' ".33.... 1.65... {.8 3%...» mum... . 2...... ...... ...... .... k 321...}. 3.1.1.)! gfixwv I i It .0 .. a ... I . N . ixarttxh . . a...) .. 6.7.12.5 ......av. «Jaguarhnh .....h... .. V. . _ 97...... Q 2 1?...23‘11 .933... .. It... . . . . p s ...“. at: . .... Etlttv ain‘t... 1.3“. . V . ., L. 7 . . : lava: - . ..\ ..b t 1.. : : .... 51": c L . 5 ‘ . . . I. ’11 :::: ... ethic}... ... I .012: o ..3.....)\. .1. .3“. . . . ... . .. .. ,. o . . . . .. x . 1.311).... .. .48.. .13.: 5|. .vr .r .. ... 1......9'. r......9£l|ki.1.. .. :v «5— . . . v .. .. 3):}; c .. 9......1 x i. 35...}; \ Lé... .. .. . 1L: 2 . , ... . . . .. . . 7......3‘1 9 2 ..v I 7 25.1.... v.33: .ybtlavqig at .... 3'2 .. f V . .. . . _ . h. .. . a... .....It Poi-L Eillz‘ 1;..l.7’&..vt.z\&. 9.... I).......?\ .. . a. .. . e _. . ’ttgar:a{xxvzi.§h§ 1 ...... 5...... 143,155.»..3'52950vlslv ... 5.5!}: u .. . . . .. . . e, .39."; . ‘1‘. ..g“..§-)of3x1 )a...‘ it. 35.: 3.... 9 .. 2.. .. . . .v . f . ...-31'...‘ a 3.3.1.... 3.... u 1.13., ...... 9.... 13%“. 2.\..%4§ . . u If . .u r. . . BI! ... . ......{uI 3!. ha... . 921:2. auxkxix ...... 35.5.. ....9 .. 3.. .... a . 1.1.}! t Vltfigvtnl 7). k. sssss N .z...&.2.-.9..Ix.....hvi....hl ’ 2...». 5a.. . . an... 3.49.... ... ..{fiuutzrtusfirawmbnuufi}....uhfi.......$i.a... hi: I7‘fi' [In N: ' I , In. . . ll IIII’EI?! ’Pdtail! 1| ... ..a ..I( .-. s :8)... .....l-iih‘iklt: army-'11:!!! ..I I van. .. is .....Iron... . .. tit. do"... 2%. . a £5191. :5; . . . . .... 7...... . . Lg. mfi. 77.... :7... famHJJT..~f.-h_ .. humming 1 K. .111. $ .8. agar}: 2...}. n 9‘ u... . .. . I. .... 9?. x I! dc» = ,1 a)... . x... . . t s . :v tit: 3:19.... cits} 3 all .' ... .... :1 SSE-II: . #9.“... . .V .. . .. , ...Mm... .... -...zfi... as}... . .7 uragkammun _ . .. .u. .r ... w...“ t A! . .. 117% w. . 35:123.. . .Rnuvau. .rg-vfikmwu .. l. . 7 1 . hr .1 r If“ Ikfl’QK II... ’33 t . . ... ...... hurl... flhh.lc§.¥§§.u3uaru. i§§.fiwn 33.5.... .. z . . .-...u. en... «#1.. .. ...... “Shh”... r ... .. . . .... . .. . . .....uvz.......:.... .. hummwfkawwv... ......nak ... dz... 5.... a. ....W... $5.... .. ...... fin. ...... ,. ., ... . . . .3 a. E m E . gmgfitnbz. 1.2%....{fiym . . . .. 1.. owl. as... r. in... x... .... .r... . J . '5- .iv‘“; ‘1“ ‘ it. of! ti...) ilii ‘ttt,eat:2..5uuxpwfl§..§f “...“.vuwk: I ).(¥u:..:a.IL\l~v-, . . . 3| , . 2K. . . . . , I 5,)ylril‘:>§A-t¢ili .. .. .. 9 a... ORV)...3.§. ‘32.}... ...))Ito.«. )9}... ..Huhh.$tn.l! two. . .3 .13).. ...x 15.... ....) y ...... 5.. .. rgh .hmfl.na«x.r.uuh.lu.urdfi$\ .. .. n‘hfivdahwwfl ”MWvg swat." murmuamsfiliklls‘l) .. ...: ... .5 .t. 30...: b! 3. 09.. -..: its . a . ’3 {Egg 3 in? ..n. .. ....q a _ .. :3. “rats... '1'.“- Iwmwfl. h.........)!l:.m.ah. ...”.xat.i......\kr2§ H”.u.r«it|t .... III...“ 20.1.”. ’33:]: stbthg‘tliivsit... «Kat-.50.. .. I . 1273...... 2‘ . 7.1.1.; .3 1 3.2.1.3... In I )3... .. . .I... .. >. . 331.. fill. I» . . . .3... . , gnu... ...... ...”. 2.3....» E. b.s....:2¥.i..i.....,.. .....zsi...:i.z§..flfi?étz.t.i..:.%vln§tti€§..nzu5§ .zi...u:........u.z.:....1n......vo..£!:u.l. -.t...............:.~.u......im.. .2H—xflflflai.$ 2‘9““ 1:19:37? 1 I...” 11 fill! ififiwzfinfi. ... hulraoffirx 945.16.035.39...“ t afigagflnvfivfiulfiril‘..ifit She’ll, 1 viii}! “Jabs”: ....”v A. islhblutlictty§\ ysihihafllofiflu )‘x;ha!|.5..3!w:;4§vlfd‘abeas.lv 5.11.} .i I . .7. . .. . .... .... : .v . .vl. . .. £71.. ...rl. . it“... x; ...7... 1.... 1... . . . . .....h. .... Ix... thaw“...- _|....... Ln... .5... G M Halitis’... ; {(3.}... ....Icll.) iii}: V .151 3:13.?41. 153.1”. ”23. a. ...L... . .33... 3 55.”?!277 .. I £3... 2.31.7425, ...-VI I9 I”. ‘55 I; . I)! v . ...“... E... ‘15....‘22. 015...... Eli... 3L 0 1‘: 1:91!!! 53‘: 1.. 9%... 43i11!§;ql 39;»... ”[135... {tiazlt‘sti . v r. .51.... ..vuvvvel. : x “.31. x w 2. 5...... .... 113’]. 8’10..\).. .1... ll... 11:14.; I.” I}!!! .5... 1"}?! ... |§2na¥ bli ... {hr {$i9ii.‘ Agib 1|!!! 1316...... 131?. 7’: d3... '1 tin—.51.. ......1..7: a ... JrulanhmV3Igmufl.’ hfifififl‘inifu». t 7! r I 3%.... 1 a 153:? ”1”... cl! '1 s aI Luv... . .... i....v.tutl739.. “mu 1.13.3. lg: . . wit-h. an”... ...‘lxivl #1. .5 tuna... u. Ni} .... Alfie)“ ..fl‘Ilafiy. I: (if: .6 .r. 23.55.. 1.! . .olcill«.. : . . . §§§t33.5§ :9! o I... VIFA It... gilt?! I ’I 1.55., . .2: lyfldpqhydnrni 3):...31.Privillxiv‘clth 37.1.0.5! u I “grit: ..-:It’b-l (flu-3.1... “ta-Eg— . 3513) 5:25}... .. A 0‘ 2355).. a: .. «vs... ...u. .. 5.1.19.3..1u\..1JII=?«4‘5'.“ . . . .n a . .. cm». In"??? htzhov!Ju§i 9!). ll Jagi‘IOgnzli. I 7.9m”. '3‘ .fl) .. (.9... z. du‘mmdfln 1.. I; {glillivriv’Vr 1A.)! '1? . gun... {Is-.1. .3 .... grin... I)? i 0‘ "‘1. £35 . h 0;. .. z 21"”?! Isfnzl’n ..... ifgak . ”w?! .EI:?IZ.MLI..I. . .... 03.3.9... ..le .315! lil||¥1rh§ll ..D. . I: [titltat .(Ma. run)... é:tk99¢§e.$ Iii;5|$|i.01,.e.lsls\:t 1th. .... LEI, ” :‘7..- A. I .- Y {I . ..D’Ii ", ' {A} :‘I ,tvél‘ ::§. L"§.i 41‘l-z‘g,él '3... 3-.. u. ’I‘ \.¢- ...-2......1)Iu6._lots. ru7tl7ilgz 7% ‘5‘... nil-plrvl.lu.fl Riki“, 5*: ...-Bug]... .. “.93.? . P‘i..§g\§ll)l\t§=\\,bt DJ thiilihuurxi. .1313? VII... {3%.flthuvldfl r 3.. 3 .1”... . it: 11.3.: gifitlgnllfizrvivlgz? V’.‘ I 3!... r... ’50 IA 05:25) a.“ I .090H..l’1i\’9l)¥i§gzgr-i‘igfrrhid E’ItliEA‘. K. ....Iaiilz v. 777» 2121;), a 3:.) .f 1’.— .f‘gl‘ 7! in}. . g 3?: 30": Egbiil‘i: 3|. Oti?1\$i|xh.t. lQ‘II-I . Plug: .r’hwflqvi‘lvvf lulu. 11v .\. it .... ’21:: . it. ..IIEI fbgjistfiélil - l )l t)..\\ I: . ”......i.-.!...,...V...unu..>...€$..$..-(an. ginmfh... {2...}... . $2.... .3125... litigii.3!§E.tri.l§fi£§yi5 1453.2... . . tug-figuahflaiidflhgafi? ......ryvl... ..Ebfr tiny.“ . .. «2;.gsi gar..- .... éilunvvahasx 3.5.1.5»...- L‘j 3......‘....:..§.££Ps 2 . u..u.LVl..l.lfilr!M€.P.s§i .{8W11IR£33N;,3 Waging... 1.0.! :2“-.. .. 1:518”, . if 3gb... zed-Wm "flak... 7th.. . 1:935»)... E.u&.fi.atlhuaa.h«ui..Ed.‘§.fl):§3§5ilbl§xnh.fs§ {Kr . ‘ . $9.12.)“ I l. L," .9 3....) l. ...) .hfll‘iig’irtiri.‘ 3.. ’33.; . 2%.)! 3.2.; #81 I35!) gill 3...: gt??? ...-”.6. fl.“’..$€!l§¥"’§§tl£3z§ ()y.tl|1?blnvr.‘iitiih ! I {5.15 . xiii}; ’9! Iris}: v.31.....lIa.vr..l':!vE7..t..tti.tr)..$§>.vl. ...... .Vahi‘zstoatiwé {Ow-.35 fl Eggs-Agni... a t r it! I! tfgylr .‘zll’vzlvi~§5. VJ?! 9 . 0' lllll hing. -..}...Zi... . ... .. . .. rr . . . n...unanhfiflufihnuz;£5.31 :srg..is.wsfi.§i|..s:¥htv.z; 53:38.»... I... ... .51... ...-.16.... .. .. ...".Nunuurzlrrvfiznilv .I?J;Dr?billy.51?v 5.07.5 . 13.57-: . . . . .4 .. , u ..l .... . : a: LP .111??? it; t§§lxfl€ctzshwl ...“...l flail} Jitiifiisfznfizl lit-hili Pia! . .15.)?! 115...... 9.5! Oil‘s-(£31009 I; .... 5 ..TKvifihnazl a).v&§.5 u ‘ . . a . .. .. o ...! rail . , . L: . V .25. in“... .... . . . . ....r:zninaiil§xgfiflufity....§§§ £35..- I.” v.ix!fi§1..&f|§‘1llfi\l§~.zaggi : .. .... a a a .. . . . .. . .. a . {23.)}.3’. . gig!!! {$.34}. 55...?th an”... C(11,1\\£i4!$tlsflan. .ln.“ gi’??? if?!‘ 2 a} Illa; ' s silaglIois I . 21.5.50 3‘»... ‘31..‘§Ii§§‘3)1t.§l§lfilllliziizzio“ \|‘ .terIl\ u .. i. , .( .... . .5 . . . . .11 \ :fitilvg if} «2‘. 113% ’irlhhq.‘ iizfiiflflmr‘uLISn 2..)3lifik1 . . ,.'S‘§zag) g9. ”“lléj§§l§§§ozihxa%§‘ . \S 1.09%; .. kt... ...... )1; . . . £911.; 3; . ’22....) . , . g h.. a Eli‘s... Exit. . fig . Q . I.£ g?! . 511:3; ..-. -5 .......;:.3.2. t . .. 2) :23} - ..- . .. ..,tu(.n-7i..i-. .224)... ...... 35.3.63 .3. . If».-. 3.33-1.2..- .1551... .-.-u..{2.p.a§...1.....ll- . ..ilsiitxy.;so.-osvf§ht .... ,§§{¥,. .52...- 1 .91: . if»???{fitttrkiéglitssIIEEBE2g’tuafuhisthstfiE5 kt. 4 E)§§fi§§¥§§g\§|§§zitttf} I A :53 ...-(.115 . . . . £75.}!!! §§.5EO$SIE\§$EI‘E££2‘ i‘tlbinl£\§|z)\t \\L\ .x A . B.$|£Ip¥l...n.l . at..- .IV 3%.; ggfzfsugt..fllnv.....JUJDHHHKHNRtWWN..3.4.,st£2(. .\. . .. . . , . . ul ...“...iiJfliKWhuihr,’ $39!... Vivallfi.1£ian .909: ads! V.- .... .. If‘i’l‘gfit balhgg .r- ..Iv. ....st,‘ ‘5: . .. a ...... .1, D: , 3? (I. ‘04! in... ...)allfvhllt. if I“ innit... fit-...! .5. D tzVZV‘A-b; . 25$ 1! .‘sgyirtt‘h 3|, . .. 1:4?v.Ilhaak.!yv\sxi,zon:t:.a§aiawwfiat§. 5.1.... , !I..\.V....s . .. ...; .I 3‘ . 1.- ...x... . . YN... a... a , w ‘ ......y. ... if . h, .|I|5\tx\... THESIS This is to certify that the dissertation entitled BEHAVIORAL AND COGNITIVE ANTECEDENTS OF READING ACHIEVEMENT IN THE ELEMENTARY GRADES: A LONGITUDINAL STUDY presented by ROBERT L . JARVIS has been accepted towards fulfillment of the requirements for Eh . I) . degree in EDEJQAIIQN 44444344 L40 I lvfior‘l'arofessor Z ; Date JULY 13 1982 Msui.,.,.un.,.. v.1 ~ - m - , - - 042771 ”Rim 4 1’ Michigan 44am 4 University MSU LIBRARIES ~.. RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES wiII be charged if book is returned after the date stamped beIow. BEHAVIORAL AND COGNITIVE ANTECEDENTS OF READING ACHIEVEMENT IN THE ELEMENTARY GRADES By Robert L. Jarvis A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Counseling, Educational Psychology, and Special Education 1982 a. gfiqgilrfj7 I m \t x 3’ ABSTRACT BEHAVIORAL AND COGNITIVE ANTECEDENTS OF READING ACHIEVEMENT IN THE ELEMENTARY GRADES By Robert L. Jarvis The purpose of the study was two—fold: 1) to identify teachers' perceptions of behavioral attributes of elementary age students which were highly and significantly related to reading achievement over time, and 2) to investigate the efficiency and utility of using teachers' ratings of student behavioral attributes to predict future reading performance. A descriptive, longitudinal design was instigated whereby behavioral ratings were evaluated independently as well as in conjunction with IQ, initial level of reading achievement, and sex of student, as to their ability to predict subsequent reading achievement after one, two, and three year intervals. Behavioral attributes found to be most highly related to reading achievement over time appeared to reflect a student's degree of adaptation to the task orientation of the classroom, degree of self-motivation and persistence, and degree of general social maturity. Appropriate behavior adjustment was found to be a generally more important factor in terms of reading success for boys relative to their female peers. All of the behavior scales developed in the study were found to be significantly related to reading achievement over time. Results further indicated that despite the strong linear relationship between the behavior scales and future reading achievement, teachers' ratings were not found to be as predictive of later achievement as were scores on a measure of general intelligence and a measure of initial level of reading achievement. Generated prediction equations which utilized only the behavior scales in estimating future achievement tended to greatly underpredict the performance of above-average readers and greatly overpredict the achievement of below—average readers. Further, despite significantly high multiple correlations, as well as high diagnostic ”hit rates," the generated prediction equations which incorporated initial level of achieve— ment, IQ, and sex of student were found to be generally lacking in terms of accuracy in predicting above—average and below-average readers over time. The findings of the study are finally presented in light of their relationship to past research, their implications for implementation of reading screening and diagnostic programs, and their implication for future research. With much love and pride I dedicate this effort to Nancy. Her on—going expression of love, patience and support continues to encourage me to reach for heights I never though possible. ii ACKNOWLEDGEMENTS Thanks to members of my committee, Dr. Harvey Clarizio, Dr. Bill Frey, Dr. Gaston Blom, and Dr. Richard Prawat, who have provided on—going support and encouragement throughout my graduate school experience. Special thanks to Dr. Frey, whose guidance and support in both academic and non—academic domains greatly expanded my experience and horizons while at MSU. Further thanks to all of my dedicated friends and relatives who labored through the preparation of various drafts of this document -— Kac, Karen, Sue, Mary, Barbara and Roberta. I am forever indebted to Bob and Mary Ann who graciously provided the gift of friendship as they opened their hearts and home to me over the past years during completion of this project. iii TABLE OF CONTENTS Page Acknowledgements ..................................................... iii Table of Contents .................................................... iv List of Tables .......................................... . ............ vi List of Figures ....... . ...................................... . ....... Viii CHAPTER I — THE PROBLEM ......... . .................................... 1 Introduction .................................................... 1 Need for the Study .............................................. 4 Purpose of the Study ............................................ 10 Overview of the Study ........................................... 11 CHAPTER II - REVIEW OF THE LITERATURE ................................ 13 Student Classroom Behavior and Reading Achievement .............. 14 Teacher Ratings of Student Behavior and Reading Achievement... 14 Observed Classroom Behavior and Reading Achievement ........... 20 Longitudinal Prediction of Reading Achievement .................. 27 Sex Differences in Reading Achievement and Classroom Behavior... 45 Sex Differences in Reading Achievement in the Elementary Grades ........... . ............................................ 48 Sex Differences in Behavior ................................... 54 Validity of Teacher Ratings of Pupil Behavior ................... 59 CHAPTER III — DESIGN OF THE STUDY .................................... 64 Population and Sample ........................................... 64 Instrumentation ................................................. 68 Definitions ..................................................... 78 Data Collection........ ......................................... 79 Statistical Hypotheses .......................................... 80 iv Data Analysis Procedures .......................................... 81 CHAPTER IV — RESULTS ................................................... 88 Hypothesis I .......................................... . ........ ... 88 Hypothesis II ..................................................... 89 Hypothesis III ...................... . ............................. 93 Hypothesis IV ..................................................... 109 Hypothesis V ...................................................... 112 CHAPTER V — DISCUSSION ......................................... . ....... 119 CHAPTER VI — SUMMARY AND RECOMMENDATIONS ............................... 126 Limitations and Delimitations of the Study ........................ 131 Recommendations ................................................... 134 REFERENCES ............................................................. 136 APPENDICES A. Breakdown of Mean SAT Standard Scores by Grade by Sex of Student ..................................................... 147 B. Breakdown of Mean Behavior Scale Raw Scores by Grade by Sex of Student .................................................. 148 C. SBCL Item Loadings with Principal Factoring with Iteration and Variance Rotation — 6 Factors ...... . .................... 149 D. Simple Correlations Between Individual Items of the SBCL and SATl ........................................................ 154 E. Simple Correlations Between Individual Items of the SBCL and SATl, SATZ, SAT3, and SAT4 (Total Group).. .............. 158 F. Simple Correlations Between Individual Items of the SBCL and SATl, SATZ, SAT3, and SAT4 (Boys and Girls) ............. 162 G. Relative Frequency of Student Behavior Attributes Based on Teachers' Ratings From the SBCL ............................. 167 H. The School Behavior Checklist — Form A2 ....................... 170 I. Permission Forms for Release of Information ................... 172 10. 11. 12. 13. 14. 15. 16. LIST OF TABLES Page Average Percentile Rank - SAT Reading for Grades 1 - 6 ..... . ..... 65 Relative Percentage and Absolute Frequency of Subjects at Each Grade Level ......... . ........................... 66 Compariosn of Relevant Sample Versus School Population Characteristics ...... .. ..... . ......................... 67 Reading Achievement-Related Behavioral Scales .................... 73 Intercorrelation of the Behavior Scales .......................... 76 Relative Base Rates of Poor, Average, and Good Readers for SAT2, SAT3, and SAT4 ................................. 86 Correlations Between the Behavior Scales and Reading Achievement.. ........................................ 89 Correlation Between Reading Achievement and the Behavior Scales Controlling for Sex of Student ........... 9O Correlation Between Reading Achievement and the Behavior Scales Controlling for IQ ....................... 91 Correlation Between Reading Achievement and the Behavior Scales Controlling for SATl ..................... 92 Correlation Between Reading Achievement and the Behavior Scales Controlling for SATl, IQ, and Sex of Student..... ............ . ......................... 93 Multiple Correlations of Reading Achievement After One Year (SAT2) with the Behavior Scales ................... 95 Multiple Correlations of Reading Achievement After Two Years (SAT3) with the Behavior Scales .................. 96 Multiple Correlations of Reading Achievement After Three Years (SAT4) with the Behavior Scales ................ 97 Multiple Correlations of Reading Achievement After One Year (SATZ) with SAT1, IQ, Sex of Student and the Behavior Scales.. ................................ 99 Multiple Correlations of Reading Achievement After Two Years (SAT ) with SATl, IQ, Sex of Student and t e Behavior Scales ........................... 100 vi 17. 18. Page Multiple Correlations of Reading Achievement After Three Years (SAT4) with SATI, IQ, Sex of Student and the Behavior Scales. .............................. 102 Diagnostic "Hit Rates" of Good, Average, and Poor Readers at SATZ, SAT3, and SAT4. ............................ 116 vii LIST OF FIGURES Multiple Correlations of Reading Achievement with a) the Behavior Scales, b) the Behavior Scales, Sex of Student, IQ, and SATI........ ......... .... ................... Absolute Correlations of Reading Achievement with the Behavior Scales, IQ, Sex of Student, and SATl................ ............ Predictive Classification of Students at SAT2 Based on the BehaVior scaleSOOOOOOOOOO ......... ......IOOOOOOOOOOIOO 0000000000 Predictive Classification of Students at SAT3, Based on the Behavior Scales................. ......... .. ................. Predictive Classification of Students at SAT4, Based on the Behavior Scales ............................................. Predictive Classification of Students at SATZ, Based on the Behavior Scales, IQ, SATl, and Sex of Student........ ....... Predictive Classification of Students at SAT3, Based on the Behavior Scales, IQ, SATl, and Sex of Student. .............. Predictive Classification of Students at SAT , Based on the Behavior Scales, IQ, SATl, and Sex of Student ............... viii Page 113 .. 115 .. 115 CHAPTER I THE PROBLEM Introduction Significant interest and concern on the part of both educators and parents regarding the reading achievement of elementary school-age children necessitates a critical examination of potentially relevant factors which may contribute to students' relative success or failure in reading. Measures of cognitive functioning, as well as sex and socio- economic status have been shown to share a significant amount of variance with the reading achievement scores of elementary age children. However, psychologists and teachers alike have concurrently been well- sensitized to the fact that "non-intellectual" factors, e.g. emotional state, personality, and overt classroom behavior, should, and most likely_do, play a vital role in a child's academic life. The relationship between social-emotional functioning and reading achievement is at present quite unclear despite a significant body of research spanning the past fifty years. While there is an implicit assumption within much of the research that social-emotional functioning is not the primary causal factor in the behavior-achievement relation- ship (Allyon and Roberts, 1974; Gates, 1968; Spache, 1976), the direction 2 of the relationship remains empirically unresolved. It is further difficult to interpret consistent trends in the relationship from past empirical efforts due to the relatively diverse conceptions and sub- sequent measurements of ”non-intellectual" functioning and academic achievement. Many studies have been further limited by narrowness in the child population sampled, especially in terms of potentially relevant variables such as IQ, sex, and age or grade level. Particularly noticeable is the relative absence of longitudinal— developmental perspective with regard to the relationship between emo- tional/behavioral factors and reading achievement in the normal popula- tion of elementary age students. Few studies have investigated the relative long-term efficiency of using "non-intellectual" factors in predicting reading achievement. The majority of studies have focused on concurrent comparisons of student characteristics and achievement. Results of the few longitudinal studies reported appear somewhat in— conclusive with regard to the relative contribution of cognitive and "non-intellectual" factors to reading attainment. However, all would suggest that emotional or behavioral factors do make a significant and unique contribution to our understanding of the nature of childrens' reading achievement in the elementary grades (Feshback et al., 1977, Kohn and Rosman, 1974; Lambert and Urbanski, 1980). Improving our understanding of the predictive relationship between student classroom behavior and reading achievement has potential value from a number of viewpoints. At a general level, one could expect that teachers, counselors, and psychologists may gain from an increased under- standing of the dynamics of classroom behavior and its relative contri- bution to reading success or failure during the early course of reading development; a point argued by Peck (1971) who called for an increased 3 sensitivity and awareness of educators to the coping effectiveness of pupils in the classroom. Along the same line, Camp and Zimet (1974) argue the need for identification of positive, achievement-enhancing behavior, and the need for clarification of whether a lack of reading achievement is related more to the absence of positive classroom be— haviors or the presence of negative, disruptive behaviors. The litera— ture is replete with studies of the negative behavioral and personality characteristics of disabled readers. Behavioral variables have been shown in many studies to make a significant, unique contribution to the variance in reading achievement in the elementary grades. Such findings would appear to be important in light of the early identification and subsequent remediation of learning and/or emotional problems (Bryan and Wheeler, 1972; Feshbach, et al., 1977; Lambert and Nicole, 1977). Kellam, et a1. (1975) for example, found that even when intelligence test scores and school achievement were included in a prediction battery, ratings of children's social adaptation status in first grade made a signifi— cant unique contribution independent of intelligence to the prediction of later need for remedial help and grade retentions in the intervening years. The work of Cobb and his associates suggests the practical rele- vance of investigating the relationships between classroom behaviors and academic achievement in terms of direct treatment of underachieve— ment. In an initial series of studies (Cobb, 1970; Cobb, 1972a) specific behavioral categories, which appeared to be significant predictors of achievement, were defined. In samples of first and fourth grade children, Cobb found a consistent relationship between performance on standardized measures of achievement and such behavioral classes as attending to teacher, compliance with teacher requests, and volunteering to answer 4 academic questions. Subsequent studies (Cobb and Hops, 1973; Hops and Cobb, 1974) involved efforts to develop these "survival skills" in underachieving pupils through the use of behavior modification proced— ures. They found that these intervention procedures were effective in increasing and maintaining survival skill behaviors and that such gains were directly related to gains in reading achievement. Need for the Study Investigators in psychology and education have worked for many years to develop methods of predicting students' levels of reading achievement. As has been indicated, despite a significant body of research spanning the past fifty years, the predictive relationship between emotional—behavioral functioning and reading achievement is far from being clarified. During this time, most of the research has focused upon deliniating specific personality traits of disabled readers. These attempts to identify a consistent personality profile of the dis- abled reader have proven essentially futile. The early conclusions of Gates in his classic 1941 review of the relationship appear to have withstood the test of time: 1. Personality difficulties are frequently but not universally associated with reading difficulties. 2. In cases where they occur together, personality difficulties may be causes, concomitants, or results of reading difficulties. The direction is not clear. 3. There is no single personality pattern characteristic of reading failure and there is no proved one-to-one relationship between type of adjustment difficulty and type of reading dis— ability. For example, feeling of insecurity resulting from 5 undue home pressure for achievement may result in low reading achievement marked by withdrawal; compulsive, anxious reading marked by frequent errors; or it may result in higher achieve— ment in reading than would be expected from mental level. Bower and Holmes (1959), nearly two decades later, reviewed over one hundred subsequent studies of the personality—reading achievement relationship, and indicated, ". . . in the main, the same general con— clusions arrived at by Gates are, but for slight modifications still applicable." They found that only a few investigations reported signif— icant differences for specific personality traits between good and poor readers, and that many such differences tended to be "nicely counter- balanced" by other studies which reported differences in exactly the opposite direction. Review of these earlier studies indicated that the clarification of the relationship between personality and achievement was signifi— cantly limited by a general lack of success in defining and measuring personality (Russell, 1953); Spache, 1976). This validity issue is critical, as the ability to establish consistent relations between "non— intellectual" factors and reading achievement depends to a significant extent upon the meaningfulness of the personality or behavioral cate— gories used. Others have argued that discovery of such relationships is essen- tially irrelevant for use by the regular classroom teachers for improving a child's reading level (Bateman, 1966; Engleman, 1967). Over the past fifteen years, however, there has been an increased emphasis upon the study of more discrete, observable classroom behaviors, which may be potentially under the teacher's control, and their relationship to reading achievement. 6 It appears that two major areas of research have contributed to this increased emphasis on the study of overt classroom behaviors rather than personality traits, which might be predictive of academic achievement. The first, as alluded to previously, was the reported inconsistency in linking specific personality traits with reading success or failure (Bower and Holmes, 1959; Cobb, 1969; Gates, 1941; and Russell, 1953). For example, Cobb (1969) reviewed ninety—one doctoral dissertations re— ported in Dissertation Abstracts between 1961 and 1969 which examined the relationship between academic achievement and such variables as self— concept, anxiety and other personality traits. He found, as had earlier reviewers, that while modest correlations were discovered in several of the studies, replications of the findings were often difficult to obtain. Concurrently, in the late 1960's, a number of researchers were in- vestigating means of altering inappropriate classroom behavior through procedures based upon social learning theory principles. The majority demonstrated marked improvements in the classroom behavior of emotion— ally disturbed children e.g. (Hewett et al., 1967; Hops, 1971; Patterson et al., 1969) and disruptive children (Hall, et al., 1968; Madsen et al., 1968). These studies indicated that children's inappropriate classroom behavior could be significantly altered, however the question remained whether changes in behavior were concommitant with positive changes in academic achievement. Indeed, there was no clear evidence in any of the aforementioned studies that achievement had improved. Similarly, efforts to improve academic achievement by treating emo— tional disorders through traditional means, e.g. psychotherapy and coun— seling, appeared not to have been exceptionally fruitful (Ashcraft, 1970; Cobb, 1972; Munger, et al., 1964). During this same period of time, Lipe and Jung (1971) suggested a need for a clear and empirically demonstrated rationale relating classroom behavior to measured achieve- ment. Promising findings began to emerge as researchers began focusing on the relationship of observable student behaviors, as opposed to person— ality dimensions, with reading and mathematics achievement in elementary school pupils. In a five year follow-up study, Meyers, et al. (1968) demonstrated significant relationships between behavior ratings in kinder— garten and achievement scores in the fifth grade. "Attention", emerged as the first and most powerful predictor, even among other verbal ability measures, of all aspects of fifth grade reading achievement, including reading words, reading comprehension, and spelling. The correlation of ”attention" to each of these subtests was .40, .43, and .38 respectively. Using direct observation procedures in sixth grade classrooms, Lahaderne (1968) similarly found significant correlations between rate of attending behavior and reading achievement on a Scott-Foresman Basic Reading Test and the Stanford Achievement Test. Correlations were .51 and .46 res— pectively for boys and .49 and .39 for girls. A review of relevant studies over the past fifteen years which have subsequently investigated more specific non-intellectual/behaviora1 cor- relates of reading achievement in elementary school children indicates rather consistent findings regardless of often diverse methodologies and instrumentation for measuring behavior and reading achievement. Three clear behavioral categories have emerged as being significantly related to early reading achievement: independent, task-oriented behavior (Coker and Lorentz, 1977; Engin, 1975; Kohn and Rosman, 1974; McKinney, et al., 1975; Soli and Devine, 1976); attending behavior (Cobb, 1970, 1972; Feshbach, et al., 1977; Forness, et al., 1977; Meyers, et al., 1968; Samuels and Turnure, 1974; Soli and Devine, 1976; Swift and Spivack, 1967); 8 and compliance with teacher and classroom demands (Camp and Zimet, 1974; Cobb, 1970, 1972; Coker and Lorentz, 1977; Lambert and Nicoll, 1977; Soli and Devine, 1976). These findings are significant in that they suggest that not all un— desirable student classroom behaviors are necessarily related to poor reading achievement. Behaviors which teachers may consider most unde- sirable because they are disruptive of classroom order, e.g. verbal or physical aggression, may in fact, be of least importance to early reading success (Blom, et al., 1980; Cobb, 1972; Lambert and Nicoll, 1977; Lambert and Urbanski, 1980). Such findings further lend credence to future stud— ies which attempt to differentiate between academically—related behav— ioral attributes or dimensions and those behaviors which may more accur— ately reflect general social adjustment or mental health issues. As Lambert and her colleagues have suggested, one of the major limitations of past research is that it has not generally focused on the "relevant" student characteristics related to reading success. As has been suggested, the prediction of reading achievement and the early identification of children who are at high risk for school failure have been of concern to educators for many years. The majority of efforts have focused upon the development of predictive indices or instruments composed of cognitive and student behavioral variables, both independent from and in conjunction with one another. These instruments have for the most part been developed for a preschool or kindergarten population with the intention of screening for eventual reading failure. Stevenson, et al., (1976) suggested, however, that an understanding of the relative utility of these various measures have been generally difficult to attain. This situation would appear to be primarily due to the fact that the majority of predictive studies cited have tended to 9 rely solely on multiple correlation as the primary statistical procedure. Although this type of analysis indicates the relative degree of associa— tion between the predictors and the criterion, it provides limited infor— mation as to the instrument's practical value. In particular, it does not indicate how many or what types of errors in prediction are made. Practical use of such screening instruments may convey unnecessary cost to schools as well as risk to an individual child if the utility of the de- vice is inadequately assessed (Satz and Fletcher, 1979). In practice, psychologists and educators frequently make important decisions concerning the classification, prognosis, and treatment of indi— vidual students, wherein treatment decisions are made on the basis of early test signs. The accuracy of these predictions, however, has been rarely addressed. As Satz and his colleagues suggest, a true error rate of pre- diction is frequently masked by obvious confoundings from misclassification of high risk subjects (false negatives) or low risk subjects (false posi- tives), making accurate judgments regarding student outcomes tenuous. The false negatives refer to subjects for whom good achievement was predicted and whose actual achievement was "not good," while false positives refer to those subjects for whom poor achievement was predicted and whose actual achievement was "not poor." The confoundings of misclassification appear to be well exemplified in the studies of Stevenson, et al. (1976) wherein relatively high and significant multiple correlations were found between various cognitive tasks in pre-kindergarten and subsequent reading achieve- ment in the first through third grades (.66 to .77), however the ability of the predictive index to discriminate among good and poor readers in the third grade was relatively deficient. The results of this and other pre- dictive studies strongly suggest the need for future research to go be- yond a purely concurrent correlation model, towards a longitudinal 10 perspective which assesses both the accuracy as well as practical utility of a predictive scale. Satz and Fletcher (1979) suggest that beyond the notion of accuracy the predictive utility of a measure implies that detection signs (e.g. valid positives and negatives) should be evaluated within the broader con— text of the prevalence estimates of the event to be predicted — in this situation the base rates of student reading success or failure. Meehl and Rosen (1967) concur and further suggest that the efficiency of the great majority of psychometric devices are essentially impossible to evaluate because: 1) Base rates are virtually never reported 2) Most devices are reported without cross—validation data 3) Results are frequently reported only in terms of significance tests for differences between groups rather than in terms of the number of correct decisions for individuals within the groups. These shortcomings of past studies must be addressed in future investiga- tions in order to gain a better understanding of the predictive relation— ship between "non-intellectual" student characteristics and subsequent achievement in reading. Purpose 9f_th§_§£udy The purpose of this study was essentially twofold: 1) to identify teachers' perceptions of behavioral attributes of elementary age students which were highly and significantly related to reading achievement over time, and 2) to investigate the efficiency and utility of using teachers' ratings of student behavioral attributes to predict future reading per— formance. The study incorporated a descriptive, longitudinal design, whereby teachers' ratings of student behavior were evaluated independently and concurrently with IQ, initial level of reading achievement, and sex 11 of student, as to their ability to predict subsequent reading achievement after one, two, and three year intervals. The design of the present study is developed in light of what have been identified to be limitations of past studies which have proported to investigate the predictive relationship between student characteristics and academic achievement. As has been suggested, past studies have gener— ally based their findings upon a single sample in which predictor and criterion variables were assessed at a single point in time, and relative amount of variance to reading achievement attributed to the cognitive and behavioral factors reported. Due to a lack of longitudinal perspective in these studies, little empirical evidence is available as to the relative importance of student classroom behavior to future reading success or failure - an important point of knowledge in terms of a psychologist's or educator's decision to target or not target an individual student for special support services, e.g. special education, remedial reading. Closely related to the aforementioned problem is the need to determine the accuracy by which future reading achievement may be assessed by he- havioral and/or cognitive predictors. Few studies have reported the prac- tical value of their instruments in terms of actual predictive ability and efficiency. This study assesses the diagnostic utility of a predictive in- strument composed of behavioral characteristics related to reading achieve- ment relative to base rates of reading ability and disability. Overview g£_the Study Chapter I has presented the background, rationale, purpose, and ob- jectives which serve as the basis for the investigation. In Chapter II the pertinent related research is reviewed. In Chapter III the research method is described, including sample characteristics, instrumentation, data collection procedures, statistical hypotheses, data analysis 12 procedures, and definition of key terms. Chapter IV includes the presen- tation of data and evaluation of the testable hypotheses. Results are discussed in Chapter V in light of past research and practical implications, considering limitations and delimitations of the study. Chapter VI offers a summary of the study and recommendations for future research. CHAPTER II REVIEW OF THE LITERATURE This chapter has been divided into four main sections which reflect the major issues and concerns in past studies which have investigated the prediction of reading achievement in elementary aged students. The first section focuses upon the reported relationship between student social- emotional/behavioral functioning and subsequent reading achievement. The research reported primarily reflects those studies which have shown a high concurrent correlation between these variables. As student behavior has been evaluated primarily by two means, teacher ratings and classroom observation, this review is similarly dicotomized. In the second section, longitudinal, predictive studies of reading achievement are reviewed. This section addresses the major issues regard— ing the accuracy and practical utility of prediction of reading achieve- ment. As sex of student has been found to be significantly related to both classroom achievement and behavior, the third section of this review investigates the reported differences between elementary aged boys and girls in both reading achievement and classroom behavior. Finally, the fourth section examines the evidence of the validity of teachers' judg- ments regarding elementary student behavior. 13 14 Student Classroom Behavior and Reading Achievement A review of studies over the past fifteen years which have subsequently investigated behavioral—affective correlates of reading achievement in ele- mentary school children indicated rather consistent findings regardless of often diverse methodologies and instrumentation for measuring behavior and reading achievement. Three clear behavioral categories have emerged as being significantly related to early reading achievement: independent, task-oriented behavior (Coker and Lorentz, 1977; Engin, 1975; Kohn and Rosman, 1974; McKinney, et al., 1975; Soli and Devine, 1976; attending behavior (Cobb, 1970, 1972; Feshbach, et al., 1977; Forness, et al., 1977; Meyers, et al., 1968; Samuels and Turnure, 1974; Soli and Devine, 1976; Swift and Spivack, 1967); and compliance with teacher and classroom demands (Camp and Zimet, 1974; Cobb, 1970, 1972; Coker and Lorentz, 1977; Lambert and Nicoll, 1977; Soli and Devine, 1976). Teacher Ratings of Student Behavior and Reading Achievement Swift and Spivack (1967) investigated the relationship between teacher ratings of pupil behavior and students' academic success or failure in grades kindergarten through six. The purpose of their project was to learn how disturbing classroom behaviors were organized in both normal and special education classrooms, and to assess the relationship between these behaviors from the Devereux Elementary School Rating Scale and report card grades. The ranges of simple correlations for each of the ten factors of the Devereux scale and teacher marks in reading across the elementary grades were: 1) Classroom Disturbance -.21 to -.37 2) Impatience -.10 to -.54 3) Disrespect/Defiance —.08 to -.27 4) External Blame -.13 to —.27 5) Achievement Anxiety .00 to —.27 6) External Reliance -.27 to -.77 15 7) Inattentive/Withdrawn —.15 to —.74 8) Irrelevant Responses —.07 to —.40 9) Creative Initiative .21 to .60 10) Need Closeness to Teacher -.02 to .50 Analysis of individual factors by grade indicated that in the first years of school the child who approaches and responds to the teacher in a warm and friendly manner is more likely to be academically successful. Ability to attend and work independently similarly were indicative of successful readers. Regarding differences between the sexes, boys were rated as presenting significantly more problems than girls on nine of the ten factors. Girls were rated as being more inclined to approach and be friendly toward the teacher, a significant predictor of early academic success. Further, as a group, boys were rated as being more disturbing and disruptive, more impatient, and more disrespectful. The authors con— cluded that boys clearly demonstrated more "achievement—impeding behaviors," and that girls are significantly more often judged as conforming to or capable of meeting classroom behavioral demands. Ratings on two teacher rating scales were compared with observations of classroom behavior and reading achievement by Camp and Zimet on the Pittsburgh Adjustment Survey Scale (PASS) and the Connors Teacher Rating Scale (CTRS). Reading achievement was derived from the reading subtests of the California Achievement Test and Metropolitan Achievement Test as well as the Gates—MacGinitie Reading Test, and Gilmore Oral Reading Test. Results indicated that classroom observation of ”off-task" behavior cor— related significantly and positively (r=.31) with the Daydreaming subscale of the CTRS and significantly but negatively (r=-.38) with the Prosocial subscale of the PASS. The measures of reading achievement were found to be significantly positively correlated with the Prosocial subscale of the PASS (r=.56 to .81), negatively with the Daydreaming subscale (r= —.59 to -.76), and negatively with the total CTRS (r= -.30 to —.58). 16 Engin (1975) examined the relative importance of academic aptitude, as measured by the Wechsler Intelligence Scale for Children, and behavioral variables as measured by teacher ratings on the Devereux Elementary School Behavior Rating Scale on classroom achievement of third and fourth grade inner-city pupils. Analysis of the regression of the reading-related sub— tests of the Stanford Achievement Test on the WISC and Devereux Scale in— dicated that both intelligence and behavioral variables constituted impor— tant, independent components of the reading achievement of inner-city third and fourth graders. Behavioral factors contributing significantly to the explained variance included external blame (r=-.26), creative initiative (r=.30), and impatience (r=—.17), a profile similar to that found by Swift and Spivak (1967). One significant anomoly in Elgin's findings was the positive contribution (r=.39) of disrespect/defiance to the prediction of the Word Meaning subscale of the Stanford. She postulated that for "inner- city" children this dimension may be reflective of an active verbal orien— tation to school may have "an enhancing rather than deleterious effect" upon aspects of reading achievement. Kohn and Rossman (1974) found in their sample of over two hundred lower and middle class boys in New York City that preschool teachers ratings of social—emotional behavior were significantly predictive of second grade achievement. A Task-Orientation dimension of the Schaefer Classroom Be— havior Inventory correlated .33 and .44 with the Word Knowledge and Reading subtests of the Metropolitan Achievement Test. This dimension tapped the child's ability to concentrate, become absorbed, and persevere in class— room activities. Harrison (1976) evaluated two hundred underachievers on the Wechsler Intelligence Scale for Children (WISC) and the Wide Range Achievement Test (WRAT), and obtained teacher ratings of pupil behavior on a Classroom Be- havior Checklist. "Distractibility" was found to be the behavior most 17 consistently associated with low reading and spelling achievement as meas— ured by the WRAT. The interaction between distractibility and reading and spelling achievement was most significant with the younger, early ele— mentary students. Twenty-seven disabled readers were rated by teachers on the twenty categories of the Burks Behavior Rating Scales (BBRS) in an investigation by Wilf (1972). The subjects were further divided into two subgroups on the basis of the severity of reading retardation. Results indicated that the severity of reading disability appeared to be significantly related to certain characteristics of maladjustment, that is "poor ego strength" and "poor attention" appeared to be much more characteristic of the severely disabled group. Feshback, et a1. (1977) conducted a five year longitudinal study which focused on the comparison of psychometric versus behavioral kinder- garten predictors of reading performance in the first through third grades. During the spring of their kindergarten year over 800 children in middle class Los Angeles Schools were administered a variety of individual tests including the Wechsler Preschool and Primary Scale of Intelligence (WPPSI), the de Hirsch-Jansky Predictive Index, and Bender Visual-Motor Gestalt Test. During the same period pupils were rated by their teachers on the Student Rating Scale (SRS). The primary criterion variables consisted of standardized achievement tests, the Cooperative Primary Reading Tests and a criterion-referenced individualized reading inventory, both admin- istered in the first through third grades. In the first grade the total SRS score proved to be the most pre- dictive of the cognitive and behavioral independent variables. For second grade reading achievement, SRS remained the best predictor with an average correlation of .50, closely followed by the de Hirsch and IQ 18 with correlations of .47 and .40 respectively. Behavioral ratings re- mained consistently high in third grade. Even with IQ partialed out, the correlation of SRS with third grade achievement was .33. Multiple regression analysis indicated that SRS and IQ accounted for 51% of the total variance in third grade achievement. Sex of the child appeared to be a significant moderating variable in the prediction of achievement by use of both behavioral and cognitive variables. Several interesting differences in the pattern of correla- tions were found between boys and girls. IQ was consistenly a better pre- dictor of both reading and mathematics achievement for boys than for girls. In contrast, while the SRS and the de Hirsch were more highly correlated with first grade reading achievement in boys than in girls, by the third grade the correlations for girls were higher than those found in boys. This pattern held for all factors of the SRS including an attention—be— havioral control factor, but excluding a language dimension which operated very much like IQ. The authors suggested that motivational variables may be more important for girls in the early primary grades than IQ, and at the same time, kindergarten behavioral factors, which reflected poorer performance by boys than girls, may decline in importance for boys with increasing maturity. Behavioral dimensions derived from the Pupil Behavior Rating Scale is a study of Lambert and Nicoll (1977) proved to be potentially useful predictors of first and second grade reading achievement. They found that an "adaptation to classroom demands" factor was a strong predictor of both first and second grade reading achievement (r=.60, r=.66), as measured by the Cooperative Primary Test. This factor included such attributes as following directions, working independently, and interest in learning tasks. These attributes, individually, were also significantly 19 related to reading achievement: i.e. difficulty following directions (r=—.53, r=—.63), dependent on teacher (r=-.51, r=-.52), no interest or enthusiasm for learning (r=-.57), 5=—.63). Distractibility was also found to be highly correlated with first and second grade reading achieve— ment (r=-.51, r=-.50). Lambert and Nicoll further found that the adaptation dimension com— bined with the demographic variables of sex, race and socioeconomic status was more predictive of reading achievement than was cognitive functioning in combination with the demographic indicators. These results were consistent with the earlier findings of Lambert (1972) and Lambert, et al. (1976) which found that these same behavioral attributes measured in the second and fifth grades were as good, and in some cases better, predictors of subsequent ninth and twelfth grade achievement as were intelligence and achievement test data from the elementary grades. Final analysis of Lambert and Nicoll's (1977) study indicated (a) the powerful relationship of the classroom adaptation dimension to early reading; (b) the weaker, but significant relationship of an intraper— sonal factor, i.e. unhappy, withdrawn, behavior with early reading; (c) the relative unimportance of acting—out, aggressive interpersonal be— havior with reading achievement. The latter finding is interesting in light of the similar findings of Elgin (1975) with inner—city pupils and Blom, et al. (1980) with lower elementary aged girls. Lambert's findings regarding the strong relationship between ”adap- tation" and achievement were further confirmed in a recent study in which second, fourth, and sixth grade subjects were assigned to twelve behavior typology groups based upon teacher's ratings from the Pupil Behavior Rating Scale (Lambert and Urbanski, 1980). A one way analysis 20 of variance procedure comparing reading scores across the twelve groups was highly significant and post hoc comparisons again suggested the strong association between those attributes of the classroom adaptation dimension and reading achievement. Differences in measures of the in- terpersonal and intrapersonal dimensions did not appreciably contribute to the observed group differences in achievement. Observed Classroom Behavior and Reading Achievement Studies over the past fifteen years, which have investigated the relationship between observed classroom behavior of students and achieve— ment, have generally found three dimensions of behavior which appear to contribute a substantial amount of variance to reading achievement: on- and off-task behavior, attending behavior, and compliance with teacher demands. In a major replication of the early correlational studies of Meyers et al., (1968) and Lahaderne (1968) which found a relatively high re- lationship between "attention" and reading achievement, Cobb (1970) de— monstrated that greater relationships could be obtained if the behavior categories of the predictors were more precisely defined. He suggested that the earlier aforementioned studies utilized ratings which were too global and which included a number of irrelevant behavior's that masked the true relationship between specific, discrete behavior responses and achievement. Cobb began a series of studies in 1969 to clarify the relationship between a combination of specific student behaviors and achievement (Cobb, 1969, 19723). In the first study, over one hundred fourth graders were observed in five classrooms from two schools during arithmetic in- struction over a ten day period. Fourteen discrete behaviors such as 21 attending to task, talking to teacher about academic material, volun- teering, compliance with teacher requests, and out of seat behavior, etc. were recorded. The rate of each child's behavior was then used as an independent variable in producing, through a step-wise procedure, multiple regression equations to predict attainment as measured by the Stanford Achievement Test. The results indicated that the specific behavior designated as attending, talk to peers about academic material, self—stimulation, out-of—chair, and compliance were significantly related to both mathematics and reading achievement, with out-of—chair and talk to peers about academic material being the two best predictors of reading achievement. In a subsequent study, Cobb (1970) used similar procedures in eight first grade classrooms where students were observed in both reading and arithmetic activities. The major predictors of reading achievement were found to be attending (r=.59) and volunteering (r=.59), and for arith— metic, looking around and compliance. Further, attending and looking around, which were as might be expected highly correlated with each other (r=—.75), were found to be significant predictors of both reading and mathematics attainment. When Cobb (1970) split his sample into male and female subgroups, it was found that observed classroom behavior was more highly predictive for boys than girls. The multiple correlation was .71 for boys versus .45 for girls. One major difference in terms of the specific predictors, was the finding that for boys compliance and noncompliance behaviors were significant predictors of achievement, whereas for girls no significant relationship for these behaviors was found. Samuels and Turnure (1974) further investigated sex differences in classroom attention and its relationship to reading achievement at the 22 first grade level. A behavioral observation schedule was used in which 74 boys and 58 girls were observed during reading instruction. Task- relevant behaviors, e.g. looking at text or teacher were considered inattentive. Interrater reliability was reported at .89. Reading achievement was measured by recognition of forty—five words randomly selected from the Dolce List of Basic Sight Words and total attending scores for each pupil were placed into four quartiles. Results of the study indicated no significant differences on a reading readiness measure between boys and girls, and these readiness scores were used a covariate in later analysis. It was found that mean word recognition scores were significantly higher for girls (30.03) than for boys (22.68), and that these scores tended to increase from the lower to the upper quartiles of attention. Further, overall mean attention scores significantly favored the girls, and significant sex differences were found for each of the four attention quartiles all favoring the girls. An overall correlation of .44 between attention and reading achievement was found which was consistent with the finding of Cobb (1972) for fourth graders (r=.45) and by Lahaderne (1968) for sixth graders (r=.51 to .39). Forness, et a1. (1977) focused on observable kindergarten behavior to determine its utility in predicting eventual educational risk. Sub- jects were forty children followed through the end of second grade. At that time, school records were examined and information on special status such as retention or referral to remedial reading or special education were noted. On the basis of this information, all subjects were divided into two groups, risk and nonrisk. Correlation coefficients were then computed between the kindergarten observation categories of positive peer 23 interaction, task attention, nonattention, and disruptive behavior and eventual second grade status. Significant differences were found be— tween the two groups with attending behavior being the highest predictor (r=.47) with eventual risk. McKinney, et al. (1975) investigated the predictive value of a number of observed behavioral patterns in regard to fall and spring aca- demic achievement for ninety second graders. Students were observed for five minute periods on each of four days during language arts instruction in the fall of the school year and again in the spring. Twenty—seven discrete classroom behaviors were systematically coded at ten second intervals for each of the students. Interrater reliability was reported at .90. Factor analysis procedures identified twelve behavioral cate- gories which were retained as independent variables for predicting achieve— ment on the California Achievement Test. Using multiple regression procedures, it was determined that the following four categories of fall behavior predicted 33 percent of the variance in fall achievement: distractible behavior, passive responding, dependency, and constructive play. High frequencies in the first three were associated with lower initial achievement and high frequencies of constructive play (appropriate self-directed activity, not part of the curriculum) were predictive of higher initial achievement. In the spring frequencies of constructive self—directed activity (independent class- work) and aggression were found to be significant predictors of spring achievement. The final multiple R for predicting spring achievement from fall behavior was .60. Analysis of sex effects failed to show significant differences in fall behavior patterns for boys and girls, however, in the spring girls 24 evidenced a higher frequency of constructive self-directed activity than did boys. The general portrait of the academically competent child attained in this study appears to be similar to the previously reported findings (Cobb, 1972; Lahaderne, 1968; Samuels and Turnure, 1974). The results of McKinney et al. indicated that the child who is attentive, independent and task—oriented in his classroom interaction with his peers is more likely to succeed academically than the child who is distractible, de- pendent on teacher for direction, and passive in peer group activities. Soli and Devine (1976) utilized the observational procedures devel- oped by Cobb (1970, 1972a) in their investigation of the relationship between observed behavior across academic subjects and reading achieve- ment as measured by the Gates-MacGinitie Reading Test. Twenty—five observations were made on each of 312 third and fourth graders. Multi- ple regression procedures indicated that total achievement among all students was best predicted by the behaviors positive peer interaction, not attending, self—stimulation, and initiation of activity with teacher, giving a final regression coefficient of .45. When the total group was divided into two achievement levels (high and low achievers), it was found that in the high group achievement was best predicted by posi- tive peer interaction, initiation of activity with the teacher, inap- propriate locale, and noisy with a final regression coefficient of .44. Achievement in the low group was also predicted with a final coefficient of .44, however, the behavioral predictors were different from those of the high group; not attending, self—stimulation, playing, complying. Coker and Lorentz (1977) examined the moderating influence of different classroom settings, Teacher—Directed versus Program Directed, 25 upon the relationship between observed pupil classroom behavior and reading achievement. Six hundred students in grades three through eight were observed during both the fall and spring of the academic year by means of the Coping Analysis Schedule for Educational Settings (CASES), and reading achievement was assessed by means of the comprehensive sub— test of the Iowa Tests of Basic Skills. Stepwise regression procedures indicated that both fall and spring other—directed, task-oriented behavior and fall and spring compliant behavior emerged as significant predictors of spring reading achieve- ment for those pupils in the teacher directed settings. Different sig- nificant behavioral predictors were evident for those students in the program directed settings. Fall ratings of low peer dependency and high inner—directed, task oriented behaviors and spring ratings of low passive, withdrawn behavior and low inappropriately self—directed be- havior were all significant spring reading achievement predictors for this group. In the Teacher-Directed setting, pupils who were identified through initial observations as passive and withdrawn were found to be least likely as a group to gain in reading, while those who exhibited other- directed, task-oriented behavior and compliant behavior tended to gain more over the course of the year. In the Program-Directed sample, a decrease in low peer dependent behavior from fall to spring, and an increase in low passive, withdrawn behavior were found to be signifi- cantly related to growth in reading achievement. The findings of the present review along with those of e.g., Kim, Anderson, and Bashaw (1968), Ross, Lacy and Parton (1965), and Kohn and Rosman, (1973) converge to define rather similar classroom behavioral 26 dimensions theoretically related to reading achievement. Lambert and Nicoll (1977) have defined these dimensions as 1) behaviors reflecting adaptation to the classroom environment e.g. following directions, volunteering, ability to work independently; 2) behaviors reflecting intrapersonal adjustment, e.g. withdrawal, attention; and 3) behaviors reflecting interpersonal adjustment, e.g. fighting, arguing. Cobb (1970, 19723) conceptualized the behaviors attending, com— pliance, volunteering, etc. as "academic survival skills" which were necessary, but not sufficient for successful academic functioning. They were not, he hypothesized, academic behaviors per se, but rather, the first components in a chain of correct academic responding; behaviors that all children, regardless of ability level, must first be taught in order to enable them to take advantage of the instructional opportuni- ties in the classroom. Cobb and his associates have subsequently demon— strated the functional relationship of such behaviors to early reading development (Cobb and Hops, 1973; Hops and Cobb, 1974), wherein signif— icant improvements were noted in both mathematics and reading achieve- ment after initiation of "survival skill" training for underachieving first graders. Particular moderating variables have been found to have a significant impact on measurement of the classroom behavior—reading achievement relationship at the elementary level. Differential behavioral predictors have been found between the sexes (Blom, et al. 1980; Cobb, 1970; Lahaderne, 1968; Samuels and Turnure, 1974; Swift and Spivack, 1967) and at different age or grade levels (Blom, et al., 1980; Coker and Lorentz, 1977; Feshbach, et al., 1977; McKinney, et a1. 1975). 27 Intelligence has been identified as an important confounding factor because of its relatively high correlations with both reading achieve- ment and many behavioral variables (Engin, 1975; Feshbach, et al., 1977; Harrison, 1975; Lahaderne, 1968; Meyers, et al., 1968; Miller, 1972). Past evidence would suggest that sex of child, age or grade level of child, and intelligence are variables which should be either con- sidered or controlled for in future analyses of the classroom behavior- reading achievement relationship. Further, as Cobb (1970) has suggested and demonstrated, the more precisely defined the classroom behavior is, the higher the relationship with achievement criteria. In addition, behavioral definition of student social—emotional status enhances fur- ther replication and clarification of research results, as well as providing useful information to educators in terms of defining non- intellectual variables which are potentially under the teacher's control. Longitudinal Prediction_gf Reading Achievement The problem of predicting reading achievement and the identification of children who are at high risk for school failure has been a concern of educators for many years. DeHirsch, Jansky, and Lankford's Predicting Reading Failure in the mid 1960's stimulated considerable interest in the development and use of various screening test batteries to predict reading failure. Efforts have most exclusively focused upon identifi- cation of "school readiness" skills in preschool and kindergarten age populations found to be significantly related to later reading success or failure. Significant emphasis has been placed upon the use of psychometric procedures which unfortunately have assessed a rather delimited set of linguistic and perceptual-motor skills (Adelman and Feshbach, 1971; Lambert and Nicoll, 1977). While most of the aforementioned 28 studies have yielded significant correlations between their predictors and subsequent achievement criteria, the relationships have generally been weak, particularly when and if they were subjected to cross- validation procedures. Further, varying degrees of predictive accuracy are reported (Cohen, 1963; DeHirsch, Jansky, and Lankford, 1966); Duffy, et al. 1972; Friedman, et al., 1980; Haring and Ridgeway, 1967; Klein, 1977; Lesser and Bridges, 1973; Meyers, et al., 1968; Satz and Friel, 1978; Stevenson, et al., 1976). Among the more successful cognitive measures used to predict achieve— ment have been those which have evaluated word and letter identifica— tion, visual and auditory discrimination, vocabulary and language de— velopment, and general intelligence. In their review, Stevenson, et al., (1976) suggested, however, that an understanding of the relative utility of the various measures have been difficult to attain due to the dif— fering characteristics of the students sampled, the relatively small number of subjects per study, and the fact that prediction has rarely been over periods longer than one to two years. Further, few studies have attempted replication or cross—validation procedures to assess the empirical and practical utility and accuracy of their predictive in— struments' (Meehl and Rosen, 1967; Satz and Fletcher, 1979). Only in a minimal number of investigations have student behavioral characteristics been used in the development of instruments for long- itudinal prediction of reading success or failure. Although such be- havioral attributes of students have long been recognized as important factors in the development of reading competency, they have for the most part been excluded from school readiness predictive scales. Des- pite recognition of the potential importance of emotional/behavioral 29 functioning, investigators such as DeHirsch, et al., (1966) have sys- tematically chosen to exclude such variables: We recognize that a variety of social, environmental, and psychological factors are significant in the acquisition of reading skills, and we concur with Fabian (1951), who maintains that learning to read requires the developmental timing and integration of both neurophysiological and psy- chological aspects of readiness. Nevertheless, we limited ourselves to the preschool child's perceptumotor and linguistic functioning because in this area we had found considerable deviation from the norm among children who subsequently failed in reading and spelling (DeHirsch, et al., 1966, p. 5). Results of longitudinal studies which have used non-intellectual variables to predict reading achievement have suggested that student be- havioral attributes may make a significant, unique contribution to the variance in later reading ability in the elementary as well as secondary grades, independent of various measures of cognitive ability, including measured intelligence. Kellam, et al., (1975) for example, found that even when intelli- gence test scores and school achievement were included in a prediction battery, ratings of students "social adaptation status" in first grade made a significant, unique contribution to the prediction of later need for remedial help and grade retentions in the intervening elemen- tary years. The primary goal of Meyers, et al., (1968) study was to assess the predictability of fifth grade achievement and intelligence from a variety of measures of cognitive ability administered in kindergarten. These measures included tests of psychomotor coordination, perceptual speed, and linguistic and figural reasoning abilities. As an aside at the time of initial testing, students were rated by the examiners on ten different behavioral characteristics. Step-wise multiple regression 30 procedures were implemented and results suggested that a picture vocab- ulary measure appeared to be the single most valid predictor among the ability measures (r=.33-.50). The authors noted, however, that, "most striking in the results is the value of the behavior ratings in an- ticipating later achievement." Examiner's ratings of students' ”attention" emerged as the single best predictor of fifth grade reading achievement, including both vocabulary and comprehension, as well as spelling achieve- ment (r=.38-.43). Other behavioral indices which entered the prediction equations at significant levels included amount of motor activity, per- formance rate, effort displayed, and cooperation. Lambert, et al., (1976) investigated the comparative ability of various "intellectual" versus "non-intellectual" indicators in predicting various aspects of effective high school functioning. As part of an earlier study data was collected on 300 second and fifth grade students which included teacher, peer, and self ratings on various behavioral dimensions, including compliance, ability to work independently, and interpersonal skills, grades, attendance records, and available health data. Fifty-six percent of the original sample were identified for the longitudinal follow-up. "Successful or unsuccessful high school status" was defined as a composite score based on such indicators as guidance referrals, referral to remedial reading classes, delinquency, partici— pation in athletics, placement on the honor roll, and referral to gifted programs. A second set of criterion measures resulted from a factor analysis of such variables as counselor ratings of social and personality characteristics, standardized achievement and ability measures, GPA, and disciplinary referrals. I 31 Multiple regression procedures were used to determine the relative degree to which status in high school could be predicted from cognitive and behavioral variables measured in elementary school. Teacher ratings of student behavior emerged as the single best predictor of successful and unsuccessful high school status (r=.42-.56). Although IQ entered significantly in the predictions, it contributed only mini- mally to the total variance after the teacher ratings were considered (rzchange=.01-.05). Unfortunately, the multiple R2's, including both teacher ratings and IQ in predicting successful or unsuccessful adjust- ment, were relatively low (chhange=.19-.36). It appears that these low correlations may have been an artifact of the relative lack of pre- cision in the definition and measurement of the criterion. This point would appear valid in that significantly higher simple and multiple correlations were obtained between teacher's ratings and IQ and ”high ' which was a composite factor score based on GPA, school scholarship,’ and standardized achievement and ability test scores. Again the teacher ratings of student behavior in elementary school proved to be the best predictor of later high school academic achievement (r=.47-.63). In— telligence was also highly related to high school academic achievement (r=.44-.58) and together with the teacher ratings accounted for ap— proximately fifty percent of the variance in "high school scholarship." One shortcoming of this study, which the authors recognized, was the fact that elementary achievement test scores were not incorporated into the prediction batteries: Overall, the best single predictor of successful status in high school was the set of teacher ratings items. While we did not analyze the efficiency of a battery which incorporated teacher rating items plus achievement scores and grades, it might have proved to 32 be the set of predictors with greatest efficiency in estimating successful status and achievement in high school (Lambert, et al., 1976, p. 116). Stevenson, et al., (1976a) further expounded on this notion as they suggested: Predicting reading achievement is a different problem before children have begun to read than after instruction in reading has been introduced. Before children learn to read, reading achievement must be predicted from tests of prereading skills and cognitive abilities. Once the child has entered elementary school, however, prediction can be made from the child's current level of achievement. For example, Henderson, Fay, Lindemann, and Clarkson (1973) gave over 700 7-year- olds a battery consisting of commonly used psycho- metric tests, including achievement tests. Scores on the test for reading achievement correlated .72 with reading achievement 1 year later. Adding scores from the remainder of the extensive battery raised the cor- relation only to .75. Thus, with children who already have begun to read, a battery of psychometric tests may add little to the predictive value of achievement tests alone. (p. 378). Feshbach, Adelman, and Fuller (1977) further investigated the relative contribution of non-intellectual versus cognitive variables to subsequent reading achievement in the early elementary grades. Their results again suggested the importance and unique contribution of student behavior to the prediction of future reading attainment. The primary purpose of their study was to evaluate and contrast the pre— dictive efficacy of the DeHirsch-Jansky Predictive Index, a measure of linguistic and perceptual-motor skills, and teacher's ratings of stu— dent cognitive and classroom behaviors as measured by the Student Rating Scale (Adelman and Feshbach, 1971). Their initial sample included 888 kindergarten age children in ten middle class Los Angeles schools. An additional 844 children were identified as a cohort sample for purposes of cross-validation of the findings. The authors suggested that the reason for limiting their population of interest to middle class children 33 was "to eliminate the powerful influence of social class variables on school performance so as to permit a clearer assessment of the role of individual predispositional factors.” Of the original sample 403 or 45% were still available for testing at the end of third grade. This final sample included all students who had taken all three reading tests, the DeHirsch, and Student Rating Scale (SRS), and whose IQ's were 90 or above. Test scores from Cooperative Primary Reading Tests were ob— tained at the end of the first, second, and third grades and served as the primary predictive criterion. Feshbach, et al. reported that for second grade reading achieve— ment normalized SRS scores proved to be the best predictor, with an average correlation of .50. Full scale IQ and DeHirsch scores were similarly high related to second grade achievement (r=.44 and .42 res- pectively). Similar correlations were obtained for the third grade reading achievement criterion. Using multiple regression procedures it was found that with IQ partialed out, the correlation of SRS with third grade reading was .33. Highest regression weights were obtained for IQ and SRS with the multiple R based on these two variables being .57. Cross—validation with the second cohort group collaborated these results as comparable relationships were found (multiple R=.50). Mul— tiple Rs for second grade achievement were similarly consistent for the initial and cohort groups (r=.57 and .61 respectively). Stevenson, et al. (1976a) further added credence to the findings of the ability of teacher ratings of pupil behavior and cognitive skills in kindergarten to predict future reading achievement in the first through third grades. In contrast to the findings of Feshbach, et al. (1977), Stevenson and his colleagues indicated that teacher ratings 34 were not found to be as predictive of later achievement as were scores on various cognitive and psychometric tasks. It would appear that though the scales and tasks differed in the two studies, the differences in results arose, not because the correlations of teacher ratings were lower in the Stevenson, et al. study, but because the correlations of the test scores were significantly higher (Multiple R's=.56—.77). This appears attributable to the relative restrictive range or homogeneity of the Feshbach, et al. sample as compared to that of Stevenson, et al. In a subsequent follow—up study Stevenson, et al. (1976b) evaluated more specifically the predictive value of the teacher ratings obtained in the original study independent of the cognitive and other psycho- metric variables. A total of 217 children were initially evaluated and rated by teachers on 13 scales which represented such attributes as intellectually independent, hardworking, reflective, follows instruction, retains information, vocabulary, attention, and social acceptance. Variables for the scales were selected on the basis of "extensive dis— cussions about characteristics of children that appear to be important for successful performance in school." The authors' report of items included in the scale suggest high face validity, however, no measures of scale reliability were reported. Further, interpretation of their results appears somewhat confounded by the fact that various formats were used in obtaining the scale scores. As in the aforementioned Stevenson, et al. studies, stepwise mul— tiple regression procedures were implemented. Teachers' ratings of students' ability to follow instructions, vocabulary development, ability to retain information, intellectual reflectiveness, intellectual in— dependence, and ease of understanding concepts (Effective Learning) all 35 entered prediction equations at significant levels at grades one through three. A simple sum was computed for the four most frequently selected variables (Effective Learning, Vocabulary, Following Instructions, and Retaining Information) and this sum subsequently correlated with read- ing achievement scores. These were as follows: first grade achieve— ment (r=.66); second grade achievement (r=.65); and third grade achieve- ment (r=.55). The values of these correlations were found to be es— sentially equal to those obtained by the stepwise regression procedures. The correlations obtained from the sum of these four ratings were further found to be highly commensurate with correlations obtained from the sum of the four most predictive cognitive and psychometric tasks in the original study (Stevenson, et al., 1976a). Of the longitudinal studies which have investigated the utility of a variety of screening indices in predicting future reading achieve— ment, some have further attempted to ascertain the accuracy and practical value of their predictive instruments. Satz and Fletcher (1979) address the relative problems of determining predictive outcomes for students as well as the implications for such procedures upon actual programmatic or placement decisions: Because of the current interest in early screening, attempts are being made to identify the high-risk child during preschool and to place him in intervention programs before his problems become more severe or refractory to remediation. However lofty these ideals may be, they portend unnecessary cost to the schools and risk to the individual child if the predictive utility of the early detection device is inadequately assessed. A true false positive and false negative error rate is frequently masked by obvious confoundings from misclassification of high-risk 33 (false negatives) or low-risk 38 (false positives), making early detection simply not possible (p. 65). 36 Those studies which have attempted to determine the relative pre— dictive accuracy of their instruments in terms of correct prognostic classification have reported varying degrees of success. It would appear that much of the variation in predictive accuracy in these studies may be attributable to 1) the degree of the relationship be— tween predictor and criterion variables, 2) the nature of the criterion variable e.g. teacher report or test score, 3) the variability and "arbitrariness" in determination of criterion cutting scores, and 4) the general methodological means by which predictive accuracy was deter— mined. Such variation in methodology will become evident as this review proceeds. Lesser and Bridges (1973) investigated the predictive effectiveness of the Metropolitan Readiness Test, the Lee-Clark Readiness Test, and the California Test of Mental Maturity in determining whether or not a student would eventually manifest a "learning problem." Students were initially evaluated on the predictor variables early in their first grade year. Criterion measure of achievement, the California Achievement Test (CAT) and a Teacher Rating (TR) of overall performance, were obtained at the end of the first and second grades. Because of the relatively high correlation between achievement test scores and teacher "combined criterion." ratings (r=.77) the investigators decided on a In order to combine the CAT score and the TR, each score was divided into categories of "no learning problem" (NLP) and "learning problem" (LP). Teachers determined that a CAT score of 134 or more, which cor- responds to a grade placement of 1.5 or higher, would be considered to be reflective of NLP, whereas a score of 133 or less indicated LP. In terms of the TR, either of the two highest ratings ("average or better" 37 and "marginal") was considered NLP, while the third rating ("clearly immature") indicated LP. On the combined criterion all subjects who had been classified as NLP on both the CAT and the TR were classified NLP. All subjects who were classified as LP on either of the two single criteria were categorized as LP. On the combined criterion, 148 children (51 percent) were classified NLP and 145 children (49 percent) were classified LP. In order to test and compare the relative predictive effectiveness of the Metr0politan, Lee—Clark, and California Test of Mental Maturity (CTMM), the number of correct predictions made with each of these in— struments were calculated. For these analyses the authors suggested that "cutting scores were used which minimized incorrect predictions of LP." These scores which were employed as the upper limits of the pre— diction of LP, were 35 on the Metropolitan, 42 on the Lee—Clark, and 34 on the CTMM. The investigators did not indicate the rationale for how these scores were determined. Results of the study suggested that the Metropolitan was the most effective test in predicting first grade performance, as it "correctly predicted a total of 18 to 19 more children than did the other two tests." The percent of students correctly identified as LP of NLP were reported for each of the predictive measures. The Metropolitan correctly pre- dicted 86% of the LP students and 87% of the NLP students for an overall "hit rate" of 86%. The Lee-Clark correctly predicted 82% of the LP students and 78% of the NLP students for an overall hit rate of 80%. Finally, the CTMM was reported to correctly identify 82% of the LP students, 79% of the NLP students, and 80% overall. Of those children available for testing at the end of second grade, 119 were found to have 38 problems and 77 performed adequately. Using the MET cutting scores of 36/35 employed in the prediction of first grade performance, 74 (91%) of the children predicted LP were classified LP and 70 (61%) of those predicted NLP did, in fact, perform adequately. The observed primary limitation of this study is that it did not address the practical application, as well as limitations of using such instruments as screening measures for making programming decisions for individual students from the predictive information. The investigators did not address the issue of those students who were incorrectly classi— fied, that is those students who were predicted to be NLP, but indeed were LP - 39% in their sample for the second grade predictions. As such, the practical utility of such measures, e.g. Metropolitan, Lee-Clark, or CTMM, may indeed be suspect for purposes of individual decision—making. Feshbach, Adelman, and Fuller (1974) examined the relative efficiency of the DeHirsch Predictive Index versus teacher ratings of students' skills and behaviors in predicting "high or low-risk" students as mea- sured by a determined cut-off score on the Gates-MacGinitie Reading Test. Students were initially evaluated on the DeHirsch and Student Rating Scale (SRS) while in kindergarten, and reading achievement data was ob- tained at the end of the students' first grade year. A step-wise multiple regression analysis was carried out such that the independent contribution of the predictors could be determined. The SRS emerged consistently as the best predictor (r=.57). Further, ad- dressing the issue of predictive accuracy, Feshbach, et al. suggested that: . . . for the purpose of early intervention, a more relevant datum than overall correlation is the efficiency in predicting those children who, by virtue of their fal— ling below a specific criterion, can be considered reading failures (p. 53). 39 A grade placement score of 1.7 and an IQ greater than 90 were selected for defining the criterion of good versus poor reading performance. For the SRS an arbitrary Z-score was selected as the cutoff point for pur— poses of prediction from the teacher ratings, whereas for the DeHirsch, a previously established cutoff criterion of less than five of nine items passed was used for predicting future performance. Relative overall accuracy of prediction was comparable for the DeHirsch, the SRS, and a combination of the measures, ranging from 72-77% correct predictions. The authors further reported that the DeHirsch generated twice as many false positives as the SRS (39% versus 17% respectfully), and that the proportion of false negative errors was relatively small for both measures. Satz and Fletcher (1979) point out that Feshbach, et a1. (1974), however, inadvertently reversed the rows and columns of their detection tables and also reversed their detection rates such that their positive signs (valid and false) each summed to unity. Satz and Fletcher sub— sequently corrected these detection rates and found that the SRS had a 30% valid—positive rate, a 70% false—negative rate, a 3% false-positive rate, and a 97% valid—negative rate. The overall hit rate for the SRS was 77%. By comparison, the DeHirsch had a 26% valid-positive rate, a 74% false-negative rate, a 7% false-positive rate, and a 93% valid- negative rate. The overall hit rate was 73%. The major errors in the original Feshbach, et a1. (1974) study were made in terms of the relative percentages and not in terms of the ab— solute values of the false—negative and false-positive errors. The most significant inaccurate conclusion in the original study was that the false—negative rate was rather small for both measures (24-25%). The corrected rates revealed that the false—negative rates were extremely high (70-74%). Satz and Fletcher further point out that both the SRS 40 and DeHirsch failed to detect over 70% of the high risk children, "a finding that may have been overlooked given the incorrect percentages reported by the authors." They suggested that if an intervention pro— gram were to be implemented based on these percentages, many potentially high-risk children would have been excluded from the prevention program. The results of Stevenson, et al. (1976a) suggest the relative drop in predictive accuracy over periods of more than one or two years. Their previously identified optimal predictor variables generated through multiple regression procedures in kindergarten sample were evaluated to see how well they identified students who actually had later reading difficulties, as indicated by their scoring in the lowest 10 or 25% of the third grade. The findings indicated overall low predictive accuracy figures relative to actual student performance. Only one third of the children in the lowest 10% in third grade reading also scored in the lowest 10% on the prekindergarten screening tasks. Again false-negative predictions were inordinately high. Stevenson, et al., concluded that in order to identify more than half the children who had serious dif- ficulty in third grade (lowest 10%) from the prekindergarten predictors, it would have been necessary to select 25% of the children on the basis of their performance on the prekindergarten tasks. Similarly, to identify students having at least moderate difficulty in reading in third grade (lowest 25%), it would have been necessary to select 40% of the children prior to kindergarten. Stevenson, et al., further found only moderate success in terms of kindergarten teachers' predictions of students' later reading difficulty. Of the twelve children named by their kindergarten teachers and for whom third grade data were available, only three were in the lowest 10% and six were in the lowest 25%. 41 Satz and Friel (1974) investigated the predictive accuracy of twenty— two various developmental and neurological tests initially administered in kindergarten in determining subsequent first grade achievement. The criterion measure of reading achievement in this study was based upon a ten item scale of reading level which was assessed by the teacher for each child at the end of the first grade. The scale ranged from No Read— iness (score=0) to Advanced Third Reader (score=9), with Primer and First Reader scored as 4 and 5, respectively. Those whose reading level was rated as below Primer (levels 0—3) were assigned to the High Risk group; those whose reading was judged to be at the Primer level or above (levels 4-9) comprised the Low Risk group. The two criterion groups (High and Low Risk) were further subdivided into an extreme and conser- vative reading subgroup. The two High Risk subgroups were Severe (levels 0-2) and Mild (level 3) and the two Low Risk groups were Average (levels 4-5) and Superior (levels 6—9). A unique methodological aspect of this study was the use of dis- criminant function analysis to ascertain the most parsimonious combina- tion of predictor variables that would maximize the differences between the identified groups (high and low risk). A composite discriminant predictor score (Zi) was computed for each student, and these scores were then averaged for each group and tested for significance. Results in— dicated that 78% of the High Risk group were correctly classified. The false-negative rate was 22%. Similarly 86% of the Low Risk group were correctly classified with a false—positive rate of 14%. The overall hit rate was 84%. The predictive classifications were further examined within cri- terion groups in order to determine whether accuracy varied as a function 42 of extreme and conservative categories in the High and Low Risk groups. Results demonstrated that accuracy was higher in the extreme categories. All of the severely disabled readers were correctly classified and 95% of the superior readers were correctly classified. Virtually all of the misclassification errors were confined to the more conservative categories, Mild High Rish and Average Low Risk (all of the false—positives and 95% of the false-negatives). Satz and Friel finally addressed the notion of the predictive utility of their battery: The concept of predictive utility implies that the detection signs (e.g., valid positives and negatives) should be evaluated within the broader context of the pre— valence estimates of the event to be predicted. . . in this case this refers to the incidence or base rate of reading disability (Satz and Fletcher, 1979, p. 66). Because of the relatively high false—positive rate obtained with the original cutoff criterion, Satz and Friel elevated the cutoff criterion in order to reduce the likelihood of misclassifying a normal child as High Risk and instituting treatment. By optimizing the valid and false positive rates, the conditional probablities were subsequently computed for the differential composite signs (+ and — ). Given an observed base rate of reading disability of 15% in the sample, the probability that a student would be high risk, given a positive composite score, was .72. Conversely, the probability that a child would be Low Risk, given a negative composite score, was .94. In a subsequent follow—up study Satz and Friel (1978) investigated whether their abbreviated test battery given at the beginning of kinder- garten could predict reading group membership at the end of second grade on a new mixed sample of kindergarten children (males, females, black, white) based on the discriminant lambda weights derived from the 43 two standardization populations of white males. At the end of second grade individual reading measures were obtained on 80% of the original sample. On the basis of those measures, students were divided into Severe, Mild, Average, and Superior reading ability groups. The Severe group was reading approximately 1.25 years behind grade level, whereas the Superior group was reading almost 2.5 years beyond grade level expectancy. Results indicated that the composite scores from the tests in kinder- garten correctly predicted 100% of the Severe and Superior groups, while misclassifying 19% of the Mild group and 37% of the Average group. The overall hit rate on this cross-validation sample was 75%. Again conditional probabilities for each of the composite test signs were calculated. The probability that a child would be high risk, given a severe high risk com— posite score in kindergarten, was .72. This composite score, which was the basis for treatment in this sample, managed to screen 90% of the severe cases and 69% of the mild cases. A much lower predictive outcome was revealed for the mild high risk composite score, .27. In contrast predictions were much higher for both low risk composite scores, .92 and .96. Like Satz and his colleagues, Friedman, et al. (1980) found that despite relative overall high predictive accuracy of the latter screening instruments, the rates of false-positive predictions appeared relatively high and posed significant question as to the practical utility of their prediction battery. Friedman, et al. (1980) assembled a multifactional battery composed of the Star Picture Vocabulary Test, the Goodenough-Harris Human Figure Drawing Test, the Berry-Butkenica Development Test of Visual— Motor Integration (VMI), and the Personal—Social Responsiveness subtest of 44 the Caldwell Preschool Inventory. This battery was initially administered to a kindergarten sample. Primary criterion variables included reading achievement test scores on the Cooperative Primary Tests, as well as teacher questionnaire data on overall academic achievement and reading proficiency. Findings were reported for three population groups for whom teacher ratings were obtained, 1) all students whose school performance was assessed at age seven years, 2) 1,251 children evaluated at age nine years, and 3) 600 children for whom ratings were available at both seven and nine years. For the first group, (those students assessed at age seven years) the Caldwell Test was the best predictor, but had significant value for girls only. However, of the girls who failed this test, 59% did not have a problem at seven years. Thus, a high false—positive rate occurred, even with markedly significant association. Combining tests did not signifi- cantly increase the predictive value of the Caldwell Test alone. In the second group (those students assessed at age nine years), for all children and for each sex separately, the VMI and the Caldwell were the best pre— dictors with the VMI having slightly higher efficiency than the Caldwell. False—positive rates were again high—62% or more. For the third group, the VMI test repeated at age seven years predicted academic achievement and reading at age nine years at a significant level for both girls and boys. The best predictor, however, was the teacher's report of academic performance at age seven years; the false-positive rate was again high at 42%. Combining the academic performance at age seven years with the seven— year VMI yielded an increased accuracy of prediction. Using these com- bined criteria, 89% of the children were correctly classified with regard to success or failure at age nine. However, this ”hit rate" included both the correct positive and the correct negative predictions, which as have 45 been discussed are usually not of equal practical importance. For ex- ample, of those children who failed the combined criteria, 65% had problems at age nine. This 65% is high when contrasted with 11% of the school . failures in the population of nine-year-olds. However, it must be kept in mind that the remaining 35% who failed both the VMI and the teacher report at age seven, did not have school problems at age nine and would therefore have been mislabled. If the consequences of misclassifying these false-positive children into a high risk group were significantly detri— mental, this rate of 35% might be considered excessive, even though there is strong statistical significance. Sex Differences in Reading Achievement and Classroom Behavior Boys comprise the vast majority of educational problems; many esti- mates put the figure at near eighty percent. Although there appears to be no sex differences in terms of general intelligence, girls as a group, especially in the early elementary years, achieve significantly higher than boys in the areas of reading, spelling, and writing. Estimates of the number of boys who experience difficulty with the reading process at some time during their academic experience are estimated to range from seventy-give to ninety percent (Gentile and McMillan, 1976; Stanchfield, 1973). Blom (1970) noted that sex ratios in studies of reading disability range from 1.3:1 to 15:1 with boys representing the higher figures. Ent- wisle (1971) estimated that over ninety percent of referrals to reading clinics were males. Furthermore, the ratio of boys to girls referred to learning disabioity classrooms appears to be fairly consistent across studies, that is approximately 4:1 (Bannatyne, 1971; Bentzen, 1966; Naiden, 1976). 46 'Along with difficulties in reading, boys as a group also experience a number of potentially related problems during the course of the ele— mentary years. Boys are consistently rated by teachers as exhibiting significantly higher rates of disturbing classroom behaviors, e.g. ag— gressive, defiance, impulsiveness, hyperactivity, low need for achievement, and inattentiveness (Goyette, Conners, and Ulrich, 1978; Miller, 1975; Peterson, 1961; Spivak and Swift, 1967). On attitude surveys boys have been found to be more negative towards school in general (Berk, Rose, and Stewart, 1970) and toward reading in particular (Neale, Gill, and Tismer, 1970). Bentzen (1966) reported that while boys were referred three times more often than girls for academic related problems in the first grade, they were referred eleven times more often for reasons of "social and emotional immaturity." Morse, Cutler, and Fink (1954) found the ratio of boys to girls in classrooms for the emotionally disturbed to be five to one. A review of relevant studies of the relationship between classroom behavior and academic achievement of elementary school children suggest that sex of student may be relevant moderating variable in this relation— ship. Differences with respect to reading achievement in the elementary grades between boys and girls (Maccoby and Jacklin, 1974; Thompson, 1975) and with respect to classroom behavior (Rubin and Balow, 1978; Werry and Quay, 1971) have been established. It may, however, be the case that a pattern of classroom behaviors may be predictive of reading achievement for girls but not boys, or it may be that the direction of the relation— ship between behaviors and subsequent achievement is different for male and female elementary pupils. While Lahaderne (1968) and Samuels and Turnure (1974) found that sex of pupil did not have an effect on the behavior-achievement relation, a number of other investigators have found support for such a hypothesis. 47 Cobb (1970) for example found a higher correlation between attention, com- pliance, and volunteering and achievement for boys than girls, and he further reported that different behaviors entered regression equations for the two sexes predicting first grade achievement. Blom, et al. (1980) found a significant sex by reading ability interaction, wherein poor reading males were rated by teachers as exhibiting much more acting-out, aggressive behavior than their good reading male peers. On the other hand, post-hoc comparisons of the aggression scores in the female sample failed to reveal a significant difference between good and poor readers, although mean differences indicated good reading girls were rated higher in ag— gressive behavior than their poor reading female counterparts. Feshbach, et al. (1977) found several interesting differences in a longitudinal study of the relationship between teacher ratings of class- room behavior and reading achievement for first through third grade boys and girls. They found that the overall and attention-behavior control scores of the Student Rating Scale (Feshbach and Adelman, 1971) corre— lations for girls were higher than those of boys. They suggested that: Motivational variables may be more important for girls in the early primary grades, thus possibly accounting for the lesser significance of IQ (as compared to boys). At the same time, kindergarten behavior factors, which re— flect poorer performance by males than females, may decline in importance for males with increasing maturity (p. 303). Analysis of behavior-achievement data generated during the course of the investigation of Blom, et al. (1980) indicated differences in the early versus later elementary years on the relationship between various behavioral dimensions and reading achievement between the sexes. For example, good reading females in the lower elementary grades were rated higher on an aggression scale than their upper elementary peers. In addition, poor reading males were rated higher in aggression in the lower 48 elementary grades were rated higher on an aggression scale than their upper elementary peers. In addition, poor reading males were rated higher in aggression in the lower elementary grades than their upper elementary counterparts. Sex Differences in Reading Achievement_in the Elementary Grades In this section the evidence of sex differences in reading achievement in the elementary grades is reviewed. To summarize, there appears to be a developmental trend regarding differences between the sexes in reading achievement, wherein, a significantly larger proportion of boys than girls make a slow beginning at learning to read, but during the later elementary years, grades four through eight, significant population differences begin to disappear. Sex differences in reading ability have been a source of concern to educators, researchers, and school administrators for many years. As early as 1909, Ayres in his book, Laggards in Our Schools, concluded: "Our schools as they now exist are better fitted to the needs and natures of the girl than the boy pupils" (p. 155). Subsequent empirical studies have confirmed the observation that boys in American elementary schools are far more likely than girls to have serious reading difficulties. Examination of investigations from several English-speaking countries, primarily the United States and Great Britain, indicates that the trend of sex differences in mean reading achievement scores for children ap— proximately ten years and younger is different from that of older ele- mentary students; girls as a group generally score significantly higher than boys in the early grades. Mean scores in both reading comprehension and reading vocabulary were found to be significantly lower for boys than girls by Carroll (1948) for six to seven year olds, and by Preston (1962) 49 and Rutter, et al. (1970) for nine year olds. In all cases the mean achieve— ment levels of the boys were significantly lower than that of the girls. 0n standardized measures of word recognition, boys again scored significant- ly lower than girls in studies reported by Southgate (1959) for six to seven year olds, and by Kellmer Pringle, et al. (1966) in a nationwide sample of seven year olds in England. Numerous studies of first grade children have confirmed the reading achievement superiority of girls during the early stages of reading in- struction. Konski (1951) evaluated the performance of boys and girls on tests of reading readiness at the beginning of the academic year, and sub- sequent reading achievement at the end of the year. Results indicated no significant differences between the sexes on the readiness measures, however girls scored significantly higher on the reading achievement measures at the end of the year. Nila (1953) similarly tested three hundred first graders during the first weeks of school on a number of individual and group administered reading readiness tests. Again it was reported that boys as a group and girls as a group were comparable in readiness to read. The students were again tested at the end of the year. Seventy-two pupils were designated ' and of these two thirds were boys. as "reading failures,‘ During the course of the 1962-63 school year in the Los Angeles City Schools, Stanchfield (1965) initiated a study of the effect of homogeneous grouping of first grade pupils by sex upon subsequent reading achieve- ment. Over five hundred students took part in the study. The analysis of reading achievement and reading growth indicated no significant differences between boys taught in all-male groups and those taught in the sexually heterogeneous groups. The girls on the whole, irregardless of group, 50 however, scored significantly higher than the boys at the year's end. During the school year 1964—65, twenty-seven first grade reading studies were facilitated by grants from the U.S. Office of Education. A large percentage of these studies compared, as did Stanchfield, the rela— tive efficiency of different methodologies and materials in the first grade. Some of these studies did not directly compare scores of boys and girls, while others sought to determine if one particular instructional approach was more effective with either boys or girls than were other approaches. A number of these studies do provide data on sex differences in reading achievement in grade one. Of significance is the fact that each of these studies utilized a number of common pre— and post-treatment measures of achievement, primarily the Stanford Achievement Primary Bat- tery. Results from these studies which compared the relative achievement of boys and girls are cited. Tanyzer and Alpert (1966) compared the mean scores achieved through the use of three basal approaches (Lippincott; ITA; and Scott—Forsman) and found that in each of the three basal systems, the girls achieved significantly higher mean scores than the boys with significant differences favoring the girls on the Spelling and Paragraph Meaning subtests. Hahn (1965) reported comparative data only for the Word Study subtest, where girls scored significantly higher than boys. Schneyer, et al. (1966) noted significant differences favoring girls on Paragraph Reading, Spelling, and Word Study Skills. Spache (1965) and Fry (1966) both noted significant differences favoring girls. Fry also noted that none of the specific methods investi- gated (ITA; Diacritical Marking; and regular basal materials) proved to be more advantageous for boys as a group. 51 Other studies of second and third graders have confirmed early fe- male superiority in reading achievement in the elementary grades. Hughes (1953) evaluated over six hundred students in the metropolitan Chicago area in grades two through eight. Utilizing the total compre— hension score of the Chicago Reading Tests, she found the greatest mean differences between boys and girls in her third grade sample. Mean grade level scores were higher for girls at all grade levels, except for fifth grade, where factors of chronological age and mental ability significantly favored the boys. Significant statistical differences favoring the girls were found in the third and fourth grade samples. Blom, et al. (1980) similarly found significant reading achievement differences favoring girls in both the second and third grades utilizing a combined Vocabulary and Comprehension score from the Gates—McGinitie Reading Achievement Test. Consistent with the results of Hughes (1953), significant differences diminished in the upper elementary grades, four through six. Measures of reading attainment discussed to this point in the review have been exclusively group administered, standardized tests. A few reports of sex differences in the early elementary grades on individually administered tests or oral reading are, however, available. The findings of Rogers (1971) and Prawat and Jarvis (1980) appear to confirm the superiority of girls in the early elementary school grades on such reading tasks. Rogers (1971) found in a sample of 500 students aged five and a half to six years that boys as a group achieved significantly lower on an individually administered oral reading task. Prawat and Jarvis (1980) in a study of first through third grade students in a large midwestern urban school setting found girls to score significantly higher on the 52 Slossen Oral Reading Test at all three grade levels. Anderson, Hughes, and Dixon (1956) investigated individual differences in reading development in elementary age pupils at the University of Michigan Laboratory School. Significant sex differences were noted in the chronological age of pupils of learning to read. The students were judged to have "learned to read" when they achieved a reading age of eighty-four months on the Gates Primary Reading Tests. Their primary find— ings were that: 1. Age of learning to read was widely distributed despite the fact that most of the children in the sample were of superior intelligence. 2. Girls tended to read earlier than boys and there were fewer ex- treme delays in learning to read among girls--a large percentage of boys actually learned to read at second grade rather than first. The average age of learning to read was 83.1 months for girls and 89.6 months for boys. 3. Age of learning to read was significantly corrolated with in- telligence quotients recorded in the first grade (.57 for girls and .54 for boys). 4. Age of learning to read was significantly corrolated with sub- sequent reading achievement in the sixth grade (.67 for girls and .65 for boys). In a second study, Anderson, Hughes, and Dixon (1957) found that after children reached the reading age of eighty-four months on the Gates Primary Reading Test, there was no difference between boys' and girls' subsequent growth rates in reading. Sex differences in the age of learning to read tended to disappear among children of high intelligence (above 100), whereas children of lower intelligence (less than 100) tended to 53 start later, and proceed generally slower than the higher IQ group. A high degree of variability in the rate of reading development was, however, observed in the lower IQ group. Sex differences were further most prom— inent in the lower IQ group with the boys tending to start later than the girls. For children in the fourth through eighth grades the consensus of the literature indicates generally nonsignificant differences between the sexes in reading achievement (Thompson, 1975). Divergent findings tend to support the notion of female superiority throughout the elementary grades. Numerous studies have indicated no significant differences in reading achievement between the sexes on group administered measures in the upper elementary grades (Blom, et al. 1980; Commins, 1928; Heilman, 1933; Heil- man, 1961; Horton, 1973; Rutter, et al., 1970; and Wozencraft, 1963). Neale (1958) and Vernon (1938) furthermore found no significant differ- ences between the sexes in the ten to fifteen year age group in individ— ually administered tests of oral reading, as did Prawat and Jarvis (1980) with fourth through sixth graders. Stroud and Lindquist (1942) found the mean achievement levels in both reading comprehension and reading vocabulary to be significantly lower for boys in the third and fourth grades in their cross-sectional study of third through twelfth grade students in Iowa. They further reported non— significant mean differences between the sexes at all other grade levels favoring the girls in reading vocabulary and comprehension, language, and word study. Assessment data from the State of California 1975-76 testing program of second, third, and sixth grade students substantiated the finding of 54 female superiority throughout the elementary grades in reading, written expression, and spelling (Gehringer and Barnes, 1978). Results of recent Michigan Educational Assessment data further substantiate the reading attainment superiority of girls from the fourth through tenth grades: Staff found that on each of the fourth grade reading objectives, the attainment rates for girls were higher than that for boys by an average of 9.7% in 1979-80, 8.8% in 1978-79, and 8.5% in 1977-78. . . Further the assessment data show that this basic pattern holds true for seventh and tenth graders with boys closing the gap slightly at the higher grades (Katz, 1980). Gates (1961) sampled over thirteen thousand pupils in grades two through eight to determine sex differences in reading ability. A total of twelve school systems in ten states participated in the study. Boys and girls were compared at each of the seven grades on reading speed, vocabulary, and comprehension. In each of the twenty-one comparisons, girls had higher mean raw scores with most of the mean differences signif— icant. Preston (1962) obtained similar results in reading comprehension in his American sample of eleven year olds. To summarize, there appears to be a developmental trend regarding differences between the sexes in reading achievement, wherein, a signif- icantly larger preportion of boys than girls make a slow beginning at learning to read, but during the later elementary years, grades four through eight, significant population differences begin to disappear. Sex DifferenCes in Behavior Elementary age boys have been shown to exhibit significantly more achievement-inhibiting, as well as disturbing classroom behaviors as girls in both normal and special class populations of students (Rubin and Balow, 1978; Spivak and Swift, 1966; Werry and Quay, 1971). Whether one looks at mental retardation (Lehrke, 1978), reading difficulty (Spache, 1976; 55 Rutter and Yule, 1977), hyperactivity (Cantwell, 1977), aggression (Mac- coby and Jacklin, 1974; Miller, 1972), or simply low grades (McCandless, et al., 1972), boys predominate in frequency and intensity of the problem. The focus of this review is upon those differences between the sexes that have been reported in regard to the prevalence of behavioral adjust- ment difficulties of elementary age children. It is beyond the scope of this review to address the possible causes for the observed differences, and the reader is referred to Eme's (1979) article for a more indepth examination of the topic. In an investigation of nearly 1000 kindergarten through third grade children, Rubin and Balow (1971) found that educationally defined be— havior problems (i.e., inability to meet school demands), were present in 41% of their sample. In all areas in which results were differentially reported by sex, boys consistently outnumbered girls, e.g. grade retention, attitude and behavior adjustment difficulties, and special class placement. In their 1978 longitudinal examination of teacher identified behavior, problems, Rubin and Balow found the proportion of boys identified as be- havior problems remained relatively constant at 35% over the seven year follow-up period from kindergarten through sixth grade for the same sample. However, considerable variability in the proportion of behavior problems identified among the girls was evidenced, ranging from a high of 26.4% at grade 1 to a low of 12.8% at grade 6. In any single year, approximately 69% to 76% of all the subjects in the study were classified as no problem. One of the more interesting findings of the study was that more than half (59%) of the 1527 studnets who had received three or more teacher ratings, were classified as a behavior problem at least once. Among the subjects who received six teacher ratings, 60% (68% of the boys and 51% of the l 56 girls) were considered a behavior problem by at least one teacher. Des— pite the relative lack of precision in definition of a behavior problem in their study, Rubin and Balow's (1978) findings do indicate that boys are indeed rated by teachers as exhibiting more adjustment difficulties than girls, consistent with Brophy and Good's (1970) observation that, "boys are generally more salient in the teacher's perceptual field than girls." Probably the most unequivocal sex difference in the literature is the male preponderance of aggressive behavior and generally higher levels of activity (Block, 1976; Eme, 1979; Maccoby and Jacklin, 1974). Despite relative developmental changes on these dimensions for both sexes, boys are rated by both parents and teachers as exhibiting significantly higher levels of aggression and "overactivity" as compared to girls. Achenbach and Edelbrock (1981) in their investigation of the pre- valence of behavior problems among normal school age children interviewed nearly 1500 parents and solicited their responses to the 112 item Child Behavior Checklist. ANCOVA procedures indicated significant gender dif- ferences on 51 of the items. Of the 51 main effects, 26 reflected higher scores for girls and 25 higher scores for boys. Nineteen of the 25 items on which boys scored higher were found to be primarily associated with what the authors conceptualized as "a broad band externalizing syn- drome," which included such behaviors as bragging, hyperactive, cruelty to others, destructiveness, disobedience, fighting, impulsivity, showing off, temper tamtrums, and swearing. By contrast, only one of the items (screams a lot) on which girls scored higher was primarily associated with the externalizing syndrome; 10 items were primarily associated with the internalizing syndrome, and the remaining 15 items were not associated 57 with either of the syndromes. Girls scored significantly higher than boys on such behaviors as being too dependent, lonely, crying a lot, feeling unloved, nightmares, headaches, self-consciousness, shyness, moodiness, and worrying. These identified behaviors for boys and girls appear highly consistent with those of MacFarlane, et al., (1954) who found that boys exhibited significantly higher levels of overactivity, demand for attention, temper tantrums, and lying, while problems for girls centered more com- monly around excessive reserve, shyness, and specific fears. Despite the clear differences between the sexes on these individual items the mean total behavior problem scores for boys and girls were virtually identical in the Achenbach and Edelbrock study and the authors suggested that: The much higher mental health and special education referral rates for boys than girls throughout childhood and their higher court referral rates in adolescence may result more from the conflicts between their behavior and official norms than from quantitative differences in problems (p. 51). Werry and Quay (1971) again studied the prevalence of behavior problems in a nonclinical sample of 1700 students aged five to eight. Their re- sults indicated that while symptoms of pyschopathology were quite preva- lent overall in this age group, boys tended to exhibit more acting out or disruptive behaviors than girls. Forty-nine percent of the boys in the sample were rated as restless, 46% disruptive, 43% short attention span, 43% inattentive, and 48% distractible. "Distructiveness" was present in 16% of the boys and 5% of the girls, while ”hyperactivity" was present in 30% of the boys and 14% of the girls. Werry and Quay concluded that there was little doubt that not only did boys have more psychopathological symp- toms but the "connotative sense of most of the symptoms more common in boys represents 'badness'" (p. 142). Similar sentiments were expressed by Schlosser and Algozzine (1979) who found that those behaviors which tend 58 to be more characteristic of boys, e.g. acting-out, aggressive, immature, are perceived by teachers as more bothersome than those behaviors char- acteristic of girls, e.g. withdrawn, self-conscious. Lapouse and Monk (1964) in a later analysis of their examination of maladjustment among a non—clinical sample of six to twelve year old child— ren (1958), similarly found boys as a group to be rated significantly higher than girls by parents as exhibiting more extreme deviations on such behaviors as bedwetting, masturbation, overactivity, temper loss, and teacher complaints. Replication of the study with a rural elementary sample indicated similar results (Schultz, et al., 1974). Spivack and Swift (1966) in their initial standardization of the Devereaux Elementary Behavior Rating Scales found significant sex differ— ences on eight of the twelve factors of the teacher rating scale. Boys were found to score higher on those dimensions that were detrimental to achievement than girls, whether in special or regular class settings. Boys scored significantly higher on those dimensions defined as Classroom Disturbance, Slowness in Work, External Reliance, Need Achievement Recog- nition, Impatience, and Disrespect-Defiance. Girls scored significantly higher on those dimensions defined as Need for Closeness with the Teacher and Creative-Initiative (Verbal). Similar differences between boys and girls were identified by Goyette, et al., (1978) in the restandardization of the Conners Parent and Teacher Ratings Scales. In a sample of 590 elementary age students, boys were rated by their parents as exhibiting more difficulties on those factors defined as Conduct Problem (e.g. destructive, quarrelsome, fights), Learning Problem (e.g. difficulty learning, distractible, easily frus— trated, immature) and Hyperactivity (e.g. restless, disturbs others, 59 excitable, impulsive). In their four year longitudinal study of the ability of teacher ratings of students' abilities, classroom skills, and personal-social skills to predict future reading achievement, Stevenson, et al. (1976a) found that girls were rated higher on various achievement—enhancing behavior dimen— sions in all comparisons. Girls were given significantly higher ratings in Overall Adjustment and Persistence, while mean ratings were also higher in Effective Learning, Hardworking, Retaining Information, Reflective, Attention, and significantly lower in Activity Level. Validity 9f_Teacher Ratings_gf Pupil Behavior Because the judgments of classroom teachers have been increasingly used as a first indicator of behavioral problems of students and may be a primary referral source to special education or remedial services, a number of researchers have attempted to systemize teachers' observations of pupil behavior in order to achieve the measurement characteristics of standardized instruments. This has resulted in the construction of num— erous inventories of child behavior in which norms referenced procedures have been used to establish base rates of problem behavior within normal and handicapped populations of school children (Bower, 1960; Chamberlain, 1966; Miller, 1972; Ross, Lacey, and Parton, 1965; Spivak and Swift, 1966; Walker, 1970; Werry and Quay, 1971; Zax, Cowen, Izzo, and Trost, 1964). Diagnostically these instruments have proven valuable in the specification of relevant areas of behavioral intervention. Past research in the study of the relationship between pupil class- room behavior and reading achievement has utilized two primary means of assessing student behavior, (1) teacher ratings, and (2) direct class- room observation. A critical question then that must be addressed is, 60 what is the relationship between teacher judgments of pupil behavior and such techniques as observation by external observers? Young-Maston (1938) appears to be the first investigator to study the accuracy of teacher judgments of child behavior through the use of obser— vational data. Observational data derived from continuous diary recording of classroom behavior did not discriminate between pupils identified by teachers as the worst behaved and averaged behaved, although teacher ratings of the students behavior produced highly significant differences between the two groups. In this study then, the accuracy of teacher judgment was not corroborated by classroom observation. Bolstad (1974), however, suggested that this study may have been confounded by the recording pro— cedures used in analyzing student behavior by the outside observers. Bolstad (1974) further investigated the relationship between teacher ratings and direct behavioral observations of children selected by their teachers as best behaved, average behaved, and least well behaved. Third and fourth grade teachers were asked to rate these pupils on three ret— rospective rating scales measuring classroom conduct. A behavioral coding system (Walker, Mattson, and Buckley, 1971) which consisted of four appropriate and four inappropriate behavior categories was used by pro- fessionally trained observers. Results of the study indicated that teachers perceived highly significant differences between the three groups on all three rating scales. Behavioral observation data also discrimi— nated among the three groups of pupils. The proportions of appropriate classroom behavior were .92, .88, and .70 for the best, average and least well behaved groups, respectively. Boys averaged .80 and girls .87 across the three groups, a difference that was also significant. These results strongly suggested that teachers in his sample were quite 61 accurate in their assessment of pupil behavior in their classes. Nelson (1971) utilized direct observational procedures to investi— gate differences between children initially classified as conduct—dis— turbed or normal based upon behavioral ratings provided by their regular education teachers. The specific purpose of this study was to confirm the usefulness of teacher ratings and direct observation by validating one technique against the other. The results indicated that the normal subjects engaged in significantly more task—oriented behavior and sig— nificantly fewer deviant behaviors than the previously identified conduct— disturbed students, and that results of the observation procedures cor— roborated the accuracy of the teachers‘ initial ratings. Jones and Cobb (1973) conducted a study in which they compared the relative usefulness of teachers' versus observers assessments of pupil behavior. Specifically the objectives of their study were to measure the convergent and discriminant validity of observational data versus teacher ratings and to evaluate the predictive validity of these two data sources against standardized achievement scores. Teachers rated eighteen cate— gories of classroom behavior on a seven point scale; observers rated the same categories using an interval recording procedure. The Stanford Achievement Test was administered to all pupils to measure reading achievement. The results indicated that teacher ratings of students' classroom behavior tended to be more predictive of reading achievement than the observations of the same behavioral categories, e.g. validity coefficients of .62 versus .39. As might be expected, however, the con— vergence between teacher ratings and behavioral observation data was weak, with coefficients of .28 and .27 for appropriate and inappropriate be— haviors respectively. Their results do suggest that when teachers are L— 62 given similar criteria for judging behavior, as are professionally trained observers, they are at least as skilled as observers in their judgments of child behavior, and more skilled in judging classroom behaviors predictive of achievement. Camp and Zimet (1974) examined the relationship between two measures of teacher ratings of student behavior, the Pittsburgh Adjustment Scale (Ross et al., 1965) and the Connors Teacher Rating Scale (Conners, 1969) with independent observations of classroom behavior and achievement test scores in two first grade classrooms. Correlations were performed between classroom observation variables and those subscales of the PASS and CTRS which did not show any significant differences between the two classrooms. Observation of off—task behavior correlated significantly and positively with the Daydreaming subscale of the CTRS, .31 and significantly by neg— atively with the Presocial subscale of the PASS, —.38. Relationships were further examined between teacher ratings and observations within each classroom. The findings indicated that the two teachers varied greatly in the extent to which their ratings correlated with the ratings of the observers. Camp and Zimet reported that one teacher had nine out of twenty—nine significant correlations, while the other teacher had nineteen out of twenty—nine. Unfortunately, none of the correlations were reported. With the small number of teachers sampled, and the high degree of var— iability between the two, the findings appear to be inconclusive regarding the relationship between teacher ratings and observational data. In relation to achievement test scores, the authors gave no indication as to which method of behavior recording, teacher judgment or direct observation, was most closely related to measured academic achievement. 63 Several reasons may account for the difference noted between he— havioral assessment by means of teacher ratings and direct observation. It appears that one significant difficulty is that the items on the teacher rating scale may not correspond to the items on the instrument used by the observer. Even when the items are comparable, the situations in which the child is observed may not. Thus, even if observations are made by well-trained, unbiased persons, the observations inevitably reflect a significantly smaller segment of behavior than that encountered by the teacher. Furthermore, both teacher and observer ratings may vary in the extent to which they report positive as opposed to negative behavior. Werry and Quay (1969) suggest that most children, whether they have be— havior problems or not, are on-task seventy-five percent of the time. Bolstad's (1974) findings appear to concur with this hypothesis. Camp and Zimet (1974) suggest that because most children are on—task for the majority of the time, positive behavior will be more likely to be ob— served by both external observer and teacher since it is more common. Conversely, if a child's behavior is truly negative or deviant and leads to explicitly stated rules over twenty—five percent of the time, it should also be readily observed by both teacher and observer. CHAPTER III DESIGN OF THE STUDY Population and Sample Subjects for the present study were 93 students from a metropolitan elementary school in mid—Michigan. The school was selected on the basis of its representative population in terms of race and socioeconomic status according to school district administrators. Support for the study was solicited from the school principal and ten teachers, nine female and one male, who voluntarily indicated a willingness to participate in a larger, related research project (Blom, et al., 1980). The school is located in a "blue collar" area of the city and district sources indicated that approximately 21% of the students' families are currently receiving aid to dependent children support and that this figure is slightly lower than the district average. Twenty—six percent of the students are of minority racial status, as compared to a district wide average of 35%. Seventy-four percent of the school population is identi- fied as Caucasian, 17% Black, 6% Latino, and 3% Asian. Specific socio- economic and racial data on individual children were unavailable. Regarding the educational status of the students at the school, school district sources indicated that 27.5% of the students are re— ceiving Title I reading and mathematics support services as compared to 64 65 a 29% district-wide average. Four percent of the school's students are receiving special education services as compared to an average of 5.4% district-wide. Table I indicates the average percentile reading score on the Stan- ford Achievement Test (SAT) for the school's students as compared to district-wide elementary totals. TABLE I Average Percentile Rank — SAT Reading for Grades 1-6 1977-78 1978—79 1979-80 1980—81 The School 45.9 49.5 48.6 49.1 District 49.6 51.3 51.9 53.4 All students in two classrooms of each grade level, two through six, were initially evaluated. That population consisted of 226 students, 124 males and 102 females. Male—female totals at each grade level ranged between 34 and 51, and it was possible to sample at least 16 boys and 15 girls at each grade level. Subjects for whom not all relevant data (behavior checklist and intelligence) was available, or students whose protocols were reported to be invalid by test examiners or test scorers, were excluded from further involvement in the study. Ninety—three students from the original group were subsequently identified for longitudinal follow—up three years after the initial eval- uations. These subjects were those (1) who were able to be identified as 66 still attending the school district after the three year interval, (2) whose parents or guardians consented to release of their child's reading achievement test scores from the school district records, and (3) whose SAT scores were available for all four years. The mean age and grade levels of the final sample were 114.8 months i 16.2 months, and 4.0 i 1.5 respectfully. Fifty—three percent of the sample were male and 47% female. The mean intelligence quotient for the sample was 104.3 f 11.8. A breakdown by grade in Table 2 suggests a rather equal representation of students at each level. TABLE 2 Relative Percentage and Absolute Frequency of Subjects at Each Grade Level Grade Frequency Relative Percentage 2 17 18.3 3 19 20.4 4 21 22.6 5 17 18.3 6 19 20.4 A review of relative demographic, cognitive, and behavioral data suggested that the 93 subjects appears to be highly representative of the original population of students evaluated. The sample was comprised of 53% (n=49) males and 47% (n=44) as compared to 54.9% males (n=124) and 45.1% (n=102) in the original population. Table 3 indicates a relative comparison of the 93 students to the original population in terms of 67 average age, grade level, IQ, and raw scores on the behavior dimensions identified for analysis in the present study. TABLE 3 A Comparison of Relevant Sample Versus School Population Characteristics Sample Population Meani S.D. Mean S.D. Age (months) 114.8 16.2 116.4 17.6 Grade Level 4.0 1.5 4.2 1.4 IQ 104.3 11.8 101.3 12.7 Behavior Scales Attention/Motivation 16.5 4.3 16.8 4.4 Disrespect/Defiance 9.5 2.2 12.4 2.8 Attention-Seeking 10.6 2.4 10.4 2.2 Academic Self—Confidence 6.8 1.2 7.1 1.4 Socialization 3.5 0.9 3.6 1.0 Externalized Aggression 9.5 2.2 9.5 2.3 Appendix A indicates a breakdown of mean Stanford Achievement Test Scores by a grade and sex of student, while Appendix B indicates a breakdown of mean behavior scale raw scores also by grade and sex of student. 68 Instrumentation Dependent variables assessed in this study were teacher ratings of student classroom behavior, sex of student, and intelligence. Reading achievement was the dependent variable. Student Behavior. The SchOol Behavior Checklist: Form A2 (SBCL) (Miller, 1972) was used to assess student classroom behavior. The SBCL is an inventory of behaviors designed to help teachers communicate their impression of children in their classrooms. This inventory covers a wide range of social and academic behaviors from moderate social deviance in- dictative of pathology to social and academic competence. Rare and extreme deviant behaviors were not included in the inventory. The SBCL provides a relatively simple procedure for acquiring teacher's impressions of their students. The SBCL is such that achievement-related versus more mental health-related behaviors could be identified. The SBCL was developed by Miller (1972) primarily from cross—valida- tion of an earlier measure, the Pittsburgh Adjustment Survey Scales (Ross, Lacy, and Parton, 1965) with the addition of fourteen learning disability items, and two manifest anxiety items. The standardization sample was comprised of 2,627 male and 2,746 female students in the city, county, parochial and private schools of Louisville and Jefferson County, Kentucky. Teachers who rated the protocols had on the average 13.8 years of experience. Form A2 consists of 96 statements regarding childrens' behavior. Teachers were asked in a forced choice format to decide if the behavior described the child being rated, and if so, check True (T), if not, False (F). Six factor scales and one clinical scale have been derived for the SBCL—A2. The factors were found to be identical for boys and girls. Ii 69 Test-retest reliability coefficients for the individual subscales range from .70 to .90 with a reliability coefficient of .93 for the Total Disability Scale. Test-retest reliability is similar with the individual subscales ranging from .70 to .89 with an overall test—retest coefficient of .89. The student behavior ratings used as independent variables in the pre- diction of future reading achievement in this study consisted of new be- havioral dimensions derived from SBCL items which were found to be signif— icantly related to reading achievement for the total sample and boys and girls independently. Appendices D through F indicate all attained simple correlations. The basic rationale behind creation of these behavior dimensions or scales was to reduce the number of behavior variables to be considered as independent variables in the prediction of later reading achievement. Scales derived from the original SBCL sample were deemed not meaningful for purposes of the present study as they appeared too heterogenous in nature, and did not reflect specifically those behavioral dimensions which have been found to be related to reading achievement. Quay (1965) has suggested that the behavior of children may be most profitably viewed as dimensions should satisfy three internal criteria: (1) that they are mutually exclusive as possible, (2) that they are objective as possible, and (3) that they are specifically demonstrated as a cluster. Kerlinger and Pedhazur (1973) and Kim and Kohout (1975) state that data—reduction procedures, such as scale development, are practically and statistically important in problems involving prediction in general, and multiple regression procedures in specific. Both sets of authors suggest that the creation of new variables, which are composite scales of highly intercorrelated items, is one solution to the problem of multicollinearity 70 (Gordon, 1968). Multicollinearity refers to the situation in which some or all of the independent variables in a prediction equation are highly intercorrelated. Prediction, and subsequent explanation of the predictive relationship between the independent variables and dependent variable is enhanced when the predictor variables of interest are essentially in— dependent or orthogonal (Kerlinger and Pedhazur, 1973). Forty-nine items from the SBCL, that were found to be significantly related to initial levels of reading achievement, formed the pool of items in the development of the new behavior scales which were to be used in the prediction of future achievement. This item pool was subsequently sub- jected to factor analytic procedures to examine the underlying statistical interrelationships of the items and to reduce the number of behavioral pre— dictor variables to a manageable number of statistically and conceptually meaningful behavioral dimensions. Four to eight factor solutions were examined through principal factoring with iteration and varimax rotation procedures and factors were further analyzed for internal consistency reliability. Extraction of six factors produced what appeared to be the most parsimonious, independent, and con- ceptually meaningful behavioral scales given the previously identified item pool. For the purpose of identifying items in the formation of the scales, the following guidelines were observed: 1. No item was to be included in a scale unless it had a factor loading greater than or equal to .30 with a loading of .40 pre- ferable. 2. No item was to be included on more than one scale. 71 3. Each scale should have a minimum of three items (a minimal re- quirement in terms of measuring internal consistency reliability). 4. The items included in each scale should be conceptually related. The first step in the development of the scales was to examine the factor loadings of each item for the four to eight factor solutions. All factor loadings of .30 or greater were identified and each item was ex- amined to determine which BEE factor the item was best represented by. Items were selected for a given factors based on the highest loading value of that item across factors. All derived factors were subsequently eval— uated in terms of their internal consistency reliability. The next step in the procedure was to examine the factors in terms of their conceptual clarity. As was indicated, the six factor solution clustered the items in what appeared to be the most conceptually meaning- ful as well as statistically parsimonious manner. All six factors contained three or more items and all were comprised of items which appeared theoretically related. Appendix C indicates the item loading values for the six factor solution. The final step in the process was to examine the items within each scale and devise a descriptive title for each. The following is a con— ceptual and statistical description of the six derived scales. SCALE I - ATTENTION/MOTIVATION Scale I is comprised of thirteen items which tend to describe a student's general adaption to school structure and expectations (e.g. works well alone, is interested in school work, finishes classroom assign- ments, does homework, is sure of self), as well as a student's ability to attend to task (e.g. is alert in class, is able to conceptrate on things, is distractable). The internal consistency reliability coefficient 72 of this scale was .94. SCALE II - DISRESPECT/DEFIANCE Scale II consists of ten items which tend to described a student's active non—compliance to authority and structure (e.g. will put up an argument when told not to do something, is infuriated by any form of discipline, is stubborn, sulks when things go wrong), as well as exter- nalized emotional lability (e.g. is easily upset by changes, has change- able moods, when angry will do things like slamming the door or hanging the desk). The reliability coefficient of this scale was .87. SCALE III - EXTERNALIZED AGGRESSION Scale III is comprised of eight items which reflect a student's ver- bal and physically aggressive interpersonal orientation with peers in the school setting (e.g. starts fighting over nothing, hits and pushes other children, teases other children, uses abusive language towards other children, threatens to hurt other children when angry). The reliability coefficient of this scale was .87. SCALE IV - ATTENTION-SEEKING Scale IV is comprised of nine items which describe a student's "immature", impulsive, active, attention-seeking behaviors (e.g. likes an audience all the time, never seems to be still for a moment, interrupts whoever is speaking, disturbs other children with boisterous behavior, does things which are normal for children much younger, does things just to attract attention). The reliability coefficient of this scale was .87. SCALE V - ACADEMIC SELF-CONFIDENCE Scale V consists of six items which tend to describe a student's more passive, reticent orientation towards academic and classroom demands (e.g. 73 approaches a task with an air of defeatism, does not volunteer to recite in class, seems dull and slow to catch on, will not ask questions even when unsure of how to do the work, has difficulty speaking when excited or upset). The internal consistency coefficient of this scale was .60. SCALE VI — SOCIALIZATION Scale VI is comprised of three items which reflect a student's social integration with his peers (e.g. is popular with classmates, prefers to play alone and be alone, is slow in making friends). The internal con- sistency coefficient of this scale was .73. Table 4 indicates the six new behavioral scales with their corres— ponding items and factor loadings in descending magnitude for each scale. TABLE 4 Reading Achievement-Related Behavioral Scales Item No. Item Description Factor Loading SCALE I: ATTENTION/MOTIVATION 22 Does homework .88 18 Fails to carry out tasks .84 19 Lacks ambition to do well in school .80 76 Finishes classroom assignments .80 36 Finds it hard to study .79 52 ”Drags feet" when requested to do some— thing .72 57 Distractible; can't concentrate .68 64 Is interested in school work .65 48 Is able to concentrate on things .61 7 Is alert in class .57 2 Tends to give up if has something hard to finish .53 27 Is sure of self .50 38 Works well alone .48 54 47 21 29 77 26 39 62 86 34 13 20 23 31 17 72 28 35 51 70 87 92 46 66 78 43 44 16 63 74 "TABLE 4 (cont'd)." SCALE 2: DISRESPECT/DEFIANCE Sulks when things go wrong Argues with the teacher Will put up an argument when told not to do something Has changeable moods Gives other children dirty looks Is easily upset by changes When angry, will refuse to speak to anyone Is stubborn When angry, will do things like slamming the door or hanging the desk Is infuriated by any form of discipline SCALE 3: EXTERNALIZED AGGRESSION Hits and pushes other children Does things to get others angry Teases other children Does not respect other people's belongings Is not considerate of others Threatens to hurt other children when angry Starts fighting over nothing Uses abusive language towards other children SCALE 4: ATTENTION-SEEKING Likes an audience all the time Tries to be the center of attention Does things which are normal for children much younger Acts in a fearless, "dare—devil" manner Disturbs other children with boisterous behavior Never seems to be still for a moment Interrupts whoever is speaking Does things just to attract attention Deliberately interrupts what is going on by asking silly questions SCALE 5: ACADEMIC SELF—CONFIDENCE Seems dull; slow to catch on Will not ask questions even when unsure of how to do the work Approaches a task with an air of defeatism Never speaks up even when there is cause to be angry .68 .64 .62 .59 .51 .51 .47 .47 .43 .38 .79 .64 .64 .56 .55 .51 .48 .48 .77 .72 .58 .53 .52 .47 .45 .45 .43 .51 .47 .46 .44 75 "TABLE 4 (cont'd)." 88 Has difficulty speaking when excited or upset .44 12 Volunteers to recite in class .37 SCALE 6: SOCIALIZATION 79 Is slow in making friends .85 69 Is popular with classmates .71 75 Prefers to be alone and play alone .33 Table 5 indicates the intercorrelations among the six identified be- havior scales. As can be seen, relatively high positive correlations are evidenced among the Disrespect/Defiance, Externalized Aggression, and Attention—Seeking Scales. One explanation of the high degree of relation— ship among these scales would appear to be that all items within the scales share common variance with reading achievement and were selected on that basis. Further, numerous scale items loaded highly on more than the single factor for which they were identified (e.g. Is stubborn, argues with teacher, does things to attract attention, and threatens other child- ren). Conceptually, such items would appear appropriate to each factor on which they loaded. The Academic Self-Confidence and Socialization Scales appear relatively independent of the other four scales. The relatively high intercorrelations of the first four scales would suggest that their contribution to the pre— diction of future reading achievement will be diminished relative to the degree of multicollinearity among the scales. 76 TABLE 5 Intercorrelation of the Behavior Scales Scale 1 2 3 4 5 6 Attention/ Motivation — Disrespect/ Defiance .42 — Externalized Aggression .38 .64 — Attention— Seeking .44 .66 .66 - Academic Self-Confidence .34 .08 .09 .08 - Socialization .47 .15 .15 .17 .23 - Reading Achievement Reading achievement was assessed by means of the reading subtests of the Stanford Achievement Test (SAT) (Madden et al., 1975). Appropriate forms of the test were administered: Primary Level II for middle second to beginning third graders, Primary Level III for third to beginning fourth graders, Intermediate Level I for fourth to beginning fifth graders, Inter- mediate Level II for fifth to sixth graders, and Advanced for seventh through ninth graders. Standardized total reading scores were used for purposes of analysis in the present investigation. Students' total scores are composed of var— ious subscales depending upon the level of the test. Only Reading Compre- hension and Word Study Skills are used to make up the Total Reading score through the Intermediate II level. The Vocabulary score is not included 77 at these lower levels as this particular subtest is dictated by the teacher and not read by the student. At the Advanced level, Vocabulary is read by the students and Word Study Skills is not included; hence, the Total Reading score of the Advanced level includes Vocabulary and Reading Compre— hension subtests. Standardized scaled scores on the SAT were obtained through a compu— terized application of Thurstone's absolute scaling procedure, "resulting in a system of interbattery standard scores, which permits the translation of raw scores at each level into scaled scores with comparability across levels for a test area," (Madden et al., 1975, p. 30). The authors point out that within a single test area, such as Reading, the scaled scores obtained by the aforementioned methods are thus directly comparable from grade to grade, battery to battery, and form to form. The SAT was selected as the primary criterion variable for reading achievement in the present study as individual students' annual scores were readily attainable from the Lansing School District, as well as the fact that the SAT is a widely—used measurement tool in school districts throughout the country. Evaluation of reported technical data indicates reasonable measurement characteristics as well as ample sampling and norming procedures. Internal consistency or split—half reliability is reported at .86-.96 across all forms and levels for the reading subtests. Intelligence. For purposes of the present study, the Otis-Lennon Mental Ability Test (Otis and Lennon, 1969) was employed as a measure of general aptitude or intelligence. This test is a widely used, group administered general mental ability measure which evaluates students' facility in dealing with abstract concepts presented in verbal, figural, or symbolic form. Emphasis is placed upon the ability to discern 78 relationships of varying complexity within these three symbol systems. Elementary Level I was used in the second and third grades and Ele- mentary Level II was used in grades four through six. Both forms have reasonable proven measurement characteristics. Alternate-form reliability coefficients are reported as ranging from .85 to .93 and internal consis- tency reliability coefficients ranged from .88 to .95. Scores are re- ported as a deviation IQ. Deviation IQ's are normalized standard scores with a mean of 100 and a standard deviation of 16. A measure of general aptitude or intelligence was included as an independent variable in the present study, as IQ has been shown to contribute to a significant amount of the variance in elementary reading achievement scores as well as re- ported student behavior (Miller, 1972). Definitions In order to determine the relative accuracy by which the selected independent variables predict reading achievement, students were classi- fied Good, Average, or Poor Readers based upon their attained standard score on the Stanford Achievement Test. A Poor Reader was defined as a student who scored less than or equal to one standard deviation below the mean on the SAT according to test norms (T-Score 40). An Average Reader was defined as a student who scored within plus and minus one standard deviation above the mean on the SAT (T-Score 60). The afore- mentioned cut—off criteria were selected as they (at the lower achievement levels) closely approximate the school district's decision rule for con- sidering a student for compensatory or remedial reading support services. An important aspect of the study was to establish the relative pre— dictive accuracy of the independent variables with students at varying achievement levels. If long range educational decisions are to be made 79 based upon the prediction batteries, it would appear imperative that errors in prediction be negligible, especially for students in the extreme groups, Good and Poor Readers, for whom special educational programs might most likely be instigated. Data Collection Data for the initial sample of second through sixth grade students were collected during the course of a larger project of which this re— searcher was intimately involved (Blom, et al., 1980). This initial re- search project was funded through an all-university research grant at Michigan State University and supported through the auspices of Departments of Psychiatry and Special Education. All instruments were administered within the first quarter of the academic school year. School had been in session approximately two months, giving the teachers adequate time to become acquainted with their students, an important aspect in terms of the reliability and validity of their judgments regarding pupil behavior on the SBCL. The Otis-Lennon Mental Ability Test was administered over the course of a two week period by a group of undergraduate special education students, who were initially solicited, briefed and trained by this researcher in the research test administration procedures. Participating teachers were given two weeks to complete the School Behavior Checklist on each of their students. Protocols were hand-scored and recorded by this researcher and the undergraduate volunteers. Reading achievement data was obtained through the Evaluation Services Department of the participatory school district. The Stanford Achievement Test is administered annually in the district, and has in the past been used as a primary screening device for identification of students, second 80 through ninth grade, who may be in need of supplemental compensatory sup— port in reading and mathematics. Reading achievement data for each of the three subsequent years after the initial study was obtained on each of those students who continued to reside in the Lansing School District, and for whom parental release of information forms could be obtained. Statistical Hypotheses There were five major hypotheses formulated and tested in the pre- sent study of the relationship between student behavioral and cognitive characteristics and reading achievement in the elementary grades. The first hypothesis was concerned with identifying those student behavioral dimensions which were related to reading achievement over time. The second hypothesis evaluated the relative contribution of the behavioral dimensions to the explanation of future reading achievement while taking into account I!, sex, and initial level of achievement. Finally, the third through fifth hypotheses were concerned with evaluating the relative accuracy and utility of the identified behavior scales independent of and in conjunction with IQ, sex, and initial level of reading achievement (SAT1) in predicting future reading achievement. Hol: There is no linear relationship between the identified behavior scales and reading achievement as measured by the Stanford Reading Achievement Test: a. After one year (SATZ). b. After two years (SAT3). c. After three years (SAT4). H02: The identified behavior scales have no linear effect on SATz, SAT3, and SAT4 adjusting for the effects of IQ, sex of student, and initial level of reading achievement (SATl). 81 H03: There is no linear relationship between initial level of reading achievement, IQ, sex of student, and the identified behavior scales and reading achievement as measured by the Stanford Reading Achieve- ment Test: a. After one year (SATZ). b. After two years (SAT3). c. After three years (SAT4). H04: The linear rule derived from the identified behavior scales does not predict the achievement scores of good, average, and poor readers: a. After one year (SATZ). b. After two years (SAT3). c. After three years (SAT4). H05: The linear rule derived from initial level of reading achievement, IQ, sex of student, and the identified behavior scales does not pre— dict the achievement of good, average and poor readers: a. After one year (SATZ). b. After two years (SAT3). c. After three years (SAT4). Data Analysis Procedures All data in the study were analyzed at the Michigan State University Computer Center using various programs from the Statistical Package for the Social Sciences (SPSS), Nie, et al. (1975). Items from the SBCL that were and were not related to reading achieve— ment were identified by means of generating a series of Pearson Product- Moment zero-order correlation coefficients. An alpha level of .01 was set as the decision criterion. 82 Those items from the SBCL which were found to be significantly related to reading achievement (SATl) for the total sample or those items which were significantly related to reading achievement (SATl) for boys or girls only, were subsequently subjected to further analysis procedures. These items formed the pool of items for the development of behavioral scales which were to be used to predict future reading achievement (SATz to SAT4). Factor analytic statistical procedures were utilized to reduce the number of individual behavioral predictor variables to a manageable number of statistically and theoretically meaningful behavioral dimensions such as to improve the precision of prediction of future achievement. Initial principal factoring with iteration allowed, (1) all factors imposed to be orthogonal or independent, an important consideration in terms of future prediction equations, (2) factors to be arranged in the order of their importance, and (3) the first factor to be a general factor, that is, having significant loading on every variable. This factoring method automatically replaced the main diagonal elements of the corre- lation matrix of all the variables with communality estimates. Initial estimates of the communalities are given by the squared multiple corre- lation between a given variable and the rest of the variables and subse— quently tended to improve the theoretically meaningful patterning of the variables (Nie, et al., 1975). For purposes of the present study, ex— traction of six factors produced the most parsimonious, independent, and conceptually meaningful behavioral dimensions or scales, given the identified item pool. These six identified behavior dimensions were further subjected to item and scale analysis procedures by means of the Reliability subprogram of SPSS. This program provides a means for evaluating multiple—item 83 scales through the computation of widely recognized coefficients of re- liability. For purposes of the present analysis, Cronbach's Alpha Co- efficient was computed for each of the identified behavioral scales, and item-scale correlations were evaluated to determine the item's relative contribution to the variance of the scale. Internal consistency (coef— ficient alpha) reliability indicates the degree of measurement error due to measurement inconsistencies of the scale items. The program allows one to observe changes in the overall reliability of a scale with in- dividual items excluded as well as included in that scale. Partial correlation coefficients were computed in order to ascertain the relative contribution of the behavior scale to future reading scores while adjusting for the effects of IQ, initial reading achievement, and sex of student, which are all postulated to all be highly and signifi— cantly related to future reading achievement. The Partial Correlation subprogram of SPSS indicates the zero—order correlation, the degree of freedom, and a measure of the statistical significance of the relation- ship. A .01 alpha level was set as the decision criterion. Multiple regression analysis was the procedure used to determine the accuracy by which future reading achievement was predicted by the behavior scales, independent from and in conjunction with IQ, initial level of reading achievement, and sex of student. Multiple regression is a general statistical technique through which one can analyze the re- lationship between a dependent or criterion variable and a set of predictor or independent variables. As a descriptive tool, the primary purpose of multiple regression analysis is, (1) to find the best linear prediction equation and evaluate its predictive accuracy, (2) to control for other confounding factors in order to evaluate the contribution of a specific 84 variable of set of variables; and (3) to find structural relations and provide explanations for seemingly complex and multivariate relationships (Kim and Kohout, 1975, p. 321). Kerlinger and Pedhazur (1973) state that multiple regression is most often the best method of analysis of non— experimental data as it is "closely related to the basic purpose of science, the explanation of natural phenomena." A significant strength of multiple regression for purposes of the present study was its ability to handle categorical as well as continuous independent variables. Further, pre- dicted reading achievement scores could be generated and compared against actual scores to help determine not only the predictive accuracy, but predictive utility of the derived linear model, or prediction equation. Stepwise multiple regression procedures were employed to determine those variables that would yield an Optimal equation for the purposes of accurately predicting future reading achievement scores. Two separate equations were generated for predicting future achievement (SATZ, SAT3 SAT4). In the first equation only the behavior scales were entered, while in the second equation the behavior scales were entered with IQ, initial level of reading achievement (RACHI), and sex of student. In utilizing the stepwise procedure, the independent variables were entered into the prediction equation such that the variable that explained the greatest amount of variance in the dependent variable entered first; the variable that explained the greatest amount of variance in conjunction with the first entered second and so on. Regression coefficients or beta weights for each of the independent variables were generated and evaluated sta- tistically in terms of an F ratio. At each step of the analysis, F ratios were computed for variables not yet in the equation. For purposes of this study, all variables were allowed to enter into the equations in order to maximize the predictive power of the equation. 85 In order to further assess the ability of the independent variables of interest in this study to accurately predict future reading achieve- ment, predicted scores (Y'), and actual scores (Y) and residuals (i.e. errors in prediction) were calculated for each student. A decision rule was then evoked to determine whether the predicted score was commensurate with the actual score. For purposes of this study, a predicted score within plus or minus one standard error of measurement of the actual score, was considered a correct diagnostic classification or ”hit", while a predicted score outside of plus or minus one standard error of measurement was considered a diagnostic "miss". These diagnostic "hits and misses" were subsequently incorporated into three by three detection tables as per Feshbach, et al., (1974) and Satz and Friel (1974). Six separate detection tables were generated, one for each prediction equation at each measurement of future reading achievement (SATz, SAT3, SAT4). The following illustration defined the nature of the detection table (the actual tables included good, average, and poor readers): Actual Outcome Predicted Poor Reader Good Reader Total Valid Positive False Positive Rate Rate Poor Reader (+) False Negative alid Negative Rate Rate Good Reader (—) Total 86 The pratical utility of the prediction equations were finally com— puted directly by the principle of inverse probability, or Bayes Theorem (Satz and Fletcher, 1979). For each detection table, base rate informa- tion was given as the total number of actual poor, average, and good readers. Table 6 indicates the base rates for the present sample. TABLE 6 Relative Base Rates of Poor, Average, and Good Readers for SATZ, SAT3, and SAT4 SAT2 SAT3 SAT4 Poor Readers 14.0% 12.9% 11.8% Average Readers 71.0% 73.1% 75.3% Good Readers 15.0% 14.0% 12.9% Based of this base rate information, conditional probability values for each of the predictive test signs were determined. For example, the probability (P) that an individual student would become a poor reader (PR), given a positive or poor sign (+) was calculated by the following Bayesian expression: _ P(PR) - P(+ PR) P(PR +) “ P(PR) - P(+ PR) + P(GR) - P(+ GR)— The practical implications of relatively high errors in prediction (i.e. high false-positive and false-negative rates) are rather straight- forward for both schools and individual students. If a preventive or early intervention program were established for students with high—risk test 87 signs (i.e. predicted poor achievement) and high false positive rates in prediction were evidenced, wherein many of the students turned out to be average to superior readers without any intervention, the sum result would represent unnecessary expense and effort to the school as well as potentially serious implications for the student who is incorrectly "labeled" and unnecessarily targeted into "lower track", remedial programs. On the other hand, the more potentially detrimental error in terms of long term implica— tion for the individual student, the school, and community as a whole may be the false negative detection rates, wherein large numbers of students may be predicted to do well, when in fact the contrary is true. As such, a high proportion of high—risk students, who might strongly benefit from early intervention or remediation, would be systematically excluded from such services based on the prediction formula. Past research and practical consideration would suggest that prediction error rates (false positives and false negatives) should be less than 10%, and the generated prediction equations in the present study were evaluated in terms of this criteria. CHAPTER IV RESULTS To summarize, the study incorporated a descriptive, longitudinal design which purported (1) to identify teachers' perceptions of behavioral attributes of elementary age students which were highly and significantly related to reading achievement over time, and (2) to investigate the efficiency and utility of using teachers' ratings of student behavioral attributes to predict future reading performance. Teachers' ratings of student behavior were evaluated independently and concurrently with IQ, initial level of reading achievement and sex of student as to their ability to predict subsequent reading achievement after one, two, and three year intervals. Results of the study are reported in relation to the stated five statistical hypotheses. Hypothesis_I: There is no linear relationship between the identified behavior scales and reading achievement after one, two, and three year intervals. All of the behavior scales were found to be highly and significantly re— lated to reading achievement at all time intervals (P .01). The Attention/ Motivation Scale proved to be the most highly related to reading achieve- ment of all of the scales at the initial as well as subsequent one year intervals (—.38 to -.49). Attained correlational ranges for the remaining five scales included Disrespect/Defiance (-.33 to —.37), Attention-Seeking (-.34 to -.36), Academic Self—Confidence (-.24 to —.36), Externalized 88 Aggression (—.30 to -.36), and Socialization (-.23 to —.39). 89 Table 7 indicates the relative simple correlations between the behavior scales and reading achievement over four years. Correlations Between the Behavior Scales and Reading Achievement TABLE 7 Scale SATl SATZ SAT3 SAT4 Attention/ Motivation - 49* -.48* —.40* —.38* Disrespect/ , * Defiance - 33* -.34K -.35 —.37K Attention— Seeking —.36* —.34* -.35* -.35* Academic Self- Confidence —.36* -.34* -.24* -.29* Externalized Aggression -.31* — 32* —.30* -.36* Socialization —.33* —.39* -.29* -.23* P (.01 Hypothesis_II: The identified behavior scales have no linear effect on SAT2, SAT3, and SAT4 adjusting for the effects of IQ, sex of student, and initial level of reading achievement (SATl). As is evident from Table 8, statistically controlling for sex of student evidenced little impact on the predictive relationship between the behavior scales and reading achievement. All correlations remained statistically significant (P (.01) with the exceptions of Academic Self— Confidence and SAT3 and Socialization and SATA. 90 TABLE 8 Correlation Between Reading Achievement (SATZ to SAT ) and the Behavior Scales Controlling for Sex of Student Scale SATZ SAT3 SAT4 Attention/ Motivation -.46* -.36* -.36* Disrespect/ Defiance —.33* —.34* -.36* Externalized Aggression -.30* —.27* -.34* Attention— Seeking —.31* -.31* -.32 Academic Self-Confidence —.33* -.22* —.28* Socialization —.37* —.27* —.21* *p( .01 Because of measured student intelligence's (IQ) high correlations with reading achievement as well as its moderate to high observed relation- ship with the behavior scales, the attained initially strong correlations with reading achievement appeared appreciably diminished while controlling for the effects of IQ. Table 9 reflects those scales which maintained low, but significant (pwfim HM >mnmw 05m Mmmw Am>amv awn: nrm wormdwow momHmm chnHme oouumHmnHOSm 0m wmmmwdm >nrwmnnmsnwos\ Zonwnnm5nHOSI mmmWHDm I.wb .mbum .moob .owoo w.m~ .om I.wo >nmmmBHn mmHmIGOSmHmmsom |.wb .mmmH .wmom .omHo N.NN .Ho 1H.Nm mxanDmHHNmm >mmwmmmwos I.wN .muwo .wNuH .oomu .mo .wm I.bm UHmHmmvmnn\ Ummwmpom I.ww .mumw .wwHo .oowm .b0 .b0 I.wH 96 H>WHm Hm >mnmfl H20 mmmflm Am>H v awn? mam wmrmnrwmnmm5nwob\ KonHnnm5nH05I mmmenm I.wm .bbmu .Homw .0b0b b.mw .0» I.»@ moowmwwmmnwos I.No .bmmm .mpwm .ONHN m.bH .HN 1H.mo UHmfimmemon\ Ummwmnom I.wm .bumm .NNoN .ooob H.0V .wo I.bm >nmmmawo mmHMInoDmHmmsnm I.Nb .bmoo .Nwmm .OOVm .ma .wm I.NN mxnmflsmwwmmm >mmnmmmwos |.wo .bmmo .Nme .opo .Hm .uo I.NN 97 chnHva oowanmnHosm om Wmmdem >nrwmmnmw arnmm xmmwm Am>abv awn: mum wmrmnnmanon\ zonHmmfimmmwob I.wo .bbom .Hooa .omuH m.bm .oH I.Vo >nmmmBHn mmwmtnobmwmmsnm I.No .bme .Nwmb .owwm w.om .om IH.bH UHmHmmemon\ UmmHmSOm I.wu .momo .wao .oHom N.wo .Hw |.mm mOOHmHHNmnHoD I.Nw .mOBV .Nmbu .QOHm .No .om I.mw >nnmb~wosl mmmwwsm I.wm .momm .Nmmm .ooom .om .mH |.Hb 98 Intelligence (IQ), sex of student, and initial level of reading achievement (SATl) were subsequently entered into the prediction equation with the behavior scales. As is quite evident in the regression summary in Table 15, inclusion of SATl subsumed nearly the total explained vari— ance in SATZ, 78% of a total of 79%, with the socialization scale con— tributing an additional 1%, and the remaining behavior scales, IQ, and sex of student contributing no further to the explained total variance of SATZ. A separate prediction equation was computed to gain some insight into the relative contribution the explained variance in SATZ of the moderating variables SATl, IQ, and sex of student_a££e£ the behavioral scales had been considered. Results again indicated that a multiple R of .58 was attained with the six behavior scales (33% of the total ex— plained variance). Inclusion of IQ into the equation resulted in a multiple R of .73 (an additional 21% to the explained variance). Subse- quent inclusion of sex of student added negligible explanation, while final inclusion of initial reading achievement (SATl) added an additional 25% to the explained variance in SAT2 for a final multiple R of .89. Table 16 summarizes the relative contribution of the behavior scales, IQ, sex of student, and SATl tothe total explained variance of reading achievement after a two year interval (SAT3). SATl again entered the prediction equation first with a simple R of .80, explaining 64% of the variance. All behavior scales, IQ, and sex contributed only 2% to the total explained variance in SAT3 after inclusion of SATl. A final multi- ple R of .81 was attained, indicating that 66% of the total variance of SAT3 was explained by SATl, the behavior scales, sex of student, and IQ. 99 H>wfim Hm Kawnwwwm nonmemnwosm om Wmmmwpm >nUHmmfimH 05m mmmw Am>amv SHH: m>HH. Ho. mmx om mncmmSH mum firm wmwmHH .mm .mmww .umom .umom wmm.ob .oo .mH Hb.wo moomeHNmnHos I.wo .mmoo .uoO» .oHom b.wm .oe IH.om meHmmvmnn\ Ummwmbnm I.wb .moom .womm .OONH .mo .wm I.HN Ho .mm .moom .uowm .ooHo .bu .mN .wo mxnmfismwwmmm >mmnmmmwob I.wm .moHN .umbm .OOON .No .mo I.NH >nnmanODI mmmwMDm |.wb .mowb .uobm .000» .Hm .uo .Ho mmx .Hm .mon .wmbu .ooom .00 .um I.wm >nmmmBHo mmHmloodemmsom I.wb .mowa .uobo .ooow .oo .mH I.mH >nannHo:\ zenwwfim Ho ECHnHme oowanmnWOSm 0m WmeHdm >orwmmnmfi H20 Kmmnm Am>amv SHH: m>HH. Ho. mmx om mncamsn mum firm wmrmHH .mo .mon .obmw .obmo How.ow .oo .VH u.©m UHmHmmvmon\ Ummwmsom I.wm .mooo .mmHo .oomn N.Hm .Hb I.wm Ho .ow .momm .mmbH .oowH .uo .wm .oo >nmmmawn mmwmlooanQmSOm I.Nb .mpom .mme .oowo .uu .wm .wo mmx .No .mHHo .omom .ooNH .mb .bu .om >nnmsnwos\ zenHwfim Hy chnHon OOHHmHmnHoDm om wmmmwsm >oermnmfl Hrwmm «mmwm Am>abv SHH: m>aw. Ho. mmx om mncamsn mod mam wmrmHH .wm .ubmH .mmmw .mmwN www.mu .00 .mo H~.wb Ho .om .uuoN .mowH .owmo m.bo .oH .Ho mxnmusmHHNmm >mmfimmmwos I.wo .umNH .mHHm .onm b.Nb .oe I.ow moowmwwumnwo: I.Nw .umwm .apbw .oowo .mo .9» .mH UHmHmmomon\ Ummwmbom I.wu .umbo .mHmu .opo .mm .mm I.No >nnmanos\ Konwdmnwo: I.wm .ummo .oHuH .oon .ww .mw .Hb >ommmawo mmHmIOOSmHQmDom I.N© .ummw .ono .ooHu .wo .me I.wb >nannHODI mmmesm I.wm .Vmum .mpou .ooom .Hm .om .Ho mmx .Hb .umub .omoo .ooou .OV .uo .ww 103 Aggression was subsequently entered adding an additional 6% to the ex— plained variance. Further inclusion of Academic Self—Confidence, and Disrespect-Defiance added an additional 5% to the explained variance in SAT4. Subsequent inclusion of sex of student after the behavior scales added only negligible amounts to the explained variance. Further in- clusion of IQ increased the multiple R to .72 (52% of the total explained variance), and final inclusion of SATl increased the multiple R to .79 (62% of the total explained variance in SAT4). Figure 1 graphically summarizes and depicts the relative contribution of teachers' ratings of student behavior to the explained variance in future reading achievement at one to three year intervals, independent from, and in conjunction with initial level of reading achievement, IQ, and sex of student. Accuracy of prediction appeared to be greatly en— hanced by knowledge of a student's initial level of achievement, as well as general abilities. Both IQ and initial level of achievement (SATl), independently, proved to be better predictors of achievement than the aggregate of the behavior scales at all time intervals. Attention/Motivation proved to be consistently the best of the behavioral predictors. Figure 2 represents the relative contribution of the behavior scales, IQ, initial level of achievement, and sex of student to the explained variance in future reading achievement at one to three year intervals. Observation of the relationship between selected behavioral attributes from the identified behavior scales and reading achievement indicated interesting trends in the predictive relationship over time for the total group and boys and girls independently. 104 90 SATl, IQ, Behavior Scales, M‘ and Sex of Student (.89, 80 M”““““~~-a.81, .79) IQ, Behavior Scales, and .//"Sex of Student (.74, .68, 70 .72) 60 \\\\\\\\\\\\\ Behavior Scales (.58, .49, ,LJI/r”‘°51) 4O SATZ SAT3 SAT 4 Figure 1. Multiple Correlation of Reading Achievement with a) Behavior Scales, b) Behavior Scales, Sex of Student, and IQ, and c) Behavior Scales, Sex of Student, IQ and SAT 1 Evaluation of selected Attention/Motivation scale items indicated that for the total group, (1) those items which tended to reflect student self- motivation and persistence were highly and significantly related to reading achievement across time (SATl to SAT4) —— e.g. Tends to give up (-.49, -.44, -.45, —.48), Finds it hard to study (-.42, -.44, -.32, —.35), and Is interested in school work (.38, .38, .29, .30), while, (2) those items which reflected student attending/concentration behaviors tended to de— crease in strength and significance of relationship with reading achieve— ment over time -- e.g. Distractible; can't concentrate (—.37, -.40, —.27, —.21), and Able to concentrate on things (-.31, —.35, —.24, -.21). 105 00 we go mo mo #0 mo No m>HH A.mm. .mH. .ucv .UIF’VLINI M \IAI 1!! d- . . HdanHHmmbom A.mm. .aw. .omv Hoan wm5mnnmnnwop\Zonwnnmswwoslmmmwwdm A.wb. .wm. .wm ll. 011 1111111. L .mxanSmHHNmm >mmHmmmHod A.wH. .wN. .wov 4.7 >omamawo mmHmIOOUmHmmunm 7w». .Nb. .Nov - moomewNmnHos A.mm. .Nw. .va IIIIIII: mmx 0m mncmmpn A.Hm. .No. .Hbv mess. 94.5 m5» wwmcnm N. >dm0~¢nm nowwmwmnwopm 0m womawsm >oerHH 106 Observation of sex differences in the relationship suggested rela— tively different trends in the relationship of selected Attention/Motivation items without achievement over the four year period. For boys, only three of eight items which were significantly related to SATl were significantly related to SAT4, and again those items tended to reflect general persis— tence, motivation, and awareness in the classroom setting —— e.g. Tends to give up (-.49, -.49, —.47, —.46), Finds it hard to study (-.45, -.50, —.28, -.36), and Is alert in class (—.39, -.40, -.31, -.34). For the most part, other items of the Attention/Motivation scale, which were in— itially highly related to reading achievement for boys for the first two years, tended to decrease sharply in strength and significance over years three and four —- e.g. Finishes classroom assignments (.30, .28, .08, .20), Does homework (.39, .45, .29, .18), Distractible; can't concentrate (-.33, -.36, —.17, -.14), and Lacks ambition to do well in school (-.35, —.33, -.14, -.16). On the contrary for girls, the relationship between initial ratings of Attention/Motivation items and subsequent reading achievement tended to strengthen over time often from initially nonsignificant levels to moderately strong, significant levels -- e.g. "Drags" feet when requested to do something (—.25, —.32, -.30, —.30, —.47), Interested in school work (.21, .32, .27, .36), Lacks ambition to do well (-.18, -.29, —.25, -.35), Does homework (.21, .26, .29, .33), and Finds it hard to study (—.24, —.24, —.30, -.31). Again like their male counterparts, those behaviors found to be highly related to SAT4 appeared to reflect a generally persistent, motivated orientation to the classroom demands. Similarly, as was the case with the boys, early high correlations between attending/concentrating behaviors tended to decrease to nonsignificant levels by SAT4 -— e.g. Dis- tractible; can't concentrate (-.37, —.41, -.35, —.26), and Is alert in 107 class (.32, .37, .39, .14). Simple correlations between selected Disrespect/Defiance items were generally found to be in the moderately low, nonsignificant range for the total group at SATl. Only the item, Sulks when things go wrong, appeared moderately and consistently related to achievement across time for the total group (—.29, -.31, —.28, -.38). The relative strength of correlations appeared to moderately increase over time, with highest cor— relations between Disrespect/Defiance items evidenced at SAT4 —- e.g. Will put up an argument when told not to do something (-.25, -.21, —.24, —.26), Gives other children dirty looks (—.16, —.18, -.21, -.32), and Argues with teacher (—.22, -.22, —.24, -.30). For boys as a group, moderately high correlations across time were consistently evidenced on those Disrespect/Defiance items; Sulks when things go wrong (—.34, —.34, -.36, -.46), Has changeable moods (—.34, —.38, -.41, -.41), and Is infuriated by any form of discipline (-.33, -.29, —.32, -.34). The latter two items (Has changeable moods and Is Infuriated by any form of discipline) evidenced negligible relation to reading achieve- ment for girls at all time samplings, —.04 to .05 and -.13 to .01 res— pectively. For girls as a group, moderately high correlations were evi— denced at SATl and SAT4 with those Disrespect/Defiance items; When angry will refuse to speak to anyone (—.32, -.18, —.26, —.39), and When angry will do things like slamming the door or hanging the desk (—.34, -.26, —.24, —.38). Correlations appeared relatively consistent across time for the total group on those items of the Attention—Seeking scale. Moderately low relationship between the majority of items and reading achievement were evidenced -— e.g. Deliberately interrupts what is going on by asking 108 silly questions (-.35, —.32, -.31, -.38), and Tries to be the center of attention (-.32, -.31, —.28, —.33), appeared to be consistently related to reading achievement over the four year period. No items from the Attention-Seeking scale were significant for girls at SATl, however, the item, Does things just to attract attention was significantly related to achievement at SAT4 (—.36). Moderately low correlations were consistently evidenced between items of the Externalized Aggression scale and reading achievement across time for the total group. The items, Teases other children (—.29, —. 31, —. 27, —.33) and Threatens to hurt other children when angry (—.23, —.28, —.24, —.32) evidenced the strongest relationship with achievement of any items in the scale. Observation of sex differences again, however, evidenced different trends of association between selected items of the Extermalized Aggression scale and reading achievement over time. Despite moderately low correlations with SATl, the following items were found to be strongly related to SAT4 for boys as a group: Threatens to hurt other children when angry (—.47), Teases other children (-.40), Hits and pushes other children (—.40), and Does things to get others angry (-.37). On the con— trary, no items from the Externalized Aggression scale were found to be significantly related to reading achievement at any time sampling for girls. Generally low, yet consistent, patterns of relationship were evidenced across time between items of the Academic Self—Confidence scale and reading achievement for the total group. The item, Seems dull; slow to catch on, was found to be the only item from the scale to be highly and consistently related to achievement across time (-.42, -.44, —.36, -.34). For boys as a group, the aforementioned item again appeared to be con— sistently highly related to achievement (—.44, -.50, —.37, —.35, while 109 for girls, that item evidenced only moderate levels of association over time (-.31, —.24, —.27, -.30). The item, Volunteers to recite in class, was significantly related to SATZ and SAT4 for girls (.32 and .31), but found to be not related to reading achievement at any level for boys (.12, .12, —.05, .05). Hypothesis_IV: The linear rule derived from the identified behavior scales does not predict the achievement scores of Good, Average, and Poor readers after one, two and three year intervals. In order to ascertain the relative efficiency of the identified in— dependent variables (behavior scales) in predicting future reading achieve— ment, the derived equations were evaluated by observing the accuracy of prediction for each student at SATZ, SAT3, and SAT4. The residuals, or differences between each student's attained reading achievement score (Y) and estimated reading achievement score (Y’), were computed. Students were defined as Good, Average, or Poor Readers based on their attained reading achievement score (Y) as well as their predicted reading achieve— ment score (Y') at each year interval and these classifications served as the means by which relative errors in prediction were evaluated. Diagnostic ”hits" were evidenced when students' attained and predicted scores fell within one standard error of estimate from each other. Six separate detection tables were then generated, one for each prediction equation at SATZ, SAT3, and SAT4. Base rates were finally computed for Good, Average, and Poor Readers at each year interval, and conditional probabilities calculated for each of the predictive test signs to determine the relative utility of the predictive equations in forecasting future student reading achievement after one, two, and three years. The conditional probabilities are re— flected by the following Baysian expressions: 110 Probability of Being a Good Reader Given a Good Reader Test Sign _P (GR) P (_—I G__R_2 -.____,_,,-_..-- P(GR! )= P(GR) p(_ IGR) + P(PR or AR) P(-IPR 0r AR) Probability of Being a Poor Reader Given a Poor Reader Test Sign _P_(_P_R_)_ P_(_+|PR) P(PR) P(HPR) + P(GR or AR) P(+;GR or AR) P(PRI+)- Probability of Being an Average Reader Given an Average Reader Test Sign )=P(AR) P(o AR) + P(GR or PR) P(o CR or PR) Figure 3 presents the relative predictive accuracy of the behavior scales in terms of correct classifications of Good, Average, and Poor Readers after one year (SATZ) with respect to the established prediction equation. The behavior scales contributed 33% to the total explained vari— ance in SAT2(multiple R = .58). Inspection of Figure 3 revealed that the overall "hit rate" in terms of correct classification was 74% (69/93). Eighty percent of the predicted Poor Readers were actually Poor Readers, 74% of the predicted Average Readers were Average, and no Good Readers were predicted by the regression statement. At the extremes, no signifi— cant errors in prediction were evidenced, that is, no actual Good Readers were predicted to be Poor Readers, and no actual Poor Readers were pre- dicted to be Good Readers. An extremely high false positive error rate of 100% (0/14) was evidenced in the prediction of Good Readers, while a high false negative rate of 69% (9/13) was evidenced for Poor Readers. Error rate was minimized within the Average group (2%). The relative predictive accuracy of an equation involving the behavior scales after two years (SAT3) is presented in Figure 4. With respect to the established prediction equation, the behavior scales contributed 24% to the total explained variance in SAT3 (multiple R = .49). Figure 4 111 Actual Good Average Poor Reader Reader Reader Total Good Reader 0 0 0 0 .U Average 3 Reader 14 65 9 88 .3. 'o m E Poor Reader 0 1 4 5 Total 14 66 13 93 Figure 3. Predictive Classification of Students at SAT2 Based on the Behavior Scales indicates that the overall "hit rate" in terms of correct classification was 74.2% (69/93). Seventy—five percent (65/74) of the predicted Average Readers were Average, 100% (1/1) of the predicted Poor Readers were Poor, no Good Readers were predicted by the equation. Again at the extremes, no significant errors in prediction were evidenced, that is, no actual Good Readers were predicted to be Poor Readers, and no actual Poor Readers were predicted to be Good Readers. However, an extremely high false positive error rate of 100% (0/14) was again evidenced in the prediction of Good Readers, while an extremely high false negative rate of 92% (11/12) was evidenced for Poor Readers. Error rate was nil for the average group. Figure 5 presents the relative predictive accuracy of the behavior scales in terms of correct classification of Good, Average, and Poor Readers after three years (SAT4). With respect to the established pre— diction equation the behavior scales contributed 26% to the total explained variance in SAT4 (multiple R = .51). Inspection of Figure 5 indicated that the overall "hit rate" in terms of correct classification was 75% 112 ‘Actual Good Average Poor Reader Reader Reader Total Good Reader 0 0 0 0 '8 Average 3 Reader 13 68 11 92 :3 m H m Poor Reader 0 O 1 1 Total 13 68 12 93 Figure 4. Predictive Classification of Students at SAT3 Based on the Behavior Scales (70/93). Only 50% of those students predicted to be Poor Readers were in in fact Poor Readers, 76% of predicted Average Readers were Average, and no Good Readers were predicted by the regression statement. Again at the extremes, no actual Good Readers were predicted to be Poor Readers. An extreme false negative error rate (100%) was again evidenced in the pre— diction of Good Readers, while an extreme false positive error rate (.91) was observed in the prediction of Poor Readers utilizing the behavior scales only. Error rate was minimized within the Average group (.01). Hypothesis V: The linear rule derived from the initial level of reading achievement (SATl), IQ, sex of student, and the identified behavior scales does not predict the achievement scores of Good, Average, and Poor readers after one, two and three year intervals. The relative accuracy of a prediction equation involving the behavior scales, IQ, SATl, and sex of student in terms of correct classification of Good, Average, and Poor Readers after one year (SATZ) is presented in Figure 6. These independent variables contributed 79% to the explained 113 Actual Good Average Poor Reader Reader Reader Total Good Reader 0 O O 0 .0 Average ff, Reader 12 69 10 91 .3 -o o 5 Poor Reader 0 1 1 2 Total 12 70 11 93 Figure 5. Predictive Classification of Students at SAT4 Based on Behavior Scales variance in SAT3 (multiple R = .89). Evaluation of Figure 6 indicated that 93.5% (87/93) of the students were correctly classified. Relative "hit rates" were high for all groups. One hundred percent of those students predicted to be Good Readers were in fact Good Readers; 92.9% of Average Readers were predicted to be Average; and 91.7% of Poor Readers were pre— dicted to be Poor. Inclusion of IQ, sex, and SAT1 into the prediction with the behavior scales greatly decreased the false positive error rate to 21% (3/14) for Good Readers, and greatly decreased the false negative error rates for Poor Readers to 15% (2/13). Figure 7 indicates the relative accuracy of a prediction equation in- volving the behavior scales, IQ, SATl, and sex of student in terms of correct classification of Good, Average, and Poor Readers after two years (SAT3). Sixty-six percent of the total variance in SAT3 was explained by the aforementioned independent variables (multiple R = .81). Inspection of Figure 7 revealed that the overall "hit rate" in terms of correct classification was 87% (81/93). Ninety percent of those students predicted 114 Actual Good Average Poor Reader Reader Reader Total Good Reader 11 0 0 11 Average '8 Reader 3 65 2 70 ti 'H 'o a Poor 9* Reader 0 1 11 12 Total 14 66 13 93 Figure 6. Predictive Classification of Students at SAT Based on the Behavior Scales, IQ, SATI, and Sex of Student to be Good Readers were in fact Good Readers; 88% of Average Readers were predicted to be Average; and 78% of Poor Readers were predicted to be Poor Readers. Despite the addition of SATl, IQ, and sex of student to the equation in addition to the behavior sclaes, a relatively high false posi— tive rate (.31) persisted in the prediction of Good Readers, as well as a high false negative error rate (.42) in the prediction of Poor Readers. The relative accuracy of a prediction equation involving the behavior scales, IQ, SATl, and sex of student in terms of correct classification of Good, Average and Poor Readers after three years (SAT4) is presented in Figure 8. Sixty—two percent of the total variance in SAT4 was explained by the aforementioned independent variables (multiple R = .79). Inspection of Figure 8 indicated an overall "hit rate" of 88% (82/93) in terms of correct classification. Eighty—nine percent of those students predicted to be Good Readers were in fact Good Readers; 88% of Average Readers were predicted to be Average; and 86% of Poor Readers were predicted to be Predicted Poor Readers. in the prediction of Good Readers, and a similarly high false positive 115 Total 10 74 93 Actual Good Average Poor Reader Reader Reader Good Reader 9 1 0 Average Reader 4 65 5 Poor Reader 0 2 7 Total 13 68 12 Figure 7. Predictive Classification of Students at SAT3 Based on the Behavior Scales, IQ, SATl, and Sex of Student A relatively high false negative error rate (.33) persisted error rate (.45) was evidenced in the prediction of Poor Readers. Predicted Total 77 93 Actual Good Average Poor Reader Reader Reader Good Reader 8 1 0 Average Reader 4 68 5 Poor Reader 1 6 Total 12 70 11 Figure 8. Predictive Classification of Students at SAT4 Based on the Behavior Scales, IQ, SATI, and Sex of Student 116 Table 18 summarizes the relative diagnostic "hit rates" by reading group based on 1) the behavior scales only, and 2) the behavior scales, IQ, SAT, and sex of student across time. TABLE 18 Diagnostic "Hit Rates" of Good, Average, and Poor Readers Predictor(s) SATZ SAT3 SAT4 * T G A P T G A P T G A P Behavior Scales .74 .00 .73 .80 .74 .00 .75 1.00 .75 .00 .76 .50 Behavior Scales, IQ, SATl, .94 1.00 .93 .92 .87 .90 .88 .78 .88 .89 .88 .86 Sex *T-Total, G-Good, A-Average, P—Poor As is quite evident, inclusion of SATl, IQ, and sex of student in the prediction, greatly enhanced the classification "hit rate" at all levels of achievement. The behavior scales independent of the aforementioned vari— ables provided only marginal predictive accuracy and utility with future reading achievement. A consistent overall hit rate of approximately 75% was observed with most extreme errors in prediction evidenced at the ex— tremes i.e. Good Readers and Poor Readers. The behavioral scales alone tended to highly underpredict achievement for Good Readers (false positive error rates of 100% at SATZ, SAT3 and SAT4), however, produced no misclassification of students who later became Poor Readers (a more seriuos false positive error). Similarly, the behavior 117 scales alone tended to highly overpredict achievement for Poor Readers (false negative error rates of 69%, 92%, and 91% for SATZ, SAT3, and SAT4 respectively), however, produced no misclassification of students who later became good readers (an even more serious false negative error). Overall observed high "hit rates" in prediction with the inclusion of SATl, IQ, and sex of student with the behavior scales, again clearly need to be observed relative to the degree of prediction error at the extremes with Good and Poor Readers. False positive errors in predicting Good Readers, and false negative errors in predicting Poor Readers again appeared prevalent, even when using prediction equations with multiple correlations as high as .89. For SATZ, the equation underpredicted achievement for 21% (3/14) of Good Readers and overpredicted achievement for 15% (2/13) of Poor Readers. For SAT3, the equation underpredicted achievement for 31% (4/13) of Good Readera, and overpredicted achievement for 42% (5/12) of Poor Readers. Calculated conditional probabilities of decision risk again suggested relative caution in interpreting the practical utility of the generated prediction equations, especially in terms of decision-making for individual students whose predicted scores fell at the extreme ranges, i.e. Good Reader or Poor Reader. When only the behavior scales were incorporated in the prediction of future reading achievement the following conditional probabilities were attained for each of the test prediction signs: Poor Reader (+), Average Reader (0), Good Reader (-). For SATZ, SAT3, and SAT4, the probabilities that a student would be a Poor Reader given Poor Reader predicted scores P(PR +) were .29, .07, and .08 respectively. Again, for SATZ, SAT3, and SAT4, the probabilities that a student would be a Good Reader given Good Reader predicted scores P(GR -) were .00 at all three intervals. Finally, 118 for SATZ, SAT3, and SAT4, the probabilities that a student would be an Average Reader given Average Reader predicted scores P)AR 0) were .99, 1.00, and .99 respectively. These results again suggest the relative instability of the prediction equation with the extreme groups. When SATI, IQ, and sex of student were incorporated with the behavior scales in the prediction of future reading achievement the following con- ditional probabilities were obtained for each of the test prediction signs. For SATZ, SAT3, and SAT4, the probabilities that a student would be a Poor Reader given Poor Reader predicted scores P(PR +) were .84, .60, and .53 respectively. Again for SATZ, SAT3, and SAT4, the probabilities that a student would be a Good Reader given Good Reader predicted scores P(GR -) were .80, .27, and .69. Finally, for SATZ, SAT3, and SAT4, the probabili— ties that a student would be an Average Reader given Average Reader pre- dicted scores P(AR 0) were all .99. Results of this final aspect of the study again indicated that: 1) the generated prediction equations which utilized only the behavioral scales in predicted SATZ, SAT3, and SAT4 tended to greatly underpre- dict reading achievement for Good Readers, to greatly overpredict reading achievement for Poor Readers, and to accurately predict reading achievement for Average Readers. 2) despite significantly high multiple correlations, as well as high "hit rates", the generated prediction equations which incorporated the behavior scales, SATl, IQ, and sex of student were further found to be reasonably lacking in practical utility in terms of accurately identi- fying Good and Poor Readers over time. CHAPTER V DISCUSSION Student behavioral attributes which appeared most highly and signifi- cantly related to reading achievement in the present study appeared to generally reflect a student's degree of adaptation to the task orientation of the classroom, ability to attend to classroom tasks, degree of self— motivation and persistence, and degree of general social maturity. The general portrait of the academically competent student appeared similar to previously reported findings (Cobb, 1972; Lambert and Nicoll, 1977; Lahaderne, 1968; McKinney et al., 1975; Samuels and Turnure, 1974). Re- sults suggested that the student who is attentive, interested, confident, persistent, and who regularly completes assignments in the classroom setting is more likely to be a competent reader than the student who is distractible, dependent on the teacher for direction, socially immature, poorly self—motivated, and who does not regularly complete assigned tasks. At a global level, the identified achievement—related behavior charac— teristics in the present study appears conceptually consistent with the achievement—related behavioral dimensions defined by Lambert and Nicoll (1977), namely Classroom Adaptation (e.g. following directions, volunteering, working independently), Intrapersonal Adjustment (e.g. withdrawn, attention to task), Interpersonal Adjustment (e.g. fighting, arguing). The be— havioral attributes, which appeared to generally reflect the intrapersonal adjustment dimension as defined by Lambert and Nicoll, tended to be most highly related to reading achievement in the present study, (e.g. Tends to give up if has something hard to finish, Is alert in class, Is sure of self, Is interested in schoolwork, and Is distractible; can't concentrate). 119 120 Behavioral attributes which reflected adaptation to classroom task ex- pectations were similarly highly related to reading achievement (e.g. Does homework, and Finishes assignments). Volunteers to recite, unlike the findings of Cobb (1972) and Lambert and Nicoll (1977), was only mar— ginally associated with reading achievement in the present study. As was the case in the studies of Lambert and Nicoll (1977) and Lambert and Urbanski (1980), behaviors reflecting an interpersonal mala— justment dimension tended to evidence generally weaker relationships to reading achievement, than those reflecting intrapersonal adjustment and adaptation to classroom demands. Behaviors reflecting a general socially immature, attention—seeking orientation (e.g. Likes an audience all the time, Deliberately interrupts what is going on by asking silly questions, Does things just to attract attention, and Does things which are normal for children much younger) proved to be more highly related to reading achievement, than those be- haviors which indicated a more aggressive, disruptive, acting out posture (e.g. Acts up when adults are not watching, Hits and pushes other children, Threatens to hurt other children when angry, Seems unconcerned when mis- behaving, and Disturbs other children with boisterous behavior). For the total group those attributes from the Attention/Motivation Scale which reflect student self-motivation and persistence were found to be highly and consistently related to achievement across time (e.g. Tends to give up, Finds it hard to study, and is interested in schoolwork), wherein those items which reflected student attending/concentration behavior tended to decrease in relative strength and significance over time. The latter finding appears most relevant in that a great amount of past re— search has reported "attending" to be generally the most significantly related behavioral attribute to achievement for elementary age students 121 (Cobb, 1970, 1972; Feshbach, et al., 1977; Forness, et al., 1977; Meyers, et al., 1968; Soli and Devine, 1976). Results of the present study sug- gested a general trend indicating either improved attending abilities over time for elementary age students with consistent reading achievement scores, or relatively consistent gain in reading achievement independent of at— tending abilities. The relationship at this time is unclear as behavioral data was unable to be attained at each successive time sampling. Explana— tion of the relationship might be enhanced should relative stability of the behavior ratings be observed. As was the case in the studies of Cobb (1970), Feshbach, et al. (1977), Samuels and Turnure (1974), and Swift and Spivack (1967), general classroom behavioral adjustment appeared to be a generally more important factor in terms of the reading success of elementary age boys relative to their female peers. What appeared to be the most striking aspect of the observed behavior-achievement relationships for boys versus girls was the moderately strong negative relationships for boys but virtually negligible relation— ship for girls between reading achievement and those behavioral attributes which reflect a student's degree of lack of behavioral self-control, emo- tional instability and social immaturity —- behaviors most likely to be observed as disruptive in the classroom setting, (e.g. Likes an audience all the time, Deliberately interrupts what is going on by asking silly questions, Is infuriated by any form of discipline, Interrupts whoever is speaking, and Approaches a task with an air of defeatism). Different trends in the relationship of the Attention/Motivation items and reading achievement over time were as expected observed between girls and boys. Attributes which were highly related to reading achievement for boys for the first two years, tended to decrease sharply in strength and significance over years three and four. On the contrary for girls, the 122 relationship between initial ratings of Attention/Motivation items tended to strengthen over time often from initially non-significant levels to moderately strong, significant levels. Such findings suggest that there may be a developmental adjustment phenomenon oCcuring wherein classroom adaptation skills and attributes become refined over time for boys as a group, and cognitive ability and past achievement may be more important contributing factors to future reading success. The fact that intrapersonal/motivation variables appeared to be the most highly related to measured reading achievement for boys as a group over time, suggested that potential remedial interventions for elementary boys who are experiencing reading difficulties should be closely scru— tinized in terms of their novel, motivational, and "attention-getting" characteristics. External reinforcers such as tangible rewards and high levels of verbal feedback and teacher praise may further be necessary components of developmental or remedial reading programs for elementary age boys. Moderately low correlations were consistently evidenced between items of the Externalized Aggression scale and reading achievement across time for the total group. Sex differences, however, were again observed in the behavior-achievement relationship with items from the scale. Despite moderately low correlation at initial level of achievement, items such as Threatens to hurt other children when angry, Teases other children, and Hits and pushes other children were found to be strongly associated with reading achievement for boys after four years. On the contrary, no items from the Externalized Aggression scale were found to be significantly related to reading achievement at any time sampling for girls. Studies, such as those of Lambert and Nicoll (1977), Lambert and Urbanski (1980), and Cobb (1972), have suggested the relative lack of relationship between 123 such externalizing, aggressive behaviors and early reading achievement. Results of the present study suggest that at least for boys, such ini- tially—observed behaviors do indeed relate to achievement over time, and as such should be considered for intervention concurrently with develop- mental or remedial reading instruction. Results of past longitudinal studies which have incorporated be- havioral attributes do make a significant, unique contribution to the variance in later reading ability, independent of various cognitive measures, including intelligence (Feshbach, et al., 1977; Meyers, et al., 1968; Stevenson, et al., 1976b). Current attained multiple correlations be— tween reading achievement and the aggregate of the behavior scales, ap- peared highly commensurate with the attained multiple R's attained be— tween the Student Rating Scale (SRS) and reading achievement as measured by the Cooperative Primary Test over a three year period -— multiple R: .50 (Feshbach, et al., 1977). Reflective of the findings of Stevenson, et al., (1976b), the results of the present study found that despite the observed strong predictive relationship between the behavior scales and future reading achievement, teacher's ratings were not observed to be as predictive of later achieve— ment as were scores on a measure of general intelligence, as well as scores on an initial reading achievement measure. These results would suggest that very little predictive information is gained by knowledge of behavioral attributes of elementary-age students beyond knowledge of initial levels of academic functioning and general intellectual ability as measured by group-administered standardized measures. Economics would dictate, especially at a gross screening level, that attaining behavioral data would probably not be worth the additional cost and effort involved. How— ever, for purposes of prediction for individual students such behavioral 124 information may be useful for more accurately assessing future achievement. The relative efficiency and utility of the derived prediction equations were finally evaluated in terms of their ability to accurately predict reading status. Results indicated that the generated prediction equations which utilized only the behavior scales to predict subsequent reading achievement tended to greatly underpredict achievement group for Good Readers, to greatly overpredict reading achievement for Poor Readers, and to accurately predict achievement for Average Readers. Further, despite significantly high multiple correlation, as well as high "hit rates", the generated prediction equations which incorporated initial level of achieve— ment, IQ, and sex of student were found to be generally lacking in terms of accurately predicting Good and Poor Readers over time. Utilization of the derived formulae over the long run tended to screen out a significant number of students who would have been in need of possible reading support services and tended to similarly screen out those students with above average reading capabilities who might have benefited from possible enrich- ment activities. The relative "cost” in terms of the number and types of errors evidenced is somewhat difficult to assess, and is highly dependent upon how one views the cost to students in terms of access to appropriate services, as well as fiscal cost to the school district. If a preventative or early inter— vention program were to be established for students with "high—risk" test signs (i.e. predicted poor achievement) and high false positive rates in prediction were evidenced, the sum result would represent unnecessary ex- pense and effort to the school, as well as potentially serious implications for the student who may be incorrectly ”labled" and unnecessarily targeted into "lower track," remedial programs. On the other hand, it would seem that the more potentially detrimental error in terms of long term 125 implications for the individual student, the school, and community as a whole may be the false negative rate, wherein large numbers of students may be predicted to do well, when in fact the contrary is true. Evaluation of the results of the present study would suggest that the derived prediction equation which incorporated initial level of achievement, IQ, sex of student, and the behavior scales would be techni- cally adequate for use as a screening tool for potential enrichment or remedial intervention. Despite the potential of the equation as a possible screening tool, the number and type of errors evidenced would suggest that it would not be acceptable as a single primary evaluation tool. As such further individual assessment should be initiated prior to any decision— making regarding an individual student. CHAPTER VI SUMMARY AND RECOMMENDATIONS To reiterate, the purpose of the study was essentially two—fold: 1) to identify teachers' perceptions of behavior attributes of elementary age students which were highly and significantly related to reading achievement over time and 2) to investigate the efficiency and utility of using teachers' ratings of student behavioral attributes to predict future reading performance. A descriptive, longitudinal design was in— stigated whereby teachers' evaluations of student behaviors were evaluated independently as well as in conjunction with IQ, initial level of reading achievement, and sex of student, as to their ability to predict subsequent reading achievement measured after one, two and three year intervals. Initially, items from the School Behavior Checklist (Miller, 1972) which were found to be significantly related to test scores on the Stanford Reading Achievement Test were identified. Scales defining student behavior attributes were subsequently later developed incorporating these achieve- ment-related items from the SBCL. Those behavior scales were defined as reflecting the dimensions Attention/Motivation, Disrespect/Defiance, Ex- ternalized Aggression, Academic Self-Confidence, and SOCialization. These identified behavior scales were next incorporated into prediction equations so as to evaluate the accuracy of using teacher ratings of student be— havioral characteristics to predict future reading achievement measured after one, two, and three year intervals, independently and in conjunction with previously identified, relevant moderating variables including initial level of reading achievement, general intelligence, and sex of student. Finally, the practical, functional utility of the derived equations in 126 127 predicting future reading achievement was evaluated by observing the relative error in prediction for each student across time. Predictive "hits" and "misses" were calculated for Good, Average, and Poor Readers, and based upon the observed base rates of reading ability or disability, conditional probability values were calculated to again help to determine the accuracy of prediction as well as the practical utility of using the derived formulae as a basis for intervention. Student behavioral attributes which appeared most highly and signifi— cantly related to reading achievement in the present study appeared to reflect a student's degree of adaptation to the task orientation of the classroom, ability to attend to classroom tasks, degree of self—motivation and persistence, and degree of general social maturity. General classroom behavioral adjustment was found to be a generally more important factor in terms of the reading success of elementary age boys relative to their female peers. What appeared to be the most striking aspect of the ob— served behavior-achievement relationships for boys versus girls was the moderately strong negative relationship for boys but virtually negligible relationship for girls between reading achievement and those behavioral attributes which reflect a student's degree of lack of behavioral self— control, emotional instability and social immaturity —— behaviors most likely to be observed as disruptive in the classroom setting. All of the derived behavior scales were found to be significantly related to reading achievement after one to three year intervals. The Attention/Motivation scale, proved to be the most predictive of reading achievement at the initial as well as subsequent one year intervals. IQ was found to be significantly related to all of the behavior scales with the exception of the Externalized Aggression scale, and was as might be expected highly and signifiCantly related to reading achievement at all 128 four year intervals. Sex of student was found to be significantly related to the Attention/Motivation and Attention-Seeking scales. Generalizable trends were evidenced between items from the generated behavior scales and reading achievement over the four year period. For the total group those items from the Attention/Motivation scale which reflect student self—motivation and persistence were found to be highly and consistently related to achievement across time, wherein those items which reflected student attending/concentration behavior tended to decrease in relative strength and significance over time. Different trends in the relationship of the Attention/Motivation items and reading achievement over time were as expected observed between boys and girls. Items which were highly related to reading achievement for boys for the first two years, tended to decrease sharply in strength and significance over years three and four. 0n the contrary for girls, the relationship between initial ratings of Attention/Motivation items tended to strengthen over time often from initially non-significant levels to moderately strong, significant levels. Like their male peers, those be- haviors found to be highly related to reading achievement after four years appeared to reflect a generally persistent, motivated orientation to classroom demands. Most Disrespect/Defiance items reflected moderately low, non—signifi— cant relationships with initial level of reading achievement for the total group with only moderate increases in strength of relationship over time. For boys, but not for girls, moderately high correlations with reading achievement were consistently evidenced on those Disrespect/Defiance items which reflect emotional lability and difficulty with authority. Similarly, items from the Attention—Seeking scale moderately low relationships were observed over time for the total group. For boys, 129 however, items such as Does things just to attract attention, Deliberately interrups what is going on by asking silly questions, and Tries to be the center of attention appeared consistently related to achievement over the four year period. No items from the Attention-Seeking scale were signifi— cantly related to initial achievement for girls. Moderately low correlations were consistently evidenced between items of the Externalized Aggression scale and reading achievement across time for the total group. Sex differences, however, were again observed in the behavior-achievement relationship with items from the scale. Despite moderately low correlation at initial level of achievement, Threatens to hurt other children when angry, Teases other children, Hits and pushes other children, and Does things to get others angry, were found to be strongly associated with reading achievement for boys after four years. On the contrary, no items from the Externalized Aggression scale were found to be significantly related to reading achievement at any time sampling for girls. Generally low, yet consistent patterns of relationship were evidenced across time between items of the Academic Self—Confidence scale and reading achievement for the total group. The item, Seems dull; Slow to catch on was found to be the only item from the scale to be highly and consistently related to achievement across time. That item appeared consistently highly related to achievement for boys, and was found to be only moderately associated with achievement over time for girls. The items, Volunteers to recite in class, was found to be significantly related to achievement after two and four years for girls, but was found to be unrelated to achievement at any level for boys. The final task of the study was to investigate the relative con- tribution of teachers' ratings of student behavior to explained variance 130 in future reading achievement and to assess the accuracy and practical ability of predictions based on the derived behavior scales, independently and in conjunction with moderating variables with known strong predictive relationships with achievement. Results of the present study indicated that despite the observed strong linear relationship between the behavior scales and future reading achievement, teachers' ratings were not found to be as predictive of later achievement as were scores on a measure of general intelligence, as well as scores on the initial reading achievement measure. With the effects of initial level of achievement controlled for, none of the behavior scales were found to be significantly related to reading achievement at any point across the follow-up three year period. When controlling for IQ, the relationship between the behavior scales and future achievement was further diminished to moderately low significant levels to nonsignificant levels. Attention/Motivation after one year, Externalized Aggression after one and three years were the only scales to maintain a significant relationship with achievement while controlling for the effects of IQ. When initial level of achievement and IQ were entered in the pre— diction formulae with the behavior scales, initial level of achievement subsumed nearly the total explained variance of achievement (78% of a total of 79%) after one year, 64% of a total 66% of the variance in achievement after two years, and 56% of a total 62% of the variance in achievement after three years. Again, inclusion of the cognitive and prior achievement variables greatly diminished the relative contribution of the behavior scales to the explained variance in reading achievement across time. The derived regression formulae were finally evaluated in terms of their relative efficiency and utility in terms of appropriate classification 131 of students as Good, Average, or Poor Readers over the three year follow- up period. Results indicated that the generated prediction equations which utilized only the behavior scales in predicting subsequent reading achievement tended to greatly underpredict achievement group for Good Readers, to greatly overpredict reading achievement for Poor Readers, and to accurately predict achievement for Average Readers. Further, despite significantly high multiple correlations, as well as high "hit rates", the generated prediction equations which incorporated initial level of achievement, IQ, and sex of student were found to be generally lacking in terms of accurately predicting Good and Poor Readers over time. Utiliza— tion of the derived formulae over the long run tended to screen out a significant number of students who would have been in need of possible reading support services and tended to similarly screen out those students with above average reading capabilities who might have benefited from enrichment activities. Limitations and Delimitations of the Study Limitations 1. Numerous threats to the internal validity of the study in terms of interpretation of the results were evidenced due to the longitudinal nature of the study (Campbell and Stanley, 1966). The accuracy by which future reading achievement status of individual students is predicted will be affected by variables which are unable to be con- trolled given the current research design. These include history and maturation -- e.g. it is not known which students may have re— ceived additional reading support services or what other intervening cognitive, emotional, or environmental variables might be influencing current reading performance; experimental mortality —- the reasons 132 for subject attrition from the original sample are not known, and it is not known why achievement data may be missing for individual stu- dents in the follow—up sample (e.g. untestable, absent, systematically excluded); and, effects gf_testing —- e.g. the reliability of initial teacher ratings, identified behavioral scale scores, SAT reading achievement scores, and Otis—Lennon IQ scores, as well as individual student reactivity to the testing situations, examiner effects, ”test- wiseness" of the students, and changes in the form and level of the SAT. Unfortunately, due to the relatively small sample size, it was not feasible to examine in more depth the predictive behavior-achievement relationship by age and/or grade level, viewing potentially relevant developmental phenomenon in the relationship. Similarly, due to the relatively small sample size, a more indepth analysis of sex of stu- dent as an important intervening variable was not possible. As such, possibly unique predictors for boys and girls, and/or younger or older elementary students were unable to be clarified. Again from a developmental perspective, it would have been meaningful to ascertain changes in teacher perceptions of student behavior over the course of the follow-up period to further clarify the behavior— achievement relationship for individual students over that three year period. Because relatively high levels of multicollinearity were observed among the behavior scales, the unique contribution of the individual behavior scales was not able to be observed. Further, because of the observed high correlations among initial level of reading achievement, IQ, and the behavior scales, future prediction of reading achievement 133 might have been enhanced had the independent variables proven to be more orthogonal. The assigned cut-off criteria for Good and Poor Readers (plus or minus one standard deviation) appears relatively stringent as compared to past predictive studies. Accuracy of correct prediction or classifi— cation of group membership is strongly related to the arbitrary nature of the cut—off scores selected. Predictive accuracy may have been en— hanced with more liberally established cut—off criteria. Delimitations 1. Due to the nonrandom selection of subjects, generalization of the results of this study are limited to populations of students who compare simi— larly on the dimensions of age, sex, IQ, SES, race, school character- istics, and geographical region. A major limitation of multiple regression procedures is the relative instability of the assigned regression weights of the predictor vari- ables when applied to subjects in a new sample. Therefore, prediction of reading achievement of students not included in the sample upon which the prediction equations were developed, would not be meaningful. Cross—validation procedures, that is, means by which the regression weights could be validated, were unable to be implemented because of the relatively small number of subjects in the present study. Assumptions In light of the aforementioned limitations and delimitations, the follow— ing assumptions are made: 1. The measurement tools employed are reliable indicators of student status (reading achievement, intelligence, and behavioral attributes). of 134 Students' performance on the reading and intelligence measures ap- proximate their actual classroom performance and abilities. Measurement errors as a result of history, maturation, experimental mortality, and testing are randomly distributed. For purposes of prediction, the criterion values (reading achievement scores) are randomly distributed. Demographic characteristics of the follow—up sample are representative of the school and district at large from which it was taken. Recommendations In light of the observed results and limitations of the present study prediction of reading achievement over time, the following suggestions are made for future research: 1. Predictive accuracy from a longitudinal perspective might be signifi- cantly enhanced by using different behavioral predictors for boys and girls than just including sex as an independent variable in the prediction equation. In order to further clarify the observed trends in the behavior/achieve- ment relationship over time in the present study, future longitudinal studies should attempt to obtain concurrent ratings of student behavior with reading achievement at each time sampling. As initial level of reading achievement, was highly and significantly related to future achievement levels, cummulative achievement scores might enhance the predictive utility of the equations, that is raising multiple correlation above .90 and minimizing errors in prediction for individual students. 135 Cross validation and/or replication studies should be instigated to evaluate the prediction equations' ability to predict achievement of students other than those in the sample on which the formulae were developed. REFERENCES Achenbach, T.M., and Edelbrock, C.S. Behavioral problems and competencies reported by parents of normal and disturbed children aged four through sixteen. Monographs of the Society for Research 13 Child Development, 1981,_£6(1), Serial No. 188. Adelman, H. and Feshbach, S. Predicting reading failure: beyond the readiness model. Exceptional Children, 1971,_3Z, 349-54. Anderson, 1., Hughes, B., and Dixon, W. The rate of reading development and its relation to learning to read, sex, intelligence, and reading achievement in the sixth grade. Journal_gf Educational Research, 1956, 49, 447-53. Anderson, 1., Hughes, B., and Dixon, W. The rate of reading development and its relation to age of learning to read, sex, and intelligence. Journal 2f Educational Research, 1957, 59, 481—94. Anthony, J. Behavior disorders. In P. Mussen (Ed.), Charmichael's Manual gf_Child Psychology. New York: Wiley, 1970. Ashcraft, C. Later school achievement of treated and untreated emotionally handicapped children. Proceedings 3f the 78th Annual Convention 2f the American Psychological Association, 1970,_5, 651—52. Ayllon, T. and Roberts, M. Eliminating discipline problems by strengthening academic performance. Journal gf_Applied Behavioral Analysis, 1974, 7’ 71-760 Ayres, L. Laggards_ig_0ur Schools. New York: Russell Sage Foundation, 1909. Bannatyne, A. Languagg, Reading, and Learning Disabilities. Springfield, Ill: Charles C. Thomas, Publisher, 1971. Bateman, B. Learning disorders. Review_gf Educational Research. 1966, 26, 93'119. Bentzen, F. Sex ratios in learning and behavior disorders. National Ele- mentary Principal, 1966, 36(2), 13-17. Berk, L., Rose, M. and Stewart, D. Attitudes of English and American children toward their school experience. Journal 2: EdUCational Psychology, 1970,_61, 33-40. Blom, G.E. Sex differences in reading disabilities. In E. Calkins (Ed.), Reading Forum, NINDS Monograph No. 11. Washington, D.C.: U.S. Printing Office, 1970, 31-46. 136 137 Blom, G.E., Frey, W.D., Prawat, R.S., and Jarvis, R.L. Sex Differences_ig Receptive Language and Behavioral Expression_§§ They Relate_£g Early Reading Achievement. Unpublished manuscript, East Lansing, MI: Michigan State University, 1980. Bower, E.M. and Holmes, J.A. Emotional factors and academic achievement. Review 2f Educational Research, 1959, 29, 529-44. Camp, B.W. and Zimet, S.G. The relationship of teacher rating scales to behavior observations and reading achievement of first grade children. Journal 2f_Special Education, Winter 1974,_§, 353—59. Carroll, M.W. Sex differences in reading readiness at the first grade level. Elementary English, 1948,_25, 370-75. Chall, J.S. and Feldmann, S.C. é Study i2 Depth_gf First Grade Reading. Cooperative Research Project No. 2728, USOE, 1966. Chamberlin, R.W. The use of teacher checklists to identify children at risk for later behavioral and emotional problems. American Journal .gf'Diseases_gf Children, 1976, 130, 141-45. Clark, M. Reading Difficulties_i§ Schools. Harmondsworth: Penguin, 1970. Cobb, J.A. The Relationship_gf Observable Classroom Behaviors £9_Achieve— ment_gf Fourth Grade Pupils. Unpublished doctoral dissertation, University of Oregon, 1969. Cobb, J.A. Survival Skills and First Grade Academic Achievement (Technical Report No. 1). Eugene, OR: Center at Oregon for Research in the Behavioral Education of the Handicapped, December 1970. Cobb, J.A. Relationship of discrete classroom behaviors to fourth grade academic achievement. Journal_gf Educational Psychology, 1972,_63, 74—78. (a) Cobb, J.A. and Hops, H. Effects of academic survival skills in training on low achieving first graders. The Journal 9f Educational Research, 1973,_67, 108—13. Cobb, J.A. Academic survival skills and elementary academic achievement. In E. Meyen, G. Vergason, and R. Whelan (Eds.) Strategies for Teaching Exceptional Children. Denver: Love, 1972. (b) Coker, H. and Lorentz, J. Growth in reading as a correlate of pupil class- room behavior. In P. David Pearson (Ed.) Reading: Theory, Research and Practice. Twenty—sixth Yearbook_gf the National Reading Conference, 1977, 118—25. Commins, W.D. More about sex differences. SchOols and Society, 1928, 28) 599—600. 138 Duffy, 0., Clair, T., Egeland, B., and Dinello, M. Relationship of in- telligence, visual-motor skills, and pyscholinguistic abilities with achievement in the third, fourth, and fifth grades: a follow-up study. Journal 2f_Educational Psychology, 1972,_§3, 358—62. Dykstra, R. and Tinney, R. Sex differences in reading readiness, first grade achievement and second grade achievement. In J.A. Figurel (Ed.), Reading and Realism. Newark Del: International Reading Association, 1969. Eme, R.F. Sex differences in childhood psychopathology: a review. Psycho— logical Bulletin, 1979, 86, 574-95. Engelmann, S. Relationship between psychological theories and the act of teaching. Journal_gf School Psychology, 1967,_2, 93—100. Engin, A. Prediction of classroom achievement using intelligence and be- havioral variables for inner-city children. Psychological Reports, 1975, 36, 67-76. Entwisle, D.R. Implication of language socialization for reading models and for learning to read. Reading Research Quarterly, 1971, 7, 11—67. Feshbach, S., Adelman, S., and Fuller, W. Early identification of children with high risk of reading failure. Journal_gf Learning Disabilities, 1974, 1, 49-54. Feshbach, S., Adelman, H., and Fuller, W. Prediction of reading and re— lated academic problems. Journal 9f Educational Psychology, 1977, 62(4), 299-308. Forness, S., Hall, R., and Guthrie, D. Eventual school placement of kinder- gartners observed as high risk in the classroom. Psychology_ig the Schools, 1977,_1§, 315—17. Friedman, R., Fuerth, J.H., and Forsythe, A. A brief screening battery for predicting school achievement at ages seven and nine years. Psychology .13 the Schools, 1980,_17, 340-46. Fry, E.B. First Grade Reading Instruction Using 3 DiaCritical Marking System, the ITA, and_s Basal Reading system. Cooperative Research Project No. 2745, USOE, 1965. Gates, A.I. The role of personality maladjustment in reading disability. Pedagogical Seminary, 1941,.59, 77—83. Gates, A.I. Sex differences in reading abililty. Elementary School Journal, 1961, 61, 431—34. Gates, A.I. The role of personality maladjustment in reading disability. In G. Natchez (Ed.), Children with Reading Problems. New York: Basic Books, 1968. 139 Gates, A.I. and MacGinitie, W.H. Technical Manual for the Gates-MacGinitie Reading Tests. New York: Teachers College Press, Columbia Univer— sity, 1965. Gentile, L.M. and McMillan, M.M. When Johnny can't read but Mary can, men can help. The Reading Teacher, 1976,_29, 771-75. Gibson, E.J. and Levin, H. The Psychology 9f_Reading. Cambridge, MA: MIT Press, 1975. Gilbert, G.M. A survey of referral problems in metropolitan child guidance centers. A Journal 9f Clinical Psychology, 1957,_13, 37-42. Glavin, J.P. Persistence of behavior disorders in children. Exceptional Gordon, R.A. Issues in multiple regression. American Journal 2f Sociology, 1968, 13, 592-616. Gove, W. and Herb, T. Stress and mental illness among the young: a com— parison of the sexes. Social Forces, 1974,_53, 256-65. Goyette, C.H., Conners, C.K., and Ulrich, R.F. Normative data on revised Conners Parent and Teacher Rating Scales. Journal gf_Abnormal Child Psychology, 1978, 6(2), 221—36. Hahn, H.T. A Study 2: the Relative Effectiveness_gf Three Methods of Teacher Reading is Grade One. Cooperative Research Project No. 2687, USOE, 1965. Hall, V., prertz, J., and Levi, A. Attention and achievement exhibited by middle- and lower—class black and white elementary school boys. Journal sf Educational Psychology, 1977,_69(2), 115-20. Hall, R.C., Lund, D. and Jackson, D. Effects of teacher attention on study behavior. Journal_gprpplied Behavioral Analysis, 1968, 1, 1-12. Harrison, P. Intelligence and Classroom Behavior_ss PredictOrs_gf Achievement is the Underachieving Child. Unpublished doctoral dis- sertation, University of Tennessee, 1975. Heilman, A.W. Sex differences in intellectual abilities. Journal 9f Educational Psychology, 1933, 23, 47-62. Heilman, A.W. Effects 2f Intensive In—Service Program 92 Teachers' Class- room Behavior and Pupils' Reading Achievement. Cooperative Research Project No. 2709, USOE, 1965. Heilman, A.W. Prinicples and PractiCes gf_Teaching Reading. Columbus, Ohio: Charles E. Merrill Books, Inc., 1967. 140 Hewett, E.M., Taylor, F.D., and Artuso, A.A. The Santa Monica Project: Demonstration and Evaluation_gqug Engineered ClassroOm Design with Emotionally Disturbed Children_ig the Public Schools, Phase I: Ele— mentary Level. Final Report, Project No. 62893, USOE, DHEW, 1967. Hoge, R.D. and Luce, S. Predicting academic achievement from classroom behavior. Review 2f Educational Research, Summer 1979, 32(3), 479- 96. Holmes, J.A. Personality characteristics of the disabled reader. Journal 3f Developmental Reading, 1961, 111—22. Hops, H. The school psychologist as a behavior management consultant in a special class setting. Journal_gf School Psychology, 1971,_9(4). Hops, H. and Cobb, J.A. Initial investigations into academic survival- skill training, direct instruction, and first grade achievement. Journal 9f Educational Psychology, 1974,_§6, 548—53. Horton, T.R. The Reading Standards_gf Children iE_Wales. Slough: NFER, 1973. Hughes, M.C. Sex differences in reading achievement in the elementary grades. _IE Clinical Studies_i§ Reading_II: Supplementary Educational Monographs N9, 11, 1953. Johnson, D.D. Sex differences in reading across cultures. Reading Re— search Quarterly, 1973-74,_9(1), 67—86. Katz, M. Personal communication. State of Michigan: Department of Edu- cation, February 27, 1980. Kellam, S.G., Branch, J.D., Agrawal, K.C., and Ensminger, M.E. Mental Health and Going E9 School. Chicago: University of Chicago Press, 1975. Kellmer Pringle, M.L., Butler, N.R., and Davie, R. 11,000 Seven-Year-Olds. First Report of the National Child Development Study. London: Long— mans, 1966. Kerlinger, F.N. and Pedhazur, E.J. Multiple Regression in Behavioral 337 search. New York: Holt, Rinehart, and Winston, Inc., 1973. Kessler, J. Psychopathology_gf Childhood. Englewood Cliffs, N.J.: Pren- tice-Hall, 1966. Khan, S.B. Affective correlates of academic achievement. Journal EEUEHI ucational Psychology, 1969,_§Q, 216—21. Kim, Jr. and Kohout, F. Multiple regression analysis: Subprogram re- gression. In Nie, H., Hull, C., Jenkins, J., Steinbrenner, K., and Bent, D. (Eds.), Statistical Package for the Social Sciences, Second Edition. New York: McGraw-Hill Book Company, 1975. 141 Kim, Y., Anderson, H.E., and Bashaw, W.L. The simple structure of social maturity at the second grade level. Educational and Psychological Measurement, 1968,_2§, 145—53. Klein, A. The validity of the screening test of academic readiness in predicting achievement in first and second grades. Educational and Psychological Measurement, 1977,_3Z, 493-99. Kohn, M. and Rosman, B. A two factor model of emotional disturbance in the young child: validity and screening efficiency. Journal 9f Child Psychology and Psychiatry, 1973, 13, 31-56. Kohn, M. and Rosman, B. Social emotional, cognitive, and demographic determinants of poor school achievement: implications for a strategy of intervention. Journal_gf Educational Psychology, 1974, 66, 267-76. Lahaderne, H.M. Attitudinal and intellectual correlates of attention: a study of four sixth grade classrooms. Journal_gf Educational Psy- - chology, 1968,_59, 320-24. Lambert, N.M. Intellectual and non-intellectual predictors of high school status. Journal_9f Special Education, l972,_§, 247-59. Lambert, N.M., Hartsough, C.S., and Zimmerman, I.L. Comparative predictive efficiency of intellectual and non-intellectual components of high school functioning. American Journal 9f 0rthopsychiatry, 1976,_46, 109-22. Lambert, N.M. and Nicoll, R.C. Conceptual model for non-intellectual behavior and its relationship to early reading achievement. Journal _gf Educational Psychology, 1977,_69(5), 481-90. Lambert, N. and Urbanski, C. Behavioral profiles of children with dif— ferent levels of achievement. Journal_gf School Psychology, 1980, 18, 58-66. Lampouse, R. and Monk, M. Behavior deviations in a representative sample of children. American Journal 2f_0rthopsychiatry, 1964,_3£, 436—46. Lehrke, R. Sex linkage: a biological basis for greater male vulnera— bility in intelligence. In R. Osbourne (Ed.), Human Variation. New York: Academic Press, 1978. Lesser, K. and Bridges, J. The prediction of learning problems in a rural setting: can we improve on readiness tests? Journal 9f Learning Disabilities, 1973, 6, 90-94r Lipe, D. and Jung, S.M. Manipulating incentives to enhance school learning. Review 2f Educational Research, 1971, 31, 249-80. Luce, S. and Hoge, R. Relations among teacher rankings, pupil—teacher in- teractions, and academic achievement: a test of the teacher expectancy hypothesis. American Educational Research Journal, Fall 1978,_15(4), 489—500. 142 Maccoby, E.E. Sex differences in intellectual functioning. In E. Maccoby (Ed.), The Development 3f Sex Differences, Stanford: Stanford Univer— sity Press, 1966, 25—55. Maccoby, E.E. and Jacklin, C.N. The Psychology_gf Sex Differences. Stan— ford: Stanford University Press, 1974. Madden, R., Gardner, R.F., Rudman, H.C., Karlson, B. and Merwin, J.C. The Stanford Achievement Test: Technical Data Report. New York: Harcourt, Brace, Jovanovich, Inc., 1975. Madsen, C.H., Becker, W.C., and Thomas, D.R. Rules, praise, and ignoring: elements of elementary classroom control. Journal 9f_Applied REF vahorial Analysis, 1968,_1, 139-50. Manning, J.C. Evaluation.2f_Level—Designed Visual-Auditory and Related Writing Methods_gf Reading Instruction_i§ Grade One. Cooperative Research Project No. 2650, USOE, 1966. McCarthy, D. Personality and learning. American Council_gg Educational Studies, 1949,_13, 93—96. McFarlane, J., Allen, L. and Honzik, M. .A_Developmental Study 9f the Behavior Problems of Normal Children Between the Ages of Twenty-One Months and FourteeETYears. Berkeley: University of Cinfornia Press, 1954. McGuide, C., Hindsman, E., King, F.J. and Jennings, E. Dimensions of talented behavior. Educational and Psychological Measurement, 1961, 31, 3-38. McKinney, J., Mason, J., Perkerson, K. and Clifford, M. Relationship between classroom behavior and academic achievement. Journal 9f Educational Psychology, 1975,_QZ, 198-203. McLaren, V.M. Socioeconomic status and reading ability: a study of in— fant reading. In Scottish Council for Research in Education, Studies .13 Reading, Vol. 2. London: University of London Press, 1950. Meyers, C.E., Attwell, A.A., and Orpet, R.E. Prediction of fifth grade achievement from kindergarten test and rating data. Journal 9f Educational Measurement, 1968,_23, 457-63. Miller, L.C. School Behavior Checklist: an inventory of deviant behavior for elementary school children. Journal_gf Consulting and Clinical Psychology, 1972, 38, 134-44. Miller, L.C. School Behavior Checklist Manual: Form A1, Ages 3-6 and Form A2, Ages 7-13. Louisville, KY: University of Louisville, 1975. Morris, J.M. Standards and Progress iE_Reading. Slough: NFER, 1966. 143 Munger, R.F., Winkler, R.C., Tergland, J.J., and Kranzler, G.D. Counseling and Guidance for Underachieving Fourth Grade Students. Cooperative Research Project No. 2196, University of North Dakota, 1964. Murphy, H.A. Reading Achievements 22_Relation_gg Growth iE_Perception gf_Word Elements is Three Types 9f Beginning Reading Instruction. Cooperative Research Project No. 2675, USOE, 1965. Naiden, N. Ratio of boys to girls among disabled readers. The Reading Teacher, 1976, 23(5), 439-42. Neale, D.C., Gill, N., and Tismer, W. Relationship between attitudes toward school subjects and school achievement. Journal_gf Educational Research, 1970,_63, 232-37. Nie, N.H., Hull, C.H., Jenkins, J.C., Steinbrenner, K., and Bent, D.H. Statistical Package for the Social Sciences, 2nd Edition, New York: McGraw-Hill Book Company, 1975. Nila, Sister M. Foundations of a successful reading program. Education, 1953,_Z3, 543-55. Otis, A.S. and Lennon, R.T. Manual for the Otis-Lennon Mental Ability Test. New York: Harcourt, Brace, and Jovanovich, 1967. Parsley, K.M., Powell, M., O'Connor, H.A. and Deutsch, M. Are there really sex differences in achievement? Journal_gf Educational Research, 1963, 57, 210-12. Patterson, G., Littman, R., and Bricker, W. Assertive behavior in children: a step toward a theory of aggression. Monographs sf the Society for Research is Child Development, 1967, 32(4), Serial No. 113. Patterson, G.R., Shaw, D.A. and Ebner, M.J. Teachers, peers, and parents as agents of change in the classroom. In F. Benson (Ed.) Modifying Deviant Social Behaviors 13 Various Classroom Settings. Eugene, OR: University of Oregon, College of Education, Monograph No. 1, 1969. Peterson, D.R. Behavior problems of middle childhood. Journal sf Con— sulting Psychology, 1961, 23, 205-09. Prawat, R. and Jarvis, R. Gender difference as a factor in teachers' perceptions of students. Journal gf_Educational Psychology, 1980, 72, 743-49. Preston, R.C. Reading achievement of German and American children. School and Society, 1962,_2Q, 350—54. Quay, H. The Structure_gf Children's Behavior Disorders. Paper read at colloquia at the University of Minnesota, 1965. 144 Richman, N., Stevenson, J., and Graham, J. Prevalence of behavior problems in three year old children. Journal gf_Child Psychology and Psychiatry, 1975,_l§, 277—87. Roberts, J. and Baird, J.T. Behavior Patterns sf Children 22 School: United States (Vital and Health Statistics, Series 11). Rockville MDL U.S. Department of HEW, Division of Health Examination Statistics, February 1972. Rogers, D.G. The C-R Word Recognition Test. Christchurch, N.Z.: Psy— chological Service, Department of Education, 1971. Ross, A.O., Lacey, H., and Parton, D. The development of a behavior checklist for boys. Child Development, 1965,_36, 1,013-027. Rubin, R.A. and Balow, B. Prevalence of teacher identified behavior problems: a longitudinal study. Exceptional Children. 1978, 45(2). Russell, D.H. Interrelationships of the language arts and personality. Child Development and the Language Arts: Bulletin of the National Conference on Research in English. Champaign, IL: National Council of Teachers of English, 1953. Rutter, M., Tizard, J., and Whitmore, K. Education, Health and Behavior. London: Longman, 1970. Rutter, M. Classification. In M. Rutter and L. Hersor (Eds.) Child Psychiatry. Oxford, England: Blackwell Scientific, 1977. Rutter, M. and Yule, W. Reading difficulties. In M. Rutter and L. Hersor (Eds.), Child Psychiatry. Oxford, England: Blackwell Scien— tific, 1977. Samuels, S.J. and Turnure, J.E. Attention and reading achievement in first grade boys and girls. Journal gf_Educational Psychology, 1974,'6§, 29-32. Satz, P. and Fletcher, J. Early screening tests: some uses and abuses. Journal sf Learning Disabilities, 1979,_l2, 65-69. Satz, P. and Friel, J. Some predictive antecedents of specific reading disability: a preliminary two-year follow—up. Journal gf_Learning Disabilities, 1974, 2, 48-55. Satz, P. and Friel, J. Predictive validity of an abbreviated screening battery. Journal_gf Learning Disabilities, 1978, 32, 20-24. Schlosser, L. and Algozzine, B. The disturbing child: he or she? Alberta Journal_gf Educational Research, 1979, 23, 30-36. 145 Schneyder, J.W., Schultz, C.B., and Cowen, S. Comparison sf Reading Achievement for First Grade Children Taught by_s Linguistic Approach .EEH.§ Basal Reader Approach. Cooperative Research Project No. 2666, USOE, 1966. Sheldon, W.D. and Lashinger, D.R. Effects_gf First Grade Instruction Using Basal Readers, Modified Linguistic Materials, and Linguistic Readers. Cooperative Research Project No. 2683, USOE, 1966. Slossen, R.L. Slossen Oral Reading Test (SORT). East Aurora, NY: Slossen Educational Publications, 1963. Soli, S. and Devine, V. Behavioral correlates of achievement: a look at high and low achievers. Journal_gf Educational Psychology, 1976, 68, 335-41. Southgate, V. Southgate Group Reading Tests. London: University of London Press, 1959 and 1962. Spache, G.D. Personality patterns of retarded readers. Journal 2: Educational Research, 1957, 39, 488-93. Spache, G.D. Investigating the Issues_gf Reading Disability. New York: Allyn and Bacon, 1976. Spache, G.D., Andress, M.C., Curtis, H.A., Rowland, M.L., and Fields, M.H. A Study_gf_s Longitudinal First Grade Reading Readiness Program. Cooperative Research Project No. 2742, USOE, 1965. Spencer, D.V. and Moquin, L.D. Individualized Reading Versus_s Basal Reader Program_s£ the First Grade Level is Rural Communities. Coop— erative Research Project No. 2673. Spivak, G. and Swift, M.S. The Devereux Elementary School Rating Scales: a study of the nature and organization of achievement—related dis- turbed classroom behavior. Journal gf_Special Education, 1966, 3, 71—90. Stanchfield, J.M. Boys' achievement in beginning reading. In J. Figurel (Ed.), Reading and Inquiry. Proceedings of the Annual International Reading Association Conference, 1965, 10, 290-93. Stanchfield, J.M. Sex Differences_ip Learning £9_Read, Bloomington: Phi Delta Kappa Educational Foundation, 1973. Stevenson, H.W., Parker, T., Wilkinson, A., Hegion, A., and Fish, E. Longitudinal study of individual differences in cognitive develop- ment and scholastic achievement. Journal_gf Educational Psychology, 1976, 33, 507-17. Stevenson, H.W., Parker, T., Wilkinson, A., Hegion, Al, and Fish, E. Predictive value of teacher ratings of young children. Journal_gf Educational Psychology, 1976,_68, 507-17. 146 Stroud, J.B. and Lindquist, E.F. Sex differences in achievement in the elementary and secondary schools. Journal_g£ Educational Psychology, 1942, 2, 657-670 Swift, M. and Spivack, G. The assessment of achievement related class— room behavior. Journal 2£_Special Education, 1967, 2(2), 137-49. Thompson, G.B. Sex differences in reading attainments. Educational_Rse search, 1975,_£§, 16-23. Vernon, P.E. The Standardization 9£_s Graded Word List. London: Univer— sity of London Press, 1938. Walker, H.M. Walker Problem Behavior Identification Checklist Manual. Western Psychological Services, 1970. Walker, H.M., Mattson, R.H., and Buckley, N.K. The functional analysis of behavior within an experimental class setting. In W.C. Becker (Ed.) AE_Empirical Basis for Change is Eduation. Chicago: Science Research Associates, 1971, 236-63. Werry, J.S. and Quay, H.C. Observing the classroom behavior of elementary school children. Exceptional Children, 1969, 33, 461-70. Wilf, V. Observable Student Behaviors_ss Indicators_g£ Emotional Malad— justment and Reading Disability. Unpublished doctoral dissertation, Arizona State University, 1972. Wozencraft, M. Sex comparisons of certain abilities. Journal 93 Educa- tional Research, 1963,_3Z, 21-27. Young, R. ‘AE_Evaluation.9£ Elementagy Categorical Programs: 1979-80. Lansing, MI: Office of Evaluation Services, Lansing School District, November 30, 1980. Young—Maston, I. Behavior problems of elementary school children: a descriptive and comparative study. Genetic Psychology, Monographs, _32, 1938, 123-81. Zax, M., Cohen, E.L., Izzo, L.D. and Trost, M.A. Identifying emotional disturbance in the school settings. American Journal_g£ Orthopsy- chiatry, 33, 1964, 447-54. APPENDICES 147 APPENDIX A Breakdown of Mean Stanford Reading Achievement Test Standard Scores By Grade By Sex of Student SATl SAT 2 SAT 3 SAT 4 Mean S.D. Mean S.D. Mean S.D. Mean S.D. Grade 2 50.6 10.3 50.7 11.0 47.8 9.6 49.8 9.5 Male 49.4 11.3 49.6 12.3 46.7 10.4 49.8 9.7 Female 52.4 9.1 52.3 9.3 50.9 7.7 49.8 10.0 Grade 3 50.4 9.0 52.7 10.2 53.5 9.5 54.3 9.4 Male 44.0 6.9 46.1 8.5 46.9 5.6 46.9 5.9 Female 55.0 7.4 56.9 9.2 57.7 9.2 59.7 7.5 Grade 4 51.2 6.6 52.5 6.3 54.4 6.7 51.5 7.5 Male 51.3 8.0 53.8 7.5 55.7 6.5 52.4 9.2 Female 51.1 5.1 51.3 4.8 53.2 6.8 50.4 5.0 Grade 5 45.8 8.2 48.0 8.3 47.6 7.9 48.1 7.2 Male 44.5 8.8 46.5 9.1 46.9 8.6 48.4 7.5 Female 49.0 6.1 52.2 3.2 49.1 6 5 47.7 7.0 Grade 6 50.9 9.6 49.3 8.4 51.5 8.9 50.5 9.7 Male 52.4 11.7 51.0 10.9 53.0 11.0 51.9 11.1 Female 49.4 7.1 47.5 4.9 50.1 6.2 49.4 8.9 Total 49.7 8.8 50.6 8.9 50.9 8.8 50.8 8.7 148 4 APPENDIX B Breakdown of Mean Behavior Scale Raw Scores B y Grade By Sex of Student Attention Disrespect Attention Academic Externalized Socialization Motivation Defiance Seeking Self-Confidence Agression Mean S.D. Mean 5 D. Mean S.D. Mean S.D. Mean 5.0 Mean S.D. Grade 2 17.6 5.4 12.3 3.1 10.5 2.1 6.6 0.9 9.8 2.5 4.1 1.1 Male 17.8 5.3 12.3 3.1 10.5 2.2 6.8 1.1 9.9 2.6 4.3 1.1 Female .3 5.9 12.1 3.2 10.4 2.1 6.1 0.1. 9.6 2.5 3.8 1.0 Grade 3 . 4.4 12.7 3.1 12.7 3.1 6.3 2.8 9.8 2.8 3.3 0.8 Male 19.9 4.1 14.6 3.4 11.0 3.4 6.0 0.8 11.4 3.0 3.4 0.7 Female 3.4 11.4 2.0 10.0 1.5 6.5 1.0 8.7 2.1 3.3 0.9 Grade 4 . 2.7 11.7 2.3 9.0 2.0 6.8 1.4 8.8 1.9 3.3 0.2 Male 14.8 3.6 11.4 1.9 10.5 2.7 6.9 1.7 9.3 2.3 3.3 0.6 Female 1.4 12.0 2.7 9.3 0.5 6.6 1.0 8.3 1.2 3.3 0.8 Grade 5 4 4.2 13.8 3.1 2.6 6.8 1.0 9.0 1.6 3.3 0.8 Male 18.3 4.3 13.9 2.9 11 3 2.9 6.6 0.9 9.2 1.8 3.4 0.9 Female .3 1.5 13.7 3.7 . 1.5 7.0 1.1 8.4 0.8 3.3 0.5 Grade 6 4.3 12.6 2.8 2.2 7.3 1.4 10.1 2.2 3.3 0.8 Male 18 3 4.6 12.6 2.9 10 6 2.7 7.2 1.9 10.3 2.3 3.6 1.0 Female 3.5 12.6 2.8 1.6 7.4 0.9 9.9 2.1 3.1 0.3 Total 16 5 4 3 12.6 2 9 10.6 2.3 6.8 1.2 9.5 2 2 3.5 0.9 149 many Hnma romaeomm Swnr magnouvmw mannonuom sen: Hnmnonwos one Hmnn H: apmmm DMNMNW 1.0uwmo .wpmpm .peowm .wooeo .ouwom Hm. nmcmm arm: nova son no no meannrwsm .Howwm LWWMWW .Nowww .Nbomm J.HNomw .oowmm 150 >mwmzuHN n Anoan.mv Hnma wmnnon H amonon N wmonon w wmnnon b mmonon m wmonon o NM. comm roamzonw .mmowu .Hmmob .HHHNM .oemmw .omuoo .Ouwum Nu. Hmmmmm OnUmH ornwmnms .oowbw .uooeu .oewmu .NNNoN .Houbw |.oo~ou No. mmmHH< cemmn cw ormumm .ommbm .mHNbN .Hwoew .oewww .HuMNH .Nmoob Nu. Hm mcnm om mmHm .mopww .Homuo 1.0Hmmm |.owbum .bwmbb .0ubaw Mm. comm mvcmnwwmzuHx o Aoosn.mv Hnma muonon H mmonon N wmnnon w muonon b wmnnon w menon m aw. woman QCHHW mHot no owned 0: .Nowmu .oonu .omw~o .Hwoma .momwo .owbub be. 20:.n mmw acmmnnosm ._wbbw 1.0mmbm .Hpmbw .OONVO .bowou .OEbob no. zmHmcmm awn: nmmnrmn ._wowb .owuwm .Nwomw .umo- 1.0umwo .OHmoa em. >UHm no nonnmsnnmnm .o~omu .OVwmb .owooo .Hmwwy swomom .wwwww My. annmm no on noonmn om mnnmnnpo: .ommom .woomm .Hooa~ .VNONH I.ow~m~ 1.0HNmN m». :unmmm moon: arms . nmacmmnmm no no moamnrnsm .uwowm .Hmomo I.omo~u .OVomb .oowbw .Hmeb— we. mCHWm .Hoowo .obomm .ooHNo .uoowu .OHmme I.o~b~u mu. cumnnmnnnvwo .ouwmw ._~mum .Howbw .Humwo .Nb~e~ .HomoV om. mncvvonu .mmwwu .bouoo .wmmmm .meNH I.~moo~ 1.0wwob aw. zmwwmzuHx o Anoan.mv Hnma wmonon p moonon N wmnnon w wmonon » moonon m mmonon a ma. comm nrnsmm no mnnnmon mnnmsnwos .Hwoou .bmwoc .wmpuw .bmbwm .-~w~ 1.000uo aw. woocpmn .wwowb .oooob .wuwmm I.ooooo .NHoOR .uomup ‘0. Como nsndmm nonamw non wocsmmn orwpanm: .ompuo I.No~w 1.0mwbm .mmwow .woowo .Nopou um. arnmmnmom onrmn ornwanms .owpmw .uboom .momwo .b~w~b I.owmou .ooaou um. wnmmmnm no onw mom to mHoso .oopow I.ouwww 1.HNNNm .Hoooo .owuoo .mwORm um. unsnmsmm mmmumsamonm .wowuw .Obbwo .ooooH .mwwwm 1.oooew .NoNom NV. nnwwmchx 0 Aoo=n.av Hnms muonon H wmnnon m wmonon w mmonon b wmnnon m mmnnon 0 mm. Unmmwncwnw mummwnnm arm: mxnwnma I.~mmpw .Nmemm .oNo~u I.~e~e~ .euuwm .oooww ow. Unmncncm onronm Sen: wouwanOCmsmmw .Hmmpw .Huomm .wmmme .wmbwu .wamo 1.0uuHm Ho. dmmmmnHmn mnnnncam .Noeoo .NVHN 1.ooomm 1.-moo .omubb .bmumo 154 APPENDIX D Simple Correlations Between Individual Items of the SBCL and the Initial Stanford Reading Achievement Test (SATl) Item Totala Boys Girlsc 1. Is friendly —.02 .14 .17 2. Tends to give up if has something hard to finish -.48 -.49 -.38 3. Interrupts whoever is speaking - 17 -.31 .09 4. Penmanship at least one grade level below age expectation -.23 -.15 -.29 5. Starts fighting over nothing - 25 - 28 - 10 6. Is a helpful child 21 19 17 7. Is alert in class 40 8. Poorly coordinated when doing things with hands, such as coloring — 15 - 17 18 9. Reading ability at least one grade level below age expectation — 65 — 66 - 59 10. Just stands around on the playground 00 01 01 11. Acts up when adults not watching - 14 - 11 —.13 12. Volunteers to recite in class 13 12 23 13. Hits and pushes other children - 21 — 23 — 12 14. Hands shake or is nervous when called on to recite - 08 - 04 - 09 15. Finds fault with what other children do - 10 - 14 - 11 16. Approaches a difficult task with an air of defeatism - 26 — 31 — 04 17. Is considerate of others 10 13 00 18. Fails to carry out tasks - 36 - 39 - 17 19. Lacks ambition to do well in school - 34 - 35 - 18 20. Does things to get others angry - 20 - 29 00 21. Will put up an argument when told not to do something - 25 - 27 - 15 22. Does homework 38 39 21 23. Teases other children - 29 - 34 - 23 24. Is afraid of making mistakes - 04 02 - 07 25. Is bossy with other children - 05 - 02 - 16 26. Is easily upset by changes — 16 - 10 - 10 27. Is sure of self ' 40 41 32 28. Uses abusive language toward other children - 25 - 24 O4 29. Has changeable moods - 22 — 34 04 30. Gives in when another child insists on doing something another way 03 10 - 17 31. Does not respect other people's belongings - 24 - 25 - 13 155 APPENDIX D (cont'd) Item Total8 Boysb GirlsC 32. Does not forget things which anger her/him -12 -17 -O8 33. Seems to be off in own world —16 -16 -09 34. Is infuriated by any form of disCipline -20 -33 -01 35. Likes an audience all the time —31 -36 -01 36. Finds it hard to study -42 —45 -24 37. Has to have everything own way -03 —13 11 38. Works well alone 27 28 19 39. When angry, will refuse to speak to anyone -17 —10 -32 40. School performance is far below capabilities -35 -36 -18 41. Has no friends 10 14 O6 42. Behind at least one school grade due to academic difficulties -54 -54 -49 43. Seems dull; slow to catch on —42 -44 -31 44. Will not ask questions even when unsure as to how to do the work -13 -23 05 45. Fights back if another child has been asking for it -05 -05 03 46. Never seems to be still for a moment -26 -26 —17 47. Argues with me -22 -29 -1O 48. Is able to concentrate on things 31 29 24 49. Boasts of own toughness -19 -15 -17 50. Seems to think of self as worthless -04 05 -19 51. Tries to be the center of attention -22 -32 05 52. "Drags feet" when requested to do something -23 -18 -25 53. Accepts adult suggestions 10 06 12 54. Sulks when things go wrong -29 -34 —14 55. Becomes frightened easily -10 -09 -08 56. Resents even the most gentle of criticism of work -06 -11 12 57. Distractible; can't concentrate -37 -33 -37 58. Able to see the bright side of things 05 01 07 59. Fights with smaller children —04 -04 —01 60. Spelling performance at least one grade level below age expectation —53 -57 -38 61. Fearful of being hurt at play 07 -01 O7 62. Is stubborn -24 -30 -08 63. Never speaks up even when there is cause to be angry -10 -28 12 64. Is interested in schoolwork 38 42 21 156 APPENDIX D (cont'd) Item Total8 Boysb GirlsC 65. Tries to get other children into trouble O4 05 00 66. Does things just to attract attention ~30 ~30 ~23 67. Never fights back, even if someone hits or pushes first 01 02 ~04 68. Prefers to be around adults rather than play with children 07 13 ~04 69. Is popular with his classmates 28 28 25 70. Does things which are normal for children much younger ~30 ~32 ~16 71. Never sticks up for self when picked on by other children 08 22 ~11 72. Threatens to hurt other children when angry ~23 ~25 ~06 73. Average or above IQ 47 41 54 74. Does not take orders when other children are in charge ~14 ~20 01 75. Prefers to be alone and play alone 15 28 ~12 76. Finishes classroom assignments 30 3O 17 77. Gives other children dirty looks ~16 ~22 ~09 78. Deliberately interrupts what is going on by asking silly questions ~31 ~35 01 79. Slow in making friends ~26 ~20 ~33 80. Seems as happy as most children 17 15 10 81. Finds fault with instructions given by adults ~04 ~04 ~01 82. Seems unconcerned when misbehaving ~20 ~20 ~17 83. Cries easily O9 12 05 84. Is afraid of strange adults 04 08 ~01 85. Is self-confident 17 13 20 86. When angry, will do things like slamming the door or banging the desk ~18 ~09 ~34 87. Acts in a "dare—devil", fearless manner ~21 ~26 12 88. Has difficulty speaking clearly when excited or upset ~12 ~26 12 89. Has a "chip on shoulder" ~07 ~14 11 90. Becomes embarrassed easily ~13 ~15 '—17 91. Bright, but doesn't apply self ~16 ~14 ~06 92. Disturbs other children with boisterous behavior ~18 ~22 01 93. Behind at least two school grades due to academic difficulties ~30 ~29 ~21 94. Arithmetic skill at least one grade level below age expectation ~42 ~41 ~38 157 APPENDIX D (cont'd) Item Totala Boysb GirlsC 95. Much anxiety~~afraid of such things as storms, school, death, injury, war ~05 ~03 ~02 96. Frequent headaches, stomach aches or other non-specific physical complaints 02 11 ~13 (a) r .23 significant at .01 level for total group (b) r .28 significant at .01 level for (c) r .31 significant at .01 level for APPENDIX E 158 Simple Correlations Between Individual Items of the SBCL and the Stanford Reading Achievement Test (Total Group) 1. Is friendly ~02 12 ~03 ~04 2. Tends to give up if has something hard to finish ~48 ~44 ~45 ~48 3. Interrups whoever is speaking ~17 ~12 ~17 ~13 4. Penmanship (handwriting) at least one grade level below age expecta- tion ~23 ~27 ~14 ~22 5. Starts fighting over nothing ~25 ~25 ~14 ~22 6. Is a helpful child 21 27 15 16 7. Is alert in class 40 41 36 28 8. Poorly coordinated when doing things with hands, such as coloring or pencil work ~15 ~19 ~03 ~02 9. Reading ability at least one grade level below. age expectation ~65 ~63 ~57 060 10. Just stands around on the playground 00 ~04 ~01 03 11. Acts up when adults not watching ~14 ~15 ~20 ~33 12. Volunteers to recite in class 29 25 18 24 13. Hits and pushes other children ~21 ~27 ~17 ~27 14. Hands shake or is nervous when called on to recite ~08 ~03 04 ~08 15. Finds fault with what other children do ~10 ~15 ~15 ~23 16. Approaches a difficult task with an air of defeatism ~26 ~20 ~19 ~19 17. Is considerate of others 10 19 10 15 18. Fails to carry out tasks (home- work assignments, seat work, etc.) ~36 ~36 ~32 ~28 19. Lacks ambition to do well in school ~34 ~34 ~21 ~24 20. Does things to get others angry ~20 ~27 ~21 ~26 21. Will put up an argument ~25 ~21 ~24 ~36 22. Does homework ' 38 42 32 24 23. Teases other children ~29 ~31 ~27 ~33 24. Is afraid of making mistakes ~04 ~03 01 ~02 25. Is bossy with other children ~05 ~07 ~10 ~14 26. Is easily upset by changes ~16 ~08 ~10 ~21 APPENDIX E (cont'd) 159 Item SATl SAT2 SAT3 SAT4 27. Is sure of self 40 38 26 34 28. Uses abusive language toward other children ~25 ~19 ~24 ~23 29. Has changeable moods ~22 ~25 ~28 ~25 30. Gives in when another child insists on doing something another way 03 02 08 07 31. Does not respect other people's belongings ~24 ~21 ~23 ~23 32. Does not forget things which anger her/him ~21 ~04 ~11 ~11 33. Seems to be off in own world ~16 ~30 ~18 ~07 34. Is infuriated by any form of discipline ~20 ~16 ~19 ~25 35. Likes an audience all the 'time ~31 ~32 ~24 ~25 36. Finds it hard to study ~42 ~44 ~32 ~35 37. Has to have everything own way ~03 ~10 ~14 ~01 38. Works well alone 27 28 25 19 39. When angry, will refuse to speak to anyone ~17 ~21 ~23 ~27 40. School performance is far below capabilities ~35 ~39 ~18 ~11 41. Has no friends 10 12 11 15 42. Behind at least one school grade due to academic difficulties ~54 ~54 ~42 ~33 43. Seems dull; slow to catch on ~42 ~44 ~36 ~34 44. Will not ask questions even when unsure as to how to do the work ~23 ~23 ~16 ~19 45. Fights back if another child _ has been asking for it ~05 ~03 ~03 ~07 46. Never seems to be still for a moment ~26 ~20 ~27 ~28 47. Argues with me ~22 ~22 ~24 ~30 48. Is able to concentrate on things 31 35 24 21 49. Boasts of own toughness ~19 ~11 ~18 ~26 50. Seems to think of self as worthless ~04 ~05 06 05 51. Tries to be the center of attention ~22 ~21 ~22 ~23 APPENDIX E (cont'd) 160 Item SATl SATz SAT3 SAT4 52. "Drags feet" when requested to do something ~23 ~27 ~13 ~15 53. Accepts adult suggestions 10 19 11 11 54. Sulks when things go wrong ~29 ~31 ~28 ~38 55. Becomes frightened easily ~10 ~09 02 ~07 56. Resents even the most gentle criticism of work ~06 ~06 ~06 ~10 57. Distractible; can't con- centrate ~37 ~40 ~27 ~21 58. Able to see the bright side of things 05 02 05 14 59. Fights with smaller children ~04 ~04 ~03 ~15 60. Spelling performance at least one grade level below age expectation ~53 ~56 ~39 ~42 61. Fearful of being hurt at play 07 18 18 18 62. Is stubborn ~27 ~21 ~18 ~23 63. Never speaks up even when there is cause to be angry ~10 ~15 ~08 ~06 64. Is interested in schoolwork 38 38 29 30 6S. Tries to get other children in trouble 04 ~12 ~04 ~02 66. Does things just to attract attention ~30 ~28 ~33 ~40 67. Never fights back, even if someone hits or pushes first ~01 ~01 ~05 ~06 68. Prefers to be around adults, rather than play with children ~17 - ~23 ~16 ~11 69. Is popular with his classmates 28 38 25 24 70. Does things which are normal for children much younger ~30 ~30 ~17 ~24 71. Never sticks up for self 08 05 09 06 72. Threatens to hurt other children when angry ~23 ~28 ~24 ~32 73. Average or above IQ ~47 ~46 ~46 ~50 74. Does not take orders when other children are in charge ~14 ~16 ~17 ~23 75. Prefers to be alone and play alone 25 16 08 18 76. Finishes classroom assign- ments 30 28 18 25 77. Gives other children dirty looks ~16 ~18 ~21 ~32 78. Deliberately interrupts what is going on by asking silly questions ~31 ~28 ~26 ~30 161 APPENDIX E (cont'd) Item SATl SATZ SAT3 SAT4 79. Slow in making friends ~26 ~35 ~22 ~18 80. Seems as happy as most children 17 23 08 05 81. Finds fault with instruc— tions given by adults ~04 ~09 ~15 ~26 82. Seems unconcerned when misbehaving ~20 ~23 ~20 ~29 83. Cries easily 09 14 10 01 84. Is afraid of strange adults 04 01 00 O6 85. Is self—confident 17 18 12 O6 86. When angry, will do things like slamming the door ~18 ~17 ~14 ~29 87. Acts in a "dare-devil", fearless manner ~21 ~23 ~14 ~19 88. Has difficulty speaking clearly when excited or upset ~12 ~13 01 ~12 89. Has a "chip on shoulder" ~07 ~16 ~09 ~17 90. Becomes embarrassed easily ~13 ~11 ~07 ~03 91. Bright, but doesn't apply self (underachiever) ~16 ~19 ~11 ~17 92. Disturbs other children with boisterous behavior ~18 ~16 ~16 ~23 93. Behind at least two school grades due to academic difficulties ~30 ~28 ~07 ~18 94. Arithmetic skill at least one grade level below age expecta- tions ~42 ~46 ~30 ~35 95. Much anxiety~~afraid of such things as storms, school, death, injury, was (considered phobic) ~05 ~10 07 00 96. Frequent headaches, stomach aches or other non-specific physical complaints 02 ~02 12 ~16 * r .23 significant at .01 level 162 APPENDIX F Simple Correlations Between Individual Items of the SBCL and the Stanford Achievement Test (SATl, SAT2, SAT3, SAT4) for Boys and Girls I t em SAT 1 SAT 2 SAT 3 SAT 4 Boys Girls Boys Girls Boys Girls Boys Girls 1. Is friendly 14 17 17 24 16 14 22 17 2. Tends to give up if has something hard to finish ~49 ~38 ~49 ~26 ~47 ~34 ~46 ~50 3. Interrupts whoever is speaking ~31 09 ~29 16 ~33 05 ~29 08 4. Penmanship at least one grade level below age expectation ~15 ~29 ~22 ~29 ~01 ~32 ~14 ~36 5. Starts fighting over nothing ~28 ~10 ~24 ~22 ~16 ~09 ~33 ~00 6. Is a helpful child 19 17 26 24 . 12 14 14 17 7. Is alert in class 39 32 4O 37 31 39 34 14 8. Poorly coordinated when doing things with hands, - such as coloring ~17 18 ~28 32 ~06 35 ~07 28 9. Reading ability at least one grade level below age expectation ~66 ~58 ~66 ~53 ~56 ~53 ~56 ~66 10. Just stands around on the playground 00 01 00 ~08 08 ~16 12 ~12 11. Acts up when adults not watching ~11 ~13 ~07 ~25 ~18 ~18 ~32 ~31 12. Volunteers to recite in class 12 23 12 32 ~05 27 05 31 13. Hits and pushes other children ~23 ~12 ~29 ~20 ~23 ~02 ~40 ~05 14. Hands shake or is nervous when called on to recite ~04 ~09 ~01 O4 16 ~15 ~07 ~08 15. Finds fault with what other children do ~14 ~11 ~16 ~20 ~20 ~17 ~27 ~22 16. Approaches a difficult task with an air of defeatism ~31 ~04 ~29 06 ~21 ~07 ~23 ~05 17. Is considerate of others 13 00 19 12 14 ~02 19 05 18. Fails to carry out tasks ~39 ~17 ~39 ~20 ~30 ~26 ~25 ~28 l9. Lacks ambition to do well in school ~35 ~18 ~33 ~29 ~14 ~25 ~16 ~35 20. Does things to get others angry ~29 00 ~33 ~16 ~30 ~03 ~37 ~09 1(33 APPENDIX F (cont'd) Item - SAT1 SAT2 SAT3 SAT4 ' Boys Girls Boys Girls Boys Girls Boys Girls 21. Will'put up an argument when told not to do something 27 15 21 18 23 23 34 36 22. Does homework 39 ~21 ~45 ~26 ~29 ~29 ~18 ~33 23. Teases other children ~34 ~16 ~34 ~23 ~38 ~09 ~41 ~21 24. Is afraid of making mis- takes 02 ~15 08 ~07 13 ~16 ~08 04 25. Is bossy with other ' . children ~02 ~13 ~03 ~16 ~11 ~10 ~03 ~27 26. Is easily upset by changes ‘ ~10 ~23 ~04 ~10 00 ~21 ~15 ~28 27. Is sure of self 41 31 39 32 20 29 35 29 28. Uses abusive language ~24 ~19 ~25 04 ~25 ~15 ~29 ~09 29. .Has changeable moods ~34 06 ~38 04 ~41 ~04 ~41 ~02 30. Gives in when another ' child insists on doing something another way 10 ~18 09 ~17 21 ~16 08 01 31. Does not respect other people's belongings ~25 ~13 ~25 ~09 ~29 ~07 ~29 ~10 32. Does not forget things which anger her/him ~17 ~08 ~13 08 ~21 ~05 ~23 01 33. Seems to be off in own world ~16 ~09 ~36 ~12 ~16 ~13 ~10 03 34. Is infuriated by any form ' of discipline ~33 01 03 ~32 ~03 ~34 ~13 35. Likes an audience all the time ~36 ~01 ~35 ~14 ~24 ~14 ~26 ~22 36. Finds it hard to study ~45 ~24 ~50 ~25 ~28 ~30 ~36 ~31 37. Has to have everything own way ~13 11 ~10 ~13 ~25 ~01 ~18 00 38. Works well alone 28 19 32 15 16 32 10 26 39. When angry, will refuse to speak to anyone ~10 ~32 ~27 ~19 ~23 ~26 ~17 ~39 40. School performance is far below capabilities ~36 ~18 ~41 ~27 ~14 ~17 ~13 . 02 41. Has no friends ~14 ~02 14 ~08 16 05 22 08 42. Behind at least one school grade due to academic difficulties ~54 ~49 ~54 ~49 ~40 ~39 ~32 ~30 43. Seems dull; slow to catch on ~44 ~31 ~50 ~24 ~37 ~27 ~35 ~30 44. Will not ask questions even when unsure as to how to do . the work ~23 ~5 ~20 02 ~10 00 ~18 02 164 APPENDIX F (cont'd) Item SATl SATZ SAT3 SAT4 Boys Girls Boys Girls Boys Girls Boys Girls- 45. Fights back if another child has been asking for it ~05 03 ~02 18 ~09 12 ~07 ~02 46. Never seems to be still for a moment ‘ -26 ~17 ~19 ~17 ~26 ~22 ~30 ~22 47. Argues with me ~29 ~10 ~28 ~11 ~29 ~18 ~31 ~28 48. Is able to concentrate on things 29 24 39 25 14 35 23 13 49. Boasts of own toughness ~15 ~17 ~13 12 ~16 ~16 ~31 ~13 50. Seems to think of self as worthless 05 ~19 04 ~26 20 ~22 00 28 51. Tries to be the center of attention ~32 05 ~30 04 ~28 ~07 ~33 ~07 52. "Drags feet" when re~ quested to do something ~18 ~25 ~22 ~32 ~02 ~30 ~08 ~47 53. Accepts adult suggestions 06 12 09 35 06 34 ~07 37 54. Sulks when things go wrong ~34 ~34 ~34 ~23 '~36 ~12 ~46 ~27 55. Becomes frightened easily ~09 05 ~08 O4 07 ~07 ~07 ~01 56. Resents even the most gentle criticism of work ~11 12 ~06 ~01 ~08 03 ~14 04 57. Distractible; can't con- centrate ~33 ~37 ~36 ~41 ~17 ~35 ~14 ~26 58. Able to see the bright side of things 01 ~07 ~02 ~10 02 ~09 16 ~12 59. Fights with smaller children ~04 ~01 ~09 12 ~04 05 ~33 04 60. Spelling performance at least one grade level below age expectation ~57 ~38 ~63 ~37 ~41 ~26 ~40 ~42 61. Fearful of being hurt at play ~01 07 O8 25 13 18 08 21 62. Is stubborn ~30 ~08 ~29 ~06 ~25 ~02 ~29 ~15 63. Never speaks up even when there is cause to be angry ~28 12 ~32 03 ~16 00 ~18 05 64. Is interested in schoolwork 42 21 39 32 27 27 26 36 65. Tries to get other children in trouble 05 00 ~09 ~18 01 ~11 02 ~09 66. Does things which are normal for children much younger ~30 ~23 ~29 ~21 ~34 ~28 ~42 ~36 67. Never fights back, even if someone hits or pushes first 02 ~04 ~03 ~02 _~05 ~09 ~07 ~09 68. Prefers to be around adults rather than play with children 13 ~04 ~01 ~09 11 ~01 O9 10 165 APPENDIX F (cont'd) Item SATl SAT2 SAT3 SAT4 Boys Girls Boys Girls Boys Girls Boys Girls 69. Is popular with his classmates 28 25 37 39 17 34 25 19 70. Does things which are normal for children much younger ~33 ~16 ~28 ~32 ~27 ~32 ~36 ~24 71. Never sticks up for self when picked on by other children 22 ~11 15 ~08 22 .~08 21 ~10 72. Threatens to hurt other ' children when angry ~25 ~06 ~32 ~09 ~32 03 ~47 ~02 73. Average or above IQ (Intelligence Quotient) 41 54 46 41 41 51 53 47 74. Does not take orders when other children are in charge ~20 01 ~18 ~09 ~21 ~06 ~27 ~17 75. Prefers to be alone and play alone 28 ~12 18 ~17 22 ~15 24 14 76. Finishes classroom ' assignments 30 17 28 20 08 26 20 28 77. Gives other children dirty looks -21 ~09 ~22 ~16 ~26 ~17 ~36 ~29 78. Deliberately interrupts what is going on by asking silly questions ~35 01 ~32 ~02 ~32 05 ~38 03 79. Slow in making friends ~20 ~33 ~33 ~39 ~14 ~33 ~16 ~18 80. Seems as happy as most children 15 10 20 23 01 14 12 28 81. Finds fault with instruc- tions given by adults ~04 ~01 ~09 ~06 ~10 ~22 ~30 ~19 82. Seems unconcerned when misbehaving ~20 ~17 ~20 ~26 ~14 ~27 ~28 ~29 83. Cries easily 12 05 19 O7 16 02 04 ~02 84. Is afraid of strange adults 08 ~01 ~01 03 04 ~03 ~12 03 85. Is self-confident 13 20 17 16 O3 22 04 05 86. When angry, will do things like slamming the door or banging the desk ~09 ~34 ~11 ~26 ~06 ~24 ~21 ~38 87. Acts in a ”dare—devil", fearless manner ~26 19 ~27 10 ~20 17 ~30 11 88. Has difficulty speaking clearly when excited or upset . ~04 ~28 ~04 ~29 08 ~09 ~05 ~19 89. Has a "chip on shoulder" ~14 11 ~23 00 ~21 11 ~33 13 166 APPENDIX F (cont'd) I t em SAT 1 SAT 2 SAT 3 SAT 4 Boys Girls Boys Girls Boys Girls Boys Girls 90. Becomes embarrassed easily ~15 ~17 ~08 ~20 ~05 ~16 ~15 O7 91. Bright, but doesn't apply self (underachiever) ~14 ~06 ~16 ~17 ~04 ~13 ~11 ~22 92. Disturbs other children with boisterous behavior ~22 01 ~20 01 ~19 ~04 ~30 ~06 93. Behind at least two school grades due to academic difficulties ~29 ~21 ~27 ~20 ~01 ~10 ~17 ~17 94. Arithmetic skill at least one grade level below age expectation ~41 ~38 ~49 ~35 ~21 ~38 ~26 ~46 95. Much anxiety~~afraid of such things as storms, school, death, injury, was (con- siderated phobic) ~03 ~05 ~09 ~08 12 ~03 01 00 96. Frequent headaches, stomach aches or other non-specific physical complaints 11 ~13 03 ~09 21 01 ~13 ~20 167 APPENDIX G Relative Frequency of Student Behavior Attributes Based on Teachers' Ratings From the School Behavior Checklist Item No Yes 1. Is friendly 2% 98% 2. Tends to give up if something hard to finish 65 35 3. Interrupts whoever is speaking 73 27 4. Penmanship at least one grade level below age expectation 78 22 5. Starts fighting over nothing 88 12 6. Is a helpful child 7 93 7. Is alert in class 23 77 8. Poorly coordinated when doing things with hands, such as coloring or pencil work 88 12 9. Reading ability at least one grade level below age expectation 68 32 10. Just stands around on the playground 96 4 11. Acts up when adults not watching 70 30 12. Volunteers to recite in class 19 81 13. Hits and pushes other children 78 22 14. Hands shake or is nervous when called on to recite 89 11 15. Finds fault with what other children do 73 27 16. Approaches a difficult task with an air of defeatism 85 15 17. Is considerate of others 19 81 18. Fails to carry out tasks 74 26 19. Lacks ambition to do well in school 77 23 20. Does things to get others angry 77 23 21. Will put up an argument when told not to do something 72 28 22. Does homework 24 76 23. Teases other children 72 28 24. Is afraid of making mistakes 79 21 25. Is bossy with other children 82 18 26. Is easily upset by changes 81 19 27. Is sure of self 38 62 28. Uses abusive language toward other children 87 13 29. Has changeable moods 57 43 30. Gives in when another child insists on doing something another way 55 45 31. Does not respect other people's belongings 92 8 32. Does not forget things which anger her/him 91 9 168 APPENDIX G (cont'd) Item No Yes 33. Seems to be off in own world 83 17 34. Is infuriated by any form of discipline 87 13 35. Likes an audience all the time 86 14 36. Finds it hard to study 74 26 37. Has to have everything own way 87 13 38. Works well alone 36 64 39. When angry, will refuse to speak to anyone 87 13 40. School performance is far below capabilities 80 20 41. Has no friends 99 1 42. Behind at least one school grade due to academic difficulties 75 25 43. Seems dull; slow to catch on 88 12 44. Will not ask questions even when unsure as to how to do the work 91 9 45. Fights back if another child has been asking for it 30 7O 46. Never seems to be still for a moment 79 21 47. Argues with me 70 30 48. Is able to concentrate on things 22 78 49. Boasts of own toughness 88 12 50. Seems to think of self as worthless 92 8 51. Tries to be the center of attention 78 22 52. "Drags feet" when requested to do something 74 26 53. Accepts adult suggestions 19 81 54. Sulks when things go wrong 65 35 55. Becomes frightened easily 96 4 56. Resents even the most gentle criticism of work 93 7 57. Distractible; can't concentrate 71 29 58. Able to see the bright side of things 6 94 59. Fights with smaller children 96 4 60. Spelling performance at least one grade level below age expectation 68 32 61. Fearful of being hurt at play 91 9 62. Is stubborn 62 38 63. Never speaks up even when there is cause to be angry 94 6 64. Is interested in schoolwork 15 85 65. Tries to get other children into trouble 92 8 66. Does things just to attract attention 74 26 67. Never fights back, even if someone hits or pushes firts 93 7 68. Prefers to be around adults, rather than play with children 94 6 169 APPENDIX C (cont'd) Item No Yes 69. Is popular with his classmates 18 82 70. Does things which are normal for children much younger 90 10 71. Never sticks up for self when picked on by other children 94 6 72. Threatens to hurt other children when angry 76 24 73. Average or above IQ 28 72 74. Does not take orders when other children are in charge 83 17 75. Prefers to be alone and play alone 91 9 76. Finishes classroom assignments 26 74 77. Gives other children dirty looks 72 28 78. Deliberately interrupts what is going on by asking silly questions 88 12 79. Slow in making friends 82 18 80. Seems as happy as most children 12 88 81. Finds fault with instructions given by adults 87 13 82. Seems unconcerned when misbehaving 78 22 83. Cries easily 91 9 84. Is afraid of strange adults 94 6 85. Is self-confident 22 78 86. When angry, will do things like slamming the door or banging the desk 84 16 87. Acts in a "dare—devil", fearless manner 86 14 88. Has difficulty speaking clearly when excited or upset 8O 20 89. Has a "chip on shoulder" 90 10 90. Becomes embarrassed easily 72 28 91. Bright, but doesn't apply self 78 22 92. Disturbs other children with boisterous behavior 88 12 93. Behind at least two school grades due to academic difficulties 89 11 94. Arithmetic skill at least one grade level below age expectation 69 31 95. Much anxiety—~afraid of such things as storms, school, death, injury, war 97 3 96. Frequent headaches, stomach aches or other non—specific physical complaints 93 7 PLEASE NOTE: Copyrighted materials in this document have not been filmed at the request of the author. They are available for consultation, however, in the author’s university library. These consist of pages: Appendix H, Pages 170-171 (School Behavior Checklist.) University Microfilms Inte rnational 300 N Zeeb Rd., Ann Arbor, MI 48106 (313) 761 -4700 _L 10. 12. 13. 14. 15. 16. 17. 18. 170 APPENDIX H The School Behavior Checklist — Form A2 School Behavior Checklist by Lovlck c. Mlller, Ph.D. Form A2 Published by Ages 7-13 WESTEIN PSYCHOLOGICAL SERVICE PUILISH E15 AND DISTIIIUYOIS ‘\ 12031 WIBHIRE IOULEVAID 10$ ANGELES. CALIFORNIA arms A DIVISION or MANSON WESTERN CORPORATION DIRECTIONS The statements on this checklist are often used by teachers to describe children's behavior. Read each statement and decide whether it describes the child being rated. If it does, blacken the space for that item under "T" on the separate answer sheet; if not. blacken the space under "F" on the answer sheet. It is important to mark EACH statement. If you are in doubt. blacken the answer that Is most characteristic of the child. STATEMENTS . Is friendly 20. Does things to get others angry . Tends to give up if has something hard to finish 21 Will put up an argument when told not to do something . Interrupts whoever is speaking 22- 0°63 homework . Penmanship (handwriting) at least one grade level below 23- Teases other children age expectation 24. Is alraid of making mistakes ‘ Starts fighting over nothing 25. Is bossy with other children - '5 a “9'9"“ Child 26. Is easily upset by changes . ls alert in class 27. Is sure of self . Poorly coordinated when using hands. such as in coloring or 28. Uses abusive language toward other children pencil work 29. Has changeable moods . Readlng ability at least one grade level below age expectation ‘ _ _ . 30. Gives in when another Chlld Insists on domg something Just stands around on the playground 30011161 way . Acts up when adults not watching 31. Does not respect other people's belongings Volunteers to recite in class 32. Does not forget thlngs which anger her/ him Hits and pushes other children 33' Seems '0 be 0" 'n own world Hands shake or is nervous when called on to recite 34' ls Infuriated by any form 0' dlsc:pllne Finds fault with what other children do 35' L'kes a” “Ud'ence 8" "‘9 "me 36. Finds it hard to stud Approaches a difficult task with an air of deleatism y g 37. Has to have everything own way Is consuderate of others _ 38. Works well alone Fails to carry out tasks (homework assngnments. seat work. 810-) 39. When angry. will refuse to speak to anyone . Lacks ambition to do well in school 40. School performance is far below capabilities Copyright 9 1977 by WESTERN PSYCHOLOGICAL SERVICES Not to be reproduced in whole or in pan without written permussmn 01 Western Psychologncal Services. All rights reserved. 1 W-l 44A 23456789 Printed In USA. 41 . 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61 . 62. 63. 64. 65. 66. 67. 68. 69. 171 APPENDIX H (cont'd) Has no friends . I Behind at least one school grade due to academic difficulties Seems dull: slow to catch on Will not ask questions even when unsure as to how to do the work Fights back if another child has been asking for it Never seems to be still for a moment Argues with me is able to concentrate on things Boasts of own toughness Seems to think of self as worthless Tries to be the center of attention "Drags feet" when requested to do something Accepts suggestions of adults Sulks when things go wrong Becomes frightened easIly Resents even the most gentle criticism of work Distractible; can't concentrate Able to see the bright side of things Fights with smaller children Spelling performance at least one grade level below age expectation Fearful of being hurt at play Is stubborn Never speaks up even when there is cause to be angry Is interested in schoolwork Tries to get other children into trouble Does things just to attract attention Never fights back. even if someone hits or pushes first Prefers to be around adults. rather than play with childern ls popular with classmates 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81 . 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. Does things that are normal for children much younger Never sticks up for self when picked on. by other children When angry. threatens to hurt other children IQ (Intelligence Quotient) average or above Does not take orders when other children are in charge Prefers to be alone and play alone Finishes classroom assignments Gives other children dirty looks Deliberately interrupts what is going on by asking silly quesfions Slow in making friends Seems as happy as most children Finds fault with instructions given by adults Seems unconcerned when misbehaving Cries easily is afraid of strange adults is self-confident When angry. will do things like slamming the door or bang- ing the desk Acts in a “dare-devil." fearless manner Has difficulty speaking clearly when excited or upset Has a ”chip on shoulder" Becomes embarrassed easily Bright. but doesn't apply self (underachiever) Disturbs other children with boisterous behavior Behind at least two school grades due to academic difficulties Arithmetic skill at least one grade level below age expec- tation Unusually afraid of such things as storms. school. death. injury (phobic) Frequent headaches. stomach aches or other non-specific physical complaints 172 APPENDIX I Permission Forms for Release of Information MICHIGAN STATE UNIVERSITY EAST LANSING . nacmcm 43m come: on moan-row - nucstN ruin. - July 15, 1981 Dear Parent(s): In the fall Of 1977 your Child's classroom.at Maple Grove Elementary School participated in a study conducted by Midhigan State University and the Lansing School District. The purpose of this study was to develop a better understanding Of how Children learn to read. Parental consent fOrms were sent to your home at that time, and you allowed us to obtain infOrmation regarding your child's progress in reading. we would like to fOllowaup on our earlier findings. To do this, we need your consent to Obtain infbrmation regarding your child's current progress in reading as well as his/her growth in reading over the past fOur years. This irdtunetion has already been Obtained by the Lansing School District. In making use Of this infOrmation anonymity will be protected: your child will not be identified by name in.any aspect Of the study. If you approve of our use of this infOrmation, please Sign this letter and return it by July 30, 1981 in the stamped self-addressed envelope. Sincerely, Robert L. Jarvis Research Assistant Richard S. Prawat Assosiate Professor Senior Research Assistant I approve of releasing my child's reading achievement infbmuetion. Parent(s): Child: 173 APPENDIX I (cont'd) MICHIGAN STATE UNIVERSITY wrunsmc~manom 48825 COT-LEG! 0! EDUCATION ; BUCKSON HALL July 29, 1981 Dear Parent(s): Within the past two weeks you received a letter requesting release of your child's reading achievement scores front the Lansing School District for purposes Of a follow-up study in conjunction with Michigan State University. Because this is a follow-up study, much of its practical importance depends upon gaining current reading achievement information on as many of the stu- dents in the original 1977 study as possible. If you have already returned your permission form, releasing your child's reading scores, we sincerely appreciate your cooperation and concern in helping us better understand those factors which contribute to reading suc- cess or failure among students in Lansing. If you have not as yet returned this form, we would again like to encourage your participation in this mean- ingful study by completing and returning the enclosed form. We stress that your child will not be identified by name in any aspect of the study and the confidentiality of your child's name and test scores will be strictly guarded. We again thank you for allowing your child's participation in our original 1977 study and encourage your support in our current efforts. Sincerely , Robert L. Jarvis Research Assistant _ - _ Richard S. Prawat Associate Professor I approve of releasing my child's reading achievement information Parent(s): Child : M'illlilllilillilllfillilililllllllilllis