveal too much of their body. 23. I wonder why some peOple wear clothes that are immodest. TABLE 3.--The results of the item analysis for the 236 boys' responses to the "aesthetic" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. 1 2 3 4 5 Index to Corre- D.I. Cor. lation 2 10 39 62 72 53 1.22 6 7 .399 3 31 36 55 59 55 1.91 2 5 .523 4 51 36 7O 51 28 2.09 1 1 .595 5 95 6O 52 20 9 1.03 8 8 .351 6 2 2 18 48 166 .61 10 10 .324 ll 9 42 103 71 1.53 5 4 .526 8 10 10 68 109 39 .92 9 6 .402 9 13 23 81 83 36 1.67 4 3 .589 10 9 33 76 63 55 .58 11 11 .174 11 51 81 71 25 8 1.69 3 2 .594 12 53 61 81 31 10 1.14 7 9 .349 For the girls' responses to the "modesty" scale the reliability coefficient was .71 which was the third highest reliability coefficient for the scales. The item analysis 65 revealed that only one item, Item 14, had poor discrimina- tory power and it had the lowest item-total correlation. The majority of the girls responded "usually" for this state- ment. All of the other items had satisfactory discrimination indices which ranged from 1.32 to 2.30. All of the other items with the exception of Item 21 had item-total correla- tion coefficients of .43 or greater (see Table 4). TABLE 4.--The results of the item analysis for the 269 girls' responses to the "modesty” scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. l 2 3 4 5 Index to Corre- D.I. Cor. lation 13 17 12 82 51 107 1.36 8.5 7 .456 14 1 9 60 124 75 .78 11 11 .360 15 9 14 37 69 140 1.36 8.5 4 .537 16 34 43 50 59 83 1.79 3 6 .520 17 41 45 104 51 28 1.42 7 9 .429 18 36 42 74 66 51 2.30 1 1 .688 19 43 63 78 54 31 1.45 6 8 .437 20 42 47 92 51 37 2.04 2 2 .649 21 33 37 105 49 45 1.32 10 10 .384 22 18 36 ‘85 80 50 1.48 5 5 .521 23 16 20 100 68 65 1.62 4 3 .578 66 The reliability coefficient for the “modesty" scale for the boys was .65 and it was tied with the "management" scale for the fifth highest reliability coefficient for the eight scales for the boys. The results of the item analysis showed that Item 17 had a low discrimination index and item- total correlation coefficient when compared with the indices and correlation coefficients for the other items. The re- sponses to Item 17 were not clustered about any specific re- sponse which indicated that the item lacked clarity, was am- biguous, or was not relevant. The other three items which had the lowest item-total correlation coefficients were 19, 21, and 14 and like Item 17 these items all pertained to the use of conservative or inconspicuous clothing while the re- maining items related to imoaesty through body exposure. This could be an indication that the boys in this sample considered the two aspects of modesty, conservatism and body exposure, separately. The remaining seven items of the scale showed satisfactory discriminatory power and item-total cor- relations (see Table 5, p. 67). 67 TABLE 5.--The results of the item analysis for the 236 boys' responses to the "modesty” scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. l 2 3 4 5 Index to Corre- D.I. Cor. lation 13 68 30 89 34 15 1.73 3 5 .520 14 8 10 85 96 37 1.09 10 8 .402 15 42 37 57 52 48 1.83 2 7 .488 16 81 51 52 22 30 1.92 1 6 .496 17 52 59 86 28 11 .91 11 11 .309 18 123 59 38 11 5 1.67 4 1 .621 19 41 48 78 52 17 1.14 9 10 .335 20 109 78 34 11 4 1.42 7 3 .573 21 31 27 100 44 34 1.34 8 9 .382 22 69 67 67 23 10 1.64 5.5 4 .539 23 44 52 99 29 12 1.64 5.5 2 .590 Interest Scale 24. My friends and I try each others clothes to see how we look in them. 25. I enjoy trying on shoes of different styles and colors. 26. I study collections of accessories in the stores to see what I might combine attractively. 27. I try on some of the newest ClOthes each season to see how I look in the styles. 28. I read magazines and newspapers to find out what is new in ClOthing. 68 29. It's fun to try on different garments and acces- sories to see how they look together. 30. 1 experiment with new or different "hair do's" to see how I will look. 31. I like to know what is new in clothing even if none of my friends care and I probably would not want to wear it anyway. 32. I try on clothes in shops just to see how I will look in them without planning to buy. 33. When I buy a new garment I try many different accessories before I wear it. 34. I am curious about why peOple wear the clothes they do. The reliability cOefficient for the girls' responses to the "interest“ scale was the highest of all the coeffi- cients for the eight scales for the girls. The estimated reliability for the responses of the 269 girls for the "in- terest" scale concerning the willingness to give attention, to investigate, manipulate or experiment with the putting together of the parts of a costume was .77. All of the items had discrimination indices of 1.21 or higher except Item 34, the theoretical item. Item 34 had a discrimination index of 1.19 and an item-tOtal correlation coefficient of .41 which indicated that this item could pos- sibly be improved (see Table 6, p. 69). The reliability coefficient for the boys' responses to the ”intereSt" scale was .81 which was the highest reli- ability coefficient obtained for all the scales. The item analysis substantiated this high reliability for all of the 69 discrimination indices were above 1.30 except for Item 24 which had a discrimination index of 1.06 and all of the item-tOtal correlation coefficients were .50 and above. The low discriminatory power for Item 24 resulted from the ma- jority, 73%, of the boys indicating that they "almost never" tried on each other's clothes to see how they looked in them (see Table 7, p. 70). TABLE 6.--The results of the item analysis for the 269 girls' responses to the "interest" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. 1 2 3 4 5 Index to Corre- D.I. Cor. lation 24 72 42 79 37 39 1.93 1 9 .514 25 1o 13 38 72 136 1.63 6 1 .628 26 13 29 61 88 78 1.73 2 3 .605 27 10 13 59 9o 97 1.63 6 2 .620 28 11 20 66 78 94 1.63 6 7 .559 29 3 2 52 104 108 1.21 10 4 .593 30 6 15 74 75 99 1.36 9 10 .481 31 12 16 76 88 77 1.47 8 8 .533 32 56 81 81 32 19 1.66 4 6 .560 33 19 58 107 59 26 1.67 3 5 .584 34 26 41 108 66 28 1.19 11 11 .410 70 TABLE 7.--The results of the item analysis for the 236 boys' responses to the "interest" scale. Frequency of Choice Discrim- Rank Item- Itenl of Each Weight ination According total No. 1 2 3 4 5 Index to Corre- D.I. Cor. lation 24 172 37 16 7 4 1.06 11 8 .537 25 66 49 53 47 21 1.70 7 7 .540 26 56 54 64 46 16 1.99 4.5 4 .627 27 48 52 59 56 21 2.16 2 2 .664 28 76 50 59 34 17 2.19 1 3 .641 29 58 67 57 33 11 2.11 3 1 .750 30 100 59 46 16 15 1.84 6 9 .547 31 56 59 7O 4O 11 1.97 4.5 5 .625 32 140 58 23 5 10 1.31 10 10 .507 33 91 78 41 23 3 1.47 9 6 .578 34 71 51 75 28 11 1.53 8 11 .502 Comfort Scale 35. The way my clOthes feel on my body is important to me. There are certain textures in fabrics that I like and especially try to buy, for example, soft, fuzzy, sturdy, smooth. I am more sensitive to temperature changes than 36. 37. 38. Others and I have difficulty being comfortable in my clOthes as a result. I wear my pants or slacks with an easy fit even if tight ones are fashionable. 71 39. I get rid of garments I like because they are not cwlfortabl e . 40. I find it difficult to buy clothes suitable to the temperatUre. 41. I would buy a very comfortable bathing suit even if it were nOt the current style. 42. I avoid garments that bind the Upper arm. 43. I am irritable if my clOthes are uncomfortable. 44. I am extremely sensitive to the texture of the fabrics in my clothing. 45. I wonder what makes some clothes more comfort- able than others. The reliability coefficient for the "comfort" scale, which referred to the use of clothing to achieve comfort as related to temperature, physical response to certain tex- tures, and acceptance of tightness or looseness in garments, was .57 for the girls. This reliability coefficient was the second lowest of all the reliability coefficients for the scales to which the girls responded. The item analysis showed that four of the statements, Items 35, 36, 40 and 42, had discrimination indices below .90 and correlation coefficients at or below .38. The lack of discriminatory power for Item 35 resulted from the major- ity or 64% of the girls indicating that the way their clOthes felt on their bodies was "almost always" important to thGMu Similarly, for Item 36, 41% of the girls "usually" tried to buy certain textures and 35% "almost always” liked and tried to buy certain textures. For Item 42, 37% "almost always" 72 and 33% "usually" avoided garments which bound the Upper arm. All of these results indicated that the majority of subjects in this sample responded in the same way and that the items were not distributing subjects along the continuum as was desirable for items of a scale. The results for Item 40 indicated that there was some ambiguity or inconsis- tency in the statement because the responses to the item were not consistent with the responses to the other state- ments of the scale. The other seven items of the "comfort" scale had discrimination indices of 1.22 or greater and correlation coefficients of .40 or greater which indicated that the sub- jects' responses to each of these items were consistent with their overall responses to the items of the scale (see Table 8, p. 73). The reliability coefficient for the boys' responses to the "comfort" scale was .61 and this coefficient was the second lowest of all the reliability coefficients for the scales to which the boys responded. The results of the item analysis showed that only two items, 37 and 40, had discrimination indices below 1.20 and item-total correlation coefficients below .40. Both of these items pertained to comfort as it related to tempera- ture. Perhaps this aspect of comfort was not considered as 73 relevant to the boys as the physical response to textures and tightness or looseness of garments as related to comfort. All the Other nine items of the scale had discrim- ination indices of 1.22 or higher and item-total correlation coefficients of .42 or above (see Table 9, p. 74). TABLE 8.--The results of the item analysis for the 269 girls' responses to the “comfort" scale. Frequency of Choice Discrim- Rank Item- Item - of Each Weight ination According total No. l 2 3 4 5 Index to Corre- D.I. Cor. lation 35 3 3 18 74 172 .74 11 8 .380 36 3 ll 50 111 94 .75 10 11 .294 37 68 74 84 27 16 1.22 7 6 .413 38 14 45 72 72 66 1.47 4 5 .471 39 29 61 114 44 21 1.33 5 4 .500 40 56 96 89 17 11 .82 9 9 .370 41 62 74 70 49 14 1.30 6 7 .404 42 8 20 52 88 101 .89 8 10 .318 43 13 28 68 91 69 1.49 2.5 3 .522 44 42 65 97 43 22 1.68 1 1 .569 45 28 63 107 50 21 1.49 2.5 2 .537 74 TABLE 9.--The results of the item analysis for the 236 boys' responses to the "comfort" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According tOtal No. 1 2 3 4 5 Index to Corre- D.I. Cor. lation 35 9 9 43 85 90 1.22 9 8 .445 36 17 27 66 87 39 1.41 7 4 .491 37 74 76 61 17 8 1.14 10 10 .380 38 32 31 65 71 37 1.69 1 2 .508 39 55 68 70 29 14 1.52 4.5 6 .459 40 58 94 66 16 2 .66 ll 11 .230 41 63 55 62 37 19 1.48 6 9 .423 42 27 27 56 76 50 1.52 4.5 7 .457 43 17 36 80 65 38 1.58 2 1 .551 44 56 79 60 28 13 1.56 3 3 .504 45 54 68 72 33 9. 1.38 8 5 .466 Special Attention Scale 46. When new fashions appear on the market, I am one of the first to own them. 47. I have clothes that I don't wear because every- one else has them. 48. I like to be considered an Outstanding dresser by my friends. 49. I try to keep my wardrobe in line with the latest styles. 50. I go to nearby cities to shop for better fashions. 51. I try to buy clothes which are very unusual. 75 52. I avoid wearing certain clothes because they do not make me feel distinctive. 53. I enjoy wearing different clothing even though I attract attention. 54. I try to buy clothes with the best labels. 55. I wear different clothes to impress people. 56. I am interested in why some people choose to wear such outlandish clothes. The reliability coefficient for the girls' responses to the "special attention“ scale which pertained to seeking prestige and status through the use of clOthes was .71 which was the same as the coefficients for the "modesty'' and "social approval" scales. 7 1 Only two of the discrimination indices for the items of the "special attention” scale were below 1.20 and only one item-totalcorrelation was below .42 which indicated that nine of the items yielded consistent responses. The two less satisfactory items were Item 56, the theoretical state- ment, which had a low discrimination index and a low item- tOtal correlation and Item149 which had a low discrimination index. The responses to Item 56 appeared to be inconsiStent with the responses to the Other items of the scale and for Item 49 the majority of the girls, 62%, "usually'l or "almost always" tried to keep their wardrobes in line wfith the latest styles (see Table 10, p. 76). TABLE 10.--The results of the item analysis for the 269 76 girls' responses to the "special attention” scale. V— Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. 1 2 3 4 5 Index Corre- D.I. Con lation 46 51 103 93 15 7 1.42 7 1 .619 47 90 87 57 24 11 1.74 1. 3 .581 48 16 26 68 94 65 1.62 4 5 .543 49 7 18 77 108 59 1.19 10 7 .507 50 52 54 85 45 33 1.60 5 8 .490 51 78 85 71 22 13 1.74 1. ' 2 .615 52 29 69 114 43 14 1.23 9 10 .428 53 53 74 67 29 12 1.40 8 9 .479 54 31 52 86 71 29 1.63 3 6 .509 55 44 54 111 42 18 1.53 6 4 .555 56 22 48 104 53 42 .96 11 11 .297 For the boys' responses to the ”special attention" scale the reliability coefficient was .77 which was the third highest reliability coefficient. The item analysis showed that all of the items except Items 56 and 47 discriminated very well and only Item 56 had a low item-total correlation. The results for the analysis of the boys' responses to Item 56 were very similar to those of the girls for the same item 77 in that the responses to the item were inconsistent with the responses to the other items of the scale. For Item 47, 87% of the boys indicated that they "seldom" or "almost never" had clOthes which they never wore because everyone else had them. All of the Other nine items of the “special attention" scale had discrimination indices of 1.55 or above and item- tOtal correlation coefficients of .51 or above(See Table 11). TABLE ll.--The results of the item analysis for the 236 boys' responses to the "special attention" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total .No. 1 2 3 4 5 Index to Corre- D.I. Con lation 46 73 83 51 22 7 1.83 2.5 1 .673 47 112 83 31 9 1 1.08 10 8 .521 48 34 46 85 42 29 1.75 4 3 .611 49 35 36 61 75 29 2.09 1 2 .643 50 93 51 59 19 14 1.83 2.5 4 .609 51 A 107 64 41 17 7 1.72 5 5 .604 52 54 59 81 27 14 1.58 8 9 .504 53 72 64 69 18 13 1.70 6 6 .571 54 41 51 75 51 18 1.55 9 10 .489 55 56 6o 73 33 14 1.61 7 7 .530 56 46 60 79 42 9 1.05 11 11 .339 57. 58. 59. (-)60. 61. 62. 63. 64. 65. 66. 67. 78 Management Scale I plan for and prepare clothes to wear several days in advance. I see that my out-of-season clothing is cleaned and stored. I look over the clothing in my wardrobe before each season so that I know what I have. I am enticed into buying garments I like with- out having anything to go with them. I enjoy trying to get the most for my money in clothing purchases. I wear a raincoat or carry an umbrella to pro- tect my clothes in rainy weather. I have something to wear for any occasion that occurs. I have long-term ideas for purchasing more ex- pensive items of clothing such as coats or suits. I carefully plan every purchase so that I know what I need when I get to a store. I am more concerned about the care of my ClOthing than my friends are about theirs. I try to find out how I can save as much time, energy and money as possible with my clothes. The reliability coefficient for the girls' responses to the "management" scale which measured attitudes toward the thoughtful and careful use of resources including the use of time, money and energy in planning, buying and using clothing was estimated to be .67 and was the sixth highest reliability coefficient for the girls. The item analysis showed that only one item, Item 60, had a low discrimination index and item-total correlation coefficient. Item 60 was negatively stated and could have been ambiguous to the sub- jects because the results of the item analysis showed that 79 the responses to this item were inconsistent with the re- sponses to the Other items. Three other items, 61, 66, and 64, had low discrimination indices because each had a clus- tering of responses to the items but they had satisfactory item-total correlation coefficients. The other seven items had discrimination indices above 1.22 and item-total corre- lation coefficients of at least .47 (see Table 12). TABLE 12.--The results of the item analysis for the 269 girls' responses to the ”management" scale. V_f Frequency of Choice Discrim- Rank Item- Item - of Each Weight ination According tatal No. 1 2 3 4 5 Index to Corre~ D.I. Con lation 57 46 47 91 57 28 1.59 3 8 .466 58 3 14 50 110 92 1.27 4.5 4 .536 59 6 14 35 74 140 1.27 4.5 3 .548 60 21 32 116 64 36 .96 11 11 .288 61 6 12 54 87 110 1.10 10 10 .439 62 47 40 59 64 59 2.01 1 2 .550 63 10 37 88 105 29 1.23 6 5 6 .491 64 14 49 113 64 29 1.16 8 9 .458 65 8 36 83 103 39 1.62 2 l .606 66 3 29 118 97 22 1.12 9 7 .486 67 6 24 88 106 45 1.23 6.5 5 .497 80 For the boys' responses to the "management" scale the estimated reliability coefficient was .65 which was the same as the reliability coefficient for the "modesty" scale. The item analysis showed that one item, Item 60, was very inconsistent with the other items of the scale because it had very low discriminatory power and a very low item-total correlation coefficient. This was the same item that was in- consistent for the girls' responses. Item 62 and Item 63 had low discriminatory power because the responses were clustered together in each of these items. For Item 62, 55% of the boys "almost never" wore a rain coat or carried an umbrella to prOtect their clOthes from the rain; for Item 63, 43% "usually" had something to wear for any occasion that occurred. The other eight items of the "management" scale had discrimination indices above 1.25 and item-total correlations at or above .43 (see Table 13, p. 81). 81 TABLE l3.~-The results of the item analysis for the 236 boys' responses to the "management” scale. v v Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According t0ta1 No. 1 2 3 4 5 Index to Corre- D.I. Con lation 57 87 59 57 19 14 1.72 3 5 .504 58 29 42 57 73 35 2.09 2 2 .639 59 43 41 66 60 26 2.19 1 1 .650 60 7 14 91 69 55 .42 11 11 .066 61 17 25 80 69 45 1.36 7 7 .458 62 129 53 30 16 8 1.06 9.5 9 .380 63 17 33 65 102 19 1.06 9.5 10 .357 64 40 69 82 37 8 1.27 8 8 .430 65 28 39 73 65 31 1.61 5 6 .503 66 26 35 108 49 18 1.53 6 3 .569 67 16 43 91 63 23 1.67 4 4 .563 Social Approval Scale 68. I check with my friends about what they are wearing to a gathering before I decide what to wear. 69. I would rather miss something than wear clothes which are n0t really apprOpriate. 70. I feel more a part of the grOUp if I am dressed like my friends. 71. I wear clothes that everyone is wearing even though they may nOt look as good on me. 82 72. I am uncomfortable when my clothes are different from all others at a party. 73. I try to dress like Others in my group so that people will know we are friends. 74. I get new clOthes for a special occasion if the clothes I have are not the type my friends will be wearing. 75. I have gone places and then wished after I got there that I had not gone because my clothes were nOt suitable. (-)76. I wear what I like even though some of my friends do not approve. 77. When I buy a new article of clothing I try to buy something similar to what my friends are wearing. 78. When someone comes to school dressed unsuitably I try to figure out why he is dressed as he is. The reliability coefficient for the 269 girls' re- sponses to the ”social approval" scale measuring attitudes toward the use Of clothing to attain a feeling Of belonging- ness or the approval Of others in a particular role was .71 which was the same as the reliability coefficients for the “modesty" and "special attention" scales. Again it was a negatively stated item, Item 76, which had the lowest item- total correlation and discriminatory power. All of the other items except Item 78 had item-tOtal correlations of .45 or greater and all of the items except Item 76 had discrimina- tion indices of 1.05 or greater (see Table 14, p. 83). The reliability coefficient for the boys' responses to the ”social approval" scale was .71 which was the fourth highest reliability coefficient for the eight scales for the 83 boys. Only one item had an index of discrimination of less than 1.20 and an item-total correlation of less than .44. This item was negatively stated Item 76 which also ranked lowest when the girls' responses were analyzed. All Of the other items solicited consistent responses from the 236 bOys (see Table 15, p. 84). TABLE 14.--The results of the item analysis for the 269 girls' responses to the "social approval'I scale. Frequency Of Choice Discrim- Rank Item- Item of Each Weight ination According total No. l 2 3 4 5 Index to Corre- D.I. Con.1ation 68 10 24 65 9O 80 1.11 8 8 .462 69 9 23 87 81 69 1.30 6 9 .451 70 6 27 62 73 101 1.95 1 1 .697 71 77 112 61 15 4 1.29 7 5 .523 72 14 28 73 79 75 1.67 3 4 .562 73 71 95 72 21 10 1.33 5 6 .480 74 32 54 112 49 22 1.70 2 2 .606 75 23 56 142 35 13 1.05 10 7 .473 76 .24 67 113 50 15 .93 11 11 .336 77 29 57 110 57 16 1.55 4 3 .578 78 32 47 109 64 17 1.07 9 10 .373 84 TABLE 15.~~The results of the item analysis for the 236 boys' responses to the "social approval" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. 1 2 3 4 5 Index to Corre- D.I. Cor.lation 68 54 63 68 39 12 1.64 3 7 .513 69 28 47 73 61 27 1.33 7 10 .439 70 11 25 60 85 55 1.50 5.5 4 .545 71 42 84 84 16 9 1.20 10 6 .513 72 29 38 71 59 39 1.66 2 9 .508 73 48 78 67 36 7 1.56 4 1 .589 74 48 7O 79 26 13 1.50 5.5 5 .541 75 39 73 94 27 3 1.30 8 8 .512 76 23 50 94 44 25 .88 11 11 .321 77 12 49 88 81 6 1.29 9 3 .557 78 52 74 62 37 11 1.73 1 2 .563 Psychologjcal Dependence Scale 79. Certain clothes make me feel more sure of myself. 80. I decide on the clOthes to wear according to the mood I'm in that day. 81. Days when I feel low I wear my gayest clothes. 82. I “dress-up" to make an Ordinary Occasion seem more exciting. 83. I am aware Of being more friendly and out going when I wear particular clOthes. 84. I feel and act differently according to whether I am wearing my best school clothes or not. 85 85. I buy clothing to boost my morale. 86. I get bored with wearing the same kind Of clothes all the time. 87. I have more self confidence when I wear my best school clothes. 88. When things are not going well I like to wear brighter colors. 89. I wonder why some clOthes make me feel better than others. The estimated reliability coefficient for the girls' responses to the "psychological dependence” scale was .75, the second highest reliability coefficient for the eight scales to which the girls responded. All of the items ex- cept Items 82 and 86 had discrimination indices Of 1.25 or greater and item-total correlations Of .48 or greater which indicated that the majority of the items were consistent with the total scale. Item 86 had low discriminatory power because 37% of the girls "usually" and 31% "almost always" got bored when they wore the same kind of clothes all the) time; Item 82 had lower discriminatory power than the re- maining nine items because 40% of the girls indicated that they "sometimes" dressed up to make an ordinary occasion seem more exciting (see Table 16, p. 86). 86 TABLE l6.-~The results of the item analysis for the 269 girls' responses to the "psychological dependence" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total NO. 1 2 3 4 5 Index to Corre- D.I. Con lation 79 3 9 42 102 113 1.25 9 8 .479 80 24 62 92 55 36 1.67 3 5 .564 81 38 73 103 39 16 1.36 7 9 .476 82 28 67 107 49 18 1.08 10 10 .403 83 12 25 84 87 61 1.78 2 2 .651 84 21 37 86 79 46 1.64 4 4 .592 85 22 47 135 48 17 1.55 5 3 .625 86 3 15 68 99 84 .95 11 11 .331 87 10 19 62 90 88 1.92 1 1 .685 88 30 59 134 34 12 1.30 8 6 .563 89 24 37 114 57 37 1.41 6 7 .491 The estimated reliability coefficient for the boys' responses to the "psychological dependence” scale was .78 which was the second highest reliability coefficient for the eight scales to which the boys responded. All of the items of this scale appeared to be very satisfactory since the lowest discrimination index was 1.27 and the lowest item- total correlation coefficient was .43 (see Table 17, p. 87). 87 TABLE l7.--The results of the item analysis for the 236 boys' responses to the "psychological dependence" scale. Frequency of Choice Discrim- Rank Item- Item of Each Weight ination According total No. l 2 3 4 5 Index to Corre- D.I. Con lation 79 15 20 84 76 41 1.27 11 10 .445 80 59 57 70 37 13 1.67 5 7 .536 81 72 79 62 18 5 1.50 7 5 .560 82 50 71 78 32 5 1.47 8 9 .517 83 22 34 100 60 20 1.84 2.5 2 .676 84 18 58 95 46 19 1.55 6 6 .547 85 62 78 58 29 9 2.00 1 1 .676 86 20 28 81 61 46 1.38 9 11 .425- 87 15 42 68 77 34 1.84 2.5 3 .647 88 50 102 62 14 8 1.28 10 8 .535 89 52 41 91 40 12 1.78 4 4 .581 Validity An attempt was made to validate the first seven of the Creekmore Scales of Eight Clothing Variables. The subject's rank ordering of seven words and phrases in a criterion measure, completed after he had responded to the eight scales, was compared with the rank ordering of the 88 importance of the first seven variables according to the summated scores that he Obtained on the scales. The Spearman rank order correlation method was used to determine the de- gree of association between the two rankings for each of the 199 girls and 151 boys who completed both the scales and the criterion measure. If the resulting correlation co- efficients were equal to or exceeded .7141 the relationship between the two rankings was considered significant at the .044 level for a One-tailed test. However, because so many rank order correlation coefficients were calculated, it was not unreasonable to expect that 4.4% of the total number calculated for the boys and 4.4% of the tOtal number calcu- lated for the girls could have been significant because of chance alone and nOt because of an aCtual relationship be- tween the variables. The results Of the calculations of the Spearman rank order correlation coefficients for the girls showed that 34 Of the 199 correlations were significant at the .044 level or above. 0f the number of significant correlations nine could have occurred by chance alone. Consequently, the re- maining 25 correlations or 12.5% of the total correlations for the girls could be considered significant relationships. 1 Edwards, Statistical Methods . . ., p. 513. 89 This low proportion Of significant results indicated that there was little association between the two rankings for the girls in this sample. When the same procedure was followed for the re- sponses of the 151 boys, 20 correlation coefficients equaled or exceeded .714 but Of this number seven could have oc- curred by chance alone. The remaining 13 correlation coef~ ficients or 8.6% of the tOtal number of rank correlation co- efficients calculated were significant. These results also indicated very little correspondence between the two rank- ings. If the criterion measure with which the scores Of the first seven scales were compared had been a truly reli- able and valid criterion, the results would have indicated that the first seven of the Creekmore Scales of Eight ClOth- ing Variables were not valid. However, such a conclusion could not be justified in this case because the reliability and validity of the criterion measure were not known. What was more likely was that the criterion measure was even less reliable and valid than the scales. As noted earlier Oppen- heim indicated that single statements were less reliable than groups Of statements for locating individuals along a 90 continuum according to their attitudes.1 A second possible explanation for the lack Of agreement between the two rank- ings was that the words and phrases considered only portions Of the variables because each of the clothing variables was. Operationally defined to include several aspects. For exam- ple, "comfortable" in the criterion measure corresponded to the "comfort“ scale which was concerned with comfort as re- lated to temperature, physical response to certain textures, and tightness or looseness in garments. With these consider- ations in mind, no conclusions could be drawn about the validity of the scores obtained from the administration of the Creekmore Scales of Eight Clothing Variables. Recommendations for Improvements One of the objectives of this study was to propose recommendations for the improvement of the Creekmore Scales of Eight Clothing Variables. The recommendations proposed in this section were based on the results of the analyses of the responses made by 269 girls and 236 boys. To the extent that the underlying assumptions were legitimate, the follow- ing suggestions should offer methods for improving the reli- ability of some of the Creekmore Scales. For the specific sample of girls whose responses were analyzed in this study, some of the items of the a V 1Oppenheim, p. 73. 91 "aesthetic" scale represented social points of view for the majority of the subjects. Therefore, although these items were related to the tOtal content of the scale, they did nOt distribute the subjects along a continuum according to their attitudes which is the purpose of the scale which measures behavioral indicants of attitudes. Rather than recommending immediate deletion of Items 2, 6, 7, 8, 9, and 10 because they represented social points of view, the “aesthetic” scale should be administered to several other samples. If the same items still fail to dis- criminate among the individuals, they should be replaced by more discriminating items so that the rank ordering of the subjects according to their responses to the items of the scale will be more consistent with the rank of the subjects according to their total summated scores. Because the reliability was low for the ”aesthetic" scale for the boys, some revisions should be made. Item 10’ should be reworded so that it does not solicit inconsistent. responses. If Item 6 continues to yield similar responses when used with other samples of boys, it should be changed or deleted so that it does not reflect a social point of view. Items 5, 8, and 12 should also be re-examined and altered so that they solicit responses which are more con- sistent with the remaining items of the scale. 92 When the "modesty“ scale is administered to other samples of girls and the results analyzed, particular notice should be paid to Item 14. If it continues to yield a clus- tering of responses it should be altered so that it produces responses which will distribute subjects along the scale in a manner consistent with the other items of the scale. For the "modesty” scale for the boys the relevance of Item 17 should be examined. For this particular sample of boys this item did not solicit responses which were consis- tent with their responses tO Other items of the scale. The possibility of dividing the "modesty" scale into two parts, one concerning body exposure and the other pertaining to conservatism in dress, should also be investigated because the boys' responses in this study showed a separation be- tween these two aspects. All of the items of the "interest" scale for girls appeared to be very satisfactory but the theoretical item, Item 34, should be reuexamined to see if a more relevant aspect of interest could be used. The only recommendation which can be made for making items more discriminating and Consistent with the total "interest" scale for the boys is that responses made by boys in other samples to Item 24 be carefully examined. If the majority of boys continue to report that they almost never 93 try on each other‘s clothes, this item should be replaced by a more discriminating item. Three of the items of the "comfort" scale, Items 35, 36, and 42, yielded clusterings of responses so that the individuals were not distributed along the continuum in a manner consistent with their responses to the other items Of the scale. If similar results occur when the scale is ad- ministered to other samples of girls, these items should be altered or replaced with more satisfactory items. The ter- minology Of Item 40 should be changed so that it is less ambiguous. The relevancy of Items 37 and 40 for the boys should be investigated because in this study the boys' re- sponses to these items were not consistent with their re- sponses to the other items of the "comfort" scale. For the "special attention" scale for the girls a more pertinent item for the theoretical aspect of the use of clothing to gain prestige and status should be formulated. If Item 49 continues to relate to a social point of view for other samples of girls it should be revised so that it be- comes more discriminating. A similar recommendation can be made for the improve- ment of the reliability of the "special attention“ scale for boys as was made for the girls. Item 56 should be revised so 94 that it solicits more discriminating responses which are consistent with the other items of the scales. If Item 47 continues to yield a cluster of responses for other samples of boys, it should be replaced by a more discriminating item. For both the boys and girls Item 60 of the ”manage- ment” scale did not yield responses which were consistent with the total scale. Perhaps if this negatively stated item were reworded it would be more effective. The responses of girls in other samples to Items 61, 64, and 66 should be carefully investigated and if their responses still tend to cluster about single responses these items should be re- placed by more discriminating items. Similarly, if most of the boys in other samples continue to respond in the same way to Items 62 and 63 these items should be revised. For both the boys' and girls' replies to Item 76 the discriminatory power and item-total correlation were low which indicates that this negatively stated item shoud be revised so that it yields more consistent responses. The only item of the “psychological dependence” scale which appeared to be inconsistent with the other items of the scale was Item 86 to which the majority of the girls responded in the same manner. If this item continues to re- late to a social point of view it should be replaced by a 95 more differentiating item. All of the items for the ”psychological dependence" scale for boys appeared to be satisfactory so that no recom- mendations were necessary. All of the recommendations which have been made in this section were based Upon the analysis of the responses of the subjects in this sample and may not be applicable or apprOpriate when the scales are administered to other sub- jects. This was the reason for the constant repetition of the idea that the responses of the subjects in other samples be analyzed before the items which represented social points of view are deleted or altered. CHAPTER VI SUMMARY AND CONCLUSIONS The study of the socio-psychological meanings in- herent in the use of clothing has been a recent development in the field of clothing. Because the development is recent the total area of attitudes toward and behaviors related to clothing is relatively unexplored and very few techniques have been refined sufficiently for obtaining data concerning attitudes toward clothing. Thus, there exists a great need for reliable and valid methods of obtaining data pertaining to attitudes toward clothing so that these attitudes can be more thoroughly investigated. One of the major objectives of a recent project at Michigan State University was the composition of eight scales measuring attitudes toward and behaviors related to clothing. The theoretical basis for these scales originated in Creekmore's investigation of the relationship between clothing behaviors and general values which resulted from the striving to satisfy needs. The researchers compiled items from studies by Creekmore, Brady and Sharpe and items 96 97 which they developed which pertained to the operationally defined clothing variables. It was the desire of the re- searchers of this project that the final "Importance of Clothing" questionnaire which was a composite of eight Likert-type summated rating scales become the basis for standardized scale for measuring attitudes toward clothing. To assess the adequacy of the Creekmore Scales of Eight Clothing Variables, the responses of 269 girls and 236 boys in grades ten, eleven and twelve of a high school located in a mid-western industrial city of the United States were analyzed. The reliability and validity of the responses were estimated and suggestions for improvements of the scales were proposed. The methods selected for the inves- tigation of the reliability and validity were Hoyt's analyn sis of variance technique for estimating reliability, item analysis consisting of discriminatory power and item-total correlations for investigating the contribution each item made to the total scale, and the Spearman rank order corre- lation for attempting to ascertain the validity of the scales. These procedures were followed separately for the responses of the boys and the girls of this sample. Certain limitations were kept in mind for the inves~ tigation of the reliability and validity of the Creekmore Scales. For the investigation of reliability the intervals 98 were assumed to be equal although the Likert method of con~ structing summated rating scales does not guarantee more than ordinal measurement. Secondly, the relationship between the weights for the subjects' responses to the individual items and their total summated scores for a scale was assumed to be linear so that the Pearson product-moment cor— relation method could be used for investigating item-total correlations. To the degree that these assumptions were justified, the results of this study are significant. Fin- ally, the results of this analysis applied only to the responses of the subjects of this sample because the sample was not randomly selected. With these limitations in mind the analysis of the responses to the Creekmore Scales was conducted. The results of the procedures concerning reliability showed that several of the scales attained or approached a satisfactory degree of reliability. The most satisfactory reliability coefficients were obtained for the "interest," "psychological dependence," "special attention," and "social approval" scales while the reliability coefficients for the "management," "comfort," and "aesthetic" scales were some— what below the acceptable level of reliability. In all of the scales but two, "modesty" and "management," the relia- bility coefficients for the boys' responses were higher than for the girls. 99 On the basis of the item analysis, revisions were suggested for improvement of some of the scales. When the items had poor discriminatory power because the majority of subjects responded in the same manner to the items indicat- ing a social point of view, revisions should not be made immediately. Instead the scales containing these items should be administered to many samples and if the items continue to solicit similar responses from the majority of the subjects, these items should be replaced by more dif— ferentiating items so that subjects can be distributed along the continuum according to their attitudes. For the other items which had low discriminatory power and low item~total correlations but did not appear to represent social points of view, suggestions were made to reword, clarify and alter these statements so that they might solicit responses more consistent with the subject's total score for the scale. All of the negative statements, Items 2, 6, 10, 60 and 76, had low discriminatory power and item-total cor- relations. Although many authorities had suggested that scales should consist of half negatively stated items and half positively stated items, in these scales the negative statements failed to produce responses which were consistent with the other items of the scales. The lack of consistency of responses to the negatively stated items could be an indication of response set or ambiguous wording of the items. 100 The wording of the negatively stated items should be care- fully revised so that the items do not represent ambiguous or double~barre1ed situations and the subjects should be encouraged to answer sincerely and honestly. The majority of the Creekmore Scales of Eight Cloth- ing Variables provided a useful method for measuring the consistency of adolescent boys' and girls' responses to statements pertaining to clothing. With revisions of the less satisfactory items of the remaining scales, these scales can be improved and become useful for obtaining information about adolescents’ reactions to statements about clothing. These scales, if properly used and interpreted, should be- come valuable techniques for the investigation of attitudes toward and behaviors related to clothing. No conclusions could be drawn about the validity of the Creekmore Scales of Eight Clothing Variables. There was only slight correspondence between the ranking of the items of the criterion measure and the ranking of the first seven variables according to the subject's summated scores for the seven scales. Because the reliability and validity of the criterion measure were not known and because the words and phrases did not cover the entire aspects of each of the operationally defined clothing variables, there was no justification for stating that the scales were not valid; 101 nor was there any justification for saying that the scales were valid. Recommendations The major recommendation for further research stud- ies which evolved from this study was that the suggestions for improvements be followed and the revised scales admin- istered to many samples of adolescents. The results obtained from these samples should be analyzed in the same manner as they were in this study and comparisons made to determine if the improvements were satisfactory. If necessary, fur~ ther revisions should be made so that the scales become very valuable techniques for measuring attitudes toward clothing. 4 The "modesty" scale should be investigated more thoroughly to ascertain if it can be divided into two parts, one for body exposure and the other for conservatism in dress. A more detailed inquiry should be conducted so that an attempt can be made to validate the scales. The existing criterion measure could be revised so that the words and phrases included all of the aspects of the operationally defined variables and placed at the beginning of the ques- tionnaire. If these revisions failed to produce satisfactory 102 results, some other criterion should be sought so that the validity of the Creekmore Scales of Eight Clothing Variables can be established. B I BL IOGRAP HY BIBLIOGRAPHY Brady, Betty L. I‘Clothing Behavior: Refinement of a Measure and Relationships with Social Security and Insecur~ ity for a Group of College Women." Unpublished Master's thesis, Department of Clothing and Tex- tiles, Pennsylvania State University, 1963. Bridgman, C. S. "The Relation of the Upper-Lower Item Dis- crimination Index D to the Bivariate Normal Corre- lation Coefficient, " Educational and Psychological Measurement, XXIV (Spring, 71964), 857- 790. 77 Creekmore, Anna Mary. "The Relationship of Clothing to Se1f~Concept and to Attitudes toward Clothing." (Research in progress for Michigan State University Agricultural Experiment Station Project 784). "Clothing Behaviors and Their Relationship to General Values and to the Striving for Basic Needs." Unpublished Ph.D. thesis, Department of Clothing and Textiles, Pennsylvania State University, 1963. Cronbach, Lee J. Essentials of Psychologjcal Testing_. New York: Harper and Brothers, 1949. Cureton, E. E. "The Upper and Lower 27% Rule,” Psycho- metrika, XXII (September, 1957), 293-296. Dickey, Lois Edith. "Projection of the Self through Judg- ments of Clothed~Figures and Its Relation to Self~ Esteem, Security-Insecurity and to Selected Clothing Behaviors." Unpublished Ph.D. thesis, Department of Clothing and Textiles, Pennsylvania State Uni- versity, 1967. Downie, N. N. and Heath, R. W. Basic Statistical Methods. New York: Harper and Row, 1965. Ebel, Robert L. MeasuringfiEducational Achievement. Engle~ wood Cliffs, New Jersey: Prentice-Hall, Inc. , 1965. 104 105 Statistical Methods for the Behavioral Edwards, Allen L. - - - - - New7Yorkx Holt, Rinehart an7d Winston, Sciences. 19 . ___. Techniques oiAttitvude ScalelConstruthiovn. New York: Appleton-Century-Crofts, Inc., 1957. Fiske, Donald W. "Some Hypotheses Concerning Test Adequacy," Educational and Psychological Measurement, XXVI, No. l (1966), 69-88. 7 Garrett, Henry E. Elementary Statistics. New York: David McKay Company, Inc., 1964. Guilford, J. P. Psychometric Methods, New York: McGraw- Hill Book Company, Incf, 1954. Hacklander, Effie Hewitt. "The Relationship of Concern for the Body and the Clothing of Adolescents." Unpub- lished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 1968. Hays, William L. Statisticsfifoersychologists. Holt, Rinehart and Winston, 1965. New York: Helmstadter, G. C. 9T1”CIP]°S.9f2PEYCh9JPEl9a]1M§95”59m305° New York: Appletonécentury—Crofts, 1964. Hoyt, Cyril J. and Krishnaiah, P.R. "Estimation of Test Reliability by Analysis of Variance Technique," Journal of Ex erimental Education, XXVIII (March, 19307, 257-259. Hoyt, Cyril. "Test Reliability Estimated by Analysis of Variance," Psychometrika, VI (1941), 153-160. Humphrey, Carolyn Andree. "The Relationship of Stability of Self Concept to the Clothing of Adolescents." Unpublished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State Univer- sity, 1967. 106 Hundley, Winifred Sue. "The Relationship of Clothing to Social Class, High School Position and Status Incon- sistency of Adolescent Boys and Girls." Unpublished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 1967. Kelley, Truman L. "The Selection of the Upper and Lower Groups for the Validation of Test Items," Journal of Educational Psychology, XXX (1939), 17- 4. Kerlinger, Fred N. Egundations of Behaviorag Researgh, New York: Holt, Rinehart and Winston, Inc., 1965. Klaasen, Mary Green. "Self Esteem and Its Relationship to Clothing.’l Unpublished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 1967. Lee, Leroy. "Better Testing through Item Analysis," American Biology Teacher, XXVII (December, 1965), 758-759. Lesgold, Alan M. "Analysis of Contingency Tables, Act II." Technical Report, No. 14. Computer Institute for Social Science Research, Michigan State University, January 12, 1968. Likert, R. A. “A Technique for the Measurement of Attitudes," Archives of Psychology, CXL (June, 1932), 4-53. McNemar, Quinn. "Opinion-Attitude Methodology," Psycholog- ical Bulletin, XLIII (1946), 289-374. . Ps cholo ical Statistics. New York: John 77Wiley and Sons, Inc., 1955. Morris, John. "Rank Correlation Coefficients." Technical Report, No. 47. Computer Institute for Social Science Research, Michigan State University, Jan- uary 5, 1967. Newcomb, Theodore M. Social Psychology. New York: Henry Holt and Company, Inc., 959. Nunnally, Jum C., Jr. Tests and Measurements. New York: McGraw-Hill Book COmpany,7IncL,7l959. Oppenheim, A. N. Questionnaire Design and Attitude Measure- ment. New or : Basic B00 5, Inéf, 1966. 77 107 Phillips, Emma Jane. "The Relationship of the Knowledge and Use of Brand-Named Clothing with Certain Personal and Social Characteristics of a Selected Group of Sorority Women." Unpublished Master's thesis, Department of Textiles and Clothing, University of Tennessee, 1966. Renn, Enma Jane. "Clothing Behavior and Relationships to Desire for Social Participation and to Reasons for Desiring Social Participation." Unpublished Master's thesis, Pennsylvania State University, 1965. Ruble, William L., et a1. "Analysis of Variance with Equal Frequency in Each Cell,‘I STAT Series Description, No. 14. Agricultural Experiment Station, Michigan State University, September, 1967. Ryan, Mary Shaw. Clothing: A Study in Human Behavior. New York: Holt, Rin7ehart an7d Winston, Inc., 1966. Selltiz, Claire et al. Research Methods in Social Relations. New York: Ho 1t, Rinehart 7and Winston, 155., 1965. Senders, Virginia L. Measurement and Statistics. New York: Oxford University P7ress, 1958. Sharpe, Elizabeth Susan. "Development of a Clothing Interest-and-Importance Scale." Unpublished Master's thesis, Ohio State University, 1963. Siegel, Sidney. Nonparametric Statistics for the Behavioral Sciences. 7New7Yoer: MWcGraw-Hill BOOk7 7C7ompany, Inc., 1956. Snedecor, George W. Statistical Methods. Ames, Iowa: The Iowa State University Press, 1965. Stanley, Julian C. MeasUrement in TodaystSchools. Engle- WOOd C11ff$3 NEW JérSGY: Prentice-Hall, Inc., 1965. Young, Mary Jane. "The Relationship of Clothing to Peer Acceptance and to Personal Appearance of Adolescents." Unpublished Master's thesis, Department of Textiles, Clothing and Related Arts, Michigan State University, 19670 APPENDIX IMPORTANCE OF CLOTHING Read the following statements and rate each according to the scale given below. Place the number corresponding to your choice in front of each statement. The statements generally refer to a school situation. Scale: 5. Almost Always - very few exceptions 4. Usually - majority of the time 3. Sometimes 2. Seldom - not very often 1. Almost Never - very few exceptions l. The way I look in my clothes is important to me. 2. When I am shopping I choose clothes that I like even if they do not look the best on me. 3. It bothers me when my shirt tail keeps coming out. 4. I consider the fabric texture with the line of the garment when choosing my clothes. 5. I use clothing as a means of disguising physical problems and imperfections thrOUQh skillful use of color, line and texture. 6. I wear clothes which have buttons or snaps missing. 7. I pay a lot of attention to pleasing color com- binations. 8. I keep my shoes clean and neat. 9. I carefully coordinate the accessories that I wear with each outfit. 10. I wear the clothing fads that are popular in our school even though they may not be as becoming to me. 11. I spend more time than others coordinating the colors in my clothes. 109 110 . A1rr71ost aTwa7ys 451130511): 7 37.7 Sometimes 2.7 S7e717do7m 7 1. Almost never vv vavfifi ii w v vvwf 12. I try to figure out why some peOple's clothes look better on them than others. 13. Unlined sheer dresses or blouses reveal too much ,of the body. 14. I select clothes that are conservative in style. 15. I feel uncomfortable when someone has forgotten to close his or her zipper. 16. The first time in the season that I go to a pub- lic beach or pool I feel exposed in my bathing suit. 17. I choose clothing with small prints, even though a larger design looks equally good on me. 18. I feel embarrassed when I see someone in too low cut a dress. 19. I select clothes which do not call attention to myself in any way. 20. I feel embarrassed when I see someone in clothes that are too tight. 21. I like dark or muted colors rather than bright ones for my clothes. 22. I hesitate to associate with those whose clothes seem to reveal too much of their body. 23. I wonder why some peOple wear clothes that are immodest. 24. My friends and I try each others clothes to see how we look in them. 25. I enjoy trying on shoes of different styles or colors. 26. I study collections of accessories in the stores to see what I might combine attractively. 111 5. Almost a7lways7747. Usually 3. 7Somet7imes77 2.7 Seldom7 27. 28. 29. 30. 31. 32. 33- 34. 35. 36. 37. 38. 39. 40. ]. AJWE§F.F¢Y9r--- - --2- _' -- --- I try on some of the newest clothes each season to see how I look in the styles. I read magazines and newspapers to find out what is new in clothing. It's fun to try on different garments and acces- sories to see how they look together. I experiment with new or different "hair do's" to see how I will look. I like to know what is new in clothing even if none of my friends care and I probably would not want to wear it anyway. I try on clothes in shops just to see how I will look in them without really planning to buy. When I buy a new garment I try many different accessories before I wear it. I am curious about why peOple wear the clothes they do. The way my clothes feel to my body is important to me. There are certain textures in fabrics that I like and especially try to buy, for example, soft, fuzzy, sturdy, smooth. I am more sensitive to temperature changes than others and I have difficulty being comfortable in my clothes as a result. I wear my pants or slacks with an easy fit even when tight ones are fashionable. I get rid of garments I like because they are not comfortable. I find it difficult to buy clothes suitable to the temperature. 112 5.7 A71mo7$t always 11. 7Us7ual 1y 3. 7Som7e7tim7es77 2.7 Se17d7om7 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 1. Almost never I would buy a very comfortable bathing suit even if it were not the current style. I avoid garments that bind the upper arm. I am irritable if my clothes are uncomfortable. I am extremely sensitive to the texture of the fabrics in my clothing. I wonder what makes some clothes more comfortable than others. When new fashions appear on the market, I am one of the first to own them. I have clothes that I don't wear because everyone else has them. I like to be considered an outstanding dresser by my friends. I try to keep my wardrobe in line with the latest styles. I go to nearby cities to shop for better fashions. I try to buy clothes which are very unusual. I avoid wearing certain clothes because they do not make me feel distinctive. I enjoy wearing very different clothing even though I attract attention. I try to buy clothes with the best labels. I wear different clothes to impress peoPle. I am interested in why some peOple choose to wear such unusual clothes. I plan for and prepare clothes to wear several days in advance. 113 Almost always 4. Usually 3'. Som7et7imes7 2? Se7ldom77 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 1. Almost ”PVQC- - y I see that my out-of-season clothing is cleaned and stored. I look over the clothing in my wardrobe before each season so that I know what I have. I am enticed into buying garments I like without having anything to go with them. I enjoy trying to get the most for my money in clothing purchases. I wear a raincoat or carry an umbrella to protect my clothes in rainy weather. I have something to wear for any occasion that occurs. I have a long-term idea for purchasing more ex- pensive items of clothing such as coats or suits. I carefully plan every purchase so that I know what I need when I get to a store. I am more concerned about the care of my clothing than my friends are about theirs. I try to find out how I can save as much time, energy and money as possible with my clothes. I check with my friends about what they are wear- ing to a gathering before I decide what to wear. I would rather miss something than wear clothes which are not really appropriate. I feel more a part of the grOUp if I am dressed like my friends. I wear clothes that everyone is wearing even though they may not look as good on me. 114 5. Almost always 74.7Usually 3} SomEtimes 72} SeldOm 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. l.yAlmestrnever I am uncomfortable when my clothes are different from all others at a party. I try to dress like others in my group so that people will know we are friends. I get new clothes for a special occasion if the clothes I have are not the type my friends will be wearing. I have gone places and then wished after I got there that I had not gone because my clothes were not suitable. I wear what I like even though some of my friends do not approve. When I buy a new article of clothing I try to buy something similar to what my friends are wearing. When someone comes to school dressed unsuitably, I try to figure out why he is dressed as he is. Certain clothes make me feel more sure of myself. I decide on the clothes to wear according to the mood I'm in that day. Days when I feel low I wear my gayest clothes. I "dress-up" to make an ordinary occasion seem more exciting. I am aware of being more friendly and out going when I wear particular clothes. I feel and act differently according to whether I am wearing my best school clothes or not. I buy clothing to boost my morale. I get bored with wearing the same kind of clothes all the time. 115 w 5. Almost always 4.7Usually 7 3.7 Sometimes 2. Seldom7 1. Almost never 2, fly .2 87. I have more self confidence when I wear my best school clothes. 88. When things are not going well I like to wear brighter colors. 89. I wonder why some clothes make me feel better than others. Rank the following items according to their importance to to you. Place a 7 in front of the most important and a 1 in front of the least. Do not use any number more than once. I prefer that my wardrobe be: comfortable unusual attractive conservative and non-revealing “similar to most others “ my own combinations of garments and accessories organized and coordinated by a plan "‘111111111111111’111111111111“