a |pm of mis v&Mmm, mwum. be a.oiAanm or mmituKft t-m m m mmm tm mmmt w Iren® It Hons* A ttwol® submitted in j*rti*l fulfillment of I'm requirements for th« degree of gtoetor of F&iloaopii^, in the Oep&rtsaont of Child Half are in the Oradoat,® College of tb* State Cnlwreiiy of low*. ProQuest Number: 10598613 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted, In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest ProQuest 10598613 Published by ProQuest LLC (2017). Copyright of the Dissertation is held by the Author. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code Microform Edition © ProQuest LLC. ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106 - 1346 ■Co d * 3" ' ACKSfOWlEdOMMIS The writer wishes to express appreciation to Dr* Qrvis G* Irwin who directed the study fob his assistance in setting up the problem* his encouragement during the progress of the study and his help in preparing the manuscript* The writer would also like to express her gratitude to Dr* Paul J. Blommere and Dr* Harold P* Bechtoldt for their assistance in working out the experimental design and in the statistical analysis of the data* The writer is deeply obligated to Mr* Charles Spiker* who not only assisted in the design for the study but also helped in securing subjects* assumed full responsibility for the onerous task of scheduling appointments* and participated in collecting the data* Perhaps the deepest obligation is to Mrs* Wanda Spiker* v/ho turned her home into a laboratory for a period of month©* enrolled subjects in the project* served as an efficient caretaker for sibs who accom* panied test subjects* and could be depended upon to handle graciously any situation which might arise* Without her help the project could not have been completed* To the parents who enrolled their children in the project thanks are also due# TABLE OF CONTENTS Chapter Page I, Introduction * * * . * * * « , * . . * * • * • * « * * * * * 1 Statement of the Problem 5 II* Methods and Procedure ? The Seales Used * . * « « * * * . ►* . • • • • « « « * 7 Subjects 10 Procedure . . « » » • ................. * * * « « < , « * » * « IS 111* Consistency of Performance • » * * , * „ , „ « * « . * < * * IS Test-retest Reliability of Kuhlmann Scale for Total Group • * * ; • • * « • * » • • • • « 18 Test-retest Reliability of Cattell Scale for Total Group « * » * « . » * • * * « * * » « 19 Gattell-Kuhlmann Consistency for Total Group » • • • SO Consistency of Test Performance as Related to Examiner * • • .....................• *. «<>* * * » 21 Kuhlmann TesWretest Reliability » » * » « • # » 21 Cattell Test-retest Reliability * * „ * * * * * 21 Cattell-Kuhlmann Consistency * « ' « + • • • • • ZZ Test~r@test Reliability as a Function of Scale Order on a Given Bay • • » * « • • • • • • 23 Summary of Findings Regarding Consistency of Performance 26 PSr. Resultsj Level of Intelligence and Factors Associated with IQ • » • • « » • » « » » * • • « » * 29 IQ Bifferences Associated with Examiner * * * * * * * 36 Kuhlmann 36 Cattell 39 Summary of Examiner Differences • * • • • » • • 41 Test Performance as a Function of Scale Order • * * * 42 Kuhlmann 43 Cattell * * * * * , * * * » * , . * * « « « « « » 43 Kuhlmann Test-retest Changes * • • ] • » • • • • • • * 45 Cattell Test-retest Changes • • • • • « » * • • • • • 48 Summary of Factors Associated with Obtained IQ's * , 50 V. Discussion » « « » « « * « • * « » * * * * » * « « » * * » « 54 VI* Summary and Conclusions o * * * * * * * * * * * , * * * * 60 Bibliography * * * * „ * , * » « « * * * » * * « * * . * . 66 Appendix A . * * * * * * * * * * * * ' * * * * * * * * * * 67 Appendix 76 i ii TABLE OF TABLES Table Fag® X. Distribution of Kuhlmann Test and Retest IQ’s According to Sex 3G II* Distribution of Cattell Test and Retest IQ’s According to Sex 31 III* Distribution of Differences Between Kuhlmann and Cattell Test and Retest IQ*s According to Age Level when Kuhlmann is Subtracted from Cattell 34 IF. Mean Kuhlmann and Cattell IQ’s and Differences Between IQ’s on Two Scales on Test and Retest According to Age Level 35 V* Analysis of Variance Associated with Kuhlmann IQ’s for Thirty-Two Cases Tested by Two Examiners 59 VI* Analysis of Variance Associated with Cattell IQ’s for Thirty-Two Cases Tested by Two Examiners 40 VII* Distribution of Differences Between Initial and Retest Kuhlmann IQ’s According to Age Travel and Sex when Original is Subtracted from Retest 46 VIII* Analysis of Variance Associated with Kuhlmann IQ’s for Total Group of Eighty Cases 47 IX* Distribution of Differences Between Initial and Retest Cattell IQ’s According to Age Level and Sex when Original is Subtracted from Retest 49 X. Analysis of Variance Associated with Cattell IQ* s for Total Group of Eighty Cases 51 XI. Basic Data for Study 76 iv X Chapter I INTRODUCTION Although considerable weight is frequently attached to a score obtained on a single test, the reliability of tests in current use for the measurement of intelligence of infants and young children has not been systematically investigated* C attell (1) and Kuhlmann (7), both of whom have constructed scales of intelligence covering the period of infancy and early childhood, represent opposing points of view* Cattell, whose scale of infant Intelligence was published in 1940, reported reliability coefficients obtained by the split-half method of *56, *88, *86, *89, *90, *85, *71 at three, six, nine, twelve, eighteen, twenty-four, and thirty months respectively* The reliability coefficient for the S tanford-Binet calculated in the same way for a sample of the same subjects at three years was *87* On a basis of these data she concluded (pp. 49-50), The reliability coefficients of the Infant Tests thus compare favorably with the Stanford-Binet except at three and thirty months* No explanation has been found for the lower reliability coefficient at thirty months unless it be chance due to the smaller number of cases* Kuhlmann, on the other hand, (p* 90) refers to the "relative unreliability of all test scores for young children.11 He implied that variable performance was to be expected on tests and did not report the reliability of his scale* He stated (p. 17), The mathematician has asked for the impossible. To get the reliability for a test result is a psychological job. And z the only way to get it is to discover the effect of the differ­ ent factors that enter an examination and then control these factors or correct the total result for the amount of their influence* To do this, we must follow the road of controlled experimental observation of these factors instead of that of the pencil and paper logician.. .We can make a test reliable so far as the influence of factors external to the child is con­ cerned* But the influence of variable factors in the child himself are less known and less controllable where known* On these an endless amount of investigation will be required for continued progress in making the tests more reliable. !e have for this and other reasons omitted any attempt to determine the so-called reliability of the tests selected for the final scale* No reports have appeared in the literature regarding the test-retest reliability of either scale nor the consistency of per­ formance on the two scales* With the exception of the 1922 Kuhlmann revision of the Binet Scale (6)* which is superseded by the 1939 revision, the 1939 Kuhlmann Tests of Mental Development and the Cattell Infant Intelligence Scale are the only commercially published scales of intelligence covering the major part of the period of infancy* ^ Moat studies of other tests purporting to measure the in­ telligence of very young children contribute little to the problem of the consistency of performance of children under two and one half years of age. Retests, sometimes with the same scale and sometimes 1* The Gesell Developmental Scales (2, 3) are not classified by the writer as tests of intelligence since Gesell emphasizes that the developmental quotient yielded by his scales involves a broader concept than that of intelligence, whereas both Cattell and Kuhlmann consistently refer to their tests as instruments for measuring intelligence or mental development. s with a different instnaaent, were mad® alter vaiying intervals of time with emphasis on the predictive value over a period of time. The reader interested in this aspect of the problem is referred to the excellent summaries by Qoodenough and Maurer (4) and Maurer (5), Two studies reported by Goodenough give some idea of the consistency of performance of preschool age children over relatively short intervals of time* The reliability of the 1922 Kuhlmann scale was investigated by intensive study of three groups of Minneapolis children aged two, three and four years respectively (4). The one hundred cases included in each age group, equally divided as to sex, were selected so as to constitute a representative sampling of Min­ neapolis population in terns of paternal occupation* Children were tested on the Kuhlmann scale and retested after an average interval of 5,9 weeks (range from four to seven weeks). All testing was done by two exminers. Within each age group, each of the examiners ad­ ministered both test and retest to one-fourth of the cases, first test to one-fourth and second test to one-fourth. Conditions under which the examinations were administered war© carefully controlled. Parents observed some but not all of the tests* The test-retest correlations appear belows Age Z Age 3 Ag© 4 Total Boys .810 .817 ,832 .820 Girls .706 .902 *809 .806 Total ,759 .869 ,821 .815 There was a mean gain from test to retest of 3.0 points for the two year olds, 3.2 points for the three year olds and 6,6 points for the four year olds* d&in was not related to whether the mother had ob­ served the test* so that coaching was not regarded as a factor in the increased scores. The consistency of performance of children on two forms of the Minnesota Preschool scale was studied during the process of stand­ ardization and in follow-up studies (5), The standardization sample was similar to that used in the study of the Kuhlmann in that it was regarded as representative of Minneapolis population in terras of paternal occupation. It comprised one hundred cases at each half yearly age, from eighteen to sixty months. Within each age group, the sexes wore equally divided and the order in which the two forms of the tests were administered was counterbalanced. In most cases the second test was administered the following day* the maximum in­ terval between tests was one week. The Minnesota Preschool scale has both a verbal and performance section, but we are reporting on only the verbal section, since the non-verbal scale does not extend below three years of age. The correlation coefficients between the verbal scores on the two forms of the scale were ,88 for the eighteen month group, and .92, ,90, .85, .80, ,82, .82, .89, and .85 for succeeding six month age levels through sixty-six months. Part of the children were also given the Meriill-Palmer Scale at ** approximately the same time1* as the Minnesota Preschool scale. The correlation between the IQ equivalents on the Minnesota Preschool verbal scale and the Merrill-Palmer scale was .28 for eighty-nine children under thirty-six months of ag© at time of test, 5 .23 for 178 children between thirty*six and forty-seven months, and *25 for 240 children over forty-seven months* Th© results of these studies suggest that under carefully controlled conditions, groups of children perform quit® consistently over short intervals of time on the same tests and on different forms of equated tests but that the degree of consistency is considerably less when entirely different scales of intelligence are used* Statement of the Problem If tests are to be used meaningfully in either a research or clinical setting, it is essential that th© user have some idea of the reliability of the measures used and th© comparability of tests pur­ porting to measure the same thing* Information is lacking in both of these areas in regard to the Kuhlmann and Cattell tests of in telli­ gence at th© younger age levels* The present research was undertaken for the primary purpose of studying the consistency of performance of children between th® ages of sixteen and thirty months on two different scales of in telli­ gence administered on th© same day and given twice within an interval of one week, as a function of a number of variables* Specific prob­ lems to be investigated in this area ares 1* Test-retest reliability of th© 1939 revision of the Kuhlmann Tests of Mental Development when th© retest is administered within a two to seven day period following th® first test* Z» Test-retest reliability of th© Cattell Scale of Infant Intel­ ligence under th© same conditions*