\q a'rloé}:.oa:a ‘3. i a. 0... E .1... 0. . .5 0... .0... o a, 0.. . . I . 1 . I’m ...: . tulip: .. 1.3.0.4.... ..hflzi.33..h..0§3.flffl~l.t.fiffig .0... ...190333 .r ll...r..l...t .. I .....S..t.i...x..? 6...... 70.... 0. ’3‘. d...u...0(.0.‘|...33.0 .0. .i.£.l00.l...n... v.3...“ .. L0: .. . ...... .830. l ...-1. .30 .... ...-.11 2.2.0. .. 103.. . . .9... .0 00A." .1.........0!.¢‘...0XE.000 -..).020. ... ... .00 . .0 :2, .9 ......vd. -0392... 02.0.0.0: Ill ..0 v0. 0...... .0. .. 0.! I 00.3.... . \.0.‘.-0¢§.000. . 1.0”0 00‘ z! 30 ....."0... ...... . J...’0.’JSI.0§0. . 04‘ . . .0. . r ...0 ...... 0" . 0.00....Io0. . ...0 .. 0 ...: 7).! q“ ‘ . .. .... . _. . . . ... . 3 .. .. LU» ...v... .. 0O 9 #00 “an“. . . :01... 1. .0 . u. 270“. ...... i.”.3.~.ul.tl 0mm; u.M. (my ... 0 .0 .. .. .... 4.. , .J . I. 0, .- . . 3 - .. . ... ...... 2.9.3. 3...... .... .........4..i..:..¢.~. . ......‘i‘. .03 \..0o...q~. 0.1.2.0., [I . S 3 0 ,0. 0 0"}... . ’30... Q (1:400... in. J .0 0.0 . u... ......-...f...... $2.13!... 0......01: 0...}... i “.0 I l a . n . . .0. 3......- . .206! .03....‘0307 0.008500... ..l....._.0{3.. .. . ...-...voshufi... . 0.0.0.30. 1 3.3.0. 63.416... 1201......00. 1...... V . . . ... . . . . I ...3 ....l... s.........:..0...o.l.1...: 2...... 121.3132}... 0 ... :0 o.....05.0.....0.0.......0....:. . ....O 1 . ...-.... .... .c 3. .8. . .0.. ..n ..Rnlznuhhuflde ......“ .. «....M.....I3.5...:..01. 2N. 2 ”run-... .hfir .31 r. L I w I. . . . . I O - 0}. . . .112, ... b.,.3...‘.“x.l..-t. . 3.0.”..«2 . . 20.00...) . 3,. . .01....1... ... .o . . «0213.... 0.30.00... 0’0...‘.) 0:110:00. O0..0.0.'.:.o..00.01 .‘Il (‘23,.03000 1 5!..1. :..7..?. 1.00:} ...... 00.0-30.0 - .o’. .0. A ..‘l...¢..3¢.c"...= 00.0... 3.2.0. A30 ..-?! It; . 3.... 3 213...: 0.30.0 10.2.13...- 300 03.0.3.0... 00 30'. It. 0.06.04 . I 0...} 3. 030........‘Vl.-I"i‘g:.0d0.i 13:00... .30.;000‘ g0 . 2.00.0.3: .03..N.0I.'0.0V....00?‘.0.0..0_ 0.3....20‘30 Q... ...... ..“...ao0.-0..0.....0 . . . L .. ...: 13.00- J . 0.~H.Q..,0}. 0. .01. v .. 0‘00.0d! 0...... ”.003... tot—[$30.30; - 0 ‘Qib. '0 0.0.0”....:. 0Ou._‘.0:.00.§... . ..avvat ‘I‘. c . .0._I...:...0.’.. . ... o. 00 .. t. a ...-4 . ......n. .....r0 . .00. .000 . 0...... 0 0. n 00.1‘.‘ . 3.2.0.0. I0... ..00..J 0.23.0.1 3.0: 0. $3... ...! .0. 3... 0.V..3... ......a.’ .0. ...... 00.26.1111 2' ... .....7. J. 3... .31.. 0 no .3. ....I 3.0.03.0. 00 {0.01 0.00.30... 20.... 000.... . ...... ....II ...-9...}... 3:30:00... ....00..-2 n. . .0. . ...... . ... . .. ... . ... ...... . ...... .33... 0.3....) ....iu.§1u.t. 0 O O. .0 ...: nit-.0. .‘ .....0. ..a i .....:.... 0?. 3:001:83...- 0.09.15]... .... otond .. .400... 5-3.0...le 0. .00 on: .. I070... .....n. 1.00.03.9030000001030700 '00.. [1.030. . ., . . . . . . . . i . . 00.. ( ...-.... ....- .Io... 0.0.....0.I.—.04 ... 00.04... . . ......JNA! . . . . . , .. 0.0 100.50. ...”.L o 0 ‘33....31. 0. . . ...1 .... 2.15.0. ». . . ... 5.....1‘. ...... 0... 00......0..C: . . . ....V...... .4. ...s. ... ..- . .0. . ..00 a 009...... .0. . u. .32: ...... 3.. .. . . 00 1.0.3 .10. f... A. .. . ....0 ... 3.00.2.0... . . . . an. .03....- 0 . i . .1. ..‘1 . .0. , . .. ‘- 03.00... . . ... .. . C...‘..,-.~.....0u.0.03. .00. on... . 3 0 ..l s... . .. .1». .v .0; .00 .o.....o 3.T..5v0f..000‘..l... 0.054....— . 1...: 0.51.... 80.11....2... ..I.I:..:.c0.3..n.!o..¢ in. ... ...-“100... 0i.‘M ...... [0‘ . . I. . I. . ...-0.... 2'. o. .“0! 01C. ...: J: 0‘ “cu-$.03 v.0..roovo.“ - ."0"... .I‘.f0..3%-u0a .0”... “flu-Pfiuuu . 0.00.00 . . {11.1134 .0..l.0.....l..u..,.‘lo . I‘M” 3.3. .vpulto.:-0...?..0-00.IQ . u. . vi.$.|0.030.n.:.’0..1.0. 00.04”: . a. . .h ..I.it0§....0.=:;v I....‘|I. 0.4.01gs:lr.l¢'&$ ..u.‘ o ”0.00. .0... . '00....3‘0000300 000...! 30.0.32... .... .1... .0000. 200.. 0a.... - .q‘ .3 ..iioo'ggcf03. z... .0000 I. 0......00. 1.0 .0'.o0.....0.9...0. J....o.00..0.. ' .0... .0. ‘00 d. 3.009.303.0030 .01 1‘20 .30....0003 00.. 0.0.0.3030...‘ ... ...-.....000‘I230 0 «.00.. 0.2.0.0..-. .0 . .0}... 0.... 020' ...-.2 ‘1'? ...: 0.3:...00010‘ .0... 10. 1.0.... ... , {0.43 33.1880... .. . . 90.0... . 0...: I . .0. 3.2.0.. in .. ,. ... 330.. 3. . o............... \00 .1..001.‘.0...rl 00.... .v. ... ..IO.Pua.da...~ ...-0013.....00r3|(2.0.000. .01.] .. 20.......‘01300.,l..n.3‘- .... .3. ...-0'9! .0- .0... .0, 0.. .00..) . :1... 0......0’2 03:60. :03. 3.... ‘1’ {000 . 0.000.3'002I. 5000'. 00010.6..! 0. 00...! 50.0.... 0 .0 02.001... 3000.... 0‘. a 1 .- .3.0.0..fo -300... 0.. . ... 00.. ’ l 0 . . .....00...... . v. .0. ..0. . .. 3.0-2.0. H’3’1 ..A0100.-.70000”.—n. ...”.Ioilnbatillib‘il izaloiznino... lizi.0’. ~.0.I‘..o0......o..l-..Nn.. . 0%.}..00. 33...... .0. act}... 9 .1.\.....-..0.....|.I..I,.‘|..... L... 90 10.0... .0... u 00:... ... .. ... .X' . 10.30.. . 3!. .0. ...-3.0. .0...00......__:, ......13. .11. its. 7...}! .40.. .22.,»- X0. .00 3...... 0'.0tll....i500.‘0:.0".00.. .1119000. .... 0.. ...00..."\. 5.0.0.".c01.03....3'1300....-53......51300 0 .10. ’00.... 0’200'201 .... . or... .... 71.0.0.8... 53...... 0.! .0. .... ...-100 .....2......,.......0.I. :0 Oil-... ...}..ch .‘VI‘.NV ...» ....030....0. 45114.. 30.0. .0 .2. 5.2.0.; 0.”; 38:1}..- .0..§80....0.. 0... 3.. ......lv‘a..0; 003.03.933.510230. . .- .. :1 o .‘In ..\.l..0.0.l I... ... 00.12....- ,..0 1.2.2.39? 3.13.43.28.33...- unfi .. E. ‘a {01...}. ’35:... 35-...(0 .0! . .éini A 0. 0.0.0.0.. .0010... 0.0 ..‘0 0.00.00.70.00'0 0.0.00“ 30...}... in. .00 I... 03.0.0- .0”. . .. 1.0..1’0... .03."...0213135120.‘.... 034 A0. . “0. 51.0.2‘ .1 ...-$10.00.". . 0” no! 31.860013l0... 3.072.103: .. u...tla$.011..10!‘:.1. 3.31100..’.... .0 ....0 ...-5300......zt‘ 0.... 3.0 .. . . . . 1. :0. .l........!........~..1.o.. t .00. ...... 01...“. .... ...03.3.2.3...3120131... . 40.0. 1 0n». 90.1 -. .0...‘.00..301.n0.8vll. 40..........0....l..9¢l. 0310......1389- _ . . . . . .... ..- I .1... .0. .... 38...... .....0. ...v. .... . .10.... 0.11.383: .....ttllun..§s4~ u .. v. .00 .13....0: ‘0‘! 0.. $.00... 0.0....I3upu3zo‘o... , . . . ., . . ... ...-31. ...-...... .0... 2...... .0..-..0..0... . -.....u......0.. ....‘afléq.§..lfiu . n... f, o .. . ... '30....00‘ .3'0000010 ' . u «.00. o 0 .09... OX...’..~...O.0-.. ..3.§.th0‘0. .....00... 1.0 ... 5 I...“ ..nnJJ . ... 0.1.1. ‘0 ...-0 .t. 30.0.0.0-4006... , .0 .0. .0009”? .000... on“: "u.. 33.3....0...’ 1.0“...) 0. . “Luv-00. 2.100 3.3.“. o v 0 1...... 0 0 a... 40.09.... .....0 .0... 00’ . .0..'0. Av...- ‘n.0..0 .0 a. . . I . . c I .0 .00... .01. .o.’ 0a‘Q’Q-I‘I0nu.td.il..w”m.3.hz .0 a . . . . . . . . . . . . . , 0.- 9.. .- oa’d‘..0c0. .u00 l.....7.w0flna.0.mw. 1, 2.. ... l 000. a... .Q \c of} .. 40‘00. . 1...... :0. . . . . . ...00...0......0. 200-... ......o‘ .110. ...-09"}...1 87ft." .020...) 0.0.7....‘3?’ 0.. 30 01.3.... .... . .. .... .. 0.0:... 2...... In... 0. . 30.1,. ...X‘c...3.a.0. To... 0-... I. 0. ...0 6000.33...“ . . . . . . . . 0000.... 0.0.I . ...-..tQ-‘acu .. 0.. . . . ..0002‘300“ .0 . u. 00». a .1 . . . . f . .. . . ., . . . . . ... 1.0.0.. :1 v . ...0 L009?! .09.... p.... . . .Q .. .... 35.01%}! .....3 u. i . . . . . . . .. . . .. . _ . . . . ...: . ......00. .... 1.2.... . .....Iin.‘ . .l’. 9... .0 .. 0.. 'ca’a. . .0.- . .— .. I . . . y . . . . .... ... ....00 ..,..000 ... 000‘..- o..0c.. .0000 . 0 0...A0. .‘x‘ififié o 0‘0 00.00 .1.‘..8\. 0. J ...... 1.301.“..0: . 0.0. .33... . ...-I. .. {...-......”nugqbd. 00.0 .90.... .0. :0. I 0.0... . .200... ....- I.30v3 . . . .... ...- uonlov...0 «.u. 3‘...'...0.fl50.\.....¢a‘ co... .irio‘..‘. 0. .. 0..- ...IV0 0...... . . ..--.90}. .0;...0 0.0.”..3I‘0- J . .£“~.000.'I ‘1 0“). l’ 0’ .‘g00.\. . It... 03...... 3.13.233... ....t....al.....1.l.. 2.3.3.. ....0 .33.... . 4 . . . ... ...... .. €1.48... .... 4...... ...... .3531}... .. ...: '31....0 .0» .1 .8. V .. 1.1-. ....lfi....00.0.l...v0.12!....7|£0.12... .00.... 0003.....00051 0.40...’o..0.03.03‘fi. ......0la 0.. ...-1...... 3... 0...... ... 00...... . . . .. 0...... .. 3.. ...... .....0 tin. u. . . ..7.00....... ....8...l....o.t. Iflbs .\ Ion... ... 0. 5...? . . 113.0 14 . 231.0630 ......3.‘ .... 07.0.0. 00 ....l....l€.l¢.d.¢l11!0 ». ......Sol. 31.100 . 30.0.0... 2 .vl....u.... ... 3 «0.0.... ...Y... . .... IE. 0.. ...0... 0.0.0. ...... ..J...2.0..0....0...3.“....1m. 900830.“ Eiflg. t. .0 v . ..3...§.0’!...v. i-$)l~.3.u0o...0..0o.’ q 01 .n .. .01 11' 4 . .. 00‘0 7.3L .00. 0:! . 00 0‘..| .....3. ...—...... u. . . ... . .0 .0 u 0. ‘10Y.!. :0. 6.00%! 5‘ .3 0 cc... , M0 ., ...... ...-00...! ...3 1.0.30. ...-0 :1... 0.2.5.1.... 300...}. 1.0.0. .f...’!...\...... .0. ‘30 ....Ilo 2.3.09... 1:0..‘93 0.0... 0. .....93903... . .. 0. $2.23....3.0.2....013212. ..n V.. 3.634.108. 7 ”’00 $30.4.w. A. . . 0.1.3. .10 3.3.1.... 3.005.230.3300}...015.331.01.331. . .00.... .0 . .85... ...... 1....8.0-n..!.l....:.o.... . . .... ...: 8......0: ... . . ... 0.0.0. 01.1230 . ..3. 1... ... ....l: .c . Cu .3 30.... H0... : 00 . .....00: . ...:\\.. . ... .0.’l...vl..>.il¢.)...1.. .. . .o. O... 8.0. 0 63.1 ....-l.....li0...~0’.0. (I. .....0. 2|... .3 30... ...-.2... 0.... . ...... U. 300.! .. .... .. ..1 .2...- ... ..A. 37.0. 3.20.: 0130.. I .1001 v 0. .33....199...‘ . a. ...-‘03.... . ...l-og .2 3 0.7.2....- v”. 70.3 3.11.1003}. ,...ol...£.lI-.l?.. 3.0....- Qu’ . $033.01 .00. ’l!2.21’...0l’.310‘...002!. 0.33.2.1... 0......) nio-fii. . . 2.133-... .... - .3: 0.. . .010. ...: V... t v . .. 3.0!... . c. ...Q. . ... ...\...0.0. ...-t. 0...... ...... ...-100‘...0 0...... .... .01 .. .310 ‘ 0.00....u30xlob: 5.. 003.9,. .0. .0... ..-.'000.0...... 0.0.40... 0.... 00.0 ...-... 3:20.116... , ...-.... ... ‘..0a'..1.... .0: . 5.301. 01?... .....n ‘0. ......30 7.0.1:... 00‘ 00000 ... 0.00.3... l..- . 0.0: 30.0.: 0‘00'0300’00001". 0.0.)... . C.I.\...0¢I‘.0¢‘ 3:00.00 i800.-. v00...0..vd.00.0" .J...0000.0.00"o "O'..00.0........0 .000... . . . . .. . ..0... 0. ...-............i...'000‘.... . 3.0,..- I0..00;. O...00.l 0-00.41... .140“, . 3. ll 3.0.0.8.... .: 331:1...) 3.8.3....sll: ...... \ 025......3‘A..‘.00...l.0 . . .0... ...-......3. 0.111801030D0I...A-cllv.v2;..\. 000.00.... ._ .30 1.0.... ..n .... ...... .... . . . . . 30......I . .13.... v. ... 3.3.0.0.. q.-.\..ln......n;«uu.I0-0. $10.00.. 300...... ...-5.0. a. 00 \ 30100.0... ... ll. 00.09.03.028!" 10.0. 31.0. 0.30.04.31.2130 ’19. 1. 3.01%.. ...-.0... 3.. .030. 3.3.3.0200. 00.0.2.0 .v‘. 0.1 c ‘32,. .. : 0...J.‘00. 4.3.9241. 0:311... l......0...?. ...!f...0£....t. ...... . 0. lo. .1... ... ..l 1.0.0.. :0. ... _ 11 4 h .l. -..-33.0. 3.3:...” .000. 3", 0.. l 0‘ In . 0......3A ...! 0.0.0.5....‘3 '0'! 00. 0.03.0013!) .. 00.90.117.903... 1.0"}..:20 I... 0.0.0.0....0u...l..¢ 0.. .0050. 00.50.. .00... 00!. .I......3’0~3a a0. ......1’..0.0c.00.9 .0 0.8.0.000 00.1.0 . $0.21... 3.30.0. ... .11.... .’. 3030......10... ....o .a ...00 .00...) .0409! ‘.~..3'.37‘2“ 1...... ...-‘0‘. 0n he O.....:0.t-Ill:.. .....0...‘...ut.~.!l.. 9... ~ 05.23.0023? ...:...Q..32......:1ll.c..0...7 1... 01.5.0.0...0333 .39... 03.08.00. ...... .0..833....i..l:. . 25:9... .... 1.. 1.20.3116... .23.. ..3. u . . a. > L - .... .vo.....:...:s;0.0..: 3.....0 . 0......_3....f..|0...¢. .00 ... . ......ikf .. .... ll. . .... . 0. .)......-..0..0....¢.0..£21.33 cu..- ot....l.0..0.s1.\0.0l.v....000f. £508.... 9. 0... 0. .3...0..'0.. )3..0.....00l~0.....-.Q,I:. 00...! 010i... 0... A ... ..v... 10.3.0.00Il.‘.00.- 1...... ......0‘..o:.. .13... ‘00. 12000 0-,. 0.230....0. . . .9. .. .J. )0 30.03.60 (-0.3... . ..223...t.-v00...0.....li§.«.oo,0 , I o..l..”.’.fo50 3...). .0t..0.....s.!.x.ll.03.3. }-.'3....0.I..Jt..: 0. l...i....o.\!0.....“ I! ...00 3...... . .1090... 19.0.01 ...... ...... .... . . 120.12.... 3. . (a... 3'0. . .. .03...J. 0... - 9.9.3.... . 420.1... 3.30.383;......nit....lo.:....o...,.;z .... .....D....‘0....S..l.. . 22101.. .... 20g .0... . .I. 4 ..A 4 .q . D . I... .4 vb» 3.033....)(0'120338 . ...! ...-’0... 5.03.0 a..cifa.7‘\...i. .06 02.0.1.1... 3‘05: 5o‘0 10.3.. ..I......-..l.¢ 00.2.1.0... ...: l..........0.0~......310.l. 30.3.. . 0...:3. . . . o4..... ...! .5513}? .2103. . .0 ~03. 4.002....rf-0. .10.. 7o?!. ...... 34:100.... {30.10.900.310}. . 0.0. .... 5...... .ioig.t.l0 .. ...-02.4.11 $33.00.!!! 9....0...‘ .....339. .1... 0...! ca...- ...-.31.]! 23.32.... 0 ......0... 3.00.30...- ....8 ... ... ......) . .1 i3... 2.0. 0.101203. ...... .703... . .J.h.0‘w.v...".... .... U. . ... .... .39. .03.. ...l . £10.... 10». o ......o ... (...-.2105... 3.0x! .0 .... 3......I0f.f . I. 003.00.000.2O...11..:3o 20... I......I . I00 . l. . 1 .III. 0.3!;33.‘:..J! u ..19 .....1. 1. 1 l p 310.. . 0.2.0. - 10...... . ...DJ ....9 . .0211... 0! .v' .u. “4v... . .....31. ......0. .5. 300.... K. 5 0.3.1 0. ....0 «- .22....10 . 3.3.... .... :1... o.....o...l....ll 0.0... 0.3.9...1.2 .01.. ‘10.-..0. ... .131203)..\0.....t.. 0.30 a 4n»....( 13......313 0.. t..’....-.l.0.. ...SS...I..... 3.1.... 8 ... . .39....0ci. 0.... 0.... ...-...... ...... ...... .. - - \ ...-28.3.... .... 2.0.. . . . . .... .33....100. 270......5. 3).... .20... . a... .' 0.19.0.3}... ...-0} ......o- .§'.l...:...-I:..0'3.4 .3001! -3070....O.0O\0.. ‘il...’ A...o00.‘.-.- ‘20. 0.... ......I!’Y!..... 0.3.. .. €5.00... .a 00.2 00.32;... '0 ... ..0..u..00.'0..30-0.‘0.0.0.7 1.11.01... .0 0. ...l .lua...:..0.£... .0 . 0.... 0.. . . . ...-6......1391... 02‘ 0.70. ..5‘0..0....0‘._..-.0‘¢ £07.70. . o c 500 ca ‘.JJJ~M§’\.:I..2§ 03...! 9 o 0 CI... 0-... 0 00...}. ..1:.¢1...00_.00 710...... 3.0... 0. a... 1709.30.02. 3.0.. . .230. l0..00.3..o.0‘.l.. 2330.80.93.13 .. S0..."0.0.....o.90. .. ... 1...}...‘0... .... ...0 0.3 2,... . . . . . .011.?».00n0...08..0’.0....-1.075... 0\. .30. .... .v u... 0-. 0.0.)...3110. .. 0‘....J¢0.op0£.$.uagdv.;’l;0fl » I}! p {8312... ...... O- ....0 . ,. . . .. . . .. . 0-01.... {...-3.1.3.3.. 60.. 4 ...... 30.0 4...... .... . . . . . . . ...:10... . 330.08.323. .9. c II...- 1‘3..$1 ......?.~fl...¢."1¢.0000003. .2 a. .. ... .....u...0l.0.£ 5}: 2.-.,318. ......O. J. 390......30032.’ \.I00.2 30...... .03....0... .2133... 88.03... ... . ...) 3103.....0... . . . . . . . .... .10... ..0...I!o..0...- .o 2.: .... . ....0... .. . _ u. 3' 0.0.3.123... ... 0 53.1.6.2...(2v... .23.... .1... 10300.... ...... 01031.13... b b. ...... r33... ... 8-0:...0533..0... 00.0... .000 n. 0 .. . . . , . . .. u . 0020.0! . . . . 110.. . ... .... . 1.. 00.1331... .. $0....lo0u.'~.a0a’ iii Q. 0.0.1 .00.- 3.05.5.7. .. .3!!! . ...... .0“... )l-nt0.-§...l0..a!:40...:I‘UJ.‘ Iii-‘0 .l. .0 . .\ ...: .l ...: . . .. 0.2.. .. 0...... ...-...... . . ....- .. Saint-50.0.“! 0.... . ... .....1... 0.3.0 .0; . 70.0.03 .1)..- 0.. .... ......O.,J.1.1..3..::.0? . 3...}.— -.10}..5‘.110l.520 . )..l.01..1..05 . . . . . . .... 2-... I. 0 .00. 0.7.. .... 30.0. n . 212231....205...... ..30 ...)..3..$...1.0 0... 0.00. .0 3.3.0....- A3 . . . . . , . , . . . 32.0.0... V1 . . .0... .. ...-.0... , . ...... 3...... ... 170.01. ...... 5...:th ‘3’. D . .. 3.1.1}! . . 2.3.“: .....1!‘0.....10'3~1... . | . ...\.... 0003.11.33... .0 11‘0“. . 000... n.§.3..’.. }....S.0.......03.. 0...... ..l..io...1.....’.. 0.0.00...OI..... A.!..7ll.u!.0..!0 ....)1 I. 1 030-0 ..l: .93.. . 3...: J...,0....o....1..... t.)§«.t...1:a0.z 29...... as. .0.“ 0. . 1o 1... I)............. 5.1.0.3320. :2... .08.. 10’... :..J§1*I.:3.¢tl.o 3 ...-... ...... 3.. .. 3.230.. 1.1 3.3.0.8). 00.8.53... .0 .. ...‘9:0 .03. .... . t ’00...- at» 3......0 Zulu... 00:00. 0-1...04.1.§.........0..1.0...02:1)..1.1._... c.3300! n. .... 0.0.30.5}... 0....1!)..I..1! ...... . . 1 .1. . V3232? . 3 1.85 a)... . .... 0:133: .30...1:....x:...:..0. 3.102.154.3230 82l)-....l.798..‘...|.0 .3021)... 37.0.3322... 3.40.510}... .. .3..3....¥00!31..5983. 0. 1.30.31.13.73-.. . . 13......2 ......n.,.0...:..........!.... . .. ! 3 1...... o .78. I .. ...... 0;... 3.20: .1\....I.. ... .‘300...0~ . 013.33.41-30... .0. e... 5 a; .0 .. . ....1......... . 1.0.6.0. .0. 00‘ ... .0}. .0. ......... 0. .00.. £70.31... 011....3..’.'.0v..31.-0..I..0I...10.0‘310.g3!..£.00:’..092.00.0300...) 0.0. ... 0 0... .13.... ...-.... a. .. .82....000. .. ...-.103!!! ... .150...) ... .. 030-. 02...: ... .. 33.03.903.04 20.3-33.1...1lanl9.....’.S. 0..3...-..0..00Q-0‘.0l ...»... .0‘1fltolo: . ......86 ...JGI: 200.0. 0.0.60.0..«000 7.0.1... .50310 0.100.033. 0 0.. 00.130103. 5330.1030..0.I. :0: K. .030 0 3.3030... .01... v.0......l..0l.00-!32000-1303.. .. ‘ 1.20.. .00...’.0..0~.8 0.000.... 01.0. .. . 09.- . 0. .20. ..s 30.305... 2. 1.2.2.“. ......“1 7.01. o.....v.10.0.0213:..itP’nodin‘u.01.6.X!S0.0‘fil§ u. ”...... r ...), .00 .1327: .0... 3.03.51.00.00 0.97.... {00.0.03 .510....0. .v .0 I 0.300.}. .17....(313I. 0.1.0.... .I...... I. J 3000...... iii... a .0. I. . ...-.0030 .0. ... .3. .. 3.1.0.. .. . . 0.0.0.0. 500.....3-0..! . .. _ .... . 3.02;... . ... .0... . ...Il.. 305‘. ...-1‘..1In0.hn..49 .z..3.1...1..o1«:.0 _. v. u 0. ..401.‘0.... ..r.’ 03"‘03 § _ “0000.10 ...... . 0.... a . .... ...a‘. ...-.13.... 0. .0! ...000 5.0.2.: .... 22.4.0.2.025 ...... 2.0. .3'3..D00 p.011-.. V009 a... ...-3 $3.35130.’ . '1. .l?....:!..! a... t 9.. .. 2.3.9.0.... 1- . . r ‘ . . .r 90...].01 ... 3.1.9; u....\0.0.0..‘.00. 2. . .0 . . ... .0. . . 00.33.081.01..00..o.3130..100 .0...",1."1.. ......0 ...00... A. i'.l.. 0.0. .. . .... 0 ‘0 .0. 0L3}... ‘VD...0_..\0OI)0'B‘.. I00...¢0...‘0-0 09-010 14 (10‘1‘0044 u I‘ 1: .00 0.! .. . 030-0... 1.3.00!.....0‘..-..I.o:3’u"0.000l§... 01 .t¢!1'. ‘39....3: ...JufUZDl..-.31 3.0. 0...... ":3“III. .V..3)I£0¢.nd..a.0.2 ! . s 1“...) 30.390.21.04 : ...-.310. 0.000: 0.0.000 ...II003 0001’; 10. ...-0.2.”... ......‘b.....v..015.; ...? 3.0!...0Pliszl. l..n0‘..0.0$.1..uu‘.l$!.tl. . .1...:!u.‘..:1 5230.73.23 .0 a. out-.50-.. 0.0400 I‘l.0...0.\_u 1.05.0.5. .. f: 00.0... 10.0.0! 10.0.0 .0 a... 10 ..v..0-0-.:33I. ...-9 .....200: .0 . 5...... a... :0...» . .. . . . . . . . . . .0.....2..........w3.00.. ...? 00.0.0.3 .0. .300)... 0.0».- .90..c0.l 013......0.’ . . I 0 ‘0... ,0 ‘1‘... 0 . I700 .123.-. ...-[0.5.0 53;... 5.0.0.6.... 3.03:2... -31....)0!I.00.061.I,..o . 2.0.0.. .001... ......0 .....0..00.9..'0... . . .... . . . . . . . . ... ...-30.11.... I! , 20.700.09‘. 0’ 00.0.1..33“‘I.‘003?t.l. 0:.- . 0...; . 02...... £60.00..- 00.9... 0‘. .38 .333035183 .... (51.0.3.0...31105000Q}. .0 . 00...“... ....0 1.0.. . . . . . 3.27.233...- 0...0.0..... ...-3.93.130... 0.3.0.0 . 1:3... «.7... . .. 0.. .. .00 . I0 .... ... ... .. 1.050....3. .4.U.!.13 0.13.0.2: )- 09‘.Q.C\I «“904. .l ..l...0.o..-..ol: 1.8)? . .3‘35..0.3...l.‘:0350...v {0.1.3.1.}...18. 1: or ... ... ... II... .00.... 0 ...-.033)...I.0..0o..0..-.. , . . . . . 90...! g. .0 :30‘3h. .08.; 01.0 . .. 05-.00’..,0.01‘0.30'Q.00.00...-. 00-0. ..—".00.00'Q0 9.0004; 0,0..00500: I'"....!.!0.‘VY 00.3.0100‘.’ l.‘0l100‘....’00.0.. 0.00-3.0; . . .... .9‘0’. an. ...-0.0 0 00 ..0..00.-00..‘.z." u§,0u\”‘.'v.‘ . . .... ....0 - ... 5.2031... A...- ..0.0?l.l)u.. u .....o..0...0\0.... 2...)... .0..o0...u.0.:.1‘....l.02s... ....l. . . ...ls‘...‘0..0.1..0.0.\0 1.. V00 0.... ..00. . . . 43.300310... 0on4 ... .... 2.700 4...}! .o .0“. 0.0. 01’...0000.....S..3 .0... !*{.39. .Ifi‘a..‘¢ ,.-00v..!n.v(~n.fivo.’0.’n “'0‘! ‘.\‘000-9‘0 0. 0’40 Y . 0. .. . .0. 0. . .. .. ....0 ..‘001 3 0 00.0. .0. . 0. 0.... ... . .0.0\.Lo .0“ 0. .0 000. s . 3.. 7 g; . 31:3... 8 ... a . . 0.. . .. 0.0. . 10.98.: ...)..IVMHJ. ta...u3.a\l¢.3.00..flflul.ur .....N .. .0 00.30... 00...: 30“... (.00. .‘.0’00 .1...rtzu. . a 5 v0. 90. 0. 0.0. . . :1. 00‘; 4.. . . .- d..-‘ 1. ..0 .0 ' ... 0. 10.. 00.1%.. .48.... 0.0. it»... I. ...‘0..3«l..3.... 91...". £0 53%” 7—0.0... ..00 10.": .0 000-0.... ..000 a .....000 .. ....’ . ... :0... .0113..0.v.o0...0.-u... «0.... 3.40. :t 0... a...“ u ...I‘. .‘l ‘0. .II‘ 5 . ...-5...... ......0...1.1......|.-. c... . 29.1.... ”.... tl‘“...tT9L..3-.10.«$.&. $00‘I0§«‘0.P$fl.2 .54.”. .... 0. I... ...0. 0-0.: ......» .1...-0a..h.0.-.,0 0...0. ...-0000.00. _ . ...... .. I. 0.1. .30... .... .0 ’0...‘ 3.00.03... . 00 00.00. .K 30.0.0.1)..1000. .- 00...... 00. I... ..--:0. ...... . I... ... I3. .0000... ...-0:. 0:00.... 00.0 .03... ... .0... .0003:l .. ...! 0. ..IO... .... ’1'...va 15": ...-.0 ..I.‘..!..‘.0..~al.. It...,0.0...l0...!¢{’ ...... :03 .00.. 0.0.3... I3!’0I’0‘.0023?0:0.0 I. .. Qt..-) . .0 20.0.0. . . . .00 .r .. 0.00.). .00 010 l. .20.... '0‘..." 01.30.30,".0ilittn‘. ......0n32 .041... v. ’00-. . ... ...-0.10.0.0... 9.. 0.9.00.0.00J 0...... .0. 0.0.30.3... .00....0'.‘ .... .1... . I... I0. .01. :00. 3.1.. a'l..0 .20... £9.00. . 1.0.0 0.1.1. ...-0.0..‘010. ....u . ‘ 0 ‘0 .0...‘0-Q. .. 1330.11.321‘03 0.,...I0.‘.1..0..." 0 ... loo-1320.23-38. :1 n a. 1 u I . 4 II I 1......0. 0.0.30. ...........3.....-5.03....2.‘331.103... .... .t.-.....c...0..l.4....0.o 0.... .llu‘iénoo 7.03.0.1: . 7.1.0.... “2.3.0 ‘0..- ..a...'...... 0.4.2.0033..." ‘30.. 13.0%?! ‘3’”: . -..........\05.\u.o.€0 ...: ...AIBI- of. . 07.103 .11)....3o :0‘0‘. . 0-1.... 90...): ..0... 020 0.....‘0 . 0.0.2.... ...-0.0 0.3.0.1317... i .10.... 0.....0 . 3.0.0.... 5. :00. .. I .... . .5 .. 3.0... o... ...... -I . ....V. 01.3.... c 010.00“. .3. 00.. 2330.2... .3300..$”..‘0)s:.021 «01... .‘o‘ . :6. v0£32’..1)0..80...0:6...:.l¢£..... 3...... 3.0.0.3 0...0.o...0).u......33 c... 0....Il...a... . ..............o: 2:33.230 0 1.150 ......0..0.?I- 0A.]... .... ...? 8.... 2...... .1. $.20...) .o .2831; 0.10.. .-...l 12:10.... ... 0.40. ...... . n “I ..10 0 ".3 o. .... 0.0‘2. ... I ...1‘11 .14 r .3331002 .0... .... 3-. g . ,3 0.. 00 (1.00.007... .‘ol c.0080. . ...? 3.00.0.3? 1.0!... I. 0.0.00.0...» . 0.0 0. - ..320...'...20.3. .. ...? a . . . .3 .... ...: .. 9.304..‘0a ....0'.. ...-a 01......1- .0520?) r0, .0! 0.05.2.0} . .2... u. ...030. u. 3.330....111002212 {3.1.0.0. .... .V....2.0a.1?3t. 9. ..l...0‘..l.0\.. 301....33. 0:1: 0...! .00.... 0:91... . :0. . .0‘50.f 3‘3: o...\. .o 2.0.... i..!.l. $5.31.... ...... 3t. .1. "0.0.. ..........0......0.al?0.!!\. ... :0. o . . .8103...§0...o.o!.......1...00 ......“- .u “5’0“.0“. W1§Nw..J0.VgVHA.-fi£ Vqsa0l'flulx11'3r‘a-i! ... -../0., 0 I... ... we... 0 . 0. 00 0- '0». 0 .. . 1.00. .. u. . 00". .... 0 00. .0. 0 ... o 04.03:. .0 v0..00u0.l O. 0- . .. ....0.’ . .. I 0. . a :00... 0 0...... 0.0 0. 0.0 o .0. .00.. .0. .30.... ... 1.00 0‘2... .0.- . 0 .l. ... . I a A. V X 0 .0. 0. ..0 0o 1 .... ..0 ... .. 0.. ... .... n . ...! . 3 . p... o . 1 03.2.3.2. ......309 '21.... .Il 43.3.3002...3.&o&.00.11§.8.3§1300..\$ 10¢ . ...I‘.‘ J'. 30‘.‘ 0. 0:150:3203 0A . .00 . . 0.00.20.01.05 0.0.7.10: 0...... 0.00 n.00‘D'00’l .I ... § .0 W330:0: 1.. .1000... .‘...!..J.. O . 040.090...’.000’.. 30...}. ..0. I .10. ..0..- ‘9' .19.. 0.8,. A 30.20 0.....1’00’“ .3000. 00.0.33... 0-000- . g1“: . 0.0 . 0 . . . .. . «0.. . . .. .....K...‘ ’..I:... I... 4.01.... ‘n0n0 ...: ..0-03... .I .... 0 - .0 .0 0.. ...-.0 Q.. ...Q. ..o .12... . . . .3 0300.. 0 ... .0 .Socb7‘I; Si.‘ft.¢2..i~.v.~.0\..‘1’t3‘. 00%‘500 000.. .. 0., . .0433! v0 .... 1........0.0:....0 t... .0... 0.. 00:0... ... . .0 .00 0...?! ...-...“...n. 9.0.0.0.... 130.300.000.01... .030... . 00. .3. 0 . 0.. 9.1:... , . . f1... 1 . . 1 fl 1 ......ol.0.00 9.1.0.1200. 4"}. “axia'fifico ’ ....)3. A... .01... .0100 02.500. 11040.0. .‘..1..0 0)........0..90 .. 43.. 00.3.3.1... 32.1.0503! 9. 32...... .03. . 0. 4.03.3... ... ... ...-0.: 3-... (00 10. .20.... 0...... .... .01.. . 32.....00000913 . 2.92:0..I,’ 3.1.3.2....- . ...??..0.0¢’.‘..J..-1 .t‘qa‘Vefo‘o...‘ .“30 £1.31}! - 1.0, 0 .0 I. .....0 . .300. TO .. ~ 000.30.100.30...‘ .. 0.. .00.... . .01 0004.01.00: .0..\a0n‘b.0....00 910000. 0I3I. 000...? ... . 333-00.... 1. . 0. 3.0.0.0.... .. 9... ......‘l...... 3.4.2.0..9. 03.003003. . .. ., c. .0 9 . . .4 .. ...-.238“. a 33.0. .. IR... .u00.10..0... 0...“): ...-... ... .. L... . ... 1... ..IV: 330.1331... 8.70.0291.-. .... . .32.091.o.3l02.o.3..0£'13 ‘0 .0!1.0..00.§09.‘0 007...... 93.28.99 ....I. .... 01t\..:’..!¢ o 30.0.80... «1.4.2... ...t..t|..0.3..0.3... .... ...9. 20.5.... 30.0.0511! ...-0.04 . X!....0.IS$%1§\9’8~$; .. .... .. . .7... ..k‘vt....0 03:5. 0.. .00 o. .0 0.0.0.00 ..J.‘.O‘I‘0¢‘.O.0‘.v‘ . ... .30.... . .0...‘..0..0. .00... 3.90.000... ...-0.00. 1300.. .. .,.0..0.¢..00..-.u...;.. .0 . 0.1....31'0.a....00...‘..1 .0. v... :0... a 3.0.0:... c a .34. $0.... ..., .0... II..00...I0..J.0‘.0£\.:0 .....1. .0 31.163.130.010.- 1..0..‘00- I .. 30.2.3.0...030 I. ..0....O.00..o....0...0:.. . . 33.......0~.....c.3...v . . .0.. 3...... 3.00.00.123 .3. ... .4 ...... ...: ... . 1 . ... . . ......‘3‘0601 .....Iioo‘Q‘.-. 0! . 0.0.0., . .9 1:11.! ...... I ...!le v.\ 3.2.3332... ...ta. 0.. 00:3.l..v.. ...... ...-5.0.0. . 020.. .... ‘20. 0 a..0.V....3..o... 3.21.. 0.37.3400. ..vn.l2.00...\!. 0. a. . 0.‘ .30.. .1 .l . u I . I: 28-30.. .02.... ...»... 3 13-. . . . . . 05:20... 3 ...... 3.31-1.00... .0. ...-0.1.33!» 0 . 0.0.108... 10‘ . :3. .2 . . 3.0... .10’3310;. . . 000 .9 o 33 10.3.. ...-:3... I .0 A- 1.03.301.....o....-.. 3......209S, .0..." .... ......3. .........|..02....0. . ....0... .q .4. .1 .. . ... 71.....1: 2.0.. . . . . n 335 ..010 . ..z .33.... ...-.50.. 3190.. 0Q 0} .0! .. ...: .4. 3.0.2.9.rlt 0...... .v .00 3.0\....;- .10.... 1.1.2.... .320... .....z... ...-248 00.0 u 3... o. . so. ..0 ll... 3 ..7’ - . .5393... 1...}?! 20.320.34.33...‘ $‘\10.. .... 11.2.0.3..3... 1.)..0 .. 0.! .QA.......(.... l .... ... 2103‘... 305.1... . .. t... . . 0.3%.... o: 00.. ...! ....‘tuaj'iais 35“.“ 3‘ ...000.......0.0.0.... 0.. 0.3.3.3.“: .0. 8.0V .0. . 0.00.00... . 0 .... .. =15..- 09.000.008.00. .... . . .00. ......‘c . . ...0. 0.0.00 I. a. 3.. ...:- J .. i' ... I”. .00 ... .- .....I..... 300 3.12.)... . 2 ...-D .530 0,9,...1,0c.13.10..230.’00.0...€€'l.030r0 . :0‘). . . . . , . .....o. .V.....:.3.o02aoi 70.14?! 3.. 0... ...}... 93... ..0 itL..O... ... ... .. c. . s o 0 4.0.0 X0. 3.1.1‘ 1...?! 3.0.3.0 ...!.o..'...a.. 3...!!!2..l.¢0£. 00.... 80......0..9 0:30.. . . . . . .. . V . _ . .I. .. .0. . . . 22.0..)- ..0‘0!.0 03’. a... 0'. d..." cac‘hdufll 100...? 1a .0 1‘. 00 a 0.02..0.0..0.l:l...lu.0u: ...:f... .... L. ... P ...! ..ru . .. Ana-0.0.. 3.03.1.5: ..00-0.’ 0.0.1... r’Ig053._-.o03.0. , .. . . . . . . .. . 1”... 0‘33... ........0~.1.0.000.2. . ‘5!!! 3...!In3‘..‘0 .0! ....I 00.0.: .0. 0 .00.... 0.. ... ..09 .... 5.0003 30...... I .... ...-c.0054.......30‘!00.0.0.0:.0J 03.0.0.0 000......3....0..I.\00-..00.....v0...0.-.0.0 tfiuual. .. . .. . . . ...... 0...}...oi0u... ...... . . 3.30.0... 0...|0..:..1 ......Avl £17....- 0..~Q.‘00.l00.0\ 390.. ‘0‘..- ..7‘015..?.ao.§ u‘g‘hll’i‘ 00000-0... . , ...: . ...... a: 0 .IY.....\¢.\.......0.0. . 3.0.2:... .. 105.109.91.11) ...-0233......023. .s.$..l‘0.’;4flu I‘m. a. 010.. 00 . . . . . . . 0.. i 1.0 1.0 0 . . .0. .1 . ..u.. .‘0. .0’. ... 0‘ .....1001. .00. JM§»¢J 0“... ‘102303... '0. .30’3 ..0000000o- .. 3’J0'0‘o‘..0"Q‘.o.0.,.020.0.00\.0 ‘00.:Q0 ....\ 0.1 o. .«0 0.- . 0.0.0.0090? 0110‘. .. J1... .‘Ol.odo:0....'00.€.03ll.01.00 ... 3... [£1.21 I....l..08.0.00.00l.. J!00..3).0.03 .0... ...... .4: .... ... .0 .33’03.‘ 0.0.0.3., t‘ 43... 0..I0..00000..0O,0-..00 00.00.... 0.00.0 ...-:00... I. w ... ...... . E .v v 0.0.. l..o0.0’1‘-.l3..‘.0" l a 11114 c . . . .. .11..-. , .. . (.1: 500..-..00Ié‘loollv. .l...‘¢....080..01. 3... Q}..0‘$300.000 ...-ll 00.... 5.0..9...’ ...00 ...-3. .. . . . . . .. vr... . ..0... I ..0 III 4 ...0’I23 - ...: .6. , ‘.Xl.0.0’0.~..-‘.J.0v..00 ...-.0...0.00...000.0.u.0 . . . . . . . .. . . . . . . , ....v .....a. ’9. . . . . 1_r000t\¢0..5'0—. ... - 0.1.0.... :00“... . ...-00 .0... 13.....01 {*1}. ...... . . . . . . . . . . . n. u . . . .. 001.00. 0:...1. .0 0! 8-01 J’... .0. .4 ...-0 3.0.!n-0...0 :0. . 0 )0 .30 . . . . . . .. . . .. . . .m . 10...... 0.... . . .5IJ)-0 . a.’.70ll‘.-0..A.....0. n01.0..0|.000.l..n0’0”¢.00000.0.3 .. . . .. . . . V .. .1. 0.0.0.. 0.0. 3.?0..00..u.)o00..00l0‘.8~ ... . 0.33,... ‘0-0... .‘l‘..n00d .0! . ...... . I... . .. . . . . . ..OOIIII 5030.33.10.93. 14 .45. I J .. . 1 b . ..1 l.......;‘.......71\02:}... 0. .0..10c2....l.. 331i.‘?.«1§ 9.0.0.3! 0.0.1.. $§3..\‘ . 4...... ...... .2: . . V .. . .. .. . . . . 33.0.0 .3. 310.0 3.000... 0.3.41.0. .00 ...}. o l. . ...... 0.0.. ... us 5.0.10.0 .30... 9390...... ..I?03.0..l.0.0q\0.34:432....33. .0 .0003. 30.1.1.0). . . ‘33.. . c _. . 050.0. 3.0! 10.000.3091130 4.0.0.0. .. 3‘.....0n0 33’ A .0.-. '3..:. 0. .v‘..." a 0" 004...... ....000-3‘QC0. . 0-0.0“... ..-»... ... ’00 .4197}; t 3130...?! ..0."§’..2 00....“ 3430......531: 0‘3‘0.“v 04" 1‘ ... .0... . ...... 7. . J f... . .. .. . {Sail--00. \$,l0. .. 2.339.083... 01.13.32. .... ...-.. 0.200.030.1503! ......x‘ ...10 2.03... .i.. . .. .... .... .....I..t... 0.1.2:... . .4... ! .1..314...$...1J,‘0~.030 $0.Il\......\!...o’070({,l 1:! ... . . .2 1.02 .105... I... . ... . . £01.... .0 .033... 00.0.0 . I}... ... o 3:00.... .0. a. ... I ‘1 ‘ L I6! .3! :8... o u... do ...!o....00-....\(0.....0,.o0_ {...-:1.- ...0 .1. ......0. 1... .Q‘....30.(.:..00 0...! ....0... l...“- 0..00.fl“‘,§..\.1.. 0..'.11.1%...!..3 1 a c .l. ...- . ...? .. 1.0, .. . . . ... ...? l. o]...- .00 :07. 0.3.0041 ...-31.1.-.. a . .. I0. 00.32..'3.0’|.0\0l...v..30000.073. 500......113... .. 1010.30)... ...‘1 ...... 0.0... 00.0....01Y0vu0 ...-... ... I . . 0.0.1.. . 0....0‘0u-A....o.3000..1.. 0.. ... .333'30300". . 0: 30.3.. .9 0.6 ...-o .01.- a . .vl..unL.I.....l.0.. (I... . .. . . . )5 30.02303: ....0...0.... 0.0:...- . .000..l......a0:-.:000.00.$03.31... ...1.... 1|: I ‘1 ... P1 . .1 ..0. .1 - 1- 4 q a . . o. ...-s. .0... .... .0 o .0. o 0-.5000‘$‘..0\0V‘.0 0.0...3‘3 01-010.". .20....0 30 .'.I0 “0.4.3.. . ... ...: .0 .. .13 $0.03 . . . . .30..)0 :2... o. 0..)..J¢0300!..8! 0.3.... It’lno‘I... 3.0.0. 1.40... .o. .1 ...:000...2o§0 I. :20... .. . 000......4. .. 0:... .0.‘.¢..I..~ .... ....0‘..... . 30. ...-10.310.00.050...) ...f Q o. : . l0 . .ov J0 ...: . ... $1313... 0... ... O. In. .... 3:13.} .‘11 00 0 V «1.. I L.0.00.‘~0. .0.I..0.~..0..‘.Ib ...! a. 3.0.0.... . 0....-.030 v ....0 0.30.. 00.0.. Q. a. .5. .‘..c .0. .- n00...» . .020... ...: .0‘ . I. 0 .00 of. u’. .a .o n n ‘ 3.190.“..r0 ..c‘o' 0.. .} ......0..|‘.,0.0.0 0. D3 .30....1‘ 0.00.. ‘. Ito-...O‘.-.:.0. 013...: no.."0-00."'00ol-9000.00-00... . ...—00.. 1'. 0.0-0.0..00 0 . .I . . ...! 00 0.. ct. 0 -.. . .0 a v. ...... 0 ... .’.:>..\v|.....0u 4.02.0'000108-00‘ 4. 0.05).... .0. . 3).....0c.;.l 00. 01.00.0003..v00.‘.3v0.¢. .10.). 3.....000’ l‘..30 V ......000 0"... 3.0}.- .. ... .- .... . 7000.. 00 0 .0 .... , .0. 57‘ . .c .J . , 0.3.00 . .0300...'\..0.. .3..l.|......v ...-......3 £0". 3.3.0.0300S'0: .9003. 0..0..I’.0¢ . I I u I ...—(9.0... ' . IA . I A 30.0. .. '0! :0 1.0.3.2... 3......(0. 0:5 000.2043: - .....‘l .100 00.0.17 .1000 oral 0.0 0:00 0 .0. . . . .0. ...... 0.....0.0f.~n..0.00.0.0.0.0|03 . .. ..- I: o .00 . . u. . . .. .. .. .. 3 02.0 .. .. ;0(900.v500a... ...-.000 00. 03‘. 1.3.3....) 0-..0...0... . 13-0,. 3.‘.I3 ....II; .0. 1:00... .000. 1 000. a} . . . , . .0!- Q ‘4.0..3:‘..‘100000..0..00 .0.. .0 ....\ frz‘a’... 10.0.‘6400. . 0.... . 14 ’. H. 4 “bud... O .. ... . ... .. . . . o. .2. .001. u‘c.’ ...-30......7’ .033...) 0’. . V . 0.0.... . . ......02 u... 00’ . .s 32.0500. .00.! .10’. 0.0.00.0 04.0....I .r. . . 0 .... - .300003001..’0.... 400-. . 0... 10004-1 .6. ..-..7 . ‘0' .0 0.000.030... .0.....2o...3.0. ......0 .0: -50.}, ...-'0': ... .0. ..."..0-0: . 0...‘ ...0. ... 0.3.]. . a. . . a . . 0v . o $.34..... .... ....I...:o).:05-.lufl. . .03.... . ...... t. 1. .....l-...\uu.a.!.o.c...!:-. .1. .0300 3.....w-000. I!!! 23.00:“): . 214...... . n....w1...... n.ll.1’..10....‘.0.’:0 -.3.7..QG.?.iAI. ‘70-... .10.. .0 ' 4. I0 1. . c ..0. ...-... 0 ...: 083). .. . ....0 0 300.030.... A ... . . . . . . . .u ...I. . . 09......H .7333701’3‘...‘ 1 . .. 0 ....n~.... .. ......0 ' .D‘~.l..!,.lt.‘7‘ififi .-333...\.0210.‘0 .. ...... I | .1..3l4 0. .. .... . .0 a... :00.) 033.91.00- , 0. X.X.00..00.00-3.0....o.0 00. 00... 00 .1 . . 10.0.: .3.” 0.0.0.. . ... u 3.. .... 7.7.0.0 0...3.v‘0.2.00.=9..~.\.0.o. )- ‘307‘§,lc0.,4‘} 0.. c. . .. 0v 0! 2 r ..il‘)‘ .25.... 0.... ‘39:... ...}... .... 5-.....1-0. ....4. . .13. .... 0.2. 1'09 .. .. ... o 9... .0.) 0...! ...... . 0.. v.25..- 0 0.... I... 130.000“. 1.000.043.1550.- .1. ”- .... . ... ..- 1.20 .. >0 .... .u 0.. ...... 3323...: I... .... ..0 .. o ... .02... 9"). . . .1 1030.....113231 ... 0.0:. .000. . ...}... 0 0.0.1.... 1: I..0::‘¢101l’:\. z... A n. v .0,v' . It . ...: : . . 013.00.... . 0 I.- .3‘: .0 ......01.‘ I! .... 000.... 3.0.0.3.... ’ .... ... .003“... I..00.. D. .500. .. .... 0 ...:i Stu... 0. . .50.... a 105.029...- 0000...;000’: .60....{l‘0 .-. {Q -X. . .00. ~.. . .31.. .11 ... a .......5:..?IO. ...... .. .0. . 0....0 7 .0. 6C...!!. ... ..0 . cu... J... . .4002... 3.... ..0. $‘0o...l.. ..Iil‘DI! .1...\,...201.«.o.. A. ... .30 3. . ... ...-00.1.10...- .....0 10...]. .00....l0......!:.. :1... .... 213...... .. . Y... .o .. 9 . . .. . <..0..A..l.....o... 12......) t:.£....!< 0.700... .2 0301-5020:...00 . 3c... . ...! :0... 0...!) 0 0‘53}: .0! ..IL 3.0.. ...-0.03.0.3! I, v .....0. ...‘....0..ZE..¢. «...-.... . «0.....010-5 10:00.0. :9... 0...... 333’s. ‘00-‘31}..15. 00.0. . 14.000.01.000 0. . .0 . ...-cu . 1 ...0l L ....0\.!l. .0320...) .. _ . .90.. .... 0.. ..~. 00...)... A ...-1.3.r.0.fi¢...0. 3...... 0- . . (a 00 . . .30... 0 o o.....l1. 0.0. ....0. 19,22 00...... .1... Q...) .0 .0‘.2I¥....a 1-10.!00 70.0.. .00.! 0...! . . . . ’...\0.0.0,»0‘ 0.4"! ‘0... 0. v..‘.,.nl§.’0.‘.3..‘. 50... 0. .- n0..0.».:0.0.00~.v...:90...ch ....0‘1. . ’ 0... ..10.0oo‘\0.s. .0000 .‘udg "200‘5o’0 . ... .00....- 0 ..V .09 3’31“. 10 030"&le O O... 0 .. . . :12... . 0.0.30.0...9. . 7.0.01.7... 0. 2.10.7... ......SI. .c. s... ... 01.70.... 00.3.0 .. 110300.... 90.0. .. .11... o 0.0 00... ...:0’.‘ , . . ii. I. 0.. 0.... o .00... .. 4I0I0 3:00. . \.4\C:0 ...0.‘.~.0& 0-000 0 n . .0. Z 1.30. 103-1005... ._ ... 4 2.3.0 .... ‘0llti. . 0.0... . Vital... n 70.1 5.0).} ...“ 3‘2... 0/- .. J0 ...l.....0.0..;..u .... 0. . ’.lv0.0..51..530...000. ......05 .v. . 00"} . 10. V. .... 01.. 0.0 .... ...-3.45.“.0 ...-00.30!- .0..0\0.....01.. i0 . 0 . .....0... 0.0.0000. .0....0O.A.0.....000.. 0!. ‘1 ........f. .0 0.). ‘0 .1. .0 0.0 o . ...l l...‘.- ' . 30.4.0... . . . .. . In. .1. c .0 $0.42.... . . .. . .. I‘v’l: 00..., «av-mm I 2' 0 L ? S 3 \ 2 ft 1. (000.110 . 00. 00"]...13‘ ..0-00...’ . ...-3...! .00....51 .080. .0 . v7 . 00:... u. l... .0. v .. . ‘. .o .‘3. .01 22.! . . 00...}... 00 . .0 00 3.00.0.1: 1.. .0. 030’...-.00.35.. . . 4 .1 . l0. . C. n. 4 0.0.0.1.... ....00’.1 .l,03.0\0000. ...O‘. ... u 0.0.. ...o. 00.. ’0. ,Q. ...: . 00.00: 00.0.0000. 0.0!. $0 . v. ‘33 4 v 010' .30. . . I .0. I. I.. .... ..00’1A.0‘:. 3.0.0. ‘0 .... 0 '0]..- 0000., .-0.0 .0... .4000... 2.30 031...... Is.‘..lv.00...l a...0|..1 1. ...... ... A... . 0!. I '0‘000 0.000 ...0 . . 0.3.0. Ilul .. 00 .... .6300... .000. 00. 1.,\0¢.00.00. _ 006.00; .A .000....0..0C.0 .....u.‘0‘fi..{..’t¢‘.00'.000 0000.0 0....‘.. 00.3.3 0. 00109 . ago I‘ “(000001.."40..00'-0¢W\u”’£3~0',3h U .. 2 . .... . .0) ... a: o .0 . . .000... . . . Y. . 0 v .. . 0 0 i . 0.00 .0 ’ . 00 .o 0 0 ll... 000 s 00 ’0 . 00 .X a .35... .00) A Q. n .0-60 ,Q“0§,.'. ’0 0.0001110 ‘0‘, 0900 0.000.530.0203. .1. ... . . I00.0l. .100. .. ....0.» . . 0.300000. . . . . .. ._ --.. .. . . .... I . ...00: ..., :0 r .0. $20.... 0. n o .3000. ’?i..f€t . . 00- . 2|. 2: .l .o ...I. Iv ..0... .....7: . ......0. ‘ . .. 000. .‘luu0. .00.. 0.. .0...'0.’.00.. ... g 0.0 '00 1... 110100 ...: 0.0.... ...I..0 ‘3 no. ‘3.".0fl§“.3.“ 1 . _ ....l 0.0.»9. ’00 ... 0“ .11.... ... ... .01.... .5)! .00. 0.. .1 . . 0003.10.10.00. .. ...... . 0:0... .030... .0.) 0:. ...‘0. .90.. ..000. .0. .0... .. .. :0: 2.10.0 5.73.... ...-v. .... ..3..l.o....ll. .lvuulunl 0.. .300..0.0.¢0. 0. 10. .00....0300125... .38.. ......10. ......3... . 0.3.5.0. 7. 0. .0 ...,000. . ..3".000’. 2... 0|. 0.2.50)... 0.00.00]. 0.003. n 0...... '.‘00. '0. 13......1 5.9.00.0 00 . ... . -8032), .305. 00:30:10.. ......0! A -...v ...-.000. h 3.3... 0...... 7‘1... . . . v . . . . y u . 0 v I . . v . . 4 . 00 . . . . ~ . . . . . . . . . . . oi}... ...§“|o0'000...\.0..1.¢.0 ‘VI.0~I‘09...0 1.30,... .0»... .03.?! ... .9. . . . . 00 . o , . .. . . . c 0 0 00 A 0.. 0. . .00 .0. }u"tleu‘§ ’..,.‘¢...‘\1.01 ,20 X o‘\.‘.\a0:. . .. .... 0.12.52? 0.35.....0o.\.(.io.1o ’2... . ..Ao . ... 03.0 ......J .0. 0‘ . 2.00 v . t ,0 0.. I u n 0 o 40 0 ..00‘. ..00 w .. _ . .. . ”01!. ”0‘00”“: w...n.un0n. 3.. .0 0. ... 5‘0 ‘0‘0.‘... x 90.0 .,.. .L..0..0...-§.300.00 35",. )3. ‘3. 0. 0 v 0. .. v . . . .3... .3... . 3...... ...): 1 ... Tittlr . . . _ . .70....313700...’ .l$$‘.l. \ \1’0.\~..-0’.0 .1 0.. 0 0.0.. 1‘... .... o. . . 0 < .. , . . . . . .... ......0’ ...0‘. ... . .0 .. .... 0000. l}. 8.0:... . .00 0....- v . . .. . . . . . . . . ‘ . ..33....¥"...3.2. 0.0....’..0 0.0 {"0 x- 0“! .‘l‘ o. 0.1. ....Y. 0.1. . . .. . . . . ... u... . .A.. .. . . ........1....\ ... . , . . .... ... :3... .0...) 2 .. .4. : .. . .. . .. . . . . .. . . . . . . . .... 01.0.1113... . 1.0.0...a:-.‘..?.x‘0..¢03fl.’00’0|l:11..\\30 . _ .. ) _ . ... _ . . .. . If . . .- J....0.0..’».. 0.0.0-. . . . .. . . . . . . . , . . .‘r 3.30.232... . o 2... . .. . . . .. .. . . .... 4'3 040 . .. .. 0.00010 n0....,.. 0.00.0.0‘....3 00.. 7.12.0.0. . . .. . . . -00.“: ...0 . . ... 0 ... ..0. I .0 00?...000‘33. (21.1... n.6....o00‘...-...~.£ ...: .030 o .0. ... c 0.0. . u , .. on ...?! . .0... . -.....o: . 0. 0. 1‘4: .04’0. .30.. a . . . . . . 1 :5 ou‘ ... 0.00:.10 0 0-. .1...‘.-.. J: I... . . ....0...0...\0 . .070. ...?0‘: . ...-0.0... ,4 0a.. .10... ‘u. . . 3. o- ..0’ .. v... 4... . 1 .1... .. . ... 00.. s00 00.. 00... 0.0 .3: . n .J ... 2. 1...... ... o. . 0 ...I. .00..’05...0.0...3 . . . o . .‘1 vo. . .c . I. ......00. ...-3....- ..0 .0. .00.. A.‘00.0..{0. . . 9.0 0 . 0 00.0. . .1... 00.0.. 0 . . .. a . .0; {253001.00- .0!0..0.:0. . O} 0.. n ....) .. . .l .. )~...vlu........-13....3 ..2... . ..-:.".... .. 0. .5. x 0 .. a. a- .. .3... , “-0 .. | lg}... .... 00 . . ....1 .4 0 .. A 00.0..l-v83u . .7 00. .0 . . . .. . 0000.. u 0 0 .00... ...0 . .. . .. . 2.010.041..- 0. Q . . .........o0. . . . .00 02...... .... .41.. . . . . . ...it’..... 0.; . . .. no. .I . . . . . ... ... 4 . . , “2.0. ...... n ...00 .. ...! .000... .0 . o. .. .. .0 .0. . .- 00!. . A. .. ...-.0 . ...: ...... .00 0. 0.00 . N". . . . . .. .. . . . ln'.‘r- . . L ._ _ . .. . u . . . v..-A. uvvww 1010 This is to certify that the dissertation entitled BUILDING RELATIONAL TRUST WITHIN COMPREHENSIVE SCHOOL REFORM MODELS: EXPLORING THE RELATIONSHIP BETWEEN TRUST AND INSTRUCTIONAL IMPROVEMENT presented by TIMOTHY GENE FORD has been accepted towards fulfillment of the requirements for the PhD degree in Curriculum, Teaching, and Education Policy bWfi/Qg’mé— Major Professor’s Signature 8/17/2010 Date MSU is an Affirmative Action/Equal Opportunity Employer LIBRARY Michigan State University -A- ......—.—.-.-v-.-.-.-.-,_.-..._._..._.-..-_._.,-‘-..‘—.-.---.—‘-.-...-.-.—.—. PLACE IN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 5/08 KIProi/Aoc&Pres/ClRC/DateDue.mdd BUILDING RELATIONAL TRUST WITHIN COMPREHENSIVE SCHOOL REFORM MODELS: EXPLORING THE RELATIONSHIP BETWEEN TRUST AND INSTRUCTIONAL IMPROVEMENT By Timothy Gene Ford A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Curriculum, Teaching, and Education Policy 2010 ABSTRACT BUILDING RELATIONAL TRUST WITHIN COMPREHENSIVE SCHOOL REFORM MODELS: EXPLORING THE RELATIONSHIP BETWEEN TRUST AND INSTRUCTIONAL IMPROVEMENT By Timothy Gene Ford Evidence continues to mount from myriad studies as to the importance of social trust as a key facilitating factor in advancing and sustaining school reform. However, the educational literature provides very little guidance on how schools—particularly chronically under-performing schools plagued by low levels of trust—ought to work on developing trust in order to improve teaching and Ieaming. Utilizing data from a sample of Accelerated Schools, Success for All schools, and comparison schools from the Study of Instructional Improvement (811), this study investigates the extent to which growth in relational trust among teachers occurs in concert with these models’ design and implementation strategies, and explores the factors related to their instructional improvement processes which are most related to change in trust among teachers. When considered in tandem with other recent SII findings, this study’s findings suggest that trust may be a sufficient but not necessary condition for the improvement of instruction and student achievement. Furthermore, several aspects of the instructional improvement process were found to be related to change in teacher-teacher trust across all samples of schools: collective responsibility, critical discourse among teachers, and climate of innovation and risk-taking. Faculty stability and depth of policy implementation were also found to be related to teacher trust in both of samples of intervention schools. COPYRIGHT BY TIMOTHY GENE FORD 2010 ACKNOWLEDGEMENTS This dissertation would have not been possible without the support of several individuals. First and foremost, I must thank my wife, Alicia, for all of her moral support and patience through what was (and is) a very challenging and mentally-taxing process. Your kindness and willingness to listen helped me more than you can possibly know—l could not have done it without you. I want to thank my parents for their support as well. I know you did not always have a frame of reference to understand what the dissertation process entailed or exactly how to help, but you were tremendously helpful as well. Finally, I would like to thank my committee, Kim Maier, Peter Youngs, Don Peurach, and Rip Correnti for their assistance and helpful feedback. In particular, however, I would like to thank Gary Sykes for his assistance throughout the process. I no longer have a doubt that helping students negotiate the dissertation is a challenging endeavor for faculty, yet you handled it with the same poise and patience that I have seen throughout my years working with you. It has been a pleasure working with you all. iv TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. vii LIST OF FIGURES .......................................................................................................... viii KEY TO SYMBOLS AND ABBREVIATIONS ............................................................... ix CHAPTER I: INTRODUCTION ........................................................................................ 1 Overview and Significance of the Current Study .......................................................... 3 CHAPTER II: LITERATURE REVIEW .......................................................................... 10 Research on Trust in Schools ....................................................................................... 10 Building and/or Developing Trust in Schools .............................................................. 15 The Role of Trust in School and/or Instructional Improvement .................................. 20 Implications for Current Study .................................................................................... 28 CHAPTER III: THEORETICAL FRAMEWORK ........................................................... 30 Comprehensive School Reform and Trust Development ................................................ 31 The Accelerated Schools Project (ASP) ....................................................................... 33 Success for All (SFA) .................................................................................................... 40 Contrasting the ASP and SF A Models ......................................................................... 47 CHAPTER IV: METHOD ................................................................................................. 51 Sample, Data Sources, and Measure Construction .......................................................... SI Sample .......................................................................................................................... 51 Instruments and Construction of Composite Measures ............................................... 53 Summary of Study Measures and Other Independent Variables ..................................... 56 Outcome Variable ........................................................................................................ 56 Time-Varying Covariates (Level 1) ............................................................................. 57 Teacher Level Variables (Level 2) .............................................................................. 60 School Level Variables ................................................................................................ 6O Analytic Approach and Procedures... .............................................................................. 6] Model Development and Specification ....................................................................... 61 CHAPTER V: TRUST GROWTH IN ACCELERA TED SCHOOLS ................................ 75 ASP Model Development ............................................................................................ 75 Results .............................................................................................................................. 78 Discussion ........................................................................................................................ 85 CHAPTER VI: TRUST GROWTH IN SUCCESS FOR ALL SCHOOLS ........................ 90 SEA Model Structure ................................................................................................... 90 Results .............................................................................................................................. 91 Discussion ...................................................................................................................... l 00 CHAPTER VII: TRUST GROWTH IN CONTROL SCHOOLS ................................... 105 Control Model Structure ............................................................................................ 106 Results ........................................................................................................................... 107 Examining Results across Models ............................................................................. 112 Discussion ..................................................................................................................... l 14 CHAPTER VIII: IMPLICATIONS FOR THEORY, POLICY, AND PRACTICE. ......1 17 Purpose of the Study and Research Questions ........................................................... 117 Discussion of Findings and Implications for Theory, Policy, and Practice ............... 1 19 Study Limitations and Suggestions for Further Research... ...................................... 129 APPENDICES ................................................................................................................ 133 REFERENCES ............................................................................................................... 145 vi LIST OF TABLES Table 4.1: Characteristics of Schools by Sample ............................................................... 53 Table 4.2: Survey Component Response Rates ................................................................. 54 Table 5.1: 3-Level HLM Growth Model of Relational Trust among Teachers in ASP Schools ...................................................................................................................... 79 Table 6.1: 3-level HLM Growth Model of Relational Trust among Teachers in SFA Schools ...................................................................................................................... 94 Table 7.1: 3-level HLM Growth Model of Relational Trust among Teachers in Control Schools ................................................................................................................ 108 Table A1: Teacher Questionnaire (TQ) and School Leader Questionnaire (SLQ) Rasch Measures Used in Study ................................................................................................... 1.35 Table Bl: Pearson Product Moment Correlations between Level 1 Predictors ............... 141 Table B2: Descriptive Statistics for the ASP School Sample (n = 2418) ........................ 142 Table B3: Descriptive Statistics for the SF A School Sample (n = 2181) ........................ 143 Table B4: Descriptive Statistics for the Control School Sample (n = 1985) ................... 144 vii LIST OF FIGURES Figure 3.1: Theory of Teacher-Teacher Trust Building within the ASP Theory of Action ............................................................................................................................ 36 Figure 3.2: Theory of Teacher-Teacher Trust Building within the SF A Theory of Action ............................................................................................................................ 43 Figure 6.1: SFA School-Level Average Trust Trajectories Controlling for Faculty Stability ............................................................................................................................. 99 Figure 7.1: Average Teacher Relational Trust Trajectories over Time by Subsample...l l3 viii ill, i0 ASP AC AIC CSR HLM SCI SFA SII SLQ TQ KEY TO SYMBOLS AND ABBREVIATIONS The Accelerated Schools Project America’s Choice Akaike Information Criterion Comprehensive School Reform Hierarchical Linear Model School Characteristics Inventory Success for All Study of Instructional Improvement School Leader Questionnaire Teacher Questionnaire ix llOTl «l ,‘a . Um. El 51 Th: 51‘ CHAPTER I: INTRODUCTION In the past thirty years, we have witnessed a remarkable transformation in our world. The shift towards a more global economy has made the importance of cultivating a highly educated and well-trained workforce even more salient than in times past. The demands of maintaining our competitive edge in this global economy have placed an even more onerous burden on our nation’s schools—all at a time when the American public seems to have lost trust in their ability to “deliver the goods” (Jacobsen, 2009; Tschannen-Moran, 2004). Increased intervention at the state and federal levels through rules and regulations designed to hold educators accountable for student achievement is perhaps the biggest byproduct of this distrust. Ironically, the key to restoring trust in schools as an institution may come through paying closer attention to what occurs within their very walls—by finding ways of cultivating stronger, more trusting relationships among the stakeholders that comprise them. Trust is an essential component in any meaningful and/or productive relationship, whether it be between school colleagues or otherwise. Philosopher Sissela Bok (1978) once wrote “Whatever matters to human beings, trust is the atmosphere in which it thrives” (p. 31). If we contend that “what matters” in education is improving our nation’s schools in ways which ensure that all children have the opportunity to succeed, then the past decade of research on trust in education in particular has brought substantial evidence to bear on Bok’s claims. In their seminal book, Trust in Schools, Bryk and Schneider (2002) make a strong case for relational trust as a critical resource for school staff embarking on ambitious school improvement efforts, demonstrating that strong ties with lid: Cllflil llu IOU DEC 5“ RH: cultivated through shared expectations and fulfillment of mutual obligations increase the likelihood of school change and gains in student performance. Indeed, evidence continues to mount from myriad studies as to the importance of social trust as a key facilitating factor in advancing and sustaining school reform (Bryk & Schneider, 2002; Kochanek, 2005; Louis, 2007; Meier, 1995, 2002; Wolf, Borko, Elliott, & Mclver, 2000), building school-wide professional community (Bryk, Cambum, & Louis, 1999; Louis, Kruse, & Associates, 1995), and improving student performance, measured in terms of scores on standardized tests (Bryk & Schneider, 2002; Forsyth, Barnes, & Adams, 2006; Goddard, 2003; Goddard, Tschannen-Moran, & Hoy, 2001; Yasumoto, Uekawa, & Bidwell, 2001). These studies and others suggest that schools that are able to make the most progress in school reform efforts are also those who develop strong, trusting relationships among colleagues (Kochanek, 2003). For teachers engaged in such reform efforts, trust can have important benefits. The “cellular structure” of American classrooms has historically led teachers to experience a significant amount of isolation from their colleagues—a fact that has resulted in a high degree of autonomy in teaching practice, little if any collaboration around instruction, lack of a common technical culture and language around instruction, and/or few shared norms and values around teaching (Fullan, 2007; Lortie, 1975). However, an adequate base of trust can provide a fertile foundation for collaboration among colleagues who have previously been isolated from one another, and can facilitate conversations and critical dialogue about instructional reform that allow teachers to push each other toward improving their practice in significant ways. These activities can lead to even further trust building. Changes within Li Cl 5: 1' {. Luiluli 1015 15:? 0" L, ‘3‘” LIL: (It? "Iii,” I. i ”I” fill-fl; "must the culture of teaching and learning such as these have been linked to increased levels of commitment among teachers, strong school—wide academic press, and collective responsibility for student learning (Bryk & Schneider, 2002). Overview and Significance of the Current Study The conclusion drawn from the bulk of trust research in education thus far could be fi'amed simply: trust is a critical resource to be leveraged for school improvement, else changes in teacher and student learning are not likely to materialize (Knapp, 1997; Smylie & Evans, 2006). Two important issues are worth noting, however. First, the educational literature provides very little guidance on how schools—particularly chronically-underperforming schools plagued by low levels of trust—ought to work on building trust in order to improve teaching and learning (Kochanek, 2003, 2005; Ford & Youngs, 2009; Louis, 2007). As some scholars have noted, there is a troubling assumption implicit in much of the school improvement literature which suggests that simply putting people together to work and/or collaborate ensures that good things will happen (Bryk, 2009). Reflecting on our own personal experiences, however, we know this not to be true. Trust takes time, effort, and nurturing—it is not bestowed, and it certainly cannot be assumed. Yet as a direct result of this assumption, we know little about the antecedents of trust in schools, and, of those, which are the most predictive of growth in trust over time. The second issue, which is directly related to the first, is that by attempting to quantify academic productivity in terms of students’ test scores, the literature has overlooked to a significant degree a potentially important mediating variable in the relationship between trust and student performance: improvement in instruction. For STHCL’ I Ill :35 I u lfit‘n In” L166 Citing _... r—C; (b 'n' l“.- lent: beill researchers studying trust between teachers and their students, this issue is perhaps less problematic, because a direct causal line can be plausibly inferred between students’ academic productivity and the collective work of students and their teachers within social space. But for those investigating trust among teachers within a school building—which, in fact, constitute a significant amount of the extant trust research in education (F orsyth, 2008)—this is much more problematic, as it is only possible for teacher trust to have an indirect effect on academic productivity. In other words, for relationships among teachers to impact student achievement, they must operate through another mechanism—in this case, likely processes related to the improvement of instruction. Many factors affect the quality of instruction which teachers provide—school leadership and opportunities for teacher learning, to name a few. To be fair, many of these factors have been studied with respect to trust, but they have largely been studied in isolation to one another, disregarding the fact that instructional improvement operates as a system which is comprised of structures, programs, and processes being implemented by multiple actors within a school. Furthermore, much prior work on these matters has taken an ethnographic or qualitative case-study perspective, thus limiting the degree with which generalizations about their relative contributions to trust growth over time can be made. This author believes that redirecting the focus of trust research to address these concerns is critical to moving further toward informing policy to address the complex and inter-related challenges many low-performing schools face with regard to improving relationships among school colleagues as well as school productivity. This is a tall task to be sure, and one that cannot be addressed in a single research study. This study therefore airs to l firsts: , l It‘lllltll I LSIT’JCIE ..I. E @0815. Ram . . tun-rig} «I... b If. ‘ E aims to take some initial steps in this line of inquiry by exploring what two instructional improvement programs—Success for All and Accelerated Schools—as well as a set of control schools can tell us about building trust among teachers and the aspects of instructional reform that are most associated with that process. Recent evidence from the Study for Instructional Improvement (811), a large- scale, quasi-experimental study of the design and implementation Of three of the most popular Comprehensive School Reform (CSR) models—Accelerated Schools Project (ASP), America’s Choice (A C), and Success for All (SFA)—Seems to support this study’s goals.l In comparing the implementation strategies of these three CSR models, both Rowan and Miller (2007) and Rowan and Correnti (2009) Observed that the Accelerated Schools model employed a strategy of “cultural controls” to promote instructional change. This strategy involved developing a normative commitment among school colleagues to a broad vision for school improvement and thus Significant cooperation among stakeholders to carry out this ambitious instructional effort. Not surprisingly, Rowan and Miller (2007) also found trust among teachers in ASP schools to be the highest—on average—of the three CSR models under study. Yet fitrther studies reported by Correnti and Rowan (2007) and Rowan, Correnti, Miller, and Cambum (2009) reported that literacy instruction and achievement in Accelerated Schools were indistinguishable from control schools. These authors concluded that ASP’s instructional design (as it existed during the 811 study) was not well-suited to producing large-scale changes in instruction and student achievement—all ' The 811 study took place from the years 1999-2004. These programs’ designs and implementation strategies have all undergone some degree of change (some more significant than others). Therefore, the accounts and descriptions of these programs will always refer to their designs at the time of the SI! study. More discussion on this point is provided in Chapter 3. spit lu' ‘i. in Of ‘n I‘P‘l "lt'll l LL"‘ despite being a program which is often successful in strengthening school culture and the relationships among stakeholders. Similarly, F orsyth, Barnes, and Adams (2005), in a study not associated with SII, identified several “patterns of trust-effectiveness,” one Of which was characterized by strong trust among colleagues but low academic performance. They explore several potential factors which explain this unusual pattern, including low school SES and teacher efficacy beliefs, but no definitive conclusion was reached. Taken together, these findings seem anomalous when we consider the multitude of studies which have linked high levels of trust in schools to increases in, most significantly, student achievement. It is clear that more research is needed which attempts to fitrther tease out the role of trust in efforts of schools to improve instruction and, more particularly, indentify the factors related to instructional improvement most associated with the development of trust among colleagues. The Accelerated Schools program is quite unique among many of the most popular CSR programs in its explicit focus on improving and strengthening school culture and the relationships among colleagues. As a result, few have considered the role of trust in whole-school instructional improvement programs and what we might potentially learn about its presence—or absence—throughout the course of program implementation. However, the improvement process specified by many CSR models entails solving many problems related to instructional management, leadership, and relationships between colleagues which may also be impeding the development of trust. It is therefore quite possible that the process of instructional improvement that some C SR programs seek to foster may also create conditions more amenable to trust formation by, for example, providing more opportunities for teacher collaboration, by inculcating a set rerlfif' stung 1 lmflllle program: contrastil models (0min clam 2hr “/1 d’éfi’lt'lemm of shared norms, values, and goals for teaching and learning, and/or providing supportive leadership around instruction. This conclusion along with the above findings suggest that CSR programs represent a potentially fruitful area for understanding the ways in which trust develops among school adults—in particular teachers—and the factors related to the instructional improvement process most directly responsible for this development. Utilizing data from the Study of Instructional Improvement (811), this research study therefore aims to address the above gaps in the literature through an examination of two of these CSR programs—Success for All and the Accelerated Schools Program—which, I argue, offer contrasting approaches to teacher-teacher relational trust building embedded within their models’ implementation strategy as it existed in the years of the 811 study (1999-2004). Complementing this examination will be a similar analysis of trust growth in a set of control schools, which, it is hoped, will provide for a means of comparison of the processes of trust growth in teacher-teacher relational trust that occur in CSR schools with that which occurs in “typical,” non-intervention schools. Framed broadly, then, this study is by its very nature exploratory. First, it seeks to understand the degree of change in relational trust among teachers in each subset of intervention schools as well as the set of control schools. In other words, to what extent does trust growth accompany school improvement efforts? This study does not assume, as others have, that trust is essential to the process of school improvement and/or enhanced student performance. As a result, this study attempts to first ascertain the degree to which these programs (some of which have demonstrated effects on student achievement and instructional improvement) are actually associated with changes in trust met Ill? factors 1 which a in Inch Stud}. 3 i0 little Imports: Rule; 31 I S". {JILILII 5 » If ILL) llOn over time. Second, this study seeks to identify, based on the relevant literature, a set of factors and/or characteristics of the ASP and SFA instructional improvement programs which are key to their effective operation, hypothesize as to their relationship to change in teacher-teacher relational trust over the four-year implementation period of the SI] study, and test these hypotheses using HLM growth modeling techniques. It is important to note that, while this is not a causal-inference research design, this study represents an important step in beginning to develop a better theoretical grounding for the role of trust development in instructional improvement. Explicitly stated, then, the research questions guiding this study are thus: 1. To what extent is there growth in relational trust among teachers over time in the 811 sample of Accelerated and Success for All schools? As models of “typical,” non-intervention schools, to what extent is there growth in relational trust among teachers over time in the S11 sample of control schools? 2. What factors related to the instructional improvement process are most associated with change in teacher-teacher relational trust in each subsample of schools? 3. Looking across CSR models and control schools, what evidence exists to support a set of common factors associated with change in trust among teachers which transcend the particular model designs? The remaining study is organized in the following manner. The next chapter reviews the literature on trust and instructional improvement and sets the stage for Chapter 3, which lays out a preliminary theoretical framework for understanding relational trust building within the two CSR models under study. Chapter 4 describes the method employed in the empirical study, as well as other important information about the data being utilized. Chapters 5, 6, and 7 report and discuss the findings of the study and Chapter 8 provides a discussion of the practical, theoretical, and policy implications of this research as well as directions for future research. [I ll“. CHAPTER II: LITERATURE REVIEW An increasing number of studies across a broad range of disciplines have recognized and, to some extent, established the importance of social trust—and the closely related concept of social capital—to coveted outcomes in their respective fields. Contributions fi‘om researchers in economics (Fukuyama, 1995), political science (Putnam, 2000), sociology (Coleman, 1988; Granovetter, 1973, 1985), organizational science (Kramer & Tyler, 1996; Lewicki & Bunker, 1995, 1996), and philosophy (Baier, 1986) all signal the ubiquity of social trust—both at the individual and group level—as an important mediating factor in the failures or successes found in many different types of human institutions. But how has trust been defined in educational research, and what does this research tell us about the role of trust and trust growth in schools engaged in instructional reform? These questions frame the review of the literature in this chapter. The first section of this chapter explores how trust has been defined in educational research. The second section reviews the scant literature in education on trust building and/or development and identifies the factors which have been found to be related to that development over time. This is followed by a final section which reviews the larger literature on school improvement to assess what is known about the relationship of trust to other aspects of the school and/or instructional improvement process. Research on Trust in Schools Schools are complex social systems with unique organizational properties that make trust a critical component to ensuring their smooth and effective operation. 10 Multiple stakeholders, grade-levels, and school subjects are just a few of the institutional and structural characteristics of schools that necessitate a division of labor with respect to school outcomes. Because the tasks required to achieve a particular school outcome such as student achievement are too numerous for any one person to accomplish alone, school adults must trust that each will do their part: Since the things we typically do care about and value include such things as we cannot single-handedly either create or sustain. . .we must allow many other people to get into position where they can, if they choose, injure what we care about, since those are the same positions that they must be in in order to help us take care of what we care about. (Baier, 1986, p. 236) These mutual dependencies exist at all levels and between all stakeholders in the educational system; attempting to reduce the vulnerabilities which result from these dependencies constitutes perhaps the most important social foundation for building trust in schools (Bryk & Schneider, 2002; Tschannen-Moran & Hoy, 2000). AS Hoy and Tschannen-Moran (1999) once wrote: “Where there is no vulnerability, there is no need for trust” (p. 186). Teachers, for example, experience a unique set of vulnerabilities between one another in carrying out their tasks within the school. Teachers not only rely on other teachers to carry out the day-to-day routines of schooling, but they are acutely vulnerable to those teachers in earlier grades whose job it is to ensure that students are learning the requisite material for passage to the next grade. Trust has the opportunity to develop as these vulnerabilities are mitigated and, in turn, this can serve to motivate teachers to more readily interact, communicate, and Share information with one another about issues 11 related to instruction (Adler & Kwon, 2002; Tschannen-Moran & Hoy, 2000). Moreover, due to asymmetric power relations, teachers are also vulnerable to principals who are responsible for decisions about hiring and firing, tenure, resource allocation, and job evaluation, yet principals also depend on teachers to provide high quality instruction (Kochanek, 2003). Two sets of researchers in particular are responsible for the lion’s share of the conceptual and empirical research on trust in the field of education: Wayne Hoy and his colleagues at Ohio State University and Bryk and Schneider (1996, 2002). Hey and his colleagues (Goddard, Tschannen-Moran, & Hoy, 2001; Hoy, Sabo, & Barnes, 1996; Hoy, Tarter, & Witkoskie, 1992; Hoy & Tschannen-Moran, 1999; Tarter, Sabo, & Hoy, 1995; Tschannen-Moran, 2004; Tschannen-Moran & Hoy, 2000) have worked from a school climate perspective to conceptualize trust primarily as a school-level attribute that is maintained as part of the school culture. This group of colleagues has defined trust as “. . .the work group’s generalized expectancy that the words, actions, and/or written statements of another individual, group or organization can be relied upon (Tarter, Bliss, & Hoy, 1989, p. 295). Thus, faculty trust exists to the extent that the group understands the reliability of the group itself and the individuals of which it is comprised to be strong. Early work of Hoy and his colleagues sought to establish relationships between teacher trust and principal authenticity and teacher empowerment, and teacher trust of the principal and supportive principal leadership (Henderson & Hoy, 1982; Hoy & Kupersmith, 1984). Later work shifted focus to investigations of relationships of trust among teachers and of the teacher and principal to measures of school climate as well as school effectiveness (Hoy, Sabo, & Barnes, 1996; Hoy, Tarter, & Witkoskie, 1992; 12 in! 35 “IL Tarter, Sabo, & Hoy, 1995). These studies would culminate in the development of a school-level construct Hoy and his colleagues would term “academic optimism.” AS a combination of academic emphasis, collective efficacy, and faculty trust of students and parents, academic optimism was found to be predictive of student achievement even after controlling for demographic factors as well as prior achievement (Hoy, Tarter, & Hoy, 2006). Extensive conceptual and theoretical work conducted primarily by Tschannen- Moran and Hoy (Hoy & Tschannen-Moran, 1999; Tschannen-Moran & Hoy, 1998, 2000) led to the description of five “faces” of faculty trust: benevolence, reliability, competence, honesty, and openness. According to their definition, benevolence is confidence in the good-will of others, reliability refers to the extent to which one can count on another to come through, and competence is the ability to come through. Honesty refers to a person’s character, integrity, and authenticity, and openness speaks to the extent to which important information is shared among parties (Hoy & Tschannen- Moran, 1999). In their study of trust in urban elementary schools in Chicago, Bryk and Schneider (2002) conceptualized relational trust as an emergent property of the everyday interactions between and among adults in the school setting. Relational trust represented a significant break with earlier conceptualizations of trust from economics and psychology which were based upon moral authority or contracts. They observed that schools, unlike other organizations, are often striving to achieve multiple, inter-related goals at the same time (high goal incongruence), while the means for achieving these multiple outcomes remain complex and diffuse (high performance ambiguity). l3 Under these conditions, contractual trust breaks down, because the transaction costs of both specifying the outcome to be delivered in sufficient detail to satisfy both parties and measuring whether or not this outcome was achieved are too high (Bryk & Schneider, 2002; Ouchi, 1980). Bryk and Schneider conclude that, the conditions of high role incongruence and high performance ambiguity which characterize schools demand frequent context-specific decision making and cooperation around local problem solving. The daily social exchanges which occur in schools therefore become the most important mechanism by which school adults are able to maintain an understanding of their role obligations as well as maintain expectations for the role obligations of others (Bryk & Schneider, 2002). Relational trust, then, at its most basic level, is grounded in the day-to-day discemments of the intentions of other school adults from within the set of role-relations characterizing the social organization of schooling (e.g., teacher-teacher, teacher- principal, etc.). But while trust originates from among these interactions, these discemments also have important consequences at the organizational level—when relational trust is high among the various role-sets, the school as an organizational entity is likely to exhibit properties of its operation that are more conducive to school improvement, such as more effective decision making and stronger social support for innovation and/or change. Individual school members discern the intentions of their colleagues simultaneously according to four distinct criteria: respect, personal regard for others, integrity, and competence. Respect in a school setting, is most directly related to how individuals interact with one another; for example, teachers who are genuinely listening l4 to one another reflect a mutual regard for each other’s worth and dignity. Closely related but distinct from respect is personal regard, which is characterized by individuals who extend themselves beyond what their role requires in order to further mitigate inherent vulnerability. Integrity characterizes a person whose beliefs closely match their actions, and competence is the ability to perform the duties associated with one’s formal role.2 As can be noted, these “facets” of trust are very Similar and overlap in many ways wi th those developed separately by Hoy and his colleagues. However, while the conceptualization of trust advanced by Hoy and colleagues can be meaningful for understanding collective levels of trust, it is less useful for understanding the presence of trust among the various role-relations within a school. It is primarily due to this limitation that the current study will focus solely on the conceptualization of trust advanced by Bryk and Schneider (2002). B u i l ding and/or Developing Trust in Schools Despite the relatively small number of researchers engaged in empirical and conceptual work on trust in education, many advances in our understanding of the nature and importance of trust in schools have been made (F orsyth, 2008). However an irrl.ITJOrtant question remains: if we know trust is important, how do we build it in schools W11 ere it may be lacking? This question is only now starting to be recognized as an IllT‘Cleveloped aspect of trust theory. As a result, our theoretical knowledge about the ways in Which trust develops and the nature of its role in school improvement remain limited. I . . . . . t‘l—IIDOrtant questions which remain largely unanswered in the literature are: What are the 2\ t W ith regard to competence, however, it is also important to note that Bryk and Schneider acknowledge a at competence is often difficult—if not impossible-—to assess due to multiple aims, variation in practice, a 5 (1 lack of good data on which methods result in achievement. Despite these obstacles, however, school Illts often make these judgments on a regular basis—particularly as it relates to incompetence. 15 antecedents of trust? What specific actions taken or structures established by school personnel as a part of the school improvement process support and/or promote trust development? Julie Kochanek (2003, 2005) remains one of the only education researchers to directly tackle these questions empirically. Much of her work builds on conceptual and empirical work in the domain of organizational science, and some of these ideas bear revisiting. Organizational theorists have been studying the foundations of trust and trust development for at least three decades. The fruits of this research are vast, but only some of this work is directly relevant to an understanding of how to trust might develop in schools implementing CSR programs. Following this brief discussion, the remainder of thi 8 section describes Kochanek’s work on trust development in the field of education, out I ining her theory of trust building and the factors She identifies as being related to its development. From a basic standpoint, organizational research notes that people in or.ganizations are more likely to form trusting relationships with those whom they share phySical and social similarities (Zucker, 1986). Though harmful, socially constructed cEltegories such as race, ethnicity, gender, and social class have long provided Frameworks through which people have judged social similarity. Particular cultural vall—Ies, attitudes, and dispositions have often been ascribed to particular groups of people, leeiCIing to the proliferation of stereotypes or otherwise superficial understandings of ‘11 iTlority groups (Mickelson, 2009). Thus, a person who meets a fellow white teacher for the first time may assume, based on the color of his or her Skin, that this new colleague 1iglds similar attitudes, beliefs, and cultural values about education and the role of 16 teachers, parents, and other stakeholders in the educative process. Despite the obvious social implications, these discemments nevertheless provide a basis for one’s judgment of the “trustworthiness” of an individual when no other information is available. Because surface assumptions can often be misleading, scholars note that social similarity alone is not likely to build trust over time unless it is accompanied by sustained, positive future social interactions (Tschannen-Moran, 2004). In their development of a theory of trust formation, Lewicki and Bunker (1996) di vide the requisite interactions needed to build trust into three qualitatively distinct stages: calculus-based trust, knowledge-based trust, and identification-based trust. In the calculus-based trust stage, individuals are in the beginning stages of interaction with others and thus know little about others’ likelihood of engaging in trustworthy behavior. These researchers stress that interactions at this stage need to be straightforward and unambiguous, and this is accomplished by ensuring that judgments regarding the s‘Jccessful fulfillment of obligations remain easy to discern. For example, a teacher might grant another’s request to join a committee to research and choose a popular school improvement program. Accepting this request represents the fulfillment of an obligation in a clear, unambiguous way and sends a message to others that this teacher is, at least at thi S stage, committed to school improvement. “Deterrence-based trust” as this stage is al SO known, is so called because motivations for performance at this stage are more likely to be based on threats of punishment rather than promise of rewards, though it is clear th at rewards are derived by those who choose to preserve the trust relationship. The. knowledge-based trust stage is characterized by parties knowing each other llt‘ficrently well to make predictions about future behavror. These predictions necessarily l7 require (and parties rely upon) knowledge of prior transactions gained through frequent communication and interaction. In this stage of trust, violations will not necessarily damage trust, as long as the other party can make sense or find predictability in the other’s violations. If a teacher knows that, based on prior interactions, a fellow teacher routinely puts the needs of school above his or her own, then sporadic violations of this role obligation will not likely result in damage to the relationship because the offended teacher can rationalize this violation as perhaps “just a bad day.” The final stage of trust development, identification-based trust, is characterized by deep identification with other’s desires, values, and intentions. At this stage, members are in the end stages of developing a collective identity where joint goals and values among members of the organization have begun to crystallize. Sharing values, norms, and intentions allows other team members to speak or represent others’ interests. Salient group identification in this stage greatly enhances regular communication and interaction and paves the way for even further trust growth, virtually ensuring that occasional violations of trust are likely to be overlooked. Some researchers have noted that schools as organizations are unique types of organizations with qualities that make trust formation in some ways more challenging (Bryk & Schneider, 2002; Ouchi, I980; Kochanek, 2003). Julie Kochanek (2003, 2005) developed a grounded theory of trust building in which the school principal is the primary agent of change. She observed that trust in schools typically develops through two main mechanisms: “the creation of positive conditions that set the stage for easing another’s sense of vulnerability and by entering into a series of successful social exchanges” (p. 32). According to Kochanek (2003, 2005), the initial stage of trust building requires 18 leadership to work to ease vulnerabilities among school staff by both a) communicating a belief system which puts students’ needs first, and b) removing problem faculty and replacing them with staff members whose beliefs and dispositions more closely match those of the other faculty as well as the overall school mission. Because goal incongruence (the degree to which individuals within the organization have differing goals for student achievement and social behavior) is often high in the typical school, Kochanek theorizes that the above actions on the part of school leadership will serve to more closely cohere school staff around common goals. Sharing common goals among staff towards putting students first is likely, according to Kochanek, to further reduce vulnerabilities inherent among staff members by providing assurances that others are working towards the same goals and that decisions which impact the school are being made with these goals in mind. Because repeated social exchanges are necessary in order to establish trust, once the initial conditions for the formation of trust are secured, the next step is to provide opportunities for interaction among colleagues. Kochanek (2003, 2005) divides opportunities for interaction into two categories: low-risk and high-risk interactions. The purpose of low-risk interactions is to further ease vulnerabilities through skillful engagement of teachers in projects which are easily accomplished and that make salient teachers’ individual roles in the success. Informal social gatherings and special school social events constitute some examples of low-risk opportunities, and these types of opportunities can lead to positive discemments of respect and personal regard. Once low-risk opportunities have laid the necessary foundation of trust among school staff, Kochanek theorized that the principal can work towards providing 19 5174C litci lliell thta ifipm liters 333621; 'ttecl: opportunities for higher risk interactions, which can facilitate positive discemments of competence and integrity. Entailed in high-risk interactions is the collaboration of school adults around the technical core of schooling or other important school improvement matters. Some possible structures and/or mechanisms for encouraging high-risk interactions are: grade level meetings, formalized committees, and increased shared decision making. Some potential activities that present opportunities for complex interactions are: efforts among school staff to build and/or develop a common vision or plan for school improvement, and/or teachers engaging in critical dialogue while collaborating on issues of curricular alignment, standards, or lesson and unit planning. Many CSR programs contain as a part of their improvement design many of these structures and activities, but whether or not these matter for the growth of trust among teacher colleagues has not been sufficiently explored. The Role of Trust in School and/or Instructional Improvement AS was established in the introduction, while the literature on trust development in education is scant, the literature which identifies trust as a critical factor in the school improvement process is substantial. This disparity in and of itself is diagnostic of the theoretical gaps in the literature which exist—gaps of which this study is decidedly attempting to address. One reason for this disparity is that the causal linkages between aspects of the school improvement process and trust growth remain in question—in large part because these linkages are likely complex and reciprocal (Forsyth, 2008). But before issues of causality and causal direction can even be reasonably addressed, more information needs to be gathered on what precisely are the aspects of the school and/or 20 , w, llbu A la put Illz’ll ll resist to 1de . llillll \‘ imprm 0i Still lllSll’uc “lli'l Cl “Elm; lies- I] instructional improvement process most related to trust growth and development. This is the purpose of the current study. In keeping with this purpose, this final section reviews the larger and more diffuse literature on school improvement and the social factors that have been identified as promoting trust. As will be seen, this literature identifies several factors of importance to trust, but because many of these factors were studied in isolation to one another or with cross-sectional and/or qualitative or ethnographic research designs, we know little about their relative contributions to trust growth over time. Therefore, the purpose of this final review is to determine and select a final set of factors which have been tentatively linked to trust and that can then be tested more thoroughly for their relationship to trust growth fi'om within the contexts of the 811 study schools. An examination of this literature revealed three broad categories or domains related to the school and/or instructional improvement process, and these categories were used to organize this review: (a) the role of school leadership in instructional guidance and management; (b) the building of instructional capacity through professional development and opportunities to collaborate with colleagues; and (c) structural features of the school community which enhance instructional reform and/or trust development. School leadership. When it comes to school improvement, not surprisingly most view the principal as the lynchpin of change. Cuban (1988) identified three roles that have comprised the job of educational leaders (in this case principals) in American educational history: administrative chief, political negotiator and facilitator, and instructional leader. He notes that the former two responsibilities have typically dominated the lives of principals, to the detriment of the latter. In the current era of 21 accountability, however, school leaders’ responsibility within the school building continues to increase as leaders juggle managerial and political tasks with their new- found roles as instructional leaders (Bryk, Sebring, Kerbow, Rollow, & Easton, 1998; Supovitz & Poglinco, 2001). The increasing burdens placed on principals as well as the advent of new approaches to school improvement—such as those spawned by the school “restructuring” movement of the late 198OS—have opened up opportunities for other members of school staff (such as experienced teachers) to take leadership roles in the improvement of teaching and learning, and to share in the school improvement decision—making process (Elmore, 1990; Heller & Firestone, 1995; Johnson & Pajares, 1996). When implemented well3, research suggests that shifts in responsibility for instructional leadership and decision-making can free up time for principals to spend in classrooms observing and guiding instruction, and carry the added benefit of Simultaneously building and tapping into the expertise of the faculty as a whole (Hart, 1995; Smylie, 1995). Teacher empowerment of this kind has the potential to lead not only to trust and a strong sense of community among colleagues (Kruse, 2001; Louis, Marks, & Kruse, 1996; Rosenholtz, 1991), but has also been Shown to be related to instructional improvement as well (Smylie, Lazarus, & Brownlee—Conyers, 1996). Research also demonstrates the importance of building and communicating a vision for school improvement on the part of school leadership (Elmore & Burney, 1998; Firestone, 1989; Firestone & Corbett, 1988; Fullan, 2007; Heller & Firestone, 1995). 3 Of course there are many criticisms of shared decision-making—namely that often it results in decision- making that is largely symbolic in nature. In these cases, teachers, parents and other stakeholders felt their participation was not genuine; in other words, that they had any real “say” at all in school decisions (see for example, Bacharach, Bauer, & Shedd, 1988; Duke, Showers, & lmber, 1980; Hargreaves, 1991). 22 Pfoi'l pTDCc‘ illliol Strut pfl‘jé 111:1." [155; 113C for Providing a vision which communicates both broad purposes as well as finer-grained procedural details involved in executing the broader vision is essential for successful innovation (Huberman & Miles, 1984; Supovitz & Poglinco, 2001). In establishing a strong vision for improvement, school leaders establish consistency and coherence, and project competence through their establishment of norms and expectations around instruction and common goals for student learning (Evans, 1996; Newmann & Associates, 1996; Smylie & Hart, 1999). These facilitative structures can then trickle down to teachers allowing for the sharing of norms and values around teaching and the fostering of collective responsibility for learning (LoGerfo & Goddard, 2008; McLaughlin & Talbert, 2006). Collectively, these efforts can create a social focal point upon which trust among colleagues can grow (Bryk, Lee, & Holland, 1993; Bryk & Schneider, 2002). Effective support and monitoring of teachers’ instructional practice is, of course, a necessary corollary to providing detailed plans about how to can'y out the broader vision for improvement (Firestone, 1989; Founder, 1998; Purkey & Smith, 1985; Spillane, Halverson, & Diamond, 2004). Principals can foster social trust by ensuring that teachers are well supported in their work, by providing materials, access to resources including professional development both within and outside of school, as well as encouragement, praise, and reinforcement in their day-to-day activities (Bryk, Lee, & Holland, 1993; Founder, 1998; Rosenholtz, 1991). In addition to effective instructional leadership practice on the part of school leaders, attention to and management of school personnel to support the instructional effort is also critical. Some studies of instructionally effective schools and/or districts 23 have alluded to the importance of attracting, selecting, and hiring key personnel to help with the instructional reform effort (Elmore, 2004; Elmore & Burney, 1998; Rosenholtz, 1991). Not surprisingly, these hiring practices can have important effects on collegial trust as well. Both Bryk and Schneider (2002) and Kochanek’s (2005) studies of relational trust in Chicago elementary schools emphasize the criticality of principals’ ability to regulate the composition of their school staff to improvements in trust over time. They reason that trust both among the teacher-principal and teacher-teacher role relations is more likely to develop under leaders with wide latitude for hiring and firing because it those who are detrimental to trust either due to their incompetence or their inability to uphold the commonly held vision of the school are replaced with faculty who are more competent and amenable to the change effort. Building teacher capacity. Teacher capacity can be most basically defined in human capital terms—that is, the knowledge and Skills a teacher possesses and brings to his or her instructional practice as acquired from education and/or teaching experience. Moreover, it also refers to a teacher’s orientation to and disposition towards Ieaming and implementing new instructional policies (Ball & Cohen, 1999; Cohen & Barnes, 1993; Cohen & Hill, 2001; Spillane & Thompson, 1997). Teachers can improve their capacity on both a personal and professional (i.e., collaborative) level (Smylie, 1996). They can seek out new information on their own by attending a professional development seminar or attaining an advanced degree, or they can learn in collaborative ways from their peers, by engaging in discussions and examinations of each others’ practice, engaging in team teaching and/or planning, and/or discussing student work, standards, or curricula (Cohen & Ball, 1999). This latter approach emphasizes the utility of social capital in building the 24 human capital of teachers (Coleman, 1988; Coleman & Hoffer, 1987; Smylie, 1996; Smylie & Hart, 1999). Professional development opportunities, because of their ubiquitous presence in the school system, constitute one of the most common structures in which collaboration among teachers is likely to occur; others are common planning time and simple physical proximity (Kruse, Louis, & Bryk, 1995). While most professional development of the past has consisted of “one shot” trainings, we know that professional development is most beneficial to school improvement when it is: on-going rather than episodic, focuses on instruction and student outcomes, and provides opportunities for collegial inquiry, help, and feedback while respecting teachers’ autonomy (Correnti, 2007; Darling- Hammond & McLaughlin, 1996; Little, 1993; Lytle & Cochran-Smith, 1994; Newmann, King, & Youngs, 2000). This understanding of professional development suggests that collaboration is of little use to instructional improvement if it does not provide teachers opportunities to both support and challenge their colleagues by engaging in critical dialogue about their practice (Borko, 2004; Hart, 1998; Putnam & Borko, 1997; Rosenholtz, 1991; Supovitz, 2002; Tschannen-Moran, 2001). An ethnographic study of six urban elementary and high schools conducted by Little (1982) found that collaboration, critical dialogue and feedback from within an atmosphere of trust and mutual respect seemed to have contributed most to teacher learning. Furthermore, Susan Moore Johnson’s (1990) study of 115 public and private schools produced similar findings, suggesting that distrust and disrespect among school colleagues can serve to undermine efforts to effectively support teacher Ieaming and change through collaboration. 25 Over time, teachers engaged in these types of activities become increasingly willing to give up their individual needs for privatized practice in order to collectively build and construct meaning together, and, through such activity, they are more likely to learn how to more productively resolve disagreements (Achinstein, 2002; Clift, Veal, Holland, Johnson, & McCarthy, 1995; Uline, Tschannen-Moran, & Perez, 2003). Moreover, the actions of teachers to initiate collegiality have been Shown to more effectively impact trust development over and above the actions of principals (Hoy, Smith, & Sweetland, 2002; Tarter, Bliss, & Hoy, 1989). Quite predictably, teachers in strong “professional communities” have reported a high level of innovativeness while teachers in weak communities tended to report strong norms of privacy, and lack of innovation and support for their Ieaming (McLaughlin, 1993; McLaughlin & Talbert, 2001).4 Structural facilitators. Research suggests that several structural features of schools may also facilitate instructional reform as well as concomitant efforts to develop trusting relationships among colleagues. School size, for example, has in recent years been increasingly scrutinized for its potential effects on student achievement and other institutional qualities and practices, yet conclusions as to its role and/or importance remain in question (for a review, see Ahn & Brewer, 2009). School size has intuitive appeal for those investigating issues of trust, however. Large schools, like other large institutions, often become synonymous with bureaucracy, and bureaucracy is often seen as antithetical to the development of relational trust (Bryk & Schneider, 2002). Increases 4 The term teacher professional community has several conceptual guises in the literature. Yet despite variation in how this term is defined, there are common threads woven throughout—most particularly, shared norms and values among teachers and staff (including academic press), collaboration (which entails some measure of deprivatized practice), and shared professional development opportunities. 26 in the complexity of an organization mean increases in the size and number of social networks within it—maintaining the regular communications and effective leadership needed to foster relational trust becomes significantly more difficult in these cases (Bryk, Cambum, & Louis, 1999). Some studies have investigated the degree to which school size is related to trust, and the findings of these studies are mixed. For example, Bryk and Schneider (2002) found small school size to be an important structural facilitator of teacher-teacher trust, but not trust among the teacher-parent and teacher-principal role relations. Similarly, in their study of collective teacher trust in relation to parents and students in urban elementary schools, Goddard, Tschannen-Moran, and Hoy (2001), also found no relationship between school Size and teacher trust of parents and students. In their study of the role of professional community in the organizational Ieaming capacity of 248 Chicago elementary schools, Bryk, Cambum, and Louis (1999) found small school size and social trust to be important structural facilitators of growth in organizational learning 'and professional community, concluding that small schools increase the likelihood that trusting relationships will form among faculty. In addition to the impact of school size on trust, school size has been hypothesized to impact the depth of implementation of CSR programs (Berends, 2000; Rowan, Cambum, & Barnes, 2004; Rowan & Miller, 2007). Depth of implementation of the CSR becomes important because research has demonstrated that schools with cooperative professional cultures (i.e., high trust schools) characterized by strong principal and staff leadership are more likely to implement reforms more deeply (Rowan, Cambum, & Barnes, 2004). Indeed, without trust among teachers within a school, it is 27 likely that momentum for change efforts will wane and thus so will their impact over time (Hargreaves, 1992). Finally, the stability of the teaching faculty over time is also implicated in instructional improvement as well as trust development. Teacher turnover can impact student Ieaming and instructional quality in many ways. First, high-tumover schools are more likely to have less-qualified teachers who may lack teaching experience and/or adequate training and certification (Rockoff, 2003; Hanushek, Kain, & Rivkin, 2004). Teacher turnover can also subvert efforts to deliver high-quality instruction, as it is likely that new teachers unfamiliar with reforms will require time to learn and become accustomed to changes in instructional patterns and/or expectations (Béteille & Loeb, 2009). AS Bryk and Schneider (2002) explain, because building strong relational trust among teachers requires positive, sustained interactions over time, high amounts of faculty turnover are likely to continually thwart this process. Lack of trust coupled with frustration due to fiagmented instruction and instructional reform likely have a mutually reinforcing effect in high turnover schools. Implications for Current Study The hard work of researchers over the past four decades has resulted in a reasonable base of knowledge about the relationship of trust to individual elements critical to instructional improvement such as collaboration, teacher Ieaming, and effective school leadership. Yet we know little about the relative relationships of these factors to trust among teachers because few studies have attempted to consider these factors collectively in a single study. Is collaboration strongly associated with change in trust among teachers? Critical discourse, or perhaps shared decision-making? By controlling 28 for the effects of other important instructional improvement processes, we can better understand which factors are most associated with change in trust among teachers. Furthermore, while the loss of federal funding for CSR interventions has reduced their overall popularity and use as an approach to school reform, there is still much we can learn about school reform and trust development by examining how these programs operated at the time of the 311 study. By design, many of these programs are quite unique and offer varied approaches to educational improvement. This raises questions about the role of trust within these reform efforts, such as: whether or not trust grew among teachers over the course of implementing these reforms, and what it is about the approach to improvement they undertook that might be associated with that change. By contrasting CSR programs which varied in the degree to which certain instructional improvement factors were present (i.e., teacher collaboration, shared decision-making, etc.), we stand to significantly increase our understanding about how trust develops and of the potential pathways to trust grth which may be pursued by schools. The next chapter outlines a theoretical framework for understanding how the factors identified above in the review of the literature relate to the instructional designs of the CSR programs under study and provides a framework for hypothesizing about how these factors might be associated with change in trust among teachers within these programs. 29 CHAPTER III: THEORETICAL FRAMEWORK In the introduction, this research study was broadly framed as an exploratory analysis of the degree of trust grth within two popular CSR programs—the Accelerated Schools program and the Success for All program—and the factors most associated with change in trust among teachers in these and a set of control schools. The purpose of this study is to develop a better theoretical grounding for the role of trust development in instructional improvement. The review of the literature on trust and instructional improvement in Chapter Two identified a set of factors and/or characteristics which in past studies have been found to be related to trust and trust growth over time. The purpose of this chapter is to select a set of these factors relevant to each of the CSR programs under study in order to develop a theoretical framework for viewing, understanding, and interpreting growth in relational trust among teachers from within these programs. Additionally, we want to also develop and test hypotheses regarding a collective set of factors associated with change in teacher trust that these programs and the set of control schools might have in common. Recall that the research questions guiding this study are: I. To what extent is there growth in relational trust among teachers over time in the $11 sample of Accelerated and Success for All schools? As models of “typical,” non-intervention schools, to what extent is there growth in relational trust among teachers over time in the 811 sample of control schools? 30 2. What factors related to the instructional improvement process are most associated with change in teacher-teacher relational trust over time in each subsample of schools? 3. Looking across CSR models and control schools, what evidence exists to support a set of common factors associated with change in trust among teachers which transcend the particular model designs? This chapter begins by providing important background information and detail on each CSR program as well as its “theory of action,” as it existed in the years of the 811 study. It then theorizes about how trust might grow and develop among teachers throughout the course of its implementation. Throughout this discussion, the chapter advances several hypotheses related to each model’s potential for building trust including the factors which might be most associated with change in trust among teachers over time. Comprehensive School Reform and Trust Development The concept of Comprehensive School Reform (CSR) coalesced with the founding in 1991 of the New American Schools Development Corporation (now New American Schools), as part of President George H. W. Bush’s America 2000 initiative. NAS founders believed that past reform efforts were too fragmented; CSR programs represented a way to “break the mold” of American schooling by providing a school- wide design integrating research based practices into a coherent set of effective approaches to teaching and Ieaming (Berends, Bodilly, & Kirby, 2002; Correnti & Rowan, 2007). Since the end of the scale-up phase of the NAS project in 1998, thousands of schools have adopted the over 600 different C SR models currently available (Desimone, 2002; Rowan, Cambum, & Barnes, 2004). However, recent loss of federal 31 funding for adopting and implementing a CSR program has led to a diminished status for these programs as an approach to school reform as well as significantly lower adoption rates. Nevertheless, for the years that CSR programs were in ascendance up to and including present day, much of value to understanding effective school reform has been gained from studying these programs in their past as well as current forms. By design, CSR models attempt to Significantly alter the overall school leadership and culture of a school, the ways in which teachers work together, as well as relationships with other stake-holders such as parents—although the degree to which any of these are altered is a function of the program implemented. Yet, as was mentioned, few studies have attempted to establish clear linkages between trust growth in CSR schools and aspects of the models’ theories of action which may have facilitated its growth. The two models which are a part of this investigation—Accelerated Schools and Success for All— each have, as configured at the time of the 811 study, unique theories of action for improving instruction, which, it is hypothesized, may lead to different patterns of trust growth in schools implementing them. The next two subsections will first provide background on these two CSR models and then lay out a theory of how trust might develop within each of these programs’ theories of action.5 It is important to note two important corollaries with respect to the configuration of these programs at the time of the 811 study which have important implications for this study. First is that the developers and implementers of these and other CSR programs are constantly Ieaming about how best to design and implement 5 It is important to note that all descriptions of the SFA and ASP programs are based on their characteristics at the time of 811 data collection (the academic years of 2000-2004), even though some citations may be of more recent material on the programs. Experts in each of the programs were consulted to ensure the historical accuracy of these accounts. 32 their models. As such, the analysis undertaken in this study acknowledges that even within the time frame of the 811 study, these programs were changing as implementers discovered, developed and improved on their models of instructional reform. Second, it is also important to acknowledge that the ideals of reform that these instructional interventions seek to foster might take considerable time to develop—as much as 6 to 8 years—and lead to significant turmoil and difficulty that can last for many years. As such, it is possible that the 4 years of the 811 study were not always sufficient in capturing all of what these programs sought to develop in the schools implementing them. Finally, it also bears reiterating here that, while the theories presented in the next section may posit a causal process of trust grth and/or development, their purpose is merely illustrative and designed to provide a framework within which to View, interpret, and understand this study’s findings. Research hypotheses have been carefully worded to reflect the study’s stated goals, objectives, and research questions. The Accelerated Schools Project (ASP) Developed by Henry Levin and his colleagues at Stanford University in 1986 to bring children in at-risk situations up to grade level by the end of sixth grade, the term accelerated refers to the need for at-risk students to learn at a faster rate so they can catch up. As of the year 2000, there were more than 1,000 schools across the US. who had adopted the ASP reform model (Bloom et al, 2001; Borman, Hewes, Overman & Brown, 2004). ASP’s underlying belief is that all Students have the ability to manage more challenging material—the type of material often found in America’s best schools or reserved for gifted and talented students—which they deem “powerful learning” (Rowan, Correnti, Miller, & Cambum, 2009). Coupled with the powerful Ieaming approach, the 33 central, interrelated aims of ASP are to create a new, supportive school culture which values and sets high expectations for all children and teachers, and to institute a process by which schools can determine their own school improvement goals and discover solutions for achieving them (Bloom et al., 2001). Therefore, the Accelerated Schools program prides itself on not being a “packaged,” prescriptive system of school improvement, but “a philosophy and a process” (Hopfenberg et al., 1993, p. 20). Guided by three principles—unity of purpose, empowerment coupled with responsibility, and building on strengths—the goal of the ASP model is to assist the entire school community (teachers, leaders, parents, support and district staff, and local community) in developing the capacity to determine and carry out their own locally-relevant school improvement efforts. In other words, a set program of instructional improvement characterized by a preferred curriculum and single instructional approach does not exist in ASP schools, but rather the school community develops it collectively based on perceived needs and congruence with the tenets of powerful Ieaming (Finnan & Swanson, 2000). This approach to instructional improvement allows for a great deal of autonomy on the part of individual teachers to develop their own pathways to powerful Ieaming for their students (Rowan & Miller, 2007) Engagement in the first steps of the “process” model can last up to an entire year and begins soon after a brief training period to familiarize staff with the program’s central aims. Perhaps the most “prescribed” part of the ASP program, the process model asks the entire school community to first: 1) take stock of current school practices, strengths and challenges; 2) develop a vision of what kind of school they would like to be (within the 34 Span of \\I change be; the on-g‘ol Pmress Cl the St‘lt’llll and ex allia- AI are being I intended ll collaborati slivl as l oltiie schc reinlorcing reinforced profession, jml’rm em. In all“? disc ll: them School lei;- lltisroom , c0mllllll‘ner based Share span of what an Accelerated School is); 3) decide on and select a set of priorities for change beginning with the most important; and 4) create governance structures to support the on-going inquiry process. Once these support structures are in place, the Inquiry Process commences and the school community begins to pursue their vision by applying the scientific method to their priority challenges: developing and implementing solutions, and evaluating the outcomes of their plan (Hopfenberg et al, 1993). At the same time schools are engaging in the process model, concomitant changes are being fostered within the school culture, and values are being fostered which are intended to guide the actions and interactions of staff, such as: trust, communication and collaboration, experimentation and risk-taking, reflection, community spirit, and the school as the center of expertise. Engagement of the process model and the development of the school culture around common norms and values co-occur and are mutually reinforcing. As staff engage in the process model, the core values of ASP are being reinforced through interaction, and, Simultaneously, these values and orientations to professional practice are being drawn upon to engage productively in the school improvement process. Trust and ASP ’s theory of instructional change. As may be surmised from the above discussion, the cultivation of trust among school staff is an explicit component of the theory of instructional change in ASP. Moreover, ASP’S initial work begins at the school level and involves laying the groundwork for a school-specific approach to classroom change. One function of the school-level work is building a normative commitment on the part of the school community to its central tenets and to the values- based shared decision making process. Curricular and instructional changes are thus 35 delayed until an evidence-based plan is designed, which ASP intends to be accompanied by a strong, capable, school community. The general trust building model is displayed in Figure 3.1. Figure 3.1. Theory of T eacher-T eacher Trust Building within the ASP Theory ofAction K \ ASP Process Model Building a Vision Values-based Shared Decision Making \ J A r Relational Trust Growth / Development of School Culture \ Climate of Innovation and Risk Taking Critical Discourse among Teachers Teacher Collaboration around Instruction \ Quality Teacher PD Opportunities j In theorizing about how trust may develop between teachers within the ASP model, it is helpful to begin with the engagement of the whole school in year one process model development. While there are four distinct steps to the process model development, there are essentially two characteristics of this process of utmost importance to trust development: engagement of the school community in vision building and evidence and values-based shared decision-making. Kochanek (2003, 2005) posits that efforts to collectively build a vision or plan for school improvement represent an ideal opportunity for high-risk interactions among staff members in which discemments 36 of competence and integrity can be made. “Small wins” are essential for ensuring that these discemments are positive at the outset, and there is much to suggest that at least some of the initial process model tasks (such as developing a shared vision, assigning staff to cadres, etc.) can be, relatively speaking, quickly and successfully accomplished. Given the instructional latitude that individual teachers have within the ASP school, the role of leadership in building a common vision for improvement seems particularly critical, if for nothing else than to keep a rein what might quickly become diverging instructional approaches. Research has shown that it is possible for school-wide professional community to flourish within an atmosphere of teacher autonomy, provided principals work to establish a Shared identity around instruction and teacher Ieaming which allows for professional individualism outside the group (Scribner, Hager, & Wame, 2002). Efforts of leadership to reiterate collective goals for improvement will serve the purpose of ensuring that teachers are able to reflect on the highly espoused norms and values of the ASP model as well as serve as a social focal point upon which collective responsibility for Ieaming and trust among colleagues can grow. Further, the inclusion of multiple stake-holders (including teachers) in decision making which will have far-reaching implications for future school success likely will ensure that such involvement is perceived as genuine, and thus may fiirther motivate staff to take ownership of the reform (Duke, Showers, & lmber, 1980; Hargreaves, 1991). Collective ownership and motivation for reform provide strong potential for growth in relational trust among teachers, and the abundant confidence in oneself and others as well as a “can-do” attitude which result from this level of participation may, in turn, further 37 enhance the shared decision-making process (Johnson & Pajares, 1996). This discussion leads to the first set of hypotheses about trust growth in ASP schools: Hypothesis 1: Given the latitude ASP teachers have in choosing their approach to powerful learning, the eflorts of school leadership to build, maintain, and communicate a vision for school improvement are predicted to be positively related to change in relational trust among ASP teachers. Hypothesis 2: Shared decision making among ASP faculty is positively related to change in trust among teachers. AS was mentioned before, concurrent with the execution of the early stages of the process model is the development and dissemination of school-wide values necessary to support and sustain school improvement efforts. The bottom box of Figure 2 highlights the values and activities that present significant opportunities for trust building within the ASP model. Teaching has historically been isolated, personal work (Lortie, 1975, Fullan, 2007), but while ASP teachers have a degree of latitude in developing their own instructional approaches, teachers within ASP are also encouraged to break away from traditional norms of isolation and autonomy toward greater collaboration. High risk interactions around curriculum and instruction in which teachers engage in critical dialogue and examination of practice are those which consensus suggests have the greatest likelihood of transforming practice (Borko, 2004; Tschannen-Moran, 2001, 2004). Engagement in the ASP process model framework in the early stages of implementation works to inculcate an orientation towards collective critical examination of school matters while ensuring that teachers (as well as other staff) are building base 38 levels of trust which they then can draw upon in their later collaborations with colleagues around instruction (Hopfenberg et al., 1993). One value which further supports and encourages collaboration is the belief that ASP teachers are both facilitators as well as models of ongoing, lifelong Ieaming. In order to optimize the classroom as the center of expertise, teachers must be actively engaged in their professional communities, seeking out their colleagues to grow as “teacher-leamers” both individually and collectively (F innan & Swanson, 2000). ASP teachers can draw upon their peer support teams, study groups and cadres for the pursuit of Ieaming which supports their efforts in the classroom (Hopfenberg et al., 1993; St. John, Meza, Allen-Haynes, and Davidson, 1996). Risk taking and innovation as core ASP values go hand-in-hand with these efforts, and, with a supportive school culture and community, the vulnerabilities associated with implementing a new curriculum or instructional strategies are likely not as great as they would be in a school without those supports. This discussion leads us to our final hypotheses about trust growth in ASP schools: Hypothesis 3: Change in relational trust among ASP teachers will be positively related to teachers ’ collaboration and engagement in critical dialogue around school improvement matters. Hypothesis 4: The ASP model ’s emphasis on continuous teacher learning and innovation encourages teachers to seek out new learning from fellow teachers. Thus innovation and risk taking is predicted to be related to change in relational trust among ASP teachers. 39 Success for All (SFA) One of the most popular school reform models in the United States, Success for All was developed by Robert Slavin and Nancy Madden at Johns Hopkins University in the 19808 (Slavin, Madden, Chambers, & Haxby, 2009). SFA is an elementary-based CSR program which, since its initial development and implementation in a single Baltimore school in 1987, has been adopted by over 1,200 schools across 46 states. Its substantial growth in this period is due, in large part, to extensive evidence of its robust effects On student achievement (see, for example, Borman et al., 2004; Borman et al., 2007). Part and parcel of its demonstrated success is its singular focus on helping all children succeed through a multifaceted approach to the prevention of early academic deficits, particularly in literacy. SFA features a school-wide curriculum for grades K-8 based on the latest scientific research on effective literacy and mathematics instruction. Ensuring early reading success is accomplished through the Reading Roots and Wings program—a carefully staged and sequenced literacy development process. The goal of Reading Roots is to quickly establish strong fundamental reading skills primarily through the use of teacher-directed activities, while gradually increasing student-led reading activities as students’ reading skills strengthen. Reading Wings builds on Roots successes through the use of cooperative Ieaming techniques, which research suggests increase students’ motivation and engagement (Slavin, 1995). Literacy activities such as vocabulary building, decoding practice, and story-related writing are accomplished through work in teams, which is initiated and supported by teacher-led instruction (Slavin & Madden, 2001; Slavin, Madden, Chambers, & Haxby, 2009). 40 Aside from instruction, the program provides for specially trained tutors to work one-on-one with struggling students in first through third grades, a quarterly assessment and regrouping system to place students with peers of the same reading level, a ‘solutions team’ focused on parent education and involvement, and a program facilitator responsible for on-site instructional coaching, management of the assessment system, and ensuring all staff are communicating with one another (Peurach, in press; Slavin & Madden, 2001; Slavin, Madden, & Datnow, 2007). Further, schools are required to hold “component meetings” that enable feedback loops and ongoing dialogues between on-site staff and SFA foundation staff as well as provide monthly opportunities for school staff to interact with one another as a part of “teacher Ieaming communities” (Harris, 2003; Slavin, Madden, & Datnow, 2007). Schools and/or districts who adopt Success for All as a school-wide program are provided with materials, extensive support in the form of training and professional development, and detailed instructions on how to implement and sustain the model. Perhaps the most widely recognized aspect of the SF A model, however, is its promotion of instructional change through the establishment of procedural controls, in other words, instructional “scripts,” detailed pacing guides, and assessments (Rowan & Miller, 2007). At least in the initial phases of implementation, teachers are required to follow a scripted instructional plan which guides teaching activities through a 90-minute reading lesson period and a weekly lesson sequence. One of the program facilitator’s primary responsibilities is to ensure that teachers are becoming proficient in the instructional model, and they do so by monitoring teachers’ instruction and providing feedback, consulting assessment data, teaching demonstration lessons, and organizing 41 other professional development activities.6 Because of their potential impact on instructional change, facilitators are often seen as the “linchpin” of successful implementation of the program. As such, the ideal facilitator is one who is an experienced classroom teacher with good interpersonal Skills, and one who has the respect of his or her colleagues (Slavin, Madden, Chambers, & Haxby, 2009). Trust and SFA ’s theory of instructional change. As was mentioned in the previous sections, the purpose of this discussion is to make explicit the theory of teacher- teacher trust building embedded within SFA’S process of instructional change. Generally speaking, while building trust among students is a more explicit component of the SF A model in terms of student engagement in cooperative Ieaming strategies, it is my contention that a model for trust development can be nevertheless be discemed from within the SFA model of instructional change. In explicating this model, it is useful to break up the SFA change process into its distinct phases. In doing so, I will employ the four stages of group development (Forming, Storming, Norming, and Performing) framework outlined in Tuckman (1965) and the Success for All Foundation (2004). The general process model described in this section is shown in Figure 3.2. As Smith and her colleagues (1998) note, SF A’s approach to instructional change is characterized by getting right down to business. This fact alone is likely to temporarily heighten vulnerabilities among staff until teachers and leaders have time to adjust to what are likely new ways of doing things. Because the implementation of SFA’S instructional and cooperative Ieaming routines requires teachers and leaders to make a substantial shift 6 It is important to note that some research has found some variation among SFA schools around their recourse to procedural or professional controls, where some SFA schools continued implementing more of a procedural model and resisted moving to a more professional mode of implementation. Some schools, on the other hand, continued to evolve toward a professional model. In Short, it is acknowledged that not all SFA schools can necessarily be characterized as wholly procedural in their instructional approach. 42 away from familiar instructional practices and relationships with their colleagues, the model assumes that little is to be gained from delaying the instructional change effort. Figure 3.2. Theory of Teacher-Teacher Trust Growth within the SFA Theory of Action Forming Storming Norming Performing School Leaders (Principal/Coaches) Teachers Building a Vision Teacher /Leader Hiring Quality Teacher PD Opportunities l Supportive Instructional Monitoring } tnstructional Guidance Improvement in Instruction Shared Inst. Experience Teacher Collaboration l l l Relational Trust Growth Recall that reducing vulnerabilities which arise as a result of the school improvement process is perhaps the most critical factor in trust formation among colleagues (Bryk & Schneider, 2002; Kochanek, 2005; Tschannen-Moran & Hoy, 2000). In looking at Figure 4.2, each of the four steps of the SFA model of trust building (Forming, Storming, Norming, and Performing) has its own set of vulnerabilities and/or uncertainties that are associated with that stage of the SFA implementation process. It is argued here that part of how the SF A process builds trust is by providing opportunities to address these concerns through specific activities, most of which are an integral part of the SFA process itself. 43 The Forming phase of SF A instructional change is characterized by program Ieaming and overall adjustment to the program. Both teachers and leaders are being trained in model implementation by external providers and, as such, are Ieaming their new roles and responsibilities under the SFA program. Support for program Ieaming is critical for teachers as they are being asked to substantially change the way they teach. The support that external providers give in terms of giving teachers time and help in understanding the new method of instruction and philosophy is critical in helping to ease early vulnerabilities and uncertainties among teachers. Research demonstrates that, in early stages of interactions among parties, it behooves organizations ensure situational normality, whereby individuals feel comfortable with their roles and the roles of others (Baier, 1986; McKnight, Cummings, & Chervany, 1998). Ensuring that teachers learn their roles quickly ensures that trust may begin to develop. Moreover, the role of principals in this phase is to build a vision for improvement by establishing broad goals for school improvement efforts and communicating a set of norms and values around the SF A model. As with the ASP model, these efforts provide for a foundation on which trust can develop by providing members with a collective sense of mission and a sense of responsibility for school improvement efforts (Bryk & Schneider, 2007; Fullan, 2007; Kochanek, 2005). Furthermore, due to the uniqueness of SFA’s approach to instructional improvement, the extent to which principals are able to hire key staff with either prior experience, expertise, or a Simple willingness to give their best effort in implementing the SF A model will likely ensure a stronger school community by adding more staff committed to the improvement effort (Bryk & Schneider, 2002; Kochanek, 2003, 2005). 44 In the Storming phase, initial implementation of the instructional model begins for teachers, and, as a result, this phase is characterized by vulnerabilities related to practice. For this reason, the SFA program cautions schools to be on guard for a dip in motivation and commitment as people begin to recognize the challenges of coordinating deep- reaching school-wide improvement. Teachers are struggling to master the mechanics of the instructional program, and, as a result, initial delivery is often clunky. This results in a significant amount of discomfort for teachers, and they may begin to doubt their competence. AS a result, the role of principals, facilitators, and external support staff in this phase is to mitigate these vulnerabilities by providing supportive instructional guidance. SFA can be a highly technical instructional model to carry out effectively, and, for this reason, the role of leadership in providing detailed information, feedback, and instructional guidance from experts from within the school and without is key to ensuring teachers have the support they need (Elmore, 2004; Spillane, Halverson, & Diamond, 2004). The degree to which teachers perceive this guidance to be helpful is likely to increase teachers’ trust in the model, leadership, as well as other teachers by building teachers’ confidence in themselves and the competence of their colleagues (Bryk, Lee, & Holland, 1993; Rosenholtz, 1991). This leads to our first two hypotheses regarding trust growth in the SFA model: Hypothesis 5: The hiring of key individuals to support the SFA instructional ef/Ort likely provides for conditions more amenable to the growth of trust. It is therefore expected that teacher and leader hiring will be related to change in trust over time for teachers in SFA schools. 45 Hypothesis 6: The degree of instructional guidance that teachers receive and the supportive monitoring of practice that leaders provide will be related to change in relational trust over time for teachers in SFA schools. An iterative process of performance and monitoring characterizes the Nonning phase, and the result of this process is the formation of routines. At this stage, teachers likely need less support from leadership as well as instructional materials, and gains in student achievement and instructional change are more noticeable. This is the point at which teachers can see the results of their hard work, and SFA’S reliance on procedural controls likely make instructional change more salient than, for example, the ASP model where instructional practice and the criteria for evaluating that practice are highly variable. Recognizing and celebrating gains in important outcomes such as achievement and instructional improvement can be a significant boost to trust and confidence in oneself and colleagues, particularly if recognition is initiated by school leadership. Once routines have been established in the Norming phase, focus can turn to improving results, or outcomes of implementation characteristic of the Performing phase of implementation (Success for All Foundation, 2004). Overall, a shared experience among staff, in particular teachers, is beginning to crystallize. The common language around instruction, common curriculum, and shared purpose which are characteristic of the SFA program and created as a result of implementation serve as a framework for meaningful and productive collaboration around instruction upon which the firm base of trust created thus far can deepen and solidify (Elmore, Peterson, & McCarthey, 1996; Ford & Youngs, 2009). For these reasons, this stage of trust development in the SPA 46 model is very closely aligned with Lewicki and Bunker’s (1996) identification-based trust stage. For teachers, shared purpose and common language are reinforced through collaborative aspects of the SF A process such as the quarterly assessment, component meetings, and student regrouping process. Teachers routinely work together with other staff to look at achievement data and regroup students in reading levels based on these data (Harris, 2003). These “high-risk” interactions provide a forum for the further growth of trust, as teachers are able to make discemments of competence and integrity of their colleagues (Kochanek, 2005). From here, it is possible for change to become self- sustaining as staff members continue to work together to achieve collective impact on instructional quality and student performance. This discussion leads to our final hypotheses regarding trust growth in SF A schools: Hypothesis 7: Shared instructional experience among teachers as characterized by the common curriculum, shared learning goals, and the detailed knowledge of the role obligations of other teachers in the school building, is expected to be positively related to change in relational trust among SFA teachers. Contrasting the SFA and ASP Models Now that theories as to the development of relational trust among teachers over time have been explicated for each CSR model, the purpose of this final section is to compare and contrast them and hypothesize as to the potential of each to build trust and the existence of factors which are likely important contributors to trust across the models. To begin, both approaches clearly differ in their approach to implementation and, concomitantly, how quickly they begin to impact instructional practice. SFA begins the 47 process of classroom instructional improvement straight away while the ASP program works on school-wide culture first by training staff in a process model for the development of an instructional vision. Staff first identify challenge areas and establish governance structures in order to divide and manage the different instructional priorities. In the ASP model, improvement of classroom instruction does not typically take place until well into the third or fourth years of implementation (Bloom et al., 2001). In terms of the trajectory of trust grth between the models, one issue tO be considered in each of the models is the overall attentiveness to addressing the vulnerabilities and/or uncertainties which are an inevitable part of any school improvement process. AS has been stated elsewhere, SFA employs a system of procedural controls while ASP employs a system of cultural controls (Rowan, Correnti, Miller, & Cambum, 2009; Rowan & Miller, 2007). If implemented according to plan, the SFA’S procedural controls might help to keep a tight reign on trust-threatening situations such as inadequate instructional support, and uncertainty about one’s performance and the performance of others, for example. Again, though we must be careful about painting all SFA schools as being wholly procedurally oriented in their approach, it is theorized that maintaining a strong, steady, pace of instructional improvement along with the consistent feedback teachers receive in their performance and the performance of their students, nevertheless may set the stage for potentially bountiful growth in trust. However, some research has also noted the potential ill effects of such procedural controls on teachers’ attitudes and perceptions of SF A. Datnow and Castellano (2000, 2001), in separate studies of elementary schools implementing the SF A program reported that teachers implementing the program often resented monitoring of their practice and 48 often felt that the scripted nature of the program constrained their autonomy and creativity. ASP, on the other hand, is at its very core a trust building model. At the outset of implementation, the focus is not on improving instruction, but in preparing the school culture and community for such work in the future. Staff work together developing a vision, share in making decisions about instructional goals, and form cadres to reach those goals collaborating and engaging in critical dialogue along the way. But the potential for things to quickly move off course seems higher with ASP, given that lag times between starting the improvement process and instituting instructional change in classrooms is considerable. Teachers may wonder if what they are doing is really worthwhile as they patiently wait to get to the point where they can begin to work on instructional practice. Hypothesis 8: Because building trust represents one of the core aims of the ASP model, the average linear rate of change in relational trust among teachers over time in ASP schools will be significantly higher than that of SFA schools. Recall that the second research question of this study seeks to address the question of whether or not there is evidence to support a set of factors related to instructional improvement which are common across these CSR models and a set of control schools. After a review of the literature on instructional improvement, trust, and the CSR models under study, it seems that there are likely at least a few readily identifiable factors which are related to trust growth over time. First, both C SR models contain either structures and/or philosophies which support collaboration among colleagues. Through an emphasis on data-driven decision making, and student regrouping 49 efforts, SFA provides specific structures through which teachers are able to collaborate with one another. Similarly, ASP’s overall philosophy encourages collaboration among colleagues both around the execution of the process model and as a mechanism for ongoing teacher Ieaming and improvement. Yet the need for collaboration and engagement in critical dialogue also transcends these models in terms of its role in teacher Ieaming and trust development (Hart, 1998; Pounder, 1998; Putnam & Borko, 1997; Rosenholtz, 1991), thus collaboration is likely is also related to change in trust in control schools. Moreover, collective responsibility for improvement has been shown to be a factor highly related to trust among colleagues (Bryk & Schneider, 2002; LoGerfo & Goddard, 2008), and it is clear that implicit in both the SF A and ASP models is the emphasis on collective responsibility for student and teacher Ieaming. This discussion provides us with our final two hypotheses. Hypothesis 9: Teacher collaboration and critical discourse are predicted to be related to change in teacher-teacher relational trust across both ASP, SFA, and control schools. Hypothesis 10: Collective responsibility for school improvement is predicted to be related to change in teacher-teacher relational trust across both ASP, SF A , and control schools. 50 CHAPTER IV: METHOD Recall that this study’s research questions and hypotheses concern the degree of growth in teacher-teacher relational trust over time and the aspects of the instructional improvement process most related to change in trust among teachers. These questions were examined within a sample of SFA and ASP schools as well as a sample of control schools. In order to address these research aims, data for the current study were obtained from the Study for Instructional Improvement (811), conducted between 1999 and 2004 by researchers in the University of Michigan School of Education.7 Sample, Data Sources, and Measure Construction Sample Conducted in cooperation with the Consortium for Policy Research in Education (CPRE), $11 was a large-scale, mixed-method, longitudinal study of the design, implementation, and instructional effectiveness of three of the most widely-adopted CSR models: Accelerated Schools, America’s Choice, and Success for All (Rowan & Miller, 2009). Designed as a quasi-experiment, SII followed the progress of four demographically-matched groups of elementary schools (the 3 CSR models and a set of control schools) engaged in the process of school improvement over a four-year period from 2000 to 2004, collecting extensive survey and student achievement data on all 1 15 schools involved in the study as well as detailed qualitative case-study data on nine selected focal schools. 7 For extensive detail on all aspects of 811, including instruments, findings, as well as access to all data files, please visit the SII website at: http://wwwsii.soe.umicli.cdu. 51 The 511 study sampling technique involved selecting approximately 120 study schools (30 for each intervention and 30 control) from the over 2500 elementary schools across the US. In order to select these schools as well as balance the samples of each, study researchers first compiled a list of all elementary schools affiliated with either one of the three CSR models under study. In order to reasonably control costs associated with data collection travel, they analyzed the geographic groupings of these schools and selected 17 distinct geographic regions where concentrations of these schools were high. Concerted attempts were made to both balance the schools in terms of length of affiliation with the three programs (i.e., year of initial implementation), and demographic characteristics, including socioeconomic disadvantage, geographic location, and other school characteristics. By design, however, the final sample over-represented schools in the lowest quartile of SES in order to study instructional improvement in high-poverty schools (Rowan, Correnti, Miller, & Cambum, 2009; Rowan & Miller, 2009). Because the findings of some studies suggest that trust may be particularly important for schools serving large numbers of disadvantaged youth (see, for example, Goddard, 2003; Goddard, Tschannen-Moran, & Hoy, 2001), this aspect of the 811 study design is particularly advantageous for investigating trust building. The sample of control schools was then chosen from the 17 geographic regions using the same criteria as above, being careful to match those characteristics of selected intervention schools. This sampling procedure resulted in a final sample of 1 15 K-5 elementary schools located in 45 different school districts, 15 states, and 17 different metropolitan areas. Of the 115 total schools, 28 ASP schools, 31 AC schools, 30 SFA schools, and 26 comparison schools not implementing any CSR model participated in the 52 study (Rowan & Miller, 2009). For this study’s purposes, AC schools were omitted, leaving a total of 83 schools as part of the analysis.8 Table 4.1 displays school-level averages by intervention of some important demographic characteristics of the sample, including teachers and school leaders. Table 4.] Characteristics of Schools by Sample Total ASP SFA Control (n = 83) (n = 28) (n = 29) (n = 26) Schools Enrollment 483 .84 484.96 469.90 498.19 Proportion Free & Reduced Lunch 0.67 0.62 0.73 0.64 Pre-treatment LA Achievement 98.28 97.68 94.37 103.31 Pre-treatment Math Achievement 100.0 99.32 97.42 103.62 Faculty Stability 0.70 0.73 0.72 0.64 Teachers Female 0.87 0.89 0.86 0.87 Hispanic 0.10 0.07 0.09 0.13 Afiican American 0.18 0.15 0.20 0.18 White 0.61 0.71 0.56 0.58 Other 0.07 0.04 0.09 0.08 Years Experience 12.32 12.14 11.54 13.99 Instruments and Construction of Composite Measures The $11 data were gathered via numerous survey and assessment instruments. However, for the purposes of this analysis, three main components of the study were utilized: a teacher questionnaire (TQ) developed for the study and administered once each academic year for four years (2000-2001 to 2003-2004) to all teachers within a particular school; a school leadership questionnaire (SLQ) administered once each academic year of the study to all school leaders (e.g., principals, assistant principals, instructional coaches and program coordinators) within the building; and a school characteristics inventory 8One SFA school was omitted due to lack Ofdata for analysis. 53 (SCI), which was completed primarily by the principal of each school each academic year and contained various information about staffing, students, funding, and school-wide programs. Response rates for these surveys were generally high and consistent over the four years of the study (Rowan & Miller, 2009). Detailed information on response rates is found in Table 4.2. Table 4.2 Survey Component Response Rates SCI TQ SLQ Year ratio % ratio % ratio "/0 2000-2001 73/ 107 68 1806/ 2874 63 326 / 437 75 2001-2002 110/ 114 96 2969 / 4043 73 407 / 503 81 2002-2003 107/ 107 100 2861 / 3751 76 380/ 439 87 2003-2004 103 / 104 99 3119/3650 86 391 /434 90 Both the TQ and SLQ were the primary data sources for this investigation because they contain a wide-variety of information regarding the perceptions of teachers and leaders with respect to the instructional improvement process of their school. These surveys include questions about instructional approaches and leadership, perceptions of school climate and relationships among colleagues, and teacher and leader preparation and professional development. Many of the latent measures constructed and used in this study were selected for use based on prior research on trust and its relationship with instructional improvement as well as their establishment as valid and reliable constructs in prior SII research (see Rowan & Miller, 2007). A few items which were used are new measures developed as a result of this study; they are indicated as such in the substantive discussion of the measures later on in this chapter. A full list of the measures, the items that comprise them, as well as measure reliabilities can be found in Appendix A. These measures were developed by applying a Rasch rating-scale model (Wright & Masters, 1982) to clusters of items on the TQ and SLQ surveys using the statistical 54 program Winsteps 3.69 (Linacre, 2009). This program was used to estimate scale scores for each construct for individual teachers and school leaders separately for each year of their participation. These scores are measured in logits, or log-odds ratios, with higher numbers indicating more positive reports on a particular measure. These scores were then reassembled into the person-period (i.e., “stacked”) dataset used for the longitudinal data analysis. Along with scale scores, this program produces three other statistics which are particularly useful in guiding measure construction and determining measure validity and reliability. The first of these is item difficulty, which estimates the likelihood that a respondent will endorse the position, attitude, or behavior represented by each item in the scale. Items within the scale are placed in hierarchical order on a scale (measured in logits) with higher numbers being the more difficult items to endorse. Item fit, as measured by mean infit and outfit statistics, indicates the extent to which respondents’ answers to an item are consistent with its placement on the scale of difficulty. An infit statistic of l, for example, indicates that responses to an item are consistent with its placement on the scale. Finally, in the Rasch model, reliability is estimated for both persons and items. Analogous to the Cronbach’s alpha statistic, person separation reliability is an estimate of the consistency with which persons are placed across other items measuring the same construct (Bond & Fox, 2007). In careful consultation with the statistical output of the Winsteps program, and in concert with prior work conducted on Rasch measurement construction in the S11 data, measures were constructed. Item infit is generally relied upon more to determine proper fit of items, but acceptable values for infit vary from source to source. A generally agreed 55 upon standard for maximum value of infit (and outfit) is a mean-squared value of 1.2 (Bond & Fox, 2007). This value was the comparison criterion used for determining whether an item was kept in the measurement model or discarded. The use of longitudinal data raises one other critical consideration for the measurement of these constructs. Because teachers and leaders were surveyed at one or more time points, care has to be taken to ensure that these scales remain stable over time. This was achieved by anchoring items in subsequent years on the item parameters (i.e., item difficulties) estimated from teacher and leaders’ first year responses. Summary of Study Measures and Other Independent Variables As was indicated previously, details on all constructed measures including items comprising the measures, their locations in the survey instruments, scaling, and Rasch measure reliabilities across each year of the study can be found in Appendix A. For the theoretical reasons outlined in Chapter 3, all of the constructed measures (with the exception teacher trust which was the outcome variable) were treated as time-varying covariates in the 3-level Hierarchical Linear (HLM) Growth Models (discussed in detail later in the chapter), and were therefore entered at the repeated measures level or Level 1. These measures are discussed in more substantive terms in this section, and descriptive statistics on all measures and bivariate correlations can be found in Appendix B. Outcome Variable Teacher-teacher relational trust. (sample mean = 3.39, standard dev. = 3.96, skewness = -.164, kurtosis = -.337). This scale was developed from four items on the TQ which correspond directly to items from Bryk and Schneider’s (2002) original teacher- 56 teacher relational trust instrument. It measures a teacher’s self-reported perception of the degree to which teachers respect, care, and trust one another in the school. Time- Varying Covariates (Level 1) Critical discourse among teachers. This measure was developed in prior research by Rowan and Miller (2007) to measure teachers’ reports of the extent to which the teaching faculty engage in critical dialogue about school matters including instruction. Climate of innovation and risk taking. This measure was previously developed by Rowan and Miller (2007) and was replicated for use in this study. It measures teachers’ beliefs about the degree to which they are expected and/or encouraged to improve their teaching by Ieaming new techniques, experimenting, and taking risks in their classroom instruction. Shared instructional experience among teachers. A new scale developed for use in this study, it is designed to capture teachers’ perceptions of the degree of curricular coherence and alignment supporting instruction, shared Ieaming goals, and the extent to which this knowledge is distributed across the school’s teaching faculty. Collective responsibility for improving teaching and learning. This is also a new scale developed for this study. It measures teachers’ perceptions of the degree of collective support for each other as well as collective ownership over the improvement of teaching and Ieaming which exists among the school faculty. Teacher collaboration around instruction. This measure was previously developed by Rowan and Miller (2007) and measures teachers’ reports of the degree of collaboration that occurs among teachers around matters of instruction. 57 Instructional guidance. This is a newly developed scale, but is based on a Similar measure used by Rowan and Miller (2007), with a few additional items.9 It measures the degree to which teachers perceive there to be strong guidance and support for implementing the instructional program at the classroom level, including availability of exemplars of teaching and student work, and guidance from external CSR support staff. Quality teacher professional development opportunities. This is a new scale developed for this study which measures teachers’ perceptions of the utility and/or impact of their professional development opportunities (both informal and formal) over the past year on their efforts to improve their classroom instruction. Depth of program implementation. Depth of implementation, in the sense that it is used here, relates to the degree of “buy-in” to the adopted instructional program as well as the perceived degree of “fit” of the program with the local policy environment. Prior research has established these factors as related to the depth to which a program is likely to be implemented (Porter et al., 1988; Porter, 1994). These factors were measured by means of teachers’ perceptions of efficacy in implementing the changes, their overall belief in the program and its potential for improving teaching and Ieaming, as well as their perceptions of the sense of clarity in instructional policy within the school and the ease with which they are able to implement new programs based on past policy directives. Building a vision. This scale was developed previously by Rowan and Miller (2007), and was designed to measure school leaders’ reports of the degree to which they 9 For theoretical reasons, the original scale was modified, as there were several other items which the author felt more accurately captured instructional guidance as envisioned for this study. 58 made lm'ph‘ $532344 made a concerted effort in the past year to establish and communicate a vision for school improvement and examine progress toward that vision. Values-based shared decision making. This measure was previously developed and used in Rowan and Miller (2007). It is designed to capture school leaders’ perceptions about the degree to which the school values and engages in decision making as a school community and has a decision making process in which decisions are judged against a set of shared values about improvement. Supportive instructional monitoring. This is a new scale developed for this Study which captures school leaders’ perceptions of their efforts as a supportive instructional leader. Specifically it measures the frequency with which leaders observe and monitor their teachers’ instructional practice to ensure it reflects the overall instructional goals of the program, and the amount of praise and recognition teachers’ receive for the instructional efforts. Teacher and leader hiring for school improvement. This newly developed scale measures the emphasis school leaders placed in the past year on the hiring of new teachers and administrative staff with expertise and interests to support the instructional improvement effort. Improvement in instruction in reading from prior year. This measure captures school leaders’ perceptions about the degree of instructional improvement in language arts in their school from the previous year. Improvement in instruction in math from prior year. This measure captures school leaders’ perceptions about the degree of instructional improvement in math in their school from the previous year. 59 Teacher Level Variables (Level 2) Gender. Dummy variable coded as either 0 for ‘male’ or 1 for ‘female’. Race. Separate dummy variables coded either 0 for ‘no,’ 1 for ‘yes’ indicating whether a teacher is Hispanic, African American, White, or Other. White was used as the comparison category in all analyses. Years of experience. This variable is measured in terms of the number of years of teaching experience a teacher had at the time of her/his first survey. Missing. As a fraction of the sampled teachers lacked data on some of the above Level 2 variables, another variable was created for each of these indicating ‘missing’. In these cases the original variable was coded 0 and ‘missing’ was coded 1. School Level Measures (Level 3) School size. This is a continuous variable measured as the total enrollment of a school. Percent free and reduced lunch. This is a measure of the number of students at a school eligible a free or reduced lunch as a proportion of total enrollment. It serves as a proxy for school SES or percent poverty of a school. Student achievement in reading, prior to treatment. This is the school-level average of entering Kindergarteners on the Woodcock-Johnson language arts standardized test. Student achievement in math, prior to treatment. This is the school-level average of entering Kindergarteners on the Woodcock-Johnson math standardized test. 60 Faculty stability. This school characteristic was measured as the number of teachers within the school who had been teaching there at least 5 years as a proportion of the total teaching faculty. Analytic Approach and Procedures Growth in teacher-teacher relational trust over the four years of the S11 study and its relationship to aspects of the instructional improvement process, teacher characteristics, and characteristics of the schools implementing each of the interventions and control schools was analyzed via a series of 3—level Hierarchical Linear (HLM) growth models. This final section elucidates the specific procedures by which these analyses were carried out as well as important detail about characteristics of the models fitted. Model Development and Specification Because the longitudinal data collected by the S11 study are nested within teachers who are, in turn, nested within schools, this analysis employs Hierarchical Linear Modeling (HLM) techniques (Raudenbush & Bryk, 2002). HLM methods were developed in order to address issues inherent to nested data such as dependence among observations, underestimation of standard errors, and model misspecification due to data aggregation/disaggregation problems (Bickel, 2007; Raudenbush & Bryk, 2002). Hierarchical models are superior to OLS techniques in dealing with nested data because they allow us to partition the variance in the outcome measure into its proper subcomponents represented by each level of the structure. This is important because the observations of related clusters of people (such as students nested within classrooms) are more similar to one another than those of, in this case, students across classrooms. When 61 OLS is used, this dependence results in a smaller variance in the response variable, which leads to smaller-than-expected standard errors of the parameter estimates. Underestimated standard errors increase the likelihood of rejecting the null hypothesis and achieving statistical significance (Heck & Thomas, 2009). Multilevel growth models (also known as a multilevel model of individual change) add an additional twist when thinking about the hierarchical structure of nested data. In this case, the first level of the nested structure in a growth model is the response level, in which individual responses over time are nested within the individual. This study has a three level structure: individual responses over the four years of the study are nested within teachers which are then nested within schools. The purpose of Level 1 of the model is to model individual growth over time in trust, and thus represents the change in relational trust we would expect each member of the population to experience during the time period under study. More detail on the nature of the growth models used in this study will be discussed below and in the reporting of the results. Modeling strategy. Separate 3-level HLM models of trust change were developed and tested separately for the ASP and SFA data, and, as such, each underwent its own model development and specification process guided by the theoretical framework outlined in Chapter 3. But while the theoretical framework guided the process of model building, two essential characteristics of the modeling process were the same across C SR program data analysis. First, all variables mentioned in the previous section were considered in both models, despite the fact that theory suggested that some were likely more related to trust grth than others. Second, the general growth modeling process proceeded according to well-established procedures outlined in both Singer and Willett 62 (2003) and Hox (2010). This process as well as the form of the general models constructed in this analysis are discussed in the following few paragraphs. General consensus among the “Hox Method” and Singer and Willett approaches is to begin the growth modeling process by starting with the specification of the null model, in this case, what is referred to at the unconditional means model, or the intercept- only model. AS seen in Equation 4.1, this model is characteristic in that it contains no predictors at any level, just the grand mean of the sample yooo and an error term from each level. Y1], = yOOO + rOjk + “00;. + e1], (4.1) This model allows us to partition the variance in the outcome across the three levels, and therefore calculate the intro-class correlation coeflicients, p, which are the proportions of the total variance in the outcome that exists within person, between persons, and between schools.‘0 Running this model also allows us to establish a baseline model fitting statistic (deviance or AIC, BIC statistics) with which we will assess the fit of all subsequent models. Next, we fit the unconditional growth model, which introduces the first predictor into our model—in this case the variable TIME, which is measured in this study in years, and is centered at Year 3 (Equation 4.2).ll Unlike the means model where individual —_ m The actual calculations of ICC as well as other important statistics for the SFA and ASP samples will be 'discussed in the findings sections. This discussion is Simply illustrative of the process that was undertaken. The choice of which year to “center” the TIME variable has important consequences for interpretation, as the grand mean will reflect the “initial status” of the outcome at that year. Thus centering at Year 1 would Seem logical to understand where the S11 schools were on trust when the study began. However, while §9hools within each CSR model had begun implementation by the beginning of the 811 study, some had Initiated implementation up to 2 years before the S11 study began. This fact makes interpretations of initial Status less useful in this analysis. Therefore, centering on Year 3 seemed most logical in order to understand trust grth trajectories, while ensuring that all schools would be well into implementation. 63 slopes are constrained to be flat, the introduction of the TIME variable allows us to examine individuals’ growth trajectories with respect to the outcome. Y, = yOOO + ymo TIME + (r,,. + um, + e,,.) (4.2) Examination of the various fit criteria (discussed below) allows us to ensure that the choice to adopt a growth model over a means model was justified. The remainder Of the model fitting process consists of adding substantive predictors first at Level 1, and then Levels 2 and 3 respectively, ensuring at this point that the slopes of all explanatory variables remain fixed. For this study, the general modeling strategy with regard to choosing substantive predictors is most aptly characterized as thorough but parsimonious. Because the available variable set for this Study’s analysis is considerable and the stability of the HLM model decreases substantially with every new parameter estimated, the modeling process was initiated by means of an exploratory analysis. In this process, Level I time-varying covariates were entered one-at-a-time to understand their individual effects, thus potentially eliminating from further consideration variables which were clearly not contributing to prediction. This strategy was judiciously used, however, and sought to strike a balance between adequate testing of the study’s hypotheses and interaction effects as well as model fit. On the whole, predictors which were non-significant but nevertheless a central focus of hypotheses and/or research questions remained in the models in order that the full spectrum of their effects could be ascertained. Once a final set of explanatory variables has been selected, it is necessary to eXplore on a variable-by-variable basis whether or not any of the explanatory variables has a significant variance component between groups (including TIME). At this point, it 64 is very important to exercise care in the inclusion of random effects, because the complexity of the model goes up substantially with every random effect added, leading very quickly to serious estimation and convergence problems. Finally, statistically significant random effects can be added back in simultaneously and cross-level interaction effects can be examined by “modeling” this variance with respect to Level 2 and Level 3 predictors. Again, at each point in this process, model fit statistics will be examined to determine whether or not inclusion of any and all of the “significant” effects is justified. Organized by level, the general full model structure which resulted from the process described above is written and described in detail below.'2 The first level of the model is represented in Equation 4.4, where Y represents the teacher-teacher relational (if trust score for teacher i in school j at time t. The individual growth parameter 71301.]. represents the intercept of the true change trajectory for teacher i in school j in year 3 of the study. The individual grth parameter 71?1 9. represents the linear yearly rate of grth in trust for teacher i in school j, and the error term e is assumed to be normally fir distributed with a mean ofO and a constant variance 0'2 across schools. The Tqu X qrj segment of the equation refers to the full set of time-varying covariates that will be used at Level 1, represented here in sum notation where Q equals the total number of covariates entered. '2 This general model is presented in order to aid in the discussion of the general modeling strategy. The ASP and SFA growth models varied slightly from this overall model structure, in particular with regard to the random effects which remained a part of the final models. These deviations are discussed in their respective results sections. 65 Q Y,,).=rt0,j+rthIl\/IE +Zrtqu +e q=2 (4.4) qij It] Level 2 and Level 3 of the model become even more complex, as a result of this Level 1 model. Equations 4.5 through 4.7 represent the general structure of the Level 2 models used. As can be seen, each of the parameters estimated in Level 1 are now outcomes to be modeled in Level 2. Q n0ij = BOOj +2 [30(1) X61] + r01] (4-5) q=l nw=flw+ng Ho thy- = [3qu for q = 2, ..., Q. (4.7) In modeling the intercept from Level 1,, we get Equation 4.5. [300]. is a coefficient representing teachers’ average relational trust score in school j in Year 3 of the study. In this equation, the term X (11' signals that Q number of teacher-level variables have been included in this model. The coefficients of those predictors BOW. measure the direction and strength of association of these variables and the respective growth parameters in the model. Equation 4.6 models the slope parameter 7131 I]. from Level 1, and includes [3le, which represents the average rate of linear growth for teachers in school j, and a random- effect, r11], which allows us to examine variation in the rate of linear growth among teachers within each school. Equation 4.7 simply denotes that we model each of the other q number of Level 1 covariate parameters as a fiinction of an intercept, which represents teachers’ average score on covariate X qr'j in school j in year 3 of the study. 66 60 El SU iii \V Level 3, in turn, models the parameters which were a part of the Level 2 model equations. The Level 3 models are represented by Equations 4.8 through 4.10. s [300]. = YOOO +2: yOqs W57 + u00/' (4-8) s :1 B10) =y100+ ”le (4.9) qujzyqoo forq=2, Q. (4.10) Equation 4.8 gives us the model of the intercept at Level 2, which is represented as the sum of 'y 000, which represents the grand mean of the teacher-teacher relational trust measure in Year 3, the error term for the intercept 1100]., and the sum of the S number of school level predictors in the model. In Equation 4.9, YIOO represents the grand mean for the linear rate of change in relational trust in the sample, along with a random effect u 10]. which indicate that it is assumed that schools vary around the grand mean in their growth rate of relational trust over time. The final equation, Equation 4.10 represents the modeling of the q number of parameters estimated by Equation 4.7 in Level 2. Each is assumed fixed, with no error term, and produces a grand mean quO for each of the Q number of time-varying covariates parameters at Level 1.” Statistical packages and estimation method. Growth modeling was carried out using the statistical package HLM 6.02, using the Full Maximum-Likelihood Estimation (FML) method. In 3 level analyses with the HLM program, only F ML is available. Construction of data files and supplementary analyses such as bivariate correlations and ‘3 It is acknowledged, however, that the study design chosen here, with its use of time-varying covariates at Level 1, is but one possible way to study the relationship of instructional improvement factors to change in trust. Other model specifications such as including these factors as teacher-level characteristics at Level 2 and using these to model variation among teachers in trust growth. 67 th‘S‘C anal (All C011 hat COII tit“: COT. is if ma. 0i 1 for par mt 1i3: elf descriptive statistics were carried out with SPSS 17. More detail on supplementary analyses is contained within the discussion of the results. Model fit statistics. Model fit was assessed using the Akaike Information Criterion (AIC) statistic (Akaike, 1987). The decision to utilize this statistic instead of the more common deviance statistic was in part because of the nature of the modeling process undertaken. Both AIC and deviance are based upon the log-likelihood statistic, but they have different assumptions. Most importantly for this analysis, chi-squared test comparisons of models using deviance statistics require that the model being compared is nested within the previous model (Singer & Willett, 2003). In the case of the analyses conducted here, this was not always the case. The distinct advantage of the AIC statistic is that, as long as sample size remains constant, models do not have to be nested— making it a more appealing choice for this study (McCoach & Black, 2008). Further, the AIC statistic offers a slightly more rigorous fit comparison—it penalizes for the number of parameters estimated in the model, as can be seen by its calculation. The calculation for AIC is simply d + 2q, where d is the deviance statistic and q is the number of parameters estimated. One disadvantage, however, is that there is no formal statistical test available for comparing models using AIC. In these instances, model fit was assessed simply by considering models with larger AIC statistics to have the greater fit. The HLM program does not provide AIC statistics, so they were calculated manually. Model assumptions and diagnostics. Modeling population behavior using multilevel techniques carries with it a certain set of assumptions. These assumptions are very similar to those for standard OLS regression techniques: linear functional form, errors are independent and/or unrelated to predictors, and errors are normally distributed 68 iiilh 8 assunt files 0 Becan the fill nuH n nasci rendi Rand; gener linclt panic been Hand Very exam BSSUr reSldi with a constant variance (homoscedastic). Because HLMs have multiple levels, these assumptions have to be assessed at each level in turn. The HLM program provides output files of the residuals, fitted values, and predictors in the models at all three levels. Because it is simply not feasible to check the assumptions of the models at every phase of the modeling process, assumptions were checked at the beginning of modeling during the null model phase to detect gross violations of assumptions as well as when the full model was completed. Normality was assessed through visual inspection of the histogram of residuals at each level as well as the normal probability plot of all error terms. Homoscedasticity of residuals was assessed by generating a scatterplot of the standardized residuals of each estimated parameter against their predicted values, and generating scatterplots of the standardized residuals against each predictor in the model (including TIME). In cases where the possible presence of heteroscedasticity exists, particularly at Level 1, it may have several origins: 1) important Level 1 predictors have been omitted; 2) a random effect has been erroneously treated as fixed; or 3) the outcome data are not normal, and have heavy tails (Raudenbush & Bryk, 2002). Procedures for addressing these issues when present are to first investigate the potential causes of heterogeneity, and then seek solutions to address them. Examination of patterns in the standardized residual versus predicted plots at Level 1 for the entire sample revealed a very slight possibility of heterogeneity in errors, although the overall result of this examination was inconclusive, considering that all other normality and homogeneity assumptions were found to be met. In the case of growth modeling, there is often a greater likelihood that Level 1 residuals will be heteroscedastic and autocorrelated within person over time (Bickel, 69 2007; S 10 the s prei'iiit maybe considi anal} Si determ modcli and dc addres (2010'; unbias increa estimz 111018 ; Bryk. 513ml: lmpor 0f the Small. PTOVIC 312mm“ 2007; Singer & Willett, 2003). That is, we would not expect that one person’s responses to the same measure over time would not be in some manner related to his or her other previous responses. This suggests that the assumption of a diagonal covariance structure maybe inappropriate for longitudinal data (Singer & Willett, 2003). Taking this into consideration, alternative covariance structures were explored for the data, by means of analysis in SPSS.l4 Appropriateness of an alternative covariance structure was determined based on comparisons of model fit with fit statistics generated during modeling using the default diagonal structure. While sources of possible heterogeneity were thoroughly investigated in all cases and determined to be inconclusive, the most practical and conservative solution to addressing these concerns was to employ the use of robust standard errors. As Hox (2010) notes, the parameter estimates produced by the F ML method are consistent and unbiased, meaning that they tend to get closer to the true population values as sample size increases. However, the standard errors are likely underestimated. Robust standard errors estimated by the HLM program adjust for this underestimation, typically resulting in more accurate significance tests and confidence intervals (Hox, 2010; Raudenbush & Bryk, 2002). Raudenbush and Bryk (2002) suggest that a comparison of asymptotic standard errors and robust standard errors produced from FML estimation can provide important evidence for the potential magnitude of departures from normality. In the case of the analyses conducted here, differences in these two standard error calculations were small. Nevertheless, it was determined that employing the robust Standard errors would provide more assurance that Type 1 errors were further reduced in the analyses. More '4 The HLM 6.02 program does not provide for testing of alternative covariance structures under standard HLM3 modeling. However, the SPSS 17 mixed model routine does allow for the testing of these altemative structures, and also provides AIC statistics for model fit comparisons. 70 detail on addressing specific issues of heteroscedasticity in the models is provided in the discussion of results. Finally, and perhaps most importantly, is the assumption of linearity. There are a few issues related to the assumption of linearity in the modeling for this study that should be addressed. First, the general rule of choosing a functional form for the growth model is the more parsimonious the better. A linear functional form is the simplest form, however it may not always be the most parsimonious. There are limitations, though, to the fiJnctionaI forms which can be reasonably supported by one’s data. Generally speaking, the rule of thumb for function form is, for k measurement occasions, fitting a polynomial of at least k-l is preferred (Hox, 2010). This means that for the 4 waves of data in this study, we can reasonably fit no more than a cubic order polynomial.'5 Generally speaking, if higher-order functional forms are more favorable in comparison to a linear model, evidence can be seen in the significance of the TIME Slope parameters as well as the overall increase in the model fit statistic over the linear model. In order to more completely assess the choice of functional form at the outset of modeling, a more elaborate procedure was undertaken, however. A procedure for examining the individual growth trajectories of randomly-selected individuals in each intervention was implemented according to detailed instructions contained in Singer and Willett (2003). They specify randomly selecting 10 individuals and plotting their change in the outcome (in this case relational trust) over time, then running within-person OLS analyses of these individuals gathering their intercept and slope values, residuals, and fit statistics in a file. Using this file, they specify superimposing the “fitted line” on their '5 Some suggest an even stricter criteria. For example, for 3 waves of data you must assume a linear functional form (Singer & Willett, 2003). 71 til: 11 35 null fell VET! me 01‘ OI int 13 individual change plots, and examining this as well as residual and fit statistics to determine whether or not a linear assumption is tenable. After conducting this analysis, there was no evidence to suggest that individuals exhibited non-linear change in teacher-teacher relational trust over time. This assumption was again tested in the initial phases of growth modeling by comparing the fit statistics of null models with more complex polynomial forms to those of the linear growth model. This procedure was repeated for each separate dataset (i.e., ASP, SF A). For testing linear relationships among predictors and the outcome, scatterplots of the growth parameters versus predictors were plotted and examined for evidence of linearity of relationships. Centering of variables. The centering of explanatory variables used in a multilevel analysis—whether it be in relationship to the grand mean, group mean, or no centering at all—has important consequences. Centering a predictor around its grand mean—the most common method of centering—can serve to enhance the interpretation of data, without changing the “fit” of the model or affecting the interpretation of the slope or residual estimates (Kreft & DeLeeuw, 2006). Grand-mean centering can facilitate interpretation of estimates when it does not make sense to interpret an intercept for a value of 0 on a variable for which 0 is not a plausible value (Hox, 2010). Perhaps most importantly, however, grand-mean centering can also significantly reduce collinearity among predictors, particularly when within-level interaction effects are examined, adding statistical stability without changing the underlying model (Tabachnick & Fidell, 2007).l6 '6 Examination of the correlation tables in Appendix B demonstrates that some of the predictors used in this analysis have moderate correlations with one another. An OLS regression was performed with the raw, uncentered data in order to take advantage of the collinearity diagnostics available in SPSS. After entering all Level I predictors, VIF statistics indicated no value for a single predictor over 1.6, and all condition indexes were below 8. This, coupled with grand-mean centering, provides some assurance that collinearity will likely not become a significant issue in the analysis, particularly because not all predictors are likely to be used in a single model. 72 In t their grand \lltile it is idtintogc education factors 1R centered ‘ main fort llltlll idua only at L. teacher b missing t teaching al'OltI ll‘ll leathers Were mi COHCSpr IIIC’SQ Va 131' lable 110 more clearing In this study, all variables at each level were centered at Level 1 with respect to their grand-mean, with the exception of the TIME variable, which remained uncentered. While it is most common to grand-mean center only variables at Level 1, it can also be advantageous to grand-mean center higher-level variables (in this case race, gender, education), as this process can adjust for average differences between schools on these factors (Raudenbush & Bryk, 2002). Therefore Level 2 and Level 3 predictors were also centered with respect to their grand-mean. Missing data. In terms of longitudinal data analysis, missing data can occur in two main forms, missing data in the form of missing cases at time points, and missing data on individual variables within a case. In the HLM program, data are allowed to be missing only at Level 1. At any other level, missing data result in data for the entire school or teacher being deleted. AS a result of building the analytical files for use, there was missing data only on the characteristics of teachers at Level 2, such as gender, race, teaching experience, and educational attainment. In order to address this issue, and to avoid these cases from being lost completely, a series of dummy variables that indicated teachers were missing on these predictors was created. More Specifically, when teachers were missing on one or more of these predictors, the variable itself was coded “0” and the corresponding dummy variable was coded “1”. In essence a new category for each of these variables was created which indicated “missing.” Missing data on individual variables, which were very small in comparison to overall sample size (for each sample, no more than 10% on any single variable), and were subject to listwise deletion upon creating the data file (otherwise known as the MDM file) for use in the HLM analysis. 73 Missing data can also be discussed in terms of missing cases at time points for an individual. Within this particular sample, there are individuals who have data from all four survey points, and some who have only one. The median and mode number of time points for individuals was 3. The advantage of using HLM in concert with F ML estimation is that all participants with at least 1 observation can be included, and the results of such an analysis can be interpreted as if no missing data were present under the assumption that data are missing at random (Raudenbush & Bryk, 2002). 74 Th pop‘ll‘ll ('5 program: schools. ll of trust do chapter is conducted lRfiBHv Pu.) CHAPTER V: TRUST GROWTH IN AC CELERA T ED SCHOOLS This study is an exploratory analysis of the degree of trust growth within two popular CSR programs—the Accelerated Schools program and the Success for All program—and the factors most related to that growth in these and a set of control schools. The purpose of this study is to develop a better theoretical grounding for the role of trust development in instructional improvement and vice-versa. The purpose of this chapter is to report and discuss the findings from the analysis of trust growth which was conducted on the sample of Accelerated Schools in the 811 study. In doing so, it addresses the first two research questions as they pertain to the sample of ASP schools: 1. To what extent is there growth in relational trust among teachers over time in the S11 sample of Accelerated and Success for All schools? As models of “typical,” non-intervention schools, to what extent is there growth in relational trust among teachers over time in the S11 sample of control schools? 2. What factors related to the instructional improvement process are most related to change in teacher-teacher relational trust over time in each subsample of schools? In addressing these research question(s), this chapter’s presentation of the results is organized by the hypotheses regarding trust grth in the ASP model posited in Chapter 3. This chapter will begin with a brief discussion of ASP sample as well as the general model development process. ASP Model Development 75 In order to address questions regarding the degree of trust growth in ASP schools and the instructional improvement factors related to change in teacher trust, a 3-level HLM growth model of relational trust among teachers was developed from the ASP data. As was mentioned in the discussion of method in Chapter 4, the ASP data underwent its own model development and Specification process guided by the theoretical framework outlined in Chapter 3. However, while the theoretical framework guided the process of model building, two essential characteristics of the modeling process were the same across CSR program data analysis. First, all variables discussed in the previous chapters were considered in the models, despite the fact that theory suggested that some were likely more related to trust growth than others. AS previously stated, the general modeling strategy with regard to choosing substantive predictors could be most aptly characterized as thorough but parsimonious. Because the available variable set for this study’s analysis is considerable and the stability of the HLM model decreases substantially with every new parameter estimated, the modeling process was initiated by means of an exploratory analysis. In this process, Level 1 time-varying covariates were entered one-at-a-time to understand their individual effects, thus potentially eliminating from further consideration variables which were clearly not contributing to prediction. This strategy was judiciously used, however, and sought to strike a balance between adequate testing of the study’s hypotheses and interaction effects as well as model fit. On the whole, predictors which were non-Significant but nevertheless a central focus of hypotheses and/or research questions remained in the models in order that the full spectrum of their effects could be ascertained. 76 Second, the general modeling process proceeded according to procedures established and outlined in both Hox (2010) and Singer and Willett (2003). Briefly, this process involves first fitting an unconditional means model, which contains zero predictors and provides baseline fit statistics such as deviance and AIC, as well as baseline error variances at each level which allow for the calculation of the intraclass correlation coefficient. The unconditional growth model is then fit, which introduces the TIME predictor at the response level (Level 1). From here, model building proceeds with the judicious entering of substantive predictors first at Level I, (known as time-varying covariates), and then at Levels 2 and 3, keeping all of the Slopes of these explanatory variables fixed. Once a final set of explanatory variables has been determined, one-by- one we can then test whether or not any of these have a significant variance component, and whether or not any of the substantive predictors explain this Slope heterogeneity. Finally, interaction effects both within-level and cross-level interactions can be explored, and we can arrive at a final model. This is also the same procedure that was employed in the SFA and Control models as well. The final model which was developed and is discussed in the following section is represented here in Equations 5.1 to 5.7. You will note that this is the same model structure given in Chapter 4. Equation 5.1 displays the Level 1 model. Q = no, + n“, TIME +2: n,,,. x,,., + e, (5.1) q=2 Y tij Level Two of the ASP model is represented by Equations 5.2 to 5.4 below. Q 1:0,]. = BOO]. +Z 130‘”.qu + r00. (5.2) q=l 77 7th]. 2 + rm. (5.3) le “a = quj for q = 2, Q. (5.4) The Level 3 models are represented by Equations 5.5 through 5.7. 3 B00]: Yooo +2 7qu Wsj + ”00]- (5-5) 3 =1 B10j=Y100+ “le (5.6) [5‘10]:qu forq=2, ..., Q. (57) Results As a result of Rasch measure development and HLM file construction, the final sample for the ASP analysis was 2418 responses within 1214 teachers in 28 schools. The presentation of the 3-level HLM growth model of relational trust among ASP teachers is found in Table 5.1. Beginning with the fitting of the unconditional means model (Model 1 in the table), total variation in the outcome of the ASP sample is partitioned into its respective levels, within-teacher, between teachers, and between schools. Calculation of the intro-class correlation coefficient (ICC) from these values is straightforward and, for the purposes of illustration, is represented for Level 1 in Equation 5.8. p — 63 58 028 o§+o§+o§0 (') For the ASP sample, these calculations found that 41.4% of the total variance in teacher- teacher relational trust was found at Level 1, 41.1% was found at Level 2, and 17.5% was 78 8m. $.48- 550 8m. 83. 82m 4mm. 82. 2532: m8. :3- 23$ o5. Cod mocotoaxm (.0 83> whouomflohn— N _o>04 :38. E8 .288. 3:8 1&8. 88.8 :88. £2 8;. 50225 .3: 25 .88. we .8 1.98. 2 3 5&9 =2§8§a§ 8:9. 188. $3 3&8. 83 bzééaam 282.8 88. 8.8.8 8%: Sign 822m #8. 83 in 8. 8.8 28¢. 8. .o 855 a 88.3 so. 38.. N8. 38.- 88. 88.- 858230 553... 88. 88.8 88. :88 .88. ES ooscaxm .55 83% o8. 88.8 o8. 888 1.88. 38.8 8:850 _2o_aa§_ ...m8. 588 ...N8. :88 :8. 48.8 E 223 3:80 :48. 5.8 :38. em a .o 3:88. 83 8:285 co 35:0 :88. 48.8 :88. v8.8 :88. 84.8 3.885 526 .31,me N506 ***©V_. conga. “Intofi mow.m ***wOm. Chad ***mNm. mvwfi coo». «mop—3:— 3“: 3088:— muouomfioum ~ _o>o..— ma. 88 mm 88 ma. 88 .8 88 pa. 88 283 Ext 236ch 42825 W .252 4 382 m 382 N .252 _ .082 éoofim RWY 2% 22333 M558 ERR 32833348 Rho: SEED 33% 364%. ..m 033. 79 .68on 35$ 2: 3 Bum—=23 mm .033. 25 E 3:03.. 8m 38.5 @3285 annoy? .802 2. v a: .8. v a... .8. v a: .88. v at... £3 28 £8 £8 £8 a; 2&3 _ .93 Tm E 2 w v 38.5qu £80683 vodmm: 26mm: $53: 3.80m— co. $09 0?. vowwv: m_ . _ mm: $6M: ._ 3.03m“ 86on 8:250 mg. NE. owe. m5. floor—om 5958mm m 351— Nom. new. 5v. moo. Eon—om“: :ooéom m $54 Em. Nam. EN. ma. b:oao$£§3_ .051— wotmtacm. 3&3:wa ohzmmo‘ *Eo. owcmco mo Sam M 634 ***NN.O “linowd ***N©.o ***h©.N fin—5N mafim FE:— m _o>o..~ ......mmo. owcmzo mo 83— m 654 31.2% ***v_.m ***N©.m “lawnmé *fwomé maflm BEE N _o>o..— ***Nm.v ...,Ibvav ***Nh.v ***OM.© ***Nv.© coflom 55:5 fl _o>oA flowfim thnax owe. mood EoEo>oEo< 522 SE m S. mood- EoEo>oEo< <1— cot; ...:8; 2: .4 33% .038... So. oood- mum Begum .1.va . 2:4- £053 coon—vex do can «cocoon £8235 m 654 zmoo. omo. T 9:32: 8mm *8; oomé- wEmmmE 5980 mm». m»: .o- wEmmmE oocotoaxm .E> 68558 2 2%: 80 \\ i” of found at Level 3. From the means model, the baseline deviance statistic was also obtained (12683.90) as well as the AIC fit statistic (12691.90) which were used in subsequent model fit comparisons. Finally, this model yields the grand mean for the sample in Year 3 on the teacher-teacher relational trust measure,'y 000, which was measured as 3.845 logits. The fitting of the unconditional growth model (Model 2, Table 5.1) proceeded with addition of the predictor TIME at Level 1, which was centered on Year 3 of the study.17 All parameters remain fixed at this point, including the random effects on the rate of trust change at Levels Two and Three. The slope of the population average change trajectory,'y1 00, was estimated at .318 logits in the unconditional growth model. This means that, on average, each year of the S11 study teacher-teacher trust increased by .318 logits in ASP schools. We reject the null hypothesis for both the intercept and slope (p < .001) of teacher—teacher trust, concluding that both have an estimated value which is statistically different from zero. AIC fit statistics demonstrate that the linear change model is an improvement over the means model, although pseudo R-squared values remain slight. Model 3 introduces the first explanatory variables at Level 1, otherwise known as time-varying covariates. Model 3 represents the uncontrolled correlations of these various covariates on change in teacher-teacher trust. As was discussed, several predictors did not make it into the set of final models, but were eliminated during exploratory analysis because they were simply not predictive of relational trust growth in ASP schools. Recall that the first hypotheses posited in Chapter 3 for ASP schools predicted a positive 17Of course comparisons were made of model fit between the linear and the quadratic time parameters. Addition of the quadratic time parameter decreased model fit and was non-significant (p = .983, AIC = 12663), so the linear model was adopted in favor of the more complex model. 81 relationship of values-based shared decision making and building a vision to teacher- teacher relational trust growth. The results in Model 3 indicate some mixed findings with respect to these hypotheses. Shared decision making was not found to be significantly related to change in teacher-teacher trust (coef = .038, SE = .060, p = .526), while the efforts of the principal to build a vision for school improvement were positively related, albeit modestly, to change in teacher-teacher trust (coef = .102, SE = .057, p = .070). The final two hypotheses predicted a positive relationship between relational trust and collaboration, critical discourse, and innovation over time in ASP schools. As is evidenced in Model 3, both critical discourse and innovation have relatively large, and consequently statistically significant, effects on change in teacher-teacher trust over time in ASP schools (coef= .407, SE = .029, p < .001; coef= .202, SE = .026, p < .001, respectively), however collaboration was not found to be significantly related to trust growth over time. Examining model fit and pseudo R—squared change from the unconditional growth model, we note that model fit and variance explained have improved substantially over the previous model. Over 25 percent of Level 1 variance, 43 percent of Level 2 variance, and 66 percent of Level 3 variance in the teacher-teacher relational trust measure alone has been explained by these time-varying predictors. We also note that the slope of the population average change trajectory remains highly significant. Model 4 introduces two important Level 1 control variables—collective responsibility and policy implementation depth—to the models, and the relationships of 82 both to change in teacher-teacher trust are significant (p < .001 and .01 respectively).18 While many significant associations of time-varying predictors in Model 3 were eliminated with the addition of these controls, it is important to note that the correlations with which are hypotheses are concerned remain significant, albeit diminished. Quality teacher professional development opportunities represent one factor other than those previously mentioned whose relationship remained statistically significant with the addition of the control variables (coef = .047, SE = .022, p < .05). This finding means that, for every 1 logit increase in quality teacher professional development, growth in trust is increased .047 logits for teachers with an average score on all other time-varying predictors. Further, model fit is increased as indicated by the AIC statistic (A = 204.75). The final model, Model 5, all Level 2 and Level 3 controls are entered as well as the significant random effects, which happened to be only those associated with the rate of change in trust over time at the teacher and school levels. Again, first and foremost, even after controlling for teacher and school level characteristics, all significant associations in Model 4 remain in Model 5 including the slope of the average grth in trust. With regard to important Level 2 and 3 predictors, we see first that no teacher level characteristics are significant. At the school level, there are two important findings. First faculty stability as measured by the proportion of the overall number of faculty who have been in the school building more than 5 years, was found to be significantly related to trust. The interpretation of this finding is that, for every 1 percent increase in faculty stability, there is a corresponding 4.1 percent increase in the average trust level for ASP '8 These variables were viewed and treated as “control” variables in these and all subsequent models for two main reasons. First, because their effects on trust are relatively well-established in the literature, and second, though this study was interested in the associations of these two covariates on change in teacher- teacher trust, on the whole they were not a central aspect of this study’s analysis. In other words, they were not a part of our theoretical models of trust grth in the two CSR models. 83 schools. Not surprisingly, percent free and reduced lunch is negatively associated with levels of trust in the average ASP school (coef = -1 . 10, SE = .349, p < .01), indicating that schools with higher proportions of impoverished students have lower trust levels on average. Finally, with the inclusion of the random effects on the trust change slope parameters at Levels 2 and 3, we see that there is proportionately small but significant variation among teachers and schools in trust grth trajectories over time. Attempts to model this variation with both teacher and school level characteristics were unsuccessful, so this variation was lefi unexplained in the models. Interaction effects were also extensively investigated at this stage in the modeling process, however none were found to be significant. We see that the ‘final’ model presented here accounts nearly 92 percent of the variation in the outcome at the school level, and half of the variation in the outcome at the teacher level. In concluding the discussion of the results, discussion of HLM growth model assumptions and investigation and verification of these assumptions is essential. Because of the complex nature of both HLMs and the sheer number of predictors which were a part of this analysis, model assumption checking is a considerable undertaking, and, for this reason model assumptions were thoroughly checked at the beginning of modeling and at the final model. In the interim modeling processes, the assumption of homoscedasticity was most frequently checked, particularly with regard to the relationships of predictors to residuals. Examination of the normality of errors assumption was accomplished through a standard P-P plot and examination of scatter plots of standardized residuals versus ID 84 numbers, and histograms of residuals at all levels. Probability plots were highly normal, with no unusual data points, as well as all plots and histograms. Homoscedasticity at Level 1 was assessed by examining both scatter plots of standardized residuals versus the TIME predictor as well as scatter plots of standardized residuals versus predicted values. Scatterplots of TIME suggested equal variability at all time points. However, examination of the scatter plot of residuals versus predicted indicated the possibility of some very slight heterogeneity in errors at Level 1. While this examination was inconclusive, care was taken to thoroughly investigate the possible sources and available remedies, given the potential to bias standard error estimation. Given that the standard 3-level HLM analysis does not allow for the testing of alternative covariance structures, they were tested by means of the Mixed Model routine in the SPSS program. None of the available alternative structures (e.g., heterogeneous ARI, unstructured, etc.) provided a better fit over the default diagonal structure as measured by AIC model fit statistics. In the end, the solution which best addressed the problem was to employ the use of the robust standard errors in the analysis. As empirical growth plots of randomly selected individuals were conducted at the outset of modeling, the linearity assumption was checked primarily with regard to the relationship of Level 2 and 3 residuals to predictors. No evidence was found to suggest that any individual predictors had a curvilinear relationship to the outcome. Discussion First and foremost, the modeling of ASP schools with respect to growth in teacher-teacher relational trust over time revealed significant, positive growth over the four years of the S11 study. This is very much in line with the observation that the ASP 85 model is one which has instructional design inherently conducive to trust growth. That is, it seeks to initiate collective work on instructional improvement by establishing strong norms and values around improvement (Rowan, Correnti, Miller, & Cambum, 2009; Rowan & Correnti, 2009; Rowan & Miller, 2007), and requires the collective striving of school staff (and other stakeholders as well) to achieve this improvement (Finnan & Swanson, 2000; Hopfenberg et al, 1993). Knowing that, within this sample of ASP schools, trust growth is occurring over time, what are the factors in the ASP model most related to change in teacher-teacher trust? Revisiting our theoretical framework in Chapter 3 and comparing that to our findings with respect to theory and hypotheses, we can draw several conclusions. Overall, we see substantial support in the results for our “theory” of trust development. First, it was hypothesized that the efforts of leadership to build a vision for instructional improvement would be positively related to trust, given that the latitude that teachers have for their designing their instructional approach could lead quickly to fragmentation among the staff with respect to improvement. The findings here suggest that school leaders’ efforts in this regard might be related to increases in the social cohesion of the school over time. Due to the likelihood of it being perceived as authentic in the ASP model, shared decision-making, which was hypothesized to be related to change in teacher-teacher trust was not found to be associated with teacher trust. It may certainly be the case that authenticity is not the only important criterion to ensuring shared decision-making is successful in building relationships among colleagues. Implementation of the shared decision-making process (and the process model more generally) is complex in the ASP 86 model, and, at the time of the S11 study, not particularly well-defined. This could have potentially led to confusion about the nature of the instructional changes the ASP model was seeking to make among staff (Bloom et al., 2001). It stands to reason that shared decision-making without a great deal of focus or understanding of what is supposed to be occurring will, in the long run, result in either flat or decreased trust among staff over time. Other hypotheses regarding aspects of the school culture which would likely be related to change in teacher-teacher trust met with significant support as well, with a few exceptions. In fact, three of the four factors included in our model of trust building in the ASP theory of action were confirmed, leaving only teacher collaboration as not statistically significant. First, critical discourse among teachers was found to be strongly associated with change in teacher-teacher trust. While many past studies had more tentatively linked critical discourse to trust, this analysis further confirms that critical dialogue among teachers is related in important ways to trust. The climate of innovation and risk-taking which is present in the school was also positively related change in teacher-teacher trust. ASP is a model which encourages teachers to take risks and innovate when it comes to discovering paths to “powerful learning” for their students. Although intuitively logical, there has been little concrete support for innovation and risk-taking as an important factor in trust growth in prior research. Because trust itself is based upon mitigating risk, those teachers who deliberately take risks in their instructional practice are increasing their own vulnerability and uncertainties as well as those of the collective group with regard to the instructional effort. However, the fact that there is a strong cultural norm for innovating and risk-taking in some sense makes these 87 vulnerabilities acceptable and manageable, and, in the cases where risks and innovations pay off, it is likely that significant amounts of trust are to be gained among teachers. As the saying goes, where there is high risk, there is also the opportunity for high reward. The absence of a relationship for teacher collaboration, however, was an important finding, given that many prior studies have established the importance of collaboration among teachers to trust. One potential reason for this finding could be the measure itself and the process of improvement in ASP schools. The measure of collaboration measures teachers’ perceptions of the extent to which they clarify standards for student learning, develop thematic units, examine the scope/sequence of topics and, examine the alignment of materials and student assignments. With the exception of the first of these—clarifying standards for student learning—these areas of collaboration are not likely to be addressed in the initial years of the ASP model process. These types of collaborative activities in ASP are likely to happen later on once broader issues of instructional improvement such as standards and instructional vision are established. Further, because teachers had a certain degree of autonomy with regard to instructional decisions in their classrooms, it could be that teachers engage less in these types of activities in favor of others which were not directly measured. Collective responsibility, which has been established in prior research as related to trust (for a review, see Smylie & Evans, 2006), has further support in this study as a strong predictor of trust grth over time. Depth of policy implementation, as measured by teachers’ belief in the model and willingness to implement it well, as well as alignment of the intervention with the current policy environment, is also a significantly related to change in teacher-teacher trust. This is an important finding in the ASP model, 88 because if we assume that ASP is an instructional model conducive to trust building, we might expect that, as teachers implement the program more deeply, trust would also improve. Finally, while no teacher-level effects were found in the ASP model, two key school-level factors were significantly related to teacher-teacher trust. Most importantly, ASP modeling found an important relationship between faculty stability and average trust scores in the focal year (Year 3). Other studies have suggested that faculty stability is critical to trust development, and, in the ASP schools, it seems this is the case. This might be expected, given the complexity of the instructional intervention and the reliance of the model on cultural controls. An unstable faculty with a high degree of turnover is likely to have many difficulties establishing the committees, cadres and other aspects of the process model necessary to implement the instructional program. As new teachers arrive, they will need time to become acclimated to the ASP process and, as such, some of the critical momentum which is accumulating at the beginning of implementation is likely to be lost. Similarly, schools with higher proportions of poor students also have significantly lower trust scores among teachers. This is certainly not surprising when we consider the number of important school outcomes which are also related to SES such as achievement. 89 CHAPTER VI: TRUST GROWTH IN SUCCESS FOR ALL SCHOOLS The purpose of this chapter is to report and discuss the findings from the analysis of trust grth which was conducted on the sample of Success for All schools in the S11 study. In doing so, it addresses the first two research questions as they pertain to this subset of schools: 1. To what extent is there growth in relational trust among teachers over time in the SII sample of Accelerated and Success for All schools? As models of “typical,” non-intervention schools, to what extent is there growth in relational trust among teachers over time in the S11 sample of control schools? 2. What factors related to the instructional improvement process are most associated with change in teacher-teacher relational trust in. each subsample of schools? In addressing these research question(s), this chapter will begin with a brief discussion of the general model structure employed for the SFA analysis, and then proceed with the presentation of the results, which is organized by the hypotheses regarding trust growth in the SFA model posited in Chapter 3. SFA Model Structure In order to address questions regarding the degree of trust growth in SF A schools and the aspects of its instructional improvement process related to that growth, a 3-level HLM grth model of relational trust among teachers was developed from the SFA data. The final model which was fit is represented here in Equations 6.1 to 6.7. You will note that this is the same model structure given in Chapter 4, with one exception—the entry of 90 predictors to model variation in trajectories of trust change over time between schools (see Equation 6.6). For the sake of brevity, the interpretations of the terms in these model equations are omitted, simply because they are the same as in the previous two chapters. Equation 6.1 displays the Level 1 model. Q q =2 Level Two of the SFA model is represented by Equations 6.2 to 6.4 below. Q 71’07 : B001' +2 BOqj Xqi + r01] (62) q = l nw=fiw+hi on TCqI-j = [3qu for q = 2, ..., Q. (6.4) The Level 3 models are represented by Equations 6.5 through 6.7. S BOOj : Yooo +2 7qu Wsj + ”001' (6-5) 3 =1 S [3101' = yioo +2 yiqs Wsj + ”le (6-6) s =1 [3qu = yqoo for q = 2, Q. (6.7) Results As a result of Rasch measure development and HLM file construction, the final sample for the SFA analysis was 2181 responses within 1170 teachers in 29 schools. The presentation of the 3-level HLM grth model of relational trust among SF A teachers is found in Table 6.]. Beginning with the fitting of the unconditional means model (Model 91 l in the table), total variation in the outcome of the SF A sample is established and partitioned into its respective levels, within-teacher, between teachers, and between schools. Using the baseline means model variance statistics, calculation of the intra-class correlation coefficient (ICC) for the SFA sample found that 43.7% of the total variance in teacher-teacher relational trust was found at Level 1, 42.2% was found at Level 2, and 14.1% was found at Level 3. From the means model, the baseline deviance statistic was also obtained (11535.01) as well as the AIC fit statistic (11543.01) which were used in subsequent model fit comparisons. Finally, this model yields the grand mean for the sample in Year 3 on the teacher-teacher relational trust measure, which was estimated at 3.159 logits. The fitting of the unconditional growth model (Model 2, Table 6.1) proceeded with addition of the predictor TIME at Level 1, which was centered on Year 3 of the study. Though in the modeling of the ASP schools the rate of change in relational trust between teachers and schools remained fixed at Levels 2 and 3 until the final model, early analysis of the SFA data suggested that these parameters be allowed to vary throughout modeling.‘9 When these random effects are included, variance in further partitioned into that which indicates variance in fixed effects within teacher, between teacher, and between schools, but also in the rate of change in trust between teachers and between schools. For example, the variance in rate of change between schools is calculated by dividing the variance in rate of change between schools (Level 3) by the total variance in rate of change in the model. Carrying out these adjusted calculations for the SFA model, we had 38% of total variance within person over time, 43.4% of variance l9 . . . . . . .. . . In comparing the vanance in trust trajectory over time between teachers and schools m the SFA and ASP model 3, we see that there is substantially more variance in the SF A models. 92 in initial status between teachers, 60% of total variance in rate of change between teachers, 14.1% of variance in initial status between schools, and 40% of variance in rate of change was between schools. All of these variance components were significant (p< .001) In looking at the unconditional growth model itself, we see that the slope of the population average change trajectory was non-significant, and was so for the remainder of modeling, including in the final model. In this case, we retain the null hypothesis for the rate of change in teacher-teacher relational trust over time, concluding that this estimated value is statistically no different from zero. Unlike the ASP model, it seems clear that, on average, there is no appreciable change in teacher-teacher relational trust over time within this sample of SFA schools. AIC fit statistics demonstrate that the linear change model is an improvement over the means model. As in the previous chapter, Model 3 represents the uncontrolled relationships of the focal time-varying covariates on change in teacher-teacher trust, Model 4 introduces controls and Model 5 represents the full model. Recall that the first set of hypotheses posited in Chapter 3 for SFA schools predicted that teacher and leader hiring, instructional guidance and supportive instructional monitoring would be positively related to change in teacher-teacher relational trust. The results in Model 3 indicate some mixed findings with respect to these hypotheses. Supportive monitoring was not found to be related to change in teacher-teacher trust (coef = .124, SE = .088, p = .162), while the correlations of teacher and leader hiring on change in teacher-teacher trust were robust even after controlling for depth of implementation and collective responsibility among teachers. Once controlling for these characteristics, however, instructional guidance 93 389:. ohm. 8..» bzfifim 5:68“. .8. 88... 8... :8... 8.. ...... 8.. 5...- 8; .8225 .3: oEF 8.8. 88... ,8... .8... 58m 80.888.858.38 :....8. 8m... :88. 88... 5.5.8083. 282.8 wwo. 82.0 wctozcoE ..mE o>_to&=m Q. . . oc _ .o 8:... .wcmq 2.25.6.9... .8... ..8. N8... 88. 88... 885.588.8285 m8. :8... .88. :8... 9.8888 5.80 88. 88... 88. N8... 88. 88... 8.5.8.8.. 88888.. :88. 88... :88. 88... :88. 8.... 8888985885 88. N8... 88. ..8... :88. 8.... 8:82.58 2256 :88. 8m... ......88. 8m... :88. 8m... 8.885.856 :18. 8.8 :82. 888 :88. 8mm :88. 8.8 :18. 8.8 858.8235 8E. 38.8..— muouomfienm ~ _o>ox_ mm 88 mm 88 .8 88 .8. 88 :8. 88 8.88.... .888. 2:5va @825 m .85). .8 .85). m .85). N .802 . .85). £0930 3R. 5 338ch M29...» 85¢ 38823833 38o: £395 ESQ 3.6%-». ..o 2an 94 VON. mom. mm _. 105083-555? _ _o>oq 8.238% 88.58882 588.8% .Iszmo. ***wo_. .1182. ***mmm. owcwr—omo Sam M _o>oa— ..chwv. 532,00. “I..."EA ***h~.m ***©_.N 353m REE m _o>ou_ ***wwm. 3.3.52“. ***m_ m. ***mwm. owcmr—omo Bum N _o>oA ***m_.m ***om.m .12..— _.v filmed .....cwmvd 8588 855 N 35‘— ***h©.v ***m©.v .1335? .Icpwfi 5.3.50.0 GOmuom 55:5 _ _o>om_ 8888mm 58.38% So. vmod EoEo>oEo< 55>. SE m5. moor EoEo>oEo< <1— SE S: m8... 5.38m 53$ So. 50.- 3% Begum mg. m _ od 5255 caused a. 8.5 Enema 8806on m .055 10%. cm._- @585. comm mm; mom; $585. 8550 cam. mmor 958...: cocotomxm 8.5 3m. 2 m.. .850 53.....va vwor 502m .383. 03.. 255.8: mom. mm _ .- 28th ......moo. god oucotomxm 90 83> 80860.5 N 355 985.58 .8 288. 95 55 am _N.m$o_ adso— 0mm. 0:. mam. _w _ N C $.38— NYSE: mum. Nwo. wow. 5 ..N C wmdwwc— wmgmwofi mmm. 5v. Nwm. €3on 24: 05 3 33328 mm .2an £5 E BflomE 8a Echo 2%:me 5.53: .802 S. v m2 .8. v 3 .S. v a: .25. v 3: S _ m a _ N as 29:8 _ EB a V UonESmo flower—Sunn— NER: 5.9%: 0? 2.3V: 8.9%: 8535 owafio mo Sam M .261— mBSm 325 m 35A £58m FEE N 354 63528 3 22$ 96 was also found to be no longer significant (coef = .032, SE = .028, p = .261). The final hypothesis predicted a positive relationship between change in teacher- teacher relational trust and shared instructional experience in SFA schools. As can be seen in Models 3-5, shared instructional experience among teachers was found to be significantly associated with teacher-teacher trust in SFA schools (coef = .086, SE = .028, p < .01). Examining model fit statistics in comparison to the unconditional model, we see that 17 percent of Level 1 variance, 38 percent of Level 2 variance, and 46 percent of Level 3 variance in the teacher-teacher relational trust measure has been explained by these time-varying predictors alone. Model 4 demonstrate that the associations of both collective responsibility and policy implementation depth are significant (p < .001 and .10 respectively), and diminish some of the effects of time-varying predictors in the uncontrolled model.20 As in the ASP model, significant relationships for both critical discourse and innovation were also found (coef = .306, SE = .020, p < .001; coef = .082, SE = .034, p < .05 respectively). Quality teacher professional development opportunities, while only marginally significant in Model 3, was non-significant after controls were added. Examining the AIC fit statistic and pseudo R-squared statistics, it is clear that Model 4 is a substantial improvement over Model 3, with an overall change of 194.16. Moreover, from Model 3 to 4, the proportion of variance explained in Level 2 and 3 initial status increased 12.3% and 22% respectively. 2° It bears repeating here that these two variables were viewed and treated as “control” variables in these and all subsequent models for two main reasons. First, because their effects on trust are relatively well- established in the literature, and second, though this study was interested in the effects of these two covariates on trust growth, on the whole they were not a central aspect of this study’s analysis. In other words, they were not a part of our theoretical models of trust growth in the two CSR models, but are nevertheless related to trust and therefore must be included. 97 In the final model, Model 5, all Level 2 and Level 3 controls were entered. We see in Model 5 that, even after controlling for teacher and school level characteristics, all significant effects in Model 4 remain in the final model. At the teacher level, teacher experience was found to be significantly related to trust, though its effect appears proportionately small (coef = .021, SE = .003, p < .01). With White teachers as the comparison group, Hispanic and African American teachers were also found to have average initial statuses on trust at Year 3 significantly lower than those of their White counterparts, after adjusting after adjusting for differences among schools in faculty composition. It should be noted, however, that these effects should be treated with caution, given that the effect for missing on race was also significant. At the school level, we see that no significant effects are found. Faculty stability, which was highly significant in the ASP model on average initial status, was not so in the SF A model. Free and reduced lunch as well as school size were not found to be associated with trust in this sample of SFA schools. Finally, with the inclusion of the random effects on the trust change slope parameters at Levels 2 and 3, we see that there is a significant amount of variation among teachers and schools in trust growth trajectories over time, and attempts to model this variation with both teacher and school level characteristics led to an important discovery. As can be seen in Model 5, school level heterogeneity in growth trajectory was partially explained by faculty stability (coef = 1.579, SE = .499, p < .01). So whereas in the ASP model, faculty stability was related to variation in initial status between schools, faculty stability in the SFA model was significantly related to variation in rate of change between schools. Figure 6.1 displays the school-level average variation in trust trajectories after 98 adjusting for faculty stability. As can be seen, once faculty stability is controlled for, there is considerable variation in the school-level average trust trajectories in SFA schools, as compared to the overall average trajectory for SFA schools (seen as the shaded trendline). Figure 6.1 SFA School-Level Average Trust Trajectories Controlling for Faculty Stability 4.25 .-,-_____,,-_ ~— —— — 3.75 3.25 . 2.75 2.25 g Teacher-Teacher Relational Trust (logits) 1.75 Time (Years) Interaction effects were also extensively investigated at this stage in the modeling process, however none were found to be significant. In examining the AIC and pseudo R- squared change statistics, we see that all models improve in fit and variance explained. The ‘final’ model, Model 5 presented here accounts nearly 78 percent of the variation in the outcome at the school level, and over half of the variation in the outcome at the teacher level. Investigation and verification of HLM assumptions was accomplished again by examining the unconditional growth model for gross violations and then more thoroughly once the final model was completed. Examination of standard P-P plots of residuals, 99 scatter plots of standardized residuals versus ID numbers, and histograms of residuals at all levels found no evidence of non-normality. Probability plots were highly normal, with no unusual data points, as well as all plots and histograms. Homoscedasticity at Level 1 was assessed by examining both scatter plots of standardized residuals versus the TIME predictor as well as scatter plots of standardized residuals versus predicted values. Scatterplots of TIME suggested equal variability at all time points. Overall, the examination of the scatter plot of residuals versus predicted values was inconclusive, though there was very slight evidence of some heterogeneity in errors at Level 1. Again, several alternative covariance structures were explored, but none of the available alternative structures (e.g., heterogeneous AR], unstructured, etc.) provided a better fit over the default diagonal structure as measured by AIC model fit statistics. In light of the fact that all other assumptions were reasonably satisfied, and that the evidence suggesting the presence of heterogeneity was inconclusive, robust standard errors were used as a conservative measure in order to further curtail potential Type 1 errors. As empirical grth plots of randomly selected individuals were conducted at the outset of modeling, the linearity assumption was checked primarily with regard to the relationship of Level 2 and 3 residuals to predictors. No evidence was found to suggest that any individual predictors had a curvilinear relationship to the outcome. Discussion The key finding in the analysis of SF A schools with respect to growth in teacher- teacher relational trust over time is that, on average, there is no growth in trust among teachers over time in this sample of SF A schools. Furthermore, the average level of trust across SF A schools in Year 3 is substantially lower (13 percent lower) than that of ASP 100 schools at the same point in time. However, while there was no average growth in teacher trust over time in the SF A model, there was substantial evidence to suggest that schools varied considerably in their trust trajectories over time once faculty stability was taken into account. Indeed, it is noteworthy that, within the ASP model, faculty stability was associated with variation in initial status, while in the SFA model it is associated with variation in trust trajectories over time. This disparity in findings across the two models may simply be a result of differences in the amount of variation which existed in the two samples with regard to trust trajectories. Generally speaking, ASP had very small amounts of variation in the rate of change in trust over time relative to SFA, so much so that in the ASP modeling process, leaving these random effects fixed did very little to affect the standard error estimates. In SF A however, fixing these effects made the models very unstable.” Regardless, these results suggest that SF A could, in fact, have strong potential for building trust among teachers over time if, all else being equal, the stability of the faculty is adequately maintained. Despite the fact that, within this sample of SFA schools, average trust growth was negligible over time, we can nevertheless interpret those factors which are related to change in teacher trust. Revisiting our theoretical framework in Chapter 3 and comparing that to our findings with respect to theory and hypotheses, we can draw several conclusions. Overall, we see some support in the results for our “theory” of trust development and the factors which were identified as likely related to change in trust over time in SFA schools. First, school leaders’ emphasis on teacher and leader hiring (Hypothesis 5) was found to be positively related to change in trust among teachers in SFA schools. This 2'When comparing the estimates of the asymptotic standard errors to the robust standard error estimates. 101 finding does not seem particularly surprising when you consider the uniqueness of the SFA program, and the typically heightened responses and strong opinions many have toward the instructional program itself. Because of the scripted nature of the program, its constraints on teacher autonomy, and pervasive monitoring of practice which have been documented in other studies (see Datnow & Castellano, 2000, 2001)22, there is certainly a higher likelihood of dissatisfaction and other general discord among some faculty who disagree fundamentally with the programs’ approach. We must also acknowledge again that research has shown that considerable variation may exist in the extent to which SF A schools employ the procedural system of controls. Therefore, the discussion of these findings acknowledges that all SFA programs vary in the degree to which these attributes are present. Moreover, with the adoption of any new instructional program, there are bound to be a few teachers who are reluctant to change, and this can result in discord as well. As some prior studies of trust have shown, it only takes one or perhaps a few discontented teachers to undermine faculty trust as a whole (Bryk & Schneider, 2002; Kochanek, 2003, 2005). In these instances, the hiring of new personnel provides an opportunity for school leaders to shape their faculty in such a way as to ensure that a greater percentage of the overall faculty are willing and committed to the instructional program. Furthermore, shared instructional experience among teachers was an important factor associated with change in teacher-teacher trust. Theory suggested that structural aspects of the SFA program such as the common core curriculum, common language and 22These findings and conclusions are not necessarily the opinion of the author and remain an area of debate among researchers. They are nevertheless common criticisms often leveled against the SFA program. We must also acknowledge again that research has shown that considerable variation may exist in the extent to which SFA schools employ the procedural system of controls. Therefore, these results must be interpreted with caution. 102 expectations around instruction, and shared Ieaming goals may provide a strong framework upon which teachers can be drawn together to work on the business of instruction. Moreover, having detailed knowledge of what other teachers are doing, their role obligations, accompanied by a strong, steady pace of instruction can create conditions by which teachers can begin to trust one another to do their work. Teachers is later grades can, given these structures, be reasonably confident that students are being prepared properly for the following grades. This fact alone can serve to significantly reduce vulnerabilities among teachers and make the business of instruction a truly collective effort within the school. Again, it is worth noting that critical discourse, innovative climate and risk- taking, collective responsibility, and policy implementation depth were all positively related to change in teacher-teacher trust. However, two key pieces of the SF A program which were hypothesized to be related to trust grth over time were not found to be— instructional guidance and supportive instructional monitoring. In the case of instructional guidance, it seems that it has some potential to be related to teacher trust, however, other effects are simply more robust. This is to say that, when comparing theoretical relationships to trust growth among teachers, instructional guidance is likely weak in comparison to, for example, shared instructional experience, which relates more directly to teachers’ relationships to one another. Instructional guidance may reduce vulnerabilities that teachers themselves experience, but these feelings clearly do not translate into effects on teachers’ relationships with one another over time. With regard to the hypothesis that instructional monitoring was positively related to teacher trust seems, in retrospect, to be tenuous at best. The key to this measure was 103 the word supportive. It was theorized that receiving praise and recognition for implementing the program well might serve to mitigate some of the ill effects associated with monitoring and boost teachers’ confidence in themselves and others. It could be that supportive monitoring results in stronger trust relationships, but just not among other teachers. If monitoring is supportive, it is more likely that trust would grow within the leader-teacher role-relation, given that facilitators do the majority of the monitoring within the SFA program. Because of the limitations of the dataset, however, we are not able to assess the extent to which this is the case. 104 CHAPTER V11: TRUST GROWTH IN CONTROL SCHOOLS The purpose of this chapter is to report and discuss the findings from the analysis of trust grth which was conducted on the sample of control schools in the S11 study. In doing so, this chapter addresses all three research questions: 1.. To what extent is there growth in relational trust among teachers over time in the 811 sample of Accelerated and Success for All schools? As models of “typical,” non—intervention schools, to what extent is there growth in relational trust among teachers over time in the 811 sample of control schools? 2. What factors related to the instructional improvement process are most related to change in teacher-teacher relational trust in each subsample of schools? 3. Looking across CSR models and control schools, what evidence exists to support a set of common factors associated with change in trust among teachers which transcend the particular model designs? In addressing these research question(s), this chapter will begin with a brief discussion of the general model structure employed for the control schools analysis, and then proceed with the presentation of the results, which is organized slightly differently from the previous two chapters. Presentation of the results will begin by considering the findings within the sample of control schools. The final section of the results will briefly compare findings across samples in order to assess the hypotheses pertaining to trust trajectories as well as instructional improvement factors common to all models of trust growth. 105 Control Model Structure In order to address questions regarding the degree of trust growth in control schools and the instructional improvement factors related to teacher trust, a 3-level HLM growth model of relational trust among teachers was developed from the control school data. In the control schools modeling, as with modeling in the other subsamples, all time- varying covariates were considered in initial stages of modeling. This is important to acknowledge in the description of the control schools modeling process because, as control schools, by definition there is no instructional intervention occurring. Even though the measures being used in this study are specifically designed to capture aspects of the instructional improvement process occurring in intervention schools, these measures (instructional guidance and monitoring, critical dialogue, collaboration, etc.) also apply to processes occurring in control schools. One exception of use in this study is the measure of policy implementation depth. This measure has little theoretical relevance to control schools as it measures the motivation of teachers to implementing the CSR program and the alignment of that program to the current policy environment of the school. It was therefore omitted from the control school analysis.23 The final model which was developed and is discussed in the following section is represented here in Equations 7.1 to 7.6. You will note that this is the same model structure given in Chapter 4, with two important differences. First, trust trajectories at Level 2 (between teachers) were treated as fixed, based on a non-significant variance component. Second, trust trajectories at the school level were permitted to vary, and Level 3 predictors were used to model variation in trajectories of trust change over time between schools (see Equations 7.3 and 7.5). For the sake of brevity, the interpretations 23 . . . . . Exploratory analyses seem to support this conclusron, revealing that the measure was non-Significant. 106 of the terms in these model equations are omitted, because they are the same as in Chapters 4 and 5. Equation 7.1 displays the Level 1 model: Y =noy+an1ME+ZnW x +e,,, (7.1) q: 2 “f Level Two of the control school model is represented by Equations 7.2 and 7.3 below: Q not} = BOOj +2: BOqj Xqi + r01] (72) q=l 2% for q= Q.,. (7.3) The Level 3 models are represented by Equations 7.4 through 7.6: S B001' : yOOO +2 yOqs Wsj + ”001' (7-4) s=1 S BIO} : 7100 +2 qus W51 + ”le (7-5) s=1 Bqu = YqOO for q = 2, ..., Q. (7.6) Results As a result of Rasch measure development and HLM file construction, the final sample for the control schools analysis was 1985 responses within 1008 teachers in 26 schools. The presentation of the 3-level HLM growth model of relational trust among control school teachers is found in Table 7.1. Beginning with the fitting of the unconditional means model (Model 1 in the table), the baseline deviance statistic was obtained (11535.01) as well as the AIC fit statistic (11543.01) which were used in subsequent model fit comparisons. This model also yielded the grand mean on the 107 omo. moor wEmflE 8mm gm. :37 9.6%: $200 58%. mwm._ wEmmE. oocotoaxm .E> m5. god 850 them. 30.. x85 ......Emm. omwr 352$: mom. tumor 25:5..— .Roo. god oucotogxm mo mumo> £0868; N $25 .125. Sod :: x. EoEo>oEo< <4 SE mg. n 56 5o. wood coo. god 9 _. mm _ .o 8C» Rocco“:— EK 25% 1.88. 8% 1.58. mag bzsacoamom 268on So. fl :3 138. oofio oocotoaxm .35 eosnm ......omo. mgr ......omo. Eor ......cmo. moor 5:803:00 55on 1.5.8. $2.0 ***omo. Nwfid ..clbmo. ommd ***mmo. mmmd ...:vmo. Rmd $3.8. mgd ***ow_. 3m.m *IEN. wnmd *ifivm. mama 12.5mm. N_m.m ***wmm. NOWM :osgocfi no BEE—U 35085 323.5 8o». 3088:. .8: 3028:— muowozooi _ .301— Nm \meo mm. «Goo Mm \moo mm. \m8 Eh \moo aochm 33E 236ch @9825 m 3602 v E52 m .252 N 332 _ 382 £85m. 35:80 E anomfl $828 23¢ BaotcmeKo ESE S395 ESE ~m>m$m <5 2an 108 .68on $3: of NE Bum—:28 mm JEFF £5 E totem»: 8m Echo Eng—Sm annoy... $32 2. v a: .3. v a... .S. v Q..... is. v at... mwm: mN oméowo cmdmg coo. wt. Omm. ONN. **mmo. **o>m. ***©N..N ***wm.m *m_o. 05. mo.— so. 5.— >86 for VN_._ Sod 30.7 mg— N— wfiw—wo v0.33 mew. VNo. cNm. _NN. **vwo. ***Nwo. auxin—ON ***\im.m 32 N_ vo.wwoo vodoom m _ m. mtg. vmm. CON. ***h~_. ***©m; ***Nh.m ***Nh.m mwi mwe n. v odwmo— OSGmA: 9.3.2: ©6me— ***_h_. “3.2.—0N ***wv.N ***v~.© "1236.0 3.1.2.5 TEENS ofim 29:3 _ 354 33:53 206:5ch 02 8:350 owsmso .8 8mm m .054 mBSm 325 m _o>ox_ owcmso mo Bax N .964 £59m 3::— N 35‘— 8:om2-=£:3_ $54 EQESM $8.5:wa ennui owcaco mo Sam m 354 336 BEE m .05»— owcmso mo 8mm N $51— 336 REE N 354 cows; 55:5 _ 354 Somxxm Bottom EoEo>oEo< 522 SE EoEo>oEo< <1— SE éfism £25 35 323m £254 @833— um ooh”— 2x. $58605 m .964 @3528 E 2%: 109 €3on 424: 65 3 4642.666 3 63:? £5 :4 464.5%: 6:: £86 4:8:me 626mg .632 2. v a: .8. v a... .5. v 3.... :8. v at... mwm: mN omdowo cmbmmo coo. mun. Omm. CNN. **mmo. **©hm. ***©N..N ***wm.m *mS. 05. 34 So. S4 536 for :24 Sod v54- mwg N— va—wo vaono mew. .No. 0%. 4NN. **Vwc. ***Nwo. “Inna—0:0. “.23.,th 32 N— vowwoa vo.vooo m _ m. wnv. gm. ooN. ***h4_. ***©m4 .12."th ***Nb.m mwe mwe N. v c.0wmo_ ofiomO— gmhmo— odwmom ***_h_. ***_©.N .13..va ***v_.© “31.506 “1.3348 ***om.\i 6N6 29:3 _ 664 6me36 EgoEmsn— U4< 8:363 6w:fi6 .6 83— m 664 £58m 62:4 m 664 6w:§6 :0 834 N 664 £566 634:4 N 664 6:om3-:_5_3_ 664 8.238% hmxgwmé 036$ emcmco .6 334 m 664 £563 642:4 m 664 6w:m_.6 mo 834 N 664 36% 62:4 N 664 :86: 5:55 4 664 “SRKNM 56.38% EoEo6Eo< :62 SE EoEo6Eo< <4 pot: £53m bag: sum 32% 6:54 462634 a. 68m X. 69265 m 664 Guacuaoo E 2%: 109 teacher-teacher relational trust measure for the control school sample in Year 3, which was estimated at 3.502 logits. The fitting of the unconditional growth model (Model 2, Table 5.1) involves the addition of the predictor TIME at Level 1, which was centered on Year 3 of the study. As with the sample of SFA schools, early analysis of the control school data suggested that the rate of change in relational trust between schools be allowed to vary throughout modeling.24 When this random effect is included, calculation of the intro-class correlation coefficient (ICC) for the control school sample found that 44.9% of the total variance in teacher-teacher relational trust was found at Level 1, 38.6% was found at Level 2, and 16.4% was found at Level 3. All of these variance components were significant (p< .001). In looking at the unconditional growth model itself, we see that the slope of the population average change trajectory in the sample of control schools was non- significant, and was so for the remainder of modeling, including in the final model. In this case, we retain the null hypothesis for the rate of change in teacher-teacher relational trust over time, concluding that this estimated value is statistically no different from zero. Unlike the ASP model, it seems clear that, on average, there is no appreciable change in teacher-teacher relational trust over time within this sample of control schools. AlC fit statistics demonstrate that the linear change model is an improvement over the means model. As in the previous chapters, Model 3 represents the uncontrolled associations of the focal time-varying covariates on change in teacher trust, Model 4 introduces controls 24 Recall that there was an insignificant amount of variation in the rate of change in relational trust over time between teachers. llO and Model 5 represents the full model. Looking across these models, we see that, the relationships of critical discourse, and climate of innovation and risk-taking remained significant throughout modeling (coef= .333, SE = .025,p < .001; coef= .183, SE = .30, p < .001; coef = -.075, SE = .026, p < .01 respectively) after controlling for collective responsibility among teachers and important teacher and school-level characteristics. Once controlling for collective responsibility, however, shared instructional experience was found to be no longer significant (coef = .011, SE = .031, p = .751). As is evidenced in Models 3-5, teacher collaboration was found to have a negative relationship with change in teacher-teacher trust in control schools (coef = -.075, SE = .026, p < .01). Models 4 and 5 demonstrate that the relationship of collective responsibility itself on teacher trust is significant (p < .001). Examining the AIC fit statistic and pseudo R- squared statistics, it is clear that Model 4 is a substantial improvement over Model 3, with an overall change of 170.6. In the final model, Model 5, all Level 2 and Level 3 controls were entered. At the teacher level, teacher experience was found to be significantly related to trust, though its effect appears proportionately modest, as in the SF A model (coef = .019, SE = .007, p < .05). With White teachers as the comparison group, a negative effect for the racial categories Hispanic and African American was found. This indicates that African American and Hispanic teachers have, on average, lower initial status on trust (in Year 3) than those of their White counterparts alter adjusting for differences among schools in faculty composition. For Hispanic teachers, this translates into an average trust score of 2.42 logits (3.314 minus .890). For Black teachers, this results in an average trust score of 2.64 logits in Year 3. At the school level, we see that only one modestly significant lll relationship is found—prior school-wide mathematics achievement of the student body (coef = .037, SE = .015, p < .05). Faculty stability, which played an important role in trust growth in the ASP and SF A models, was not so in the model of control schools. Free and reduced lunch as well as school size were also not found to be associated with trust in this sample of schools. Finally, with the inclusion of the random effect on the trust change slope parameter at Level 3, we see that there is a significant amount of variation among teachers and schools in trust grth trajectories over time, and attempts to model this variation with both teacher and school level characteristics led to a significant, albeit modest, effect for pre-intervention language arts achievement of the student body (coef = .012, SE = .003, p < .01). Other interaction effects were also extensively investigated at this stage in the modeling process, however none were found to be significant. In examining the AIC and pseudo R—squared change statistics, we see that all models improve in fit and variance explained. The ‘final’ model, Model 5 presented here accounts nearly 78 percent of the variation in the outcome at the school level, and considerably more than half of the variation in the outcome at the teacher level. Examining Results across Models This section examines findings across the three subsamples of schools in order to assess queries about relative trust growth in each, and take stock of the factors related to instructional improvement common to all three models. Beginning with relative grth in the two CSR models, this study hypothesized that the ASP sample would have a significantly higher average trust growth trajectory over time than the SF A sample. Figure 7.1 provides a graphical representation of the average trust trajectories over time 112 generated from the parameter estimates for each model, after controlling for all key exogenous variables. Seen together, the differences between the models in terms of both average trust growth and initial status in Year 3 become clear. Recall that only the ASP model had significant teacher trust grth over time, and had the highest Year 3 average teacher trust score. Both SF A and control schools trust grth was negligible, and the graph demonstrates that the average teacher trust growth score in Year 3 is slightly higher in control schools, with SFA exhibiting the lowest overall score at 3.18 logits in Year 3. In order to establish whether or not ASP schools had a significantly higher average teacher-teacher trust trajectory over time, a 3-level HLM growth model of the entire sample was constructed with the instructional interventions entered as dummy codes at Level 3 on initial status and growth rate over time (SF A was omitted as comparison group). The analysis demonstrated that the ASP trust trajectory was significantly higher that SFA’s trust trajectory, but only moderately (p = .10). Figure 7.1 Average Teacher Relational Trust Trajectories Over Time by Subsample Teacher-Teacher Relational Trust (logits) 3.90 3.80 ~ 3.70 3.60 3.50 3.40 3.30 3.20 3.10 —~ 3.00 2 3 Time (Years) 113 — —ASP — - -SFA Control The third research question inquired as to the factors related to instructional improvement which are common to all three samples in this study. Recall that, in Chapter 3 two hypotheses were advanced regarding the factors likely common across these models. Based on prior research, critical discourse, collaboration, and collective responsibility were all hypothesized to be associated with change in teacher-teacher trust among teachers in all three models. What we find when looking across the models is that critical discourse, climate of innovation and risk taking, and collective responsibility among teachers were indeed so. Teacher collaboration, however, was not found to be a common factor across the models, only finding significance in the control schools sample, and was negatively—mot positively—related to change in teacher-teacher trust. Furthermore, when you consider that policy implementation depth was only pertinent to the two instructional interventions ASP and SF A, we must also consider this measure as being a common factor across models positively related to teacher trust. Discussion As with the SF A model, the control group analysis demonstrates that there was no appreciable growth in trust among teachers over time. This is not necessarily surprising if we consider that the schools selected for the control group are meant to represent “typical” non-intervention schools. Recall that, within the 811 sample, high-poverty schools were over-represented, with the average proportion of students receiving free- and-reduced lunch in the control schools at 64 percent. The struggles of high-poverty schools with respect to teacher and instructional quality, student achievement, and school c1 imate are well documented in the literature, and because trust is so intimately linked to many of these important characteristics/outcomes, this finding is perhaps expected. ll4 Interestingly, the control schools were the only group in which growth in trust was related—albeit modestly—to students’ prior achievement in math and language arts. This could be due to the fact that there was slightly more variation in the scores between control schools (Math SD. = 13, LA SD. = 17.3) than between ASP (Math SD. = 11.2, LA 5'. D. = 15.2) and SFA schools (Math SD. = 10, LA SD. = 13.1) In discussing the factors which were related to change in teacher-teacher trust, we see that many of the findings reinforce patterns of common factors across models. Critical discourse, innovation, and collective responsibility were all related to change in teach er—teacher trust, lending further support to these as factors which are likely important to trust regardless of the schools’ instructional program. Faculty stability, which played an important role in both sets of intervention schools, was surprisingly not related to teacher trust in the control schools. It is noteworthy that control schools had lower faculty stability on average than control schools, but it is possible that instructional interventions such as ASP and SF A are simply more sensitive to faculty turnover due to the collective efforts required to successfully implement the model. Furthermore, for control schools—particularly those with high amounts of school poverty—much of prior research suggests that faculty turnover is simply par for the course. TV'VO other findings are also worth mentioning. First, teacher collaboration, which was “Qt lcound to be significant in either of the CSR models, was negatively related to Change in teacher-teacher trust in control schools. Again, this does not seem so surprising when we Con sider that collaboration is not necessarily tantamount to trust. As Kochanek (2005) remil'lds us, teachers engaged in “high-risk” activities around instruction—such as thos ' . . . . . e Wh'ch examine cumcular alignment or standards for student Ieaming—Without a 115 firm base of trust are much more likely to damage trust than develop it. Much more likely for cultivating stronger trusting relationships is the ability to talk through problems and issues well, and as we saw yet again in the control schools sample, critical discourse was indeed an important factor to building trust. Ironically, however, interaction effects betvveen collaboration and critical discourse were not found. Finally, a finding of significantly lower Year 3 levels of trust for Hispanic and Black teachers as compared to White teachers is particularly noteworthy. When thinking about the origins of these lower values, two possible explanations seem possible. First, there could be appreciable differences in the propensity to trust between racial and ethnic groups- As Tschannen-Moran (2004) explains, people vary in their disposition to trust, and this disposition is often shaped by past experiences. Second, there must be consideration for the racial-ethnic positionality of individual teachers within the school. As has been stated before, people are more likely to extend trust more readily to those they perceive to be similar to themselves (Zucker, 1986). In the case of White teachers and/or teachers of color, the ability to trust their teaching colleagues may be a function of their “status” as a minority or majority within the school. While we have no empirical evidence to support these explanations, both certainly seem plausible. 116 CHAPTER VIII: IMPLICATIONS FOR THEORY, POLICY, AND PRACTICE The purpose of this final chapter is to discuss this study’s main findings in the context of their potential to inform theory, policy and practice. This chapter will begin with a brief review of the main purposes of the study and research questions. It will follow with a discussion of some important findings of this work and their implications for trust theory as well as educational policy and practice. The chapter will conclude with a discussion of the study’s limitations and directions for future research. Purpose of the Study and Research Questions The school improvement literature is replete with studies of high-performing schools which seek to gain insight into the sources of their success. Invariably, these studies point to trust among colleagues as a likely factor which has contributed to this success, but beyond this, we really know very little. How does trust grow and develop? What are the factors related to the school improvement process which are associated with this change? These are gaps in the literature this study attempted to address directly through an analysis of data from the Study of Instructional Improvement (811). By ChOOSing to analyze the grth in trust among teachers occurring within two popular, but very different, comprehensive school reform programs—Success for All and the Accelerated Schools Program—we can begin to understand the ways in which the context of school improvement matters for trust development as well as the factors related ‘0 cl'lange in teacher trust which might be common irrespective of context. The main purpose in attempting to answer these questions was to begin to develop a better theoretical grounding for the role of trust development in instructional 117 improvement—particularly for chronically-underperforming schools in which trust is likely in short supply. In doing so, two assumptions which are pervasive in the literature were challenged: first, this study challenges the notion that trust is essential to school improvement and/or increased student performance; and second, this study challenges the notion that trust development proceeds in similar ways across schools or, more spec ifically, that the factors which co-occur with trust development are similar across schools. Studying two popular and successful but very different C SR programs as well as a set of control schools adds an element of school context which, it is hoped, will facilitate our understanding of the extent to which these assumptions hold true. Three research questions were developed in order to guide the study with respect to these purposes. They were: 1 . To what extent is there growth in relational trust among teachers over time in the 811 sample of Accelerated and Success for All schools? As models of “typical,” non-intervention schools, to what extent is there growth in relational trust among teachers over time in the 811 sample of control schools? 2. What factors related to the instructional improvement process are most related to change in teacher-teacher relational trust in each subsample of schools? 3- Looking across CSR models and control schools, what evidence exists to support a set of common factors associated with trust among teachers which transcend the particular model designs? The remain der of this chapter is devoted to a discussion of the main findings of this research, the implications of these findings for trust theory, policy and practice, and the limitations of the study and directions for future research. 118 Before beginning, however, it bears reiterating here that the discussion of the findings proceeds with the acknowledgement that the configurations of these programs described here are as they were at the time of the 811 study. It is important to note two important corollaries with respect to the configuration of these programs at the time of the SII study which have important implications for this final discussion. First is that the developers and implementers of these and other CSR programs are constantly Ieaming about how best to design and implement their models. As such, the analysis undertaken in this study acknowledges that even within the time frame of the $11 study, these programs were changing as implementers discovered, developed and improved on their models of instructional reform. Second, it is also important to acknowledge that the ideals of reform that these instructional interventions seek to foster might take considerable time to develop—as much as 6 to 8 years—and lead to significant turmoil and difficulty that can last for many years. As such, it is possible that the 4 years of the 811 study were not always sufficient in capturing all of what these programs sought to develop in the schools implementing them. Discussion of Findings and Implications for Theory, Policy and Practice Growth in trust. Proceeding in order of the research questions, this section begins with a discussion of the overall findings with respect to growth in trust among teachers over time in each of the subsamples of schools. The analysis of trust growth conducted in this study revealed that gains in trust among teachers only accompanied the implementation of the Accelerated Schools model over time. For SFA and control schools, gains in trust over time were absent. Furthermore, average trust levels at Year 3 119 were lowest for SF A schools and highest for ASP schools, with control schools falling somewhere in the between. These findings, while somewhat straightforward, may have significant implications for understandings of the role of trust development in instructional improvement—particularly when they are considered in concert with other recent findings from $11 studies of the achievement and instructional patterns in ASP, SFA and control schools. To begin, however, the fact that the analysis of the 811 sample of ASP schools revealed significant gains in trust among teachers was somewhat expected. The design and “theory of action” of the ASP model is very much one in which building trust is a pervasive, if not explicit, component. ASP’s employment of “cultural controls” to promote instructional change involves developing a normative commitment among school colleagues to a broad vision for school improvement and requires significant cooperation among stakeholders to carry out the ambitious instructional effort (Rowan, Cambum, & Barnes, 2004; Rowan & Correnti, 2009). Not surprisingly, other 811 studies have also found trust among teachers in ASP schools to be the highest—on average—of the three CSR models under study (Rowan & Miller, 2007). However, in other 811 studies Correnti and Rowan (2007) and Rowan, Correnti, Miller, and Cambum (2009) reported that literacy instruction and achievement in Accelerated Schools were indistinguishable from control schools. These authors concluded that ASP’s instructional design was not well-suited to producing large-scale changes in instruction and student achievement. When considered with this study’s findings regarding growth in trust among ASP teachers, an interesting paradox seems to surface—that trust can be present and grow over time in the sample of ASP schools while 120 average instructional improvement and/or student achievement remains virtually stagnant. Furthermore, this study’s findings from the sample of SFA schools suggest something equally interesting. In the same 811 studies mentioned above, researchers also found that SF A schools produced highly distinctive patterns of literacy instruction as compared to control schools, and also high rates of academic achievement as compared to both control schools and all other schools (Correnti & Rowan, 2007; Rowan, Correnti, Miller, & Cambum, 2009). In fact, the achievement gains of students in SFA schools are well-documented in the literature. Studies of effectiveness of SFA as a program have revealed time and time again the measurable gains SFA students have made in reading achievement (Borman & Hewes, 2002; Borman, Hewes, Overman, & Brown, 2003; Borman et al., 2007). In light of this evidence, however, this study found no appreciable growth in trust among teachers over time within the SF A sample of schools. Taken together, these findings are somewhat counterintuitive when you consider what the literature currently tells us about the relationship between trust, school improvement, and student achievement. On the one hand, then, ASP schools produce significant gains in trust growth over time with no change, on average, in patterns of literacy instruction or student achievement growth over time. On the other hand, SFA schools produced, on average, no gains in trust over time among teachers, yet still manage to produce distinctive patterns of literacy instruction and student achievement growth over time. When considered together, these findings suggest that, at the very least, we exercise more caution regarding 121 claims about the relationship between trust, school improvement, and gains in student achievement. More boldly stated, these findings suggest that trust may be a sufficient but not necessary condition for the improvement of instruction and student achievement. Taking SFA as a program, for example, it is entirely possible that the set of procedural controls that are put in place as a result of program implementation effectively eliminate the need for high levels of trust. The necessity for trust among colleagues undertaking ambitious school improvement comes when, as some scholars have noted, goal incongruence and performance ambiguity are particularly high (Bryk & Schneider, 2002; Kochanek, 2003, 2005; Ouchi, 1980). In the case of SFA, when considered relative to the ASP program, both may have similar levels of goal congruence, but the procedural controls within the SFA model likely substantially reduce performance ambiguity. The detailed and steady pacing of instruction, the common curriculum, the student regrouping and assessment process, as well as feedback loops which are part and parcel of the SF A instructional design provide ample opportunities for teachers to find out how they are doing and to seek feedback on whether or not their efforts are producing results in the form of student achievement. The cultural controls which are the hallmark of the ASP program provide a different approach to instructional improvement, one which places trust at a premium. ASP requires teachers to work together extensively on issues of school improvement— particularly in the early years when important school improvement decisions are being made. Teachers during this time are likely engaging in many high-risk interactions around instruction, engaging in critical dialogue and making decisions about instruction 122 together as a team. Yet actual instructional improvement does not typically begin until these deliberations have occurred—and this delay likely increases the risk that the effort itself may lose steam and the likelihood that instructional improvement and gains in student achievement will fail to be realized. In light of $11 findings to date on the ASP program, then, at least two potential scenarios seem possible: (1) ASP begins strong in building trust among staff who become engaged in preliminary work to establish values, philosophies, and goals with respect to instructional improvement, but then that momentum fizzles as the real work of implementation begins; (2) ASP begins strong and effort remains strong throughout implementation, but the collective decisions about instruction simply do not result in any appreciable change in instruction and achievement. In the end, it is clear that there are many points at which the ASP instructional i mprovement process might fail to produce the desired results. Regardless, these findings, while preliminary, suggest that instructional improvement need not be accompanied by high levels of trust among colleagues. Furthermore, these findings suggest that the design and context of the instructional improvement program play a role in determining the need for growth in trust among Colleagues engaging in ambitious instructional reform. This being said, it is certainly ’1 ikely that a minimum level of trust among teachers is necessary to carry out the work of instructional improvement, but these findings demonstrate that growth in that trust over time may not be entirely necessary to ensure program success. Factors associated with change in trust among teachers. This section is COncemed with discussing what we have learned about the factors related to the i n Structional improvement process most related to change in trust among teachers. What 123 we find when looking across the models is that critical discourse, climate of innovation and risk-taking, and collective responsibility among teachers all co-vary with respect to trust grth over time, and across all models. These findings provide further support for past literature which has linked each of these factors to trust, but take conclusions one step further—this study establishes these factors as being related to change in teacher trust. This may be a small caveat to add, but it is significant when considering what little we know about how trust might be built in schools and the factors most responsible for its development. As an individual factor, it was somewhat interesting to find that teacher collaboration was not related to change in teacher-teacher trust within the intervention schools, and was, in fact, negatively related to trust in control schools. The finding of a negative relationship between trust change and collaboration within control schools is not necessarily surprising when we consider the past literature. In the absence of good conflict management skills and critical discourse among faculty, teacher collaboration is more likely to undermine trust among teachers than promote it (Borko, 2004; Hart, 1998; Putnam & Borko, 1997; Rosenholtz, 1991; Tschannen-Moran, 2001 , 2004; Uline, Tschannen-Moran, & Perez, 2003). Yet a possible alternative explanation also might explain the absence of an relationship for collaboration in the two CSR models, and this explanation relates to the actual measure itself. It could be that what was measured as “collaboration” within the 811 study was not wholly representative of what collaboration among teachers in each model looks like. For example, within SFA, measuring collaboration in terms of C larifying standards, and examining/changing the scope of curricular materials (among 124 other things) might not necessarily be compatible with the types of collaborations SFA needs and/or requires. It is certainly likely that SFA does not require a lot of discussion or effort to be devoted to establishing or clarifying standards or the alignment of curricular materials since the procedural controls which have been established generally take care of these things. It is far more likely that teachers collaborate on things like student regrouping and data analysis, for example. This is certainly one limitation of the dataset, to be sure, but it bears keeping in mind that it is also highly inefficient from a research standpoint and problematic from an external validity standpoint to develop understandings of collaboration which might be more CSR specific. When looking at factors related to trust grth unique to specific models, we see several note-worthy findings. Most particularly, within the ASP model, efforts of school leadership to establish and build a vision of school improvement were found to be related to change in trust among teachers. Given the latitude that teachers have in the ASP model for designing their own path to “powerful learning” for their students, it is quite possible that fragmentation among the staff with respect to improvement could develop quite rapidly. The findings here suggest that school leaders’ efforts to maintain cohesion with respect to school mission and vision provide opportunities for teachers to maintain and further build their relationships. Within the SF A model, both teacher and leader hiring and shared instructional experience among teachers were found to be related change in trust among teachers. This finding does not seem particularly surprising when you consider the uniqueness of the SFA program, and the typically heightened responses and strong opinions many have toward the instructional program itself. This fact is likely to result in an increased 125 likelihood of dissatisfaction and discord among some faculty who disagree fiindamentally with the programs’ approach. In these instances, the hiring of new personnel provides an opportunity for school leaders to shape their faculty in such a way as to ensure that a greater percentage of the overall faculty are willing and committed to the instructional program. Indeed this finding supports recent findings that leaders’ efforts to reshape the faculty to support the instructional effort were key to building trust over time (Kochanek, 2003, 2005). Furthermore, shared instructional experience among teachers was an important factor related to change in trust among teachers. This finding suggests that structural aspects of the SFA program such as the common core curriculum, common language and expectations around instruction, and shared Ieaming goals may provide a strong framework upon which teachers can be drawn together to work on the business of instruction. This finding provides at least some preliminary evidence in support of the final stage of trust development, identification-based trust, proposed by Lewicki and Bunker (1996). Recall that this is the final stage of three in the development of trust among work colleagues. It is characterized by deep identification with other’s desires, values, and intentions. Members at this stage are well on their way to developing a collective identity where joint goals and values among members of the organization have begun to crystallize. Sharing values, norms, and intentions allows other team members to speak or represent others’ interests, and thus regular communication and interaction is greatly enhanced and paves the way for even further trust growth. The findings here certainly seem to suggest that the shared work that teachers engage in under SFA over 126 time works to ensure that trust continues to develop, even as other improvement activities are taken into account. Taken together, the above findings suggest that the implicit yet pervasive assumption in the literature that trust develops in much the same way across schools seems incomplete. Guided primarily by three case-studies of Chicago schools, Kochanek’s (2003, 2005) theory of trust building suggested that all school principals work to reshape their faculty, and then get to work on ensuring that teachers have first low-risk opportunities to interact, followed by more hi gh-risk opportunities such as those around curriculum and instruction. Clearly the above findings suggest that teacher hiring may be highly dependent on the type of instruction which is occurring in a particular building. For schools implementing CSR models which call for a strong intervention in the technical core like SFA, teacher hiring is likely more important because it allows those teachers unwilling to change what they are doing to leave and to be replaced by more willing teachers. For ASP, whose instructional intervention is likely much more variable across schools, teacher hiring may be substantially less important to building trust. Even though teachers have to work together initially to agree upon standards for Ieaming, etc., they know that as soon as this work is done, they will return to their own classrooms to do what they wish. Trust in the case of ASP schools might actually devolve into some form of what Tschannen-Moran (2004) terms “cordial hypocrisy,” whereby people have developed more of a “surface” trust, and thus one which might easily break down if challenged in any way. As we know, trust runs deep among teachers who work with one another diligently on the business of instruction. Yet the ASP model, at least as it existed 127 at the time of this study, allows teachers to operate with substantial autonomy with regard to actual instruction, and thus virtually free of challenge from their colleagues. This being said, evidence from this study certainly provides some support for a set of factors which should be present in schools, yet evidence also suggests that some factors may be more context-specific. By studying trust growth in two very different comprehensive school reform models as well as a set of control schools, this study was able to view more explicitly the role that context might play in the development of trust. At the very least, the findings of this study suggest that fiiture research seeking to investigate trust development pay more attention to the instructional and policy contexts of the schools they are studying. Finally, it is important to discuss the structural aspects of the schools under study that were related to trust growth. The effects of school size on trust have been mixed in the educational research, and unfortunately, this study does little to provide more definitive answers in this regard. Across all subsets of schools, school size was not found to be related to change in trust among teachers. However, one factor which was important for both interventions was faculty stability, confirming the suspicions of other researchers that this is an important structural facilitator of trust (Bryk & Schneider, 2002). That this factor was not related to changes in trust in control schools suggests that ambitious instructional interventions such as ASP and SF A might be more sensitive to faculty tumover given the size and scope of the changes they are seeking to foster. Finally, this study has provided further evidence of a relationship between social trust and depth of policy implementation (Coburn & Stein, 2006; Desimone, 2002; Elmore, Peterson, & McCarthey, 1996; Frank, Zhao, & Borman, 2004; Scribner, Hager, 128 & Wame, 2002; Smylie & Evans, 2006; Spillane & Thompson, 1997), demonstrating that growth in relational trust among teachers in both CSR programs was related to teachers’ perceptions of efficacy in implementation as well as the alignment of the program with prior practices. These findings support the inclusion of similar measures as control variables in future studies of trust, particularly those implementing school-wide instructional reform models such as ASP and SFA. Stuay Limitations and Suggestions for Future Research This study would certainly not be complete, however, without a discussion of its limitations and the opportunities these limitations present for fisture research. At the outset of this study, it was made very clear that the research design employed here is not one in which causal inferences about the relationship of trust growth to instructional improvement process can be made. Given the complex nature of the relationship between trust and the various processes of instructional improvement studied here, a causal design would certainly have provided more detail on the specific nature of these relationships. As was stated, however, the purpose of this study was to take some preliminary steps in first establishing if there was trust grth in this CSR models over time, and, if so, what was related to that trust growth. Given what this study has provided in the way of identifying specific factors related to trust growth over time, future studies now have the opportunity to shift focus toward teasing out more definitively the causal relationships between these factors and trust growth over time. Speaking of trust growth, another limitation of this study was the manner in which trust growth was measured, specifically with regard to the focal instructional improvement factors which were an important part of the study. The use of time-varying 129 covariates at the response level in place of models which more directly modeled variation in trust trajectories between teachers and schools over time limits the conclusions we can reach with regard to trust growth and/or development over time. The only conclusions which are directly amenable to interpretation as related to trust growth over time are the findings of faculty stability in SFA schools and the finding of prior achievement in control schools. Further studies can further explore these relationships by moving the measures of instructional improvement used in this study up a level in the analysis in order to model variation between teachers in trust grth over time. Further, this study breaks little new ground on what we know about trust growth at less-studied levels of schooling and between less-studied role-relations. Studies of trust have predominantly focused on the elementary level and, of the role-relations, trust among teachers has been the most extensively investigated. While this study breaks no new ground in our understandings of trust, for example, at the secondary level or among teachers and parents, what we know about how trust develops is thin across all of these contexts. In essence, this study breaks new ground—even within these more extensively studied areas—by looking at trust development over time. Future studies should branch out and examine how trust develops in secondary schools and of the factors which give rise to trust grth in other role relations such as principals and teachers, parents and teachers, and teachers and students. With regard to research method and design, one limitation of this study was the lack of a measure of the trust between teacher, principal, and other school leaders. A few trust studies have had access to a measure of teacher-principal trust, and this can often assist in capturing more precisely the nature of all of the trusting relationships within the 130 school building. Some of these studies have found a significant relationship between teacher-principal trust and teacher-teacher trust, and having had access to such a measure in the current study may have potentially led to more accurate model specification. Future studies of this type should endeavor to include more measures of trust among different role-relations to more fully ascertain their relationships to one another, and more specifically their relationship to growth in teacher trust over time. Future studies in this vein should also explore how the power structure of the school affects trust growth, as influenced, for example, by configurations of teacher experience and seniority, as well as other important factors. Moreover, as a study of trust growth over time as it relates to the implementation of two different CSR programs, this study was clearly limited in: l) the degree to which all schools implemented the reforms at the same time; and 2) in the number of years into implementation these schools were studied. Because all of the schools began the implementation within 2 years of each other, this study was limited in what it could definitively say about trust growth in these schools over time. By centering the TIME variable on Year 3, this study was able to adjust adequately for these differences, but perhaps a future study could be designed which examined the implementation of these (or other) CSR models for schools who began at the same time. The measures generated from growth modeling such as initial status and average rate of change would be more precise and would lend themselves well to interpretation. Moreover, it must be acknowledged that the centering of TIME on Year 3 for this study may not have been fully adequate in capturing the intended effects hoped for by this study. As discussed earlier, CSR developers and providers discovered early on that 131 the ideals of reform that these instructional interventions seek to foster take considerable time to develop—sometimes as much as 6 to 8 years—and likely lead to significant turmoil and difficulty that can last for many years. Because the 811 study lasted for four years and the schools that were chosen for the study were, at most, in Year 2 of implementation when the study began, this means that the most seasoned schools were only in Year 5 of implementation for this study’s analysis. And while it is hard to criticize any study which includes such extensive data on such a large set of schools over 4 years, a study which included perhaps 2 more waves of data on trust would allow researchers to more definitively explore trust growth over time. While there was no evidence to suggest that trust growth was anything but linear over time, more waves of data would allow researchers to examine more fully potential non- linear relationships. Further, more waves of data would also allow for the more extensive investigation of trust growth in CSR programs like ASP in which instructional improvement does not really begin until Years 3 or 4. 132 APPENDICES 133 APPENDIX A RASCH MEASURES USED IN STUDY I34 @5664 :65 665:: 3 3.2: 8.8 8 66586 66 £266“; mEoo:mmm_o :65 :_ ~656on 8 692386 66 5266;. 662 >6: So 66m 56 :66— b6=:::oo 8 466696 $666.4. 24.9 _ _4.9 24.9 owd .: #9 :49 £49 and .: d9 :49 £u~9 who :49 :49 £49 w; 66ch on??? 2 66.3.5 omega .668 569$ wfixfl. 4mg 5:6 :o_6>o::_ .40 665:0 ”0... 6:65 56 6:256 £63 @285 wEv—E co no“ coow a ow 5266:. £63 @6598 :0 656.5 9 wEEB 66 96.66% $562: 6 $63 6666.406 666 36% €266“ >62 24.9 :4.9 24.9 who 249 :49 289 Ed 2&9 :49 249 Ed 249 :49 249 who 63%» bung: 2 spammed» $35.26 .668 “509$ @66on w:oE< 8.5085 665.5 ”Orr 46:66:45 626m :4 96— 65 843 on? 26664 650 636: £266... 650 .66 Sons 666 .56: 696m 45 :4 5266.4. 650 66 $3.: 628 $5 :4 £666... ~66 :65 :_ tomxo 66 0:3 666:8 636: m666F 24.9 24.9 24.9 24.9 is 249 249 2-8:. 249 Rd 249 2.49 249 2&9 as 249 249 £49 249 33 666.6 3%:th 8 66.3.5 $38.5. .668 4:694; .038”— w:oE< 42:... ”OF :6: v 66> m 66> N 66> _ 66> 55% E .63 6.536: «6.8% «GHQ mxmcttccamzm 6364 396% 58 «92 mhctzchmmau 65665 4< 25d. 135 08058080 0:00am 0:0 0.050005 :060580 :0 60:69.0 05 w:_:_::0xm 00:00: 8605.50 056000 :0 00001086008 05 w:_w:0:0\w:::_::0xm 000:0 :060580 80:00 :050385 w:50:w0::: .86: 050505 w:_00_0>0Q w:_::00_ E0008 :0: 00:00:80 w:_©_:0_u 2.3.9 2.3.9 2.3.9 2.3.9 8.0 2 049 2 049 239 2 W49 :20 2.049 2849 2849 2.049 8.0 839 2:39 £39 839 8.0 605.: Q N :0:: 0:05 0: 0>0: .068 660.3 6500.8:— :0 :050:0:0:0U :0:000:. ”00. 60:8 05 :_ w:_:000: :0 3:000 05 w::>0:0::_ :0: 0:550:88: 00:8 80:80:. 60:8 0:5:0 05 E 630:0: 5:00am 0>5800 5850:: 06: 80:80:. :03 00 :050:0 0:0 w::0_0: :0: 5568:0000: 00—8 0:0:0000. 004.9 004.9 84.9 2.0.0 o~49 £49 849 8.0 o~49 £49 «~49 2:: 0~49 £49 849 2:: 620600: 20 3:00: 0: E0600: 0: .208 860-3 86:00.: 0:0 w:_:000:. w::>0:0::: :0: 005580000: 02:00:00 :0:. 00050:: :0::8 8: 0:0 5:00:00 0500 :0>00 8:00am 3:05:50 505 80:80:. 806 b: E 00502 060008 50:3 30:: 0: 80:80: :050 :0: >000 0.: 3000380 00502 0>0: 8:00am 00:3 :0 800—305: 02800 0>0: : 80:80: :050 :0 65008:: 0:0 00:0>00 0:05:00 :0 8000505— 005800 0>0: _ 2.4.9 2.4.9 2.4.9 2.4.9 :0 2.49 2.49 2.49 2.49 020 2.49 2.49 2.49 2.49 Ed 2.49 2.49 2.49 2.49 3.0 60:08 $090.8 0: 0020.85: $38.00 .208 0:60-00 0:0:0000. w:0::< 00:0500xm 6:050:58: 00:0:m :0:. $822.8 2 2205 I36 8888—0 05 E me5 Bo: >5 8 0E 34 >83 26: 8 E @5883 b: “:38 x55 8 08 to: 08:8 8 8:82 .8588 Eot =28~E£E 85228 So 88 8 2: to: E0888? 2: E was: 83 _ meE 83038: 3 :2828 homo—o 8: 0E £82 @5388 b: “:28 xomnvofl 38m: 53> 2: 3235 2:: .3 :28: 82868 .8 8m :8an 8 co m:o£ 9 2: 3322 600883 b: E 2: 8 .390: 032322 5:5 2: 3235 was—382 b: mo 38:8 co x53 8 mo:_::tomao 58.: 2: 080 .5158 :33 £888 8389 :38 8%:th 98-58 £39 23 .933 film? 58:88 88-88 31mg 28-88 $391 $33 23 .5188 :38 £39 83189 :38 83-88 fi 38 «$1.83 88 .339 539 €38 8839 $39 339 £38 833 8.8 83%: REES: 8 mmxmafiu bmghm 6.80m “Eomév 82.83888 «cacao—050 858838.: 853,—. @530 ”OF 8:08.”: 809583 305:: 8 38:08.532 82>th :8:on Mmu no $85 wcfiog 8888? Mo @2985 ~8th5 3 0E 38:6 8; :289958: >2 x83 :8me 80 83:85 :8pr5 8 2: 3898 m8: =28E£t3 >2 @250: D820 08 3.88:8 8.: 280m 8:28:55 353:0 £820 08 EoEo>8m§ 808883 80an 8 88m 2: woos—0:38 E8 comma 35880 08 530885 wE>oE~E .8“ 8on of. c2828.: w:_>o.aE_ Sm .83 @288: 8 E 82:. 8 815,8 8:98 33:59 8:99 98.89 £888 £899 88 83.88 $.88 02.88 8318.8 031mg “593 83.8.8 88 88:88 88188 3:99 38:89 518p gamma «$.89 93 881:9 :38 0:38 339 £39 9:39 «2.55 8.0 33%: bmgha 8 mmxmgfi $355. 68% 869$ 8.82:0 8:28ng ”OH 6258.50 _< 285 137 “5&0 03:: w:_>0_:00 0: 33:00 0:0 w:_>_0m E0303 ncm :0:0w:m0>:_ 900.5 43:: 30:8 :0: 00:02 wEbSEE :0:: 00000:: :00Q0_0>0:-=0>> 0 000: _0o:0m 02,: 200m :23 03:03:00 0:0 Ewsob 2:0:woa 000E080»: 00000.0 3030: 5.030 < 3:888: 35.0.50 $33.0 0:0 00>:0E020 .we—06:00: wfixmfi 0:0,:0m mt0b0 0:0E0>0EE_ _00:0m 02% :0:: m020> 33:0 («0 H00 0 3: 30:00 0:: 5130 05130 minim neufim manta 3.0 cauflm xeuflm 02:30, 321%; «213m £0 celfim gfuim 5-3m 58-30 oenfim 02-30 neufim £2-30 “Elma 33% 3.0 30 3me: 33:33 00 00xm~0€0 3050:: .208 0500.3 @302 :0E00D 00:0:m :0m0m-m0:_0> ”Ow—m 00:08:85: 2500000 2:00:00 :0: 0:03—0:90 :0 20930098 w::©t£0 «58389:: 6:032:35 :0: m0::0:5 :0:—0x0 w:_:0m 200w E08323: 30:00 m: 2030: mm0&0:q mEEmem “:0E0>0a:: 30:8 :0: £00m 000:: w::00_:3:~:00\w:::8m £130 £13m £45m «0:30 :3 cwumqm owumqm awnmqm mwlmqm 05:0 £13m 230m owufim £130 £13m £40m £13m £15m 2.0 $0 Awmttotm 00h: 33$: M09050 0: b.2030. 0: .2000 “509$ :20; 0 wEESm ”04m game: 0:: x: :00 00:00 00w:0:0 0:0 0:_0> _ E0E0>0_:00 2:20:00 90: Samoa 0:: x: :0: 00:00 00w:0:0 88000:: 0:: 0n: :0: 00:00 00w:0:0 0:0 mat—0:: 00 030900 80 _ 900—000 00:03.: 2056505 E000 30:8 00 0002500 E0 _ 020:0: 3:05.505 $3053 0028:: 0:: 50:0 050:: .w:_:0000 30:0 5:05:85 0:: :0 :0 50 A0805: 303:0 :0 no 30 8005020 5 0: 0: 00:3 w:_000:0 £0056 0>0: _ 30:00 00:05.5 b9068300 5:0 0:0 :000: 200:0 _ 30: 30:0 020:0: 3%-th 2300 3300 839 £32 £300 £300 £33 «$.00: «$-80 330: £30: 3.5: 3-8: 3:80 3-50 03000 omumg £180 £49 p39 :39 3-80 £150 «39 2.80 .3130 £48 a; w; 8.0 8.0 A0038 $000600 0: 0300030 ©0395. .2000 “509$ 35858038: anoi m0 5&0Q ”OH 625080 _< 2.30 138 $22000 8080>088_ _00:0m 3.8003 85.698 30:? 80:03“ 30: waif .308 8 600008 080 028098 5; ts; “80038050888800 30: warm: nomnvqm nomlmqm DONINAm an_|_0m momnv0m momumqm monumqm owd owd mud mh_|~0m VBAV 6:008 83% 0 00 :0 E 00: .208 “809$ 8080>088~ _00:0m 00.0 chI 000001— 0—5 00:003. ”04m .308 80095 8:080 3:200:me 000:3 808008 oficwoofl 30:88 6880 _ mtoto 8083.088 80:8 00 @8200 388:0 808:0 wEm: 8050000 Baa—«>0 ~ @883— 0; 3009.8 :83 08888 00 m8000mm£0 8 03030 _ @0000 038‘ 000—000 >05 :2: 00m 00 020080 380300.008 8008320 000808 _ ow_:¢0m cwfllvqm ow_|vgm pw_nvqm owa. ow_nm0m uw_lm0m nw_nm0m mw_im0m :«o ow_-~0m uw_:mqm pw_u~0m ww_:~0m @580 om_|_0m um_u_0m omfin_0m nm_l_0m 0580 3002 x00. 00% N 20$ 0.88 00 L96: .2000 0800-3 9:008:02 3:050:55 03000005 ”04m 888:8 Z 2%: 139 APPENDIX B DESCRIPTIVE STATISTICS 140 ......NN. NOV 5. ‘15—. 5.- 2:300. .315. fine. ..imo. ..imo. tumo- fuwo. ......oo. **mo. :52 E EoEo>O~QE~ .2 ..... Ike. ......MO. .I. _ m. ......2. *mo. 3me. ......vo. ......mo. ......mo- 5.- .IRO. **No. ......xo. <1— 5 EoEo>OEE~ .3 ..... :8. :2. :2. S. 8. :8. 8. :8. :8. S. :8. :20. 8E: 508-. 05 33$ .2 ..... :2. :3. :8.- _o.- S. 5.- 5.. :mo. :3.- Su 5. 0:028: .55 02:95:0- .2 ..... ......oN. *LZ. ......mo. **v_. “Ibo. tug. ..S:. .1an... two—- ......2. mac—ME cofimoofl “go-Era .: ..... S.- 8. *8. ...No. 8. 8. :2. 8. 3.- 82> a 05230 .2 ..... :2. :mm. :2. :3. :00 :mm. : _ m. :9... 80885208 .505 .o ..... ...... 3. :8. :8. :8. :8. :8. :3. 05502050 005 28¢ .w ..... :R. :3. :0». :8. :2. :8. 852:0 .223ng .2 ..... :8. :2. :mm. :5. :m. _. 8088230 56:; .0 ..... :2. :3. :fi. :3 £822.88 308:8 .m ..... :mm. :2. :8. 85038 83% .v ..... ...... _ m. :3. 953 002 23 8:985 .m ..... :00 85800 .8000 .m ll- “ma-w accused 00:080- . _ E m_ N_ I o. o w n o m 0 m N _. 0.50002 02020200.: N 2000-N 200300 020.250.2200 220220: 202.00-i 20.23% Am 0308 I41 M000 _ 3.. $0.0 $0.0- de 0w.~ £02 808030008— 3:000:80:— m00.0 _ m _ .0 000.0 0: .0 00.0 00.~ 0:010 8083208— 08000800:— 0000 0 _ 0.0- 000.0 30.0- 00.0 00.: 00:8 .0080 000 000030 000.0 00nd $00.0 nmmd- 0m._ mmd w8:0::02 3:053:05 030-00095 000.0 000.0- 000.0 000.0- 00: 00.. 0205 0 050000 000.0 002 000.0 000.0 00: 00.: 00202 02280 0220 00000-0020> 000.0 000.0 000.0 .000 00: 00: .0000 00030052050 00:00 000.0 000.0- :00 0:0- 00.0 00.0 00000000000 2:00:00 000.0 00 _ .0 000.0 00 _ .0 00.0 00: 800200 30002505 000.0 002 000.0 0: .0 00.0 0: 8200:0000 00 000030 0:000 000.0 :00 2.0.0 02 .0 00.0 0 0 .0 8005000 0220 000.0 000.0 000.0 000.0- 0 0.0 00.0 0030:. 008 000 0009005 000.0 000.0- 000.0 000.0- 0 0.0 00: 00008000 .8000 000.0 00 _ .0 000.0 02.0 00.0 00.0- 808002.00 000.0 000.0- 2.0.0 0000. 00.0 00.0 020 00:00:00 50080 .m.m 00000.5! .m.m 000:30xm :000300 8002 0000090 000 u 5 03.50.0- 000.00. 0.00.. 05 00:00:08.0 0000.580 .00. 2000 I42 000.0 0m0.0 $00.0 mead- 00.0 000.- 502 8080;088— 8:000:00:— 0000 03.0- 000.0 3000- mm .0 N0.N 0:03 808300080 8:000:00:— 0000 02 .0- 000.0 000.0 A00.0 00.: 00E: .0080 000 .2080 wood mm _ .0- 000.0 30.0 2 .0 5.0 w8000802 8:000:55 0>0:-.0005 000.0 0: .0- 000.0 :00- 00.: 0 _ ._ 020$ 0 050000 000.0 000.0 000.0 30.0 02 00.0 00202 02080 02000 00000-000_0> 000.0 000.0 000.0 000.0 000 00.: 5000 000255208 00:00 000.0 0000- 000.0 E _.0- _0.0 00.0 00000000000 0:08:00 000.0 0. _.0 000.0 00_ .0 00.0 00.0 800050 00002505 000.0 000: 000.0 0:0 00.0 0: 0205:0000 00 000080- 0:000 0000 000.0 000.0 :00 00.0 3.0 802.0000 0220 000.0 000.0- 000.0 000.0- 00.0 000 0:200 000 0:0 0000.505 00.0 0 _ 0.0- 000.0 000.0- 00.0 03 005800 .8000 0000 000.0 000.0 0000 0:0 00.0. 0002000000 000.0 00 0.0- 000.0 0_ _.0- 00.0 00.0 0.2.0 080200 000080 .m.m 0600:6— .m.m 000858—m :0_00_>0Q 8002 0000090 :0 _ 0 n 5 0150.0 08000. 0.00. 20 000800020- 00000.2080 .00 2000 I43 02.0 000.0- 000.0 52 .0- 00.0 00.0 :32 20:55:95 3:000:55 02 .0 000. _ ~000.0 000.0- 500 0.0 3:65 0:253:06— 3:000:55 02 .0 0050- 000.0 00.0.0- 000- 00._ 0:53 6000-— 0:: 500000. >20 30.0 000.0 000.0- 3..— mv0 05:02:02 3:003:35 03050090 02.0 0005 000.0 000.0- 00.. 00._ 535 a 050::m 02 .0 0005 000.0 000.0- 00.2 00.0 053.2 :03600 00.300 00030-m0:3> 02.0 :00- 0000 N000- 00.0 00.0 3:53:0Qmom 03:00:00 2 _.0 000.0 000.0 000.0 mmxn 00.0 332:0 3:003:35 02.0 000.0 000.0 08.0 00.0 2..— 80535000 Q0 500000- 053:0 02.0 000.0 000.0 000.0 00.0 00.0 costumxm 00300 02.0 00 _ .0- 000.0 02.0- 5.0 2 .0 05030- 032 0:3 :0_3>0:5 02.0 000.0- 000.0 0 _0.0 00.0 mm; 85085 305.6 00 _ .0 0050 000.0 30.0 0~.N 00.0- 5:803:00 00 2 .0 000.0- 000.0 n2 .0- v0.0 00. m 5?:- 3:0_33.m 023300- .m.0 m_m0t:v_ .m.0 30:38.0 553060— :82 0:00:80 302 H :0 03.50% 300.00. 50.0200 2: SKEWBCSM 0:25.8me .vm 073.0 144 REFERENCES 145 REFERENCES Achinstein, B. (2002). Conflict amid community: The micropolitics of teacher collaboration. Teachers College Record, 104(3), 421-455. Adams, C. M. (2008). Building trust in schools: A review of the empirical evidence. In W. K. Hoy & M. F. DiPaola (Eds.), Improving schools: Studies in leadership and culture (pp. 29-54). Charlotte, NC: Information Age Publishing. Adler, P. S., & Kwon, S-W. (2002). Social capital: Prospects for a new concept. Academy of Management Review, 27(1), 17-40. Ahn, J., & Brewer, D. J. (2009). What do we know about reducing class and school size? In G. Sykes, B. Schneider, & D. N. Plank (with T. G. Ford) (Eds.), Handbook of educational policy research (pp. 426-437). Washington DC: AERA. Baier, A. (1986). Trust and antitrust. Ethics, 96, 231-260. Ball, D. L., & Cohen, D. K. (1999). Developing practice, developing practitioners. In L. Darling-Hammond & G. Sykes (Eds.), Teaching as the learning profession: Handbook of policy and practice (pp. 3-32). Berends, M. (2000). Teacher-reported effects of New American School designs to teacher background and school context. Educational Evaluation and Policy Analysis, 22, 65-82. Berends, M., Bodilly, S. J., & Kirby, S. N. (2002). Facing the challenges of whole school reform. New American schools after a decade. Santa Monica, CA: RAND. Béteille, T., & Loeb, S. (2009). Teacher quality and teacher labor markets. In G. Sykes, B. Schneider, & D. N. Plank (with T. G. Ford) (Eds.), Handbook of educational policy research (pp. 596-612). Washington DC: AERA. Bickel, R. (2007). Multilevel analysis for applied research: It 's just regression! New York: Guilford Press. Bloom, H. S., Ham, 8., Melton, L., O’Brien, J., Doolittle, F. C., & Kagehiro, S. (2001). Evaluating the Accelerated Schools approach: A look at early implementation and impacts on student achievement in eight elementary schools. New York: Manpower Demonstration Research Corporation. Bok, S. (1978). Lying: Moral choice in public and private life. New York: Pantheon. 146 Bond, T. G., & Fox, C. M. (2007). Applying the Rasch model: Fundamental measurement in the human sciences (2nd ed.). Mahwah, NJ: Lawrence Erlbaum. Borko, H. (2004). Professional development and teacher Ieaming: Mapping the terrain. Educational Researcher, 33(8), 3-15. Borman, G. D., & Hewes, G. M. (2002). The long-term effects and cost-effectiveness of Success for All. Educational Evaluation and Policy Analysis, 24, 243-266. Borman, G. D., Hewes, G. M., Overman, L. T., & Brown, S. (2003). Comprehensive school reform and achievement: A meta-analysis. Review of Educational Research, 73(2), 125-230. Borman, G. D., Hewes, G. M., Overman, L. T., & Brown, S. (2004). Comprehensive school reform and achievement: A meta-analysis. In C. Cross (Ed.), Putting the pieces together: Lessons from Comprehensive School Reform Research (pp. 109- 150). Washington, DC: National Clearinghouse for Comprehensive School Reform. Borman, G. D., Slavin, R. E., Cheung, A. C., Chamberlain, A. M., Madden, N. A., & Chambers, B. (2007). Final reading‘outcomes of the national randomized field trial of Success for All. American Educational Research Journal, 44, 701-731. Bryk, A. S. (2009). Support a science of performance improvement. Phi Delta Kappan, 90(8), 597-600. Bryk, A., Cambum, E., & Louis, K.S. (1999). Professional community in Chicago elementary schools: Facilitating factors and organizational consequences. Educational Administration Quarterly, 35, 751-781. Bryk, A. S., Lee, V. E., & Holland, P. B. (1993). Catholic schools and the common good. Cambridge, MA: Harvard University Press. Bryk, A. S., & Schneider, B. L. (2002). Trust in schools: A care resource for improvement. New York: Russell Sage Foundation. Bryk, A. S., Sebring, P. B., Kerbow, D., Rollow, S., & Easton, J. Q. (1998). Charting Chicago school reform: Democratic localism as a lever for change. Boulder, CO: Westview Press. Clifi, R., Veal, M. L., Holland, P., Johnson, M., & McCarthy, J. (1995). Collaborative leadership and shared decision making: Teachers, principals, and university professors. New York: Teachers College Press. Coburn, C. E., & Stein, M. K. (2006). Communities of practice theory and the role of teacher professional community in policy implementation. In M. I. Honig (Ed.), 147 New directions in education policy implementation: Confronting complexity (pp. 25-46). Albany, NY: SUNY Press. Cohen, D. K., & Ball, D. L. (1999). Instruction, capacity, and improvement (CPRE Research Report Series RR-43). Philadelphia, PA: Consortium for Policy Research in Education. Cohen, D. K., & Barnes, C. A. (1993). Pedagogy and policy. In D. K. Cohen, M. W. McLaughlin, & J. E. Talbert (Eds.), Teaching for understanding: Challenges for policy and practice (pp. 207-239). San Francisco, CA: Jossey-Bass. Cohen, D. K., & Hill, H. (2001). Learning policy: When state education reform works. New Haven, CT: Yale University Press. Coleman, J. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, 895-8120. Coleman, J. S., & Hoffer, T. (1987). Public and private high schools: The impact of communities. New York: Basic Books. Correnti, R. (2007). An empirical investigation of professional development effects on literacy instruction using daily logs. Educational Evaluation and Policy Analysis, 29(4), 262-295. Correnti, lR., & Rowan, B. (2007). Opening up the black box: Literacy instruction in schools participating in three comprehensive school reform programs. American Educational Research Journal, 44, 298-338. Cuban, L. (1988). The managerial imperative and the practice of leadership in schools. Albany, NY: SUNY Press. Darling-Hammond, L., & McLaughlin, M. W. (1996). Policies that support professional development in an era of reform. In M. W. McLaughlin & I. Oberman (Eds.), Teacher learning: New policies, new practices. New York: Teachers College Press. Datnow, A., & Castellano, M. E. (2000). Teachers’ responses to Success for All: How beliefs, experiences, and adaptations shape implementation. American Educational Research Journal, 3 7, 775-799. Datnow, A., & Castellano, M. E. (2001). Managing and guiding school reform: Leadership in Success for All schools. Educational Administration Quarterly, 37, 219-249. Desimone, L. (2002). How can comprehensive school reform models be successfully implemented? Review of Educational Research, 72, 433-479. 148 Duke, D. L., Showers, B. K., & lmber, M. (1980). Teachers and shared decision making: The costs and benefits of involvement. Educational Administration Quarterly, 16(1), 93-106. Elmore, R. F. (1990). Restructuring schools: The next generation of educational reform. San Francisco, CA: Jossey—Bass. Elmore, R. F. (2004). School reform from the inside out: Policy, practice, and performance. Cambridge, MA: Harvard University Press. Elmore, R. F ., & Burney, D. (1998). School variation and systemic instructional improvement in Community District #2, New York City. Philadelphia, PA: Consortium on Policy Research in Education. Elmore, R. F ., Peterson, P. L., & McCarthey, S. J. (1996). Restructuring in the classroom: Teaching, learning, and school organization. San Francisco, CA: Jossey-Bass. Evans, R. (1996). The human side of change: Reform, resistance, and the real-life problems of innovation. San Francisco, CA: Jossey—Bass. Finnan, C., & Swanson, J. D. (2000). Accelerating the learning of all students: Cultivating culture change in schools, classrooms, and individuals. Boulder, CO: Westview Press. Firestone, W. A. (1989). Using reform: Conceptualizing district initiative. Educational Evaluation and Policy Analysis, 11, 151-165. Firestone, W. A., & Corbett, H. D. (1988). Organizational change. In N. Boylan (Ed.), Handbook of research on educational administration (pp. 321-341). White Plains, NY: Longman. Ford, T. G., & Youngs, P. A. (2009). How Success for All promotes trust in a high performing urban Midwestern district. Paper presented at the annual meeting of the American Educational Research Association, San Diego, CA. Forsyth, P. B. (2008). The empirical consequences of school trust. In W. K. Hoy & M. F. DiPaola (Eds.), Improving schools: Studies in leadership and culture (pp. 1- 27). Charlotte, NC: Information Age Publishing. Forsyth, P. B., Barnes, L. L. B., & Adams, C. M. (2006). Trust effectiveness pattems in schools. The Journal of Educational Administration, 44(2), 122-141. Frank, K., Zhao, Y., & Borman, K. (2004). Social capital and the diffusion of innovations within organizations: The case of computer technology in schools. Sociology of Education, 77, 148-171. 149 Fukuyama, F. (1995). Trust: Social virtues and the creation of prosperity. New York: Free Press. Fullan, M. (2007). The new meaning of educational change. (4th ed.). New York: Teachers College Press. Goddard, R. D. (2003). Relational networks, social trust, and norms: A social capital perspective on students’ chances of academic success. Educational Evaluation and Policy Analysis, 25, 59—74. Goddard, R. D., Tschannen-Moran, M., & Hoy, W. K. (2001). A multilevel examination of the distribution and effects of teacher trust in students and parents in urban elementary schools. The Elementary School Journal, 102, 3-17. Granovetter, M. S. (1973). The strength of weak ties. American Journal of Sociology, 78, 1360-1380. Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91, 481-511. Grodsky, E., & Gamoran, A. (2003). The effects of professional development on professional community in American schools. School Effectiveness and School Improvement, 14 (1), 1-29. Hanushek, E., Kain, J. F., & Rivkin, S. (2004). Why do public schools lose teachers? Journal of Human Resources, 39, 326-354. Hargreaves, A. (1991). Contrived collegiality: The micropolitics of teacher collaboration. In J. Blase (Ed.), The politics of life in schools: Power, conflict, and cooperation (pp.46-72). Thousand Oaks, CA: Corwin. Hargreaves, A. (1992). Restructuring restructuring: Postmodernity and the prospects for educational change. Unpublished manuscript. Harris, A. (2003). Behind the classroom door: The challenge of organizational and pedagogical change. Journal of Educational Change, 4, 369-382. Hart, A. W. (1995). Reconceiving school leadership: Emergent views. The Elementary School Journal, 96, 9-28. Heller, M. F., & Firestone, W. A. (1995). Who’s in charge here?: Sources of leadership for change in eight schools. The Elementary School Journal, 96, 65-86. Henderson, J. E., & Hoy, W. K. (1982). Leader authenticity: The development and test of an operational measure. Educational and Psychological Research, 3(2), 63-75. 150 Hopfenberg, W. S., Levin, H. M., Chase, C., Christensen, S. G., Moore, M., Soler, P., et a1. (1993). The Accelerated Schools resource guide. San Francisco, CA: Jossey- Bass. Hox, J. J. (2010). Multilevel analysis: Techniques and applications (2nd ed.). New York: Routledge. Hoy, W. K., & Kupersmith, W. (1984). Principal authenticity and faculty trust: Key elements in organizational behavior. Planning and Changing, 15(2), 80-88. Hoy, W. K., Sabo, D., & Barnes, K. (1996). Organizational health and faculty trust: A view from the middle level. Research in Middle Level Education Quarterly, Spring, 21-39. Hoy, W. K., Smith, P. A., & Sweetland, S. R. (2002). The development of the organizational climate index for high schools: Its measure and relationship to faculty trust. The High School Journal, 86(2), 38-49. Hoy, W. K., Tarter, C. J ., & Hoy, A. W. (2006). Academic optimism of schools: A force for student achievement. American Educational Research Journal, 43(3), 425- 446. Hoy, W. K., Tarter, C. J., & Witkoskie, L. (1992). Faculty trust in colleagues: Linking the principal with school effectiveness. Journal of Research and Development in Education, 26, 38—58. Hoy, W. K., & Tschannen-Moran, M. (1999). Five faces of trust: An empirical confirmation in urban elementary schools. Journal of School Leadership, 9, 184 208. Huberman, M. J ., & Miles, M. B. (1984). Innovation up close. New York: Plenum. Jacobsen, R. (2009). The voice of the people in education policy. In G. Sykes, B. Schneider, & D. N. Plank (with T. G. Ford) (Eds.), Handbook of educational policy research (pp. 307-318). Washington DC: AERA. Johnson, M. J ., & Pajares, F. (1996). When shared decision works: A 3-year longitudinal study. American Educational Research Journal, 33, 599-627. Johnson, S. M. (1990). Teachers at work: Achieving success in our schools. New York: Basic Books. Knapp, M. S. (1997). Between systemic reforms and the mathematics and science classroom: The dynamics of innovation, implementation, and professional Ieaming. Review of Educational Research, 67(2), 227-266. 151 Kochanek, J. R. (2003). Building trust: The process of developing relationships in elementary schools. Unpublished doctoral dissertation, University of Chicago, Chicago, IL. Kochanek, J. R. (2005). Building trust for better schools: Research based practices. Thousand Oaks, CA: Corwin Press. Kramer, R. M., & Tyler, T. R. (1996). Trust in organizations: Frontiers of theory and research. Thousand Oaks, CA: Sage. Kruse, S. D. (2001). Creating communities of reform: Continuous improvment planning teams. Journal of Educational Administration, 39, 359-383. Kruse, S. D., Louis, K. S., & Bryk, A. S. (1995). An emerging framework for analyzing school-based professional community. In K. S. Louis, S. D. Kruse, & Associates (Eds.), Professionalism and Community: Perspectives on Reforming Urban Schools (pp. 23-44). Thousand Oaks, CA: Corwin Press. Lewicki, R. J ., & Bunker, B. B. (1995). Trust in relationships: A model of development and decline. In B. B. Bunker, J. Z. Rubin, & Associates (Eds.), Conflict, cooperation, and justice: Essays inspired by the work of Morton Deutsch (pp. 133-173). San Francisco, CA: Jossey-Bass. Lewicki, R. J ., & Bunker, B. B. (1996). Developing and maintaining trust in work relationships. In R. M. Kramer & T. R. Tyler (Eds.), Trust in organizations: Frontiers of theory and research (pp. 114-139). Thousand Oaks, CA: Sage. Linacre, J. M. (2009). Winsteps Rasch measurement. Little, J. W. (1982). Norms of collegiality and experimentation: Workplace conditions of school success. American Educational Research Journal, 19, 325-340. Little, J. W. (1993). Teachers’ professional development in a climate of educational reform. Educational Evaluation and Policy Analysis, 15(2), 129-51. LoGerfo, L., & Goddard, R. (2008). Defining, measuring, and validating teacher and collective responsibility. In W. K. Hoy & M.F. DiPaoIa (Eds.), Improving schools: Studies in leadership and culture (pp. 73-97). Charlotte, NC: Information Age Publishing. Lortie, D. (1975). Schoolteacher: A sociological study. Chicago: University ofChicago Press. Louis, K. S. (2007). Trust and improvement in schools. Journal of Educational C hange, 8, 1-24. 152 Louis, K. S., Kruse, S. D., & Associates (Eds). (1995). Professionalism and community: Perspectives on reforming urban schools. Thousand Oaks, CA: Corwin Press. Louis, K. S., & Marks, H. M. (1998). Does professional community affect the classroom?: Teachers’ work and student experiences in restructuring schools. AmericanJournal of Education, 106(4), 532-75. Louis, K. S., Marks, H. M., & Kruse, S. (1996). Teachers’ professional community in restructuring schools. American Educational Research Journal, 33, 757-798. Lytle, S. J ., & Cochran-Smith, M. (1994). Inquiry, knowledge, and practice. In S. Hollingsworth & H. Sockett (Eds.), Teacher research and educational reform: Ninety-third yearbook, National Society of Education. Chicago: University of Chicago Press. McCoach, D. B., & Black, A. C. (2008). Evaluation of model fit and adequacy. In A. A. O’Connell & D. B. McCoach (Eds.), Multilevel modeling of educational data (pp. 245-272). Charlotte, NC: Information Age Publishing. McKnight, D. H., Cummings, L. L., & Chervany, N. L. (1998). Initial trust formation in new organizational relationships. Academy of Management Journal, 38, 473-490. McLaughlin, M. W. (1993). What matters most in teachers’ workplace context? In J. W. Little, & M. W. McLaughlin (Eds.), Teachers’ work: Individuals, colleagues, and contexts. New York: Teachers College Press. McLaughlin, M. W., & Talbert, J. E. (2001). Professional communities and the work of high school teaching. Chicago: University of Chicago Press. McLaughlin, M. W., & Talbert, J. E. (2006). Building school-based teacher learning communities: Professional strategies to improve student achievement. New York: Teachers College Press. Meier, D. (1995). The power of their ideas: Lessons for America from a small school in Harlem. Boston, MA: Beacon Press. Meier, D. (2002). In schools we trust. Boston, MA: Beacon Press. Mickelson, R. (2009). Race, ethnicity, and education. In G. Sykes, B. Schneider, & D. N. Plank (with T. G. Ford) (Eds.), Handbook of educational policy research (pp. 240-257). Washington DC: AERA. Newmann, F. M., & Associates. (1996). Authentic achievement: Restructuring schools for intellectual quality. San Francisco: Jossey-Bass. 153 Newmann, F . M., King, M. B., & Youngs, P. A. (2000). Professional development that addresses school capacity: Lessons from urban elementary schools. American Journal of Education, 108, 259-299. Ouchi, W. G. (1980). Markets, bureaucracies, and clans. Administrative Science Quarterly, 25, 129-141. Peurach, D. (in press). Seeing complexity in public education: Problems, possibilities, and Success for All. New York: Oxford University Press. Porter, A. C. (1994). National standards and school improvement in the 19905: Issues and promise. American Journal of Education, 102, 421-449. Porter, A. C., F loden, R., Freeman, D., Schmidt, W., & Schwille, J. (1988). Content determinants in elementary school mathematics. In D. Grouws & T. Cooney (Eds.), Perspectives on research on effective mathematics teaching. Reston, VA: National Council of Teachers of Mathematics. Pounder, D. G. (1998). Teacher teams: Redesigning teachers’ work for collaboration. In D. G. Pounder (Ed.), Restructuring schools for collaboration: Promises and pitfalls (pp. 65-88). Albany, NY: SUNY Press. Purkey, S. C., & Smith, M. S. (1985). School reform: The district policy implications of the effective schools literature. The Elementary School Journal, 85, 353-3 89. Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community. New York: Simon and Schuster. Putnam, R., & Borko, H. (1997). Teacher learning: Implications of new views of cognition. In B. Biddle, T. L. Good, & I. F. Goodson (Eds.), The international handbook of teachers and teaching (pp. 1223-1296). Dordrecht, Netherlands: Kluwer. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical Linear Models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage. Rockoff, J. (2003). The impact of individual teachers on student achievement: Evidence from panel data. The American Economic Review, 94, 247-252. Rosenholtz, S. J. (1991). Teachers’ workplace: The social organization of schools. New York: Teachers College Press. Rowan, B., Cambum, E., & Barnes, C. (2004). Benefiting from Comprehensive School Reform: A review of research on CSR implementation. In C. Cross (Ed), Putting the pieces together: Lessons from Comprehensive School Reform Research (pp. 1- 154 52). Washington, DC: National Clearinghouse for Comprehensive School Reform. Rowan, B., & Correnti, R. (2009). Interventions to improve instruction: How implementation strategies affect instructional change. In (W. K. Hoy & M. DiPaola (Eds.), Studies in school improvement (pp. 45-74). Charlotte, NC: Information Age Publishing. Rowan, B., Correnti, R., Miller, R. J ., & Cambum, E. (2009). School improvement by design: Lessons fiom a study of comprehensive school reform programs. In G. Sykes, B. Schneider, & D. N. Plank (with T. G. Ford) (Eds.), Handbook of educational policy research (pp. 637-651). Washington DC: AERA. Rowan, B., & Miller, R. J. (2007). Organizational strategies for promoting instructional change: Implementation dynamics in schools working with Comprehensive School Reform providers. American Educational Research Journal, 44, 252-297. Rowan, B., & Miller, R. J. (2009). S11 Multi-component survey data files user ’3 guide. Ann Arbor, MI: Consortium for Policy Research in Education. Scribner, J. P., Hager, D., & Wame, T. (2002). The paradox of professional community: Tales from two high schools. Educational Administration Quarterly, 38(1), 130 160. Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. London: Oxford University Press. Slavin, R. E. (1995). Cooperative learning: Theory, research, and practice. (2nd ed.). Boston: Allyn & Bacon. Slavin, R. E., Madden, N. A. (Eds.). (2001). Success for All: Research and reform in elementary education. Mahwah, NJ: Lawrence Erlbaum. Slavin, R. E., Madden, N. A., & Datnow, A. (2007). Research in, research out: The role of research in the development and scale-up of Success for All. In S. H. Fuhrman, D. K. Cohen, & F. Mosher (Eds.), The state of education policy research (pp. 261-279). Mahwah, NJ: Lawrence Erlbaum. Slavin, R. E., Madden, N. A., Chambers, 8., & Haxby, B. (2009). 2 million children: Success for All. (2nd ed.). Thousand Oaks, CA: Corwin. Smith, L., Ross, 8., McNelis, M., Squires, M., Wasson, R., Maxwell, 8., et al. (1998). The Memphis restructuring initiative: Analysis of activities and outcomes that affect implementation success. Education and Urban Society, 30, 296-325. 155 Smylie, M. A. (1995). New perspectives on teacher leadership. The Elementary School Journal, 96(1), 3-7. Smylie, M. A. (1996). From bureaucratic control to building human capital: The importance of teacher learning in education reform. Educational Researcher, 25, 9-1 1. Smylie, M. A., & Denny, J. W. (1990). Teacher leadership: Tensions and ambiguities in organizational perspective. Educational Administration Quarterly, 26, 235-259. Smylie, M. A., & Evans, A. E. (2006). Social capital and the problem of implementation. In M. I. Honig (Ed.), New directions in policy implementation: Confronting complexity (pp. 187-208). Albany, NY: SUNY Press. Smylie, M. A., & Hart, A. W. (1999). School leadership for teacher Ieaming and change: A human and social capital development perspective. In J. Murphy and K. S. Lewis (Eds.), AERA handbook of research on educational administration (pp. 421-441). San Francisco: Jossey-Bass. Smylie, M. A., Lazarus, V., & Brownlee-Conyers, J. (1996). Instructional outcomes of school-based participative decision making. Educational Evaluation and Policy Analysis, 18(3), 181-198. Spillane, J. P., Halverson, R., & Diamond, J. B. (2001). Investigating school leadership practice: A distributed perspective. Educational Researcher, 23-27. Spillane, J. P., Halverson, R., & Diamond, J. B. (2004). Towards a theory of leadership practice: A distributed perspective. Journal of Curriculum Studies, 36(1), 3-34. Spillane, J. P., & Thompson, C. L. (1997). Reconstructing conceptions of local capacity: The local education agency’s capacity for ambitious instructional reform. Educational Evaluation and Policy Analysis, 19(2), 185-203. St. John, E. P., Meza Jr., J., Allen-Haynes, L., & Davidson, B. M. (1996). Building communities of inquiry: Linking teacher research and school restructuring. In C. Finnan, E. St. John, J. McCarthy, & S. Slovacek (Eds.), Accelerated schools in action: Lessons fiom the field (pp. 124-138). Thousand Oaks, CA: Corwin. Success for All Foundation (2004). Success for All leadership guide. Towson, MD: Success for All Foundation. Supovitz, J. (2002). Developing communities of instructional practice. Teachers College Record, 104(8), 1591-1626. Supovitz, J ., & Poglinco, S. M. (2001). Instructional leadership in a standards-based reform. Philadelphia, PA: Consortium for Policy Research in Education. 156 Tarter, C. J., Bliss, J. R., & Hoy, W. K. (1989). School characteristics and faculty trust in secondary schools. Educational Administration Quarterly, 25 , 294-308. Tarter, C. J ., Sabo, D., & Hoy, W. K. (1995). Middle school climate, faculty trust, and effectiveness. Journal of Research and Development in Education, 29, 41 -49. Tschannen-Moran, M. (2001). Collaboration and the need for trust. Journal of Educational Administration, 39(4), 308-331. Tschannen-Moran, M. (2004). Trust matters: Leadership for successful schools. San Francisco, CA: Jossey-Bass. Tschannen-Moran, M., & Hoy, W. K. (1998). Trust in schools: A conceptual and empirical analysis. Journal of Educational Administration, 36, 334-352. Tschannen-Moran, M., & Hoy, W. K. (2000). A multidisciplinary analysis of the nature, meaning, and measurement of trust. Review of Educational Research, 70, 547 593. Tuckman, B. W. (1965). Developmental sequence in small groups. Psychological Bulletin, 63, 384-399. Uline, C. L., Tschannen-Moran, M., & Perez, L. (2003). Constructive conflict: How controversy can contribute to school improvement. Teachers College Record, 105, 782-816. Wolf, S. A., Borko, H., Elliott, R. L., & Mclver, M. C. (2000). ‘That dog won’t huntl”: Exemplary school change effort within the Kentucky reform. American Educational Research Journal, 3 7(2), 349-93. Wright, B. D., & Masters, G. N. (1982). Rating scale analysis. Chicago: MESA Press. Yasumoto, J. Y., Uekawa, K., & Bidwell, C. E. (2001). The collegial focus and high school students’ achievement. Sociology of Education, 74, 181-209. Zucker, L. G. (1986). The production of trust: Institutional sources of economic structure, 1840-1920. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 8, pp. 55-111). Greenwich, CT: JAI Press. 157 ”11111171 111111311! [1' liljiflllllfillflllllfl