ࡱ> q Nbjbjt+t+ "AAJ:]8$8,&(NNN).Wk $\w))wwNNwhNNw)`N4( @Steps to Produce Raw Score to Theta Estimates for Global Physical Health Items 6/19/2009 Score Tscore SD 4 16.2 4.8 5 19.9 4.7 6 23.5 4.5 7 26.7 4.3 8 29.6 4.2 9 32.4 4.2 10 34.9 4.1 11 37.4 4.1 12 39.8 4.1 13 42.3 4.2 14 44.9 4.3 15 47.7 4.4 16 50.8 4.6 17 54.1 4.7 18 57.7 4.9 19 61.9 5.2 20 67.7 5.9 STEPS: 1) put4.sas (page 6 and 7): SAS program to read main PROMIS wave 1 dataset (n=21,133) and create ASCII file of 4 physical health items (GLOBAL03, GLOBAL06, GLOBAL07R, and GLOBAL08R) for subsequent steps (phyglob4.dat, n=21,133); double check min/max on item frequencies for next step. Note that GLOBAL07R and GLOBAL08r are reverse scored to be in the same direction as GLOBAL03 and GLOBAL06 (better physical health). GLOBAL07R also has some of its categories collapsed. 2) phyglob4.mlg (page 8): Run MULTILOG GRM on all 21,133 cases. Parameters are saved to the file phyglob4.par: 2.31116 -2.11448 -0.89035 0.28769 1.54358 2.99151 -2.79626 -1.77578 -1.04365 -0.39605 1.74443 -3.87320 -1.80644 -0.67460 0.99560 1.90220 -3.23917 -1.87531 -0.35804 1.16723 -1.00000 0.00000 1.00000 First column refers to the slope, 2nd column refers to b(1), 3rd to b(2), 4th to b(3), and 5th to b(4); first record refers to GLOBAL03, 2nd to GLOBAL06, 3rd to GLOBAL07r, and 4th to GLOBAL08r. 3) scorephy4.mlg (page 9): Score the 21,133 records from step 1 (phyglob4.dat) using saved parameters from step 2 (phyglob4.par). Output is scorephy4.out. Here are the first few lines: THETAHAT S.E. ITER ID FIELD 0.831 0.466 2 84 1.744 0.537 6 85 0.492 0.436 3 86 0.850 0.499 2 87 -1.195 0.348 6 88 -2.854 0.426 6 89 ... Save the output from scorephy4.out to a new file, scorephy4.out.dat, editing it down so it includes only the actual data (no header information). 4) ph4merge4.sas (page 10 and 11): Merge scored data from step 3 (scorephy4.out.dat n=21,133), with Honghus file (n=13,250) indicating the cases in the general population that match the 2000 census, keeping only the 5,239 cases in norming sample (flag_ns=1). Merge in the main PROMIS file to be able to check for missing data and drop cases that do not have at least one non-missing item (requirement of Seungs R program used in next step). There were 3 cases that did not have at least one non-missing item. Save items to a CSV file after converting missing values to blanks (Seungs R-program requires blanks to be used rather than SAS .s). input=scorephy4.out.dat, scalingsample2, promis_wave1_final. output=ph4data.csv (n=5,236) 5) Run Seung Chois R transformLatentMeanSD3.r program to transform parameters. Program is located on the PROMIS website (details on program and obtaining it are on page 12 in appendix). input=ph4data.csv from step 5, and phyglob4PAR.csv (a formatted version of parameters output from step 2 [note: take out imbedded blanks]) phyglob4PAR.csv: 2.31116,-2.11448,-0.89035,0.28769,1.54358, 2.99151,-2.79626,-1.77578,-1.04365,-0.39605, 1.74443,-3.87320,-1.80644,-0.67460,0.99560, 1.90220,-3.23917,-1.87531,-0.35804,1.16723, output=tpar_SingleGroup.csv (will contain the transformed item parameters): 2.22569030443809 -2.30679808650395 -1.03565974117415 0.187618682140617 1.49173665696048 2.8808800786746 -3.01475941502212 -1.95509152138162 -1.19484668074680 -0.522377913184874 1.67991871517807 -4.13305545353067 -1.98692890929614 -0.81162463215395 0.922713440084965 1.83185417586932 -3.47467779340515 -2.05844361593912 -0.48290827474289 1.10093427791179 (Note: actual output contains commas between values taken out here for readability) From the graph produced (page 13), it shows mu to be .10701 and sigma to be .96302. Hence: The multiplicative constant = k1 = 1/sigma = 1/.96302 = 1.0384 and the additive constant = k2 = -k1*mu = -1.0384*.10701 = -.11112 and t_slope=slope/k1 t_b1=(b1*k1)/k2 t_b2=(b2*k1)/k2 t_b3=(b3*k1)/k2 t_b4=(b4*k1)/k2 As a check (results from SAS code is chkph4convert.sas on page 15): slope b1 b2 b3 b4 t_slope t_b1 t_b2 t_b3 t_b4 2.31116 -2.11448 -0.89035 0.28769 1.54358 2.22569 -2.30680 -1.03566 0.18762 1.49173 2.99151 -2.79626 -1.77578 -1.04365 -0.39605 2.88088 -3.01476 -1.95509 -1.19485 -0.52238 1.74443 -3.87320 -1.80644 -0.67460 0.99560 1.67992 -4.13305 -1.98693 -0.81162 0.92271 1.90220 -3.23917 -1.87531 -0.35804 1.16723 1.83186 -3.47467 -2.05844 -0.48291 1.10093 6) Run David Thissens IRTScore program (http://www.unc.edu/~dthissen/dl.html). data input=tpar_SingleGroup.csv from step 6, but take out the CSV part, save to tpar.dat: 2.22569030443809 -2.30679808650395 -1.03565974117415 0.187618682140617 1.49173665696048 2.8808800786746 -3.01475941502212 -1.95509152138162 -1.19484668074680 -0.522377913184874 1.67991871517807 -4.13305545353067 -1.98692890929614 -0.81162463215395 0.922713440084965 1.83185417586932 -3.47467779340515 -2.05844361593912 -0.48290827474289 1.10093427791179 Parameters are input at screen (saved to file ph.out in this case page 16). Relevant output is: Score EAP SD Prop. 0 -3.38 0.48 0.00011 1 -3.01 0.47 0.00067 2 -2.65 0.45 0.00217 3 -2.33 0.43 0.00530 4 -2.04 0.42 0.01032 5 -1.76 0.42 0.01775 6 -1.51 0.41 0.02802 7 -1.26 0.41 0.04082 8 -1.02 0.41 0.05576 9 -0.77 0.42 0.07326 10 -0.51 0.43 0.09444 11 -0.23 0.44 0.12008 12 0.08 0.46 0.14435 13 0.41 0.47 0.14909 14 0.77 0.49 0.12414 15 1.19 0.52 0.08496 16 1.77 0.59 0.04875 Then, 4 was added to the Score because the actual lowest possible score is 4 (not 0). Then, transform the EAP estimates to T-scores: sd=10*sd; tscore=50+(eap*10); Score Tscore SD 4 16.2 4.8 5 19.9 4.7 6 23.5 4.5 7 26.7 4.3 8 29.6 4.2 9 32.4 4.2 10 34.9 4.1 11 37.4 4.1 12 39.8 4.1 13 42.3 4.2 14 44.9 4.3 15 47.7 4.4 16 50.8 4.6 17 54.1 4.7 18 57.7 4.9 19 61.9 5.2 20 67.7 5.9 Code is phconvert.sas and is on page 17. APPENDIX: 1. put4.sas: (creates global mental health extract as well) options ls=130 ps=52 nocenter; libname t "c:\projects\PROMIS\FinalWave1Data\Data\Combo052207"; *************************************************; proc format; value poorex 1='1: poor' 2='2: fair' 3='3: good' 4='4: very good' 5='5: excellent'; value notcom 1='1: not at all' 2='2: a little' 3='3: moderately' 4='4: mostly' 5='5: completely'; value zeroten 0='0: no pain' 10='10: worst pain imaginable'; value nonevry 1='1: none' 2='2: mild' 3='3: moderate' 4='4: severe' 5='5: very severe'; value nevalw 1='1: never' 2='2: rarely' 3='3: sometimes' 4='4: often' 5='5: always'; run; *----------------------------------------------------; /* format global01 poorex. global02 poorex. global03 poorex. global04 poorex. global05 poorex. global06 notcom. global07 zeroten. global08 nonevry. global09 poorex. global10 nevalw.; */ DATA temp; SET t.promis_wave1_final(keep=caseid global01 global02 global03 global04 global05 global06 global07 global08 global09 global10); array missx global01 global02 global03 global04 global05 global06 global07 global08 global09 global10; do over missx; if missx<0 then missx=.; end; global10r=global10; if global10=1 then global10r=5; else if global10=2 then global10r=4; else if global10=3 then global10r=3; else if global10=4 then global10r=2; else if global10=5 then global10r=1; global08r=global08; if global08=1 then global08r=5; else if global08=2 then global08r=4; else if global08=3 then global08r=3; else if global08=4 then global08r=2; else if global08=5 then global08r=1; global07r=global07; if global07=10 then global07r=1; else if global07=7 or global07=8 or global07=9 then global07r=2; else if global07=4 or global07=5 or global07=6 then global07r=3; else if global07=1 or global07=2 or global07=3 then global07r=4; else if global07=0 then global07r=5; label global10r="global10(reversed)" global08r="global08(reversed)" global07r="global07(recoded/reversed)"; ***************************************************; run; proc sort; by caseid; run; ******************************************; *--------------------------------------------------; TITLE "phyglob4 items"; run; proc freq data=temp; tables global03 global06 global07r global08r; run; data _null_; file 'phyglob4.dat'; set temp; put @1 caseid @10 (global03 global06 global07r global08r) (1.); run; *--------------------------------------------------; TITLE "menglob4 items"; run; proc freq data=temp; tables global02 global04 global05 global10r; run; data _null_; file 'menglob4.dat'; set temp; put @1 caseid @10 (global02 global04 global05 global10r) (1.); run; *--------------------------------------------------; 2. phyglob4.mlg: MULTILOG for Windows 7.00.2327.2 Created on: 2 July 2008, 8:39:04 >PROBLEM RANDOM, INDIVIDUAL, DATA = 'C:\Projects\PROMIS\FinalWave1Data\Data\Combo052207\GHSFtool\PH\phyglob4.dat', NITEMS = 4, NGROUPS = 1, NEXAMINEES = 21133, NCHARS = 7; >TEST ALL, GRADED, NC = (5,5,5,5); >EST NCYC=150; >SAVE; >END ; 5 12345 1111 2222 3333 4444 5555 (7a1,2x,4A1) 3. scorephy4.mlg: MULTILOG for Windows 7.00.2327.2 Created on: 2 July 2008, 08:56:51 >PROBLEM SC, INDIVIDUAL, DATA = 'C:\Projects\PROMIS\FinalWave1Data\Data\Combo052207\GHSFtool\PH\phyglob4.dat', NITEMS = 4, NGROUPS = 1, NEXAMINEES = 21133, NCHARS = 7; >TEST ALL, GRADED, NC = (5,5,5,5); >EST NCYC=200; >START ALL PARMS= 'C:\Projects\PROMIS\FinalWave1Data\Data\Combo052207\GHSFtool\PH\phyglob4.par'; >END ; 5 12345 1111 2222 3333 4444 5555 (7a1,2x,4A1) 4. phymerge4.sas: options ls=130 ps=54 nocenter; TITLE1 "Physical global health-4"; run; *--------------------------------------------; libname n 'C:\projects\PROMIS\FinalWave1Data\Data\USNorm'; data normid; set n.scalingsample2; if flag_ns7=1; run; *--------------------------------------------; data calibrate; infile 'scorephy4.out.dat'; input thetahat se iter caseid; run; *--------------------------------------------; data temp_cal; merge calibrate(in=in1) normid(in=in2); by caseid; if in1 & in2; run; *--------------------------------------------; libname t "c:\projects\PROMIS\FinalWave1Data\Data\Combo052207"; DATA temp; SET t.promis_wave1_final(keep=caseid global01 global02 global03 global04 global05 global06 global07 global08 global09 global10); array missx global01 global02 global03 global04 global05 global06 global07 global08 global09 global10; do over missx; if missx<0 then missx=.; end; global10r=global10; if global10=1 then global10r=5; else if global10=2 then global10r=4; else if global10=3 then global10r=3; else if global10=4 then global10r=2; else if global10=5 then global10r=1; global08r=global08; if global08=1 then global08r=5; else if global08=2 then global08r=4; else if global08=3 then global08r=3; else if global08=4 then global08r=2; else if global08=5 then global08r=1; global07r=global07; if global07=10 then global07r=1; else if global07=7 or global07=8 or global07=9 then global07r=2; else if global07=4 or global07=5 or global07=6 then global07r=3; else if global07=1 or global07=2 or global07=3 then global07r=4; else if global07=0 then global07r=5; label global10r="Global10(reversed)" global08r="Global08(reversed)" global07r="Global07(recoded/reversed)"; nmiss4=nmiss(of global03 global06 global07r global08r); keep caseid global03 global06 global07r global08r nmiss4; ***************************************************; run; proc sort; by caseid; run; ******************************************; data temp_cal; merge temp_cal(in=in1) temp(in=in2); by caseid; if in1 & in2; run; proc freq data=temp_cal; tables nmiss4; run; data temp_cal; length g3 g6 g7 g8 $1.; set temp_cal; if nmiss4 ne 4; * file cannot contain .'s for missing values - change to blanks; array conver global03 global06 global07r global08r; array out g3 g6 g7 g8; do over conver; out=put(conver,$1.); if out="." then out=" "; end; run; data _null_; file 'ph4data.csv'; set temp_cal; put @1 g3 $1. ',' g6 $1. ',' g7 $1. ',' g8 $1. ; run; 5. Seung Chois scale transformation program transformLatentMeanSD3.r Program is located on the PROMIS website: CORE Projects > PROMIS > SCC Analysis > Software  HYPERLINK "https://www.coreprojects.org/PROMIS/analysis/default.aspx?RootFolder=%2fPROMIS%2fanalysis%2fPublic%20Documents%2fPsychometrics%20and%20Statistics%2fSoftware&FolderCTID=0x012000301B0BA5B921E949B4B96AC6C330D3E8&View=%7bE086E309%2d8A59%2d4B3F%2dB0AB%2d341CD00" https://www.coreprojects.org/PROMIS/analysis/default.aspx?RootFolder=%2fPROMIS%2fanalysis%2fPublic%20Documents%2fPsychometrics%20and%20Statistics%2fSoftware&FolderCTID=0x012000301B0BA5B921E949B4B96AC6C330D3E8&View=%7bE086E309%2d8A59%2d4B3F%2dB0AB%2d341CD00  Some notes from Seung Choi about ZIP file transformLatentMeanSD3.ZIP containing his program and example: Enclosed in the attached zip file you should find the scale transformation program (transformLatentMeanSD3.r), a sample item parameter file (par_SingleGroup.csv) and a sample data file (dat_ScalingSample.csv). I estimated the mean and SD of the latent distributions using a variation of the method introduced by Mislevey (1984: Estimating Latent Distributions, Psychometrika, 49, 359-381). The method I used was taken from the IRT book by Baker & Kim (2004) on pages 269-273. It treats the item parameters as known and estimates the population mean and SD of a given sample, assuming that the latent distribution is normal. Basically, it aggregates the individual likelihood functions for a sample of examinees, takes the expectation over the normal distribution, and iteratively refines the estimates. Here are the steps I took (and the program uses) to transform the item parameters to center the scale based on the scaling sample: Obtain item parameter estimates from the single-group calibration using the calibration sample (~14,900 cases for ED domains). Extract the scaling sample subset identified by Honghu (~3,000 for ED domains) from the calibration sample. Estimate the latent mean and SD of the scaling sample using the method described above. Based on the estimated mean and SD from Step 3, derive a linear transformation equation to set the mean and SD to 0.0 and 1.0, respectively, and apply the equation to the item parameter estimates from Step 1 Repeat Steps (2) and (3) based on the transformed item parameters to confirm that the mean and SD are actually 0.0 and 1.0. I have added some directions to run in the header part of the program. It took around 20 seconds on my computer. The example invocation lines at the bottom of the program are hard coded for my domains, so there are a couple of lines you will need to modify if your items have more than 5 categories. 5. continued  5. continued  5. continued chkph4convert.sas: program to take mu and sigma and check transformations: data temp; k1= 1/.96302; k2 =-(k1)*.10701; ; run; data one; input slope b1 b2 b3 b4; cards; 2.31116 -2.11448 -0.89035 0.28769 1.54358 2.99151 -2.79626 -1.77578 -1.04365 -0.39605 1.74443 -3.87320 -1.80644 -0.67460 0.99560 1.90220 -3.23917 -1.87531 -0.35804 1.16723 ; run; data one; set one; if _n_=1 then set temp; t_slope=(slope/k1); t_b1=(b1*k1)+k2; t_b2=(b2*k1)+k2; t_b3=(b3*k1)+k2; t_b4=(b4*k1)+k2; run; proc print noobs; var slope b1 b2 b3 b4 t_slope t_b1 t_b2 t_b3 t_b4; run; 6. ph.out: Test has 4 items. Using parameters from tpar.dat: Item Ncat Parameters: 1 5 2.23 -2.31 -1.04 0.19 1.49 2 5 2.88 -3.01 -1.96 -1.19 -0.52 3 5 1.68 -4.13 -1.99 -0.81 0.92 4 5 1.83 -3.47 -2.06 -0.48 1.10 Category EAPs and Wts: Category Item 0 1 2 3 4 1 -1.627 -0.959 -0.251 0.489 1.283 1.640 2.191 2.437 2.355 1.803 2 -2.259 -1.652 -1.109 -0.617 0.458 1.817 2.519 3.109 3.463 1.634 3 -1.613 -1.194 -0.667 0.025 0.868 1.092 1.452 1.816 1.828 1.516 4 -1.674 -1.277 -0.620 0.153 0.985 1.199 1.569 1.893 1.973 1.593 Score EAP SD Prop. 0 -3.38 0.48 0.00011 1 -3.01 0.47 0.00067 2 -2.65 0.45 0.00217 3 -2.33 0.43 0.00530 4 -2.04 0.42 0.01032 5 -1.76 0.42 0.01775 6 -1.51 0.41 0.02802 7 -1.26 0.41 0.04082 8 -1.02 0.41 0.05576 9 -0.77 0.42 0.07326 10 -0.51 0.43 0.09444 11 -0.23 0.44 0.12008 12 0.08 0.46 0.14435 13 0.41 0.47 0.14909 14 0.77 0.49 0.12414 15 1.19 0.52 0.08496 16 1.77 0.59 0.04875 6. continued phconvert.sas: program to convert EAP and SDs to T-score and SE: data temp; input Score EAP SD Prop; cards; 0 -2.89 0.46 0.00179 1 -2.49 0.41 0.00442 2 -2.16 0.39 0.00874 3 -1.87 0.37 0.01566 4 -1.62 0.37 0.02460 5 -1.37 0.37 0.03544 6 -1.12 0.36 0.04873 7 -0.89 0.36 0.06437 8 -0.65 0.36 0.07978 9 -0.42 0.36 0.09148 10 -0.17 0.37 0.10067 11 0.08 0.37 0.10748 12 0.33 0.37 0.10740 13 0.60 0.38 0.09580 14 0.90 0.39 0.08129 15 1.25 0.42 0.06966 16 1.76 0.53 0.06268 ; run; data temp; set temp; score=score+4; sd=10*sd; tscore=50+(eap*10); run; proc print noobs; var score tscore sd; run;     Page  PAGE 7 of  NUMPAGES 1 l7788899999C:D:;;;;b<l<o=p=q=r=====u>v>>>:?;? @ @@@@@@@AAAAAANNNNNNNNNù jUjiCJOJQJUj CJOJQJU>*B* CJOJQJB* CJOJQJ jU0JCJOJQJjCJOJQJUjCJOJQJU CJOJQJ CJOJQJ=YZ[\]v %>Wp89:;<=YZ[\]v %>Wp89:;<=EFGe!"= K p r . / U V }  ; ? @ wx@ms5@AH$%k8e=EFGe!"= K p r . / U V }  ; ? @ @ wx@ms5@AH0H$%k8j$%A08j$%Afg6Pj :;<0Ib{*C\u01P  $ % 2 D T f t !!!,!9!G!X!e!t!y!!!!!efg6Pj :;<0I0Ib{*C\u01P  $ % % 2 D T f t !!!,!9!G!X!e!t!y!!!!!!!!!" "2"D"U"g"j"k"v""""F#c#r#######$:$[$\$p$$$$%%%&%:%`%%%#&D&E&L&k&&&&&&'4'5'j''''''(<(A(v((((())F)K))))))))*V*l********++++#+(+-+2+7+F+Y+Z+{++e!!!!" "2"D"U"g"j"k"v""""F#c#r#######$:$[$\$p$p$$$$%%%&%:%`%%%#&D&E&L&k&&&&&&'4'5'j'''''''(<(A(v((((())F)K))))))))*V*l*********++++#+(+-+2+7+F+Y+Z+{+++++*,@,W,u,,,,,,,?-F-++++*,@,W,u,,,,,,,?-F-H-N-S-X-]-b-g-u------:.^.m.r..... /b/p/u////q0r00001 1 11C1h111111 222W2|222223Z3333334*4b444445\5j5o55555555%6Y6p6666666eF-H-N-S-X-]-b-g-u------:.^.m.r..... /b/p/u////q0q0r00001 1 11C1h111111 222W2|222223Z33333334*4b444445\5j5o55555555%6Y6p666666666677e7f77779999C:D:;;p=r===v>>;? @@@@@AAAAAAAAA5BhBBBC4CgCCCD3DfDDDD2EeEEEE1FdFFFFFGGBGiGGGGGGGH.HVH~HHHHIFIGI`IzIIIIII             Z677e7f77779999C:D:;;p=r===v>>;? @@@@@A & F$AAAAAAAAA5BhBBBC4CgCCCD3DfDDDD2EeEEEE1F1FdFFFFFGGBGiGGGGGGGH.HVH~HHHHIFIGI`IzIIIIIIIJ0JJJdJ~JJJJJKK(K)KkKlKKKKKKL"LwxM,d,-B<,4Rw%Q߰4az.yv#\'RL&X_Ye} LѨ%B4\v>Y:?afwK5پ6R.#qC˸@ry0@aKq=ė a|CT4±pI^(k0E={\(}&{$a k_7XJڳkWAX{ڷ\d1kߧڿ` j/c jW(,k?j^ bTj;hv jgV{*7kHA폰 j?}CY{GHfڧ*}kWY{4/f j_W+=u=7+}+kߡX"^ڏ+$k?3[+7*"k`Y3Fjy=5&g\cESW?߹諱rYh_mJ_}}p_5C_5WW U+ \S謯hp_B_W۸v(WyxtWZyz׺NnrsBs;1yw~.)KhWSyxQϻo̓OYgs:,nT́.=́&@97n'9@L0:S-s $r=́9p[8Or瀯1:6yV'~)x[Y)*cU2Ur+}}p_YUUw*_*_8﫿_}"W[+' wi׻}'{r?4[KkIٚ]ZX8Hmڀ.t9%YQkZ?$KW]mWIm:KuaWh m{}KS`Sv1^{#|nstuv?=]M= Yu~j %\se'~ iutt1ΥzhׇǺy7S̹ܫ!sd&)!nhxY_zs.k{/_(hhz\D4;[:#1k2-)W}b 2nݽm>\>Gػ2gK:,˳rg$!ZYMԾPNussb{97Oj>gcg.rTG7>K.D;زWۮc}xʭ|XKuؐv8ccGLrµmDgkT)=_`cۛ.r뱏cˢ~m;r4بp{D=6=vǺ2jq#7 ;Vj$njGr=MG =a-1Vw=Q6}b}wdLFu="tQo"y!itO2{M?z'FDdky$ԉmu5k645W~?:awcUc5lGk(F<'(ߦɳzaǯ?/}yv77ؿ}/m_Sa_w0ϯki xoAlA^!>  @q3?$cB<B/A<)'@ A,Cq.Ļ!!K!.sC| b:` us{xuwS8U-RLL֞9|uU;_ck_7Y{a*}kQP֞}?fԾTPTAe\A_ e^~_RPO*PM;ͯK}ڭ jTP{oa8kJAY0pb j+=OUP4>CAci j_ԾVP{&k[AX{ joVPV־CA= j/`E j/f+WPI~JAgX jWP{ koTPE~YAWYq݄H;C_s_cۍ <xY>"K_|5LkF9|']^a%?i^ x#&_ifxoA<xAxp#>x4q|_< ુg_|-7v;|/"/~ O~xj7 *pǿ][-'`9c:9; 8'8cO> x9_| t+ | <ہ x>? 1'? YUo2i`> >=/|瀇<hG<,I|%Ӂx&5_|#[o./^ 0ǀ^4 gW^ x+plt9hj6=Xpݴ\R8Dd*!!<  C A"yRTOl9>4 D@=|yRTOl9>4 65BRRJx pT]I HV~ (h "EF3 4`M lq`JlPRQj(%#(%LB)UBϷ ?2>=Mk(^K;ܡ6֋y[RK_)e+ub>YR4H^Sϫ$f4맗0nAu||O3'J_|:95kְ֯np#c$RvM =Cz9xW*Gv=y̖f@S;T&.hި4o󎖌4 N^7s#F/ona88(r<3:23S]mPc) i]w]Hc=#1,띛y>$Փ^92.˭zL1ܵvG:z+=9ꗡb5yW8_ϱ}Q%U72ts@g^z%WA :t!ݠK@K߀> yЍ; @Gz0葠ǀ~t,ɠSAg^z@.A.]4:@Czo'}K_63n˶v3e}i JD,,>Wt) }@/Usz)F'^ vS?b ){/0{/k^! ~Z3V@?wj@v{@a.{7K콏@za>J(>V'{@d3Oc3zz ޗ =&z@owzo}KbW@s޿`_ ~'Jj 9~QKz7a֍ k]]1V9ǘpI }OX_XSB**XǪ@X5X5zyCx)4V XM9vڞ4Xdk`X͊ U)41rG̪u|r>}z"S{Qa>svYTQYSlfqYc j.g>Q#s jӥ6ٰt?lM䜝9CR|,6؉fn.cOu6,:T7X*Y-c>Q:i|gu\ضGo󙱿iIƆ5D{{f36-϶8cv?3vh؜}fllK23#c>QFujӥ]8vZ cZƞ Zwگ>c>9NmZkiU>3`?g,ӟRSKmmpόܒu^{ƚ} ԦNq@ƶ 3cnIF5D=נ6o}ggl;CcoSYm; =+5>jiFC۵ψnm[Fęts͈suMn3hyʱBxl>QԦ5#{-#:n|(B7Q7Р5q 3Lѥjv6X%;5T[LM>͏pn;#Ǥ/,A{zu0N>5g@^-P+ k?Z2Mfzڴf a|!g@j z9_dˀH[~8yUj# Ht3OUY,5k2*kInwx:w1mlk3j%9Tz\|F6MɛPb~> дxomYa.]w-Aޓz 3z/%JW{@ﯳ7zz zw #Gc?eef+z?O ~ ~_N HN%!w@{/{>>D(ޣXޟ`O`zg>}{O}>{_${{_. Zl@[=M.ދ^K'z%q@߳*ޫ{@E/ wy̕yI4?n>orF$O |6Ӂ/ˁx6|<[oEw ?1O?  ppA+xwཀ  |(>XG|xS'O>\_ | W_ <Zo x-|"ແ^R^ '~x _~ a?6']\h?iP\W'>xdು|ֹǭs+:o| 88~x)C/~ xO?x|u`嗀puCwnx/ཁG |G>x'<TI^`;lӁ/ˁx6l'|+w/^%/~/C<:wp^y_n`3p5@Ŋۦdz [8@8 NormalCJ_HaJmH sH tH <A@< Default Paragraph Font@@  Balloon TextCJOJQJ^JaJ8O8 Char CharCJOJQJ^JaJ(O( eudoraheader,@", Header  !, @2, Footer  !.U@A.  Hyperlink >*B*phJ 579<NN3N=@ HI% !p$'*F-q036A1FILNN4678:;<>?@ACDEGHIJLM8!+6IN59=BFK345JX %02<!TT8@0(  B S  ? OLE_LINK1 OLE_LINK2ffJ  JflJRdg%1AIJMu9 A k r  " LZ179?@EFNPW+1%+ )0;BMS^dkpuz"#$$ %%)))))))*?*E*********%+-+G+M+Z+`+++ ,,x,},,,,,,-*060g0m000"1*121:1T1Z1y111111+212x2~222"3'35566777777::@A>@UUqru2CFNDDcc   8 8 T [ g h x ~ . > ]p  FFII 1Kghns9Cku%)i|MNRRSSgq",<FV`pz"*BBDGdd18PW%*$tw (2:DL^fcert  : < p r !!:!*^*`*m*p******* +$+b+d+p+s+++, ,r,w,,,,,--- -C-E-h-j------- ..2.4.W.Y.|.~.....//Z/\///////// 001070b0f000001!1\1^1j1m1o1s11111111111%2*2Y2^2p2r2222222222222233 3 3333m<p<======5>9>??4?8?g?l?@@@@AA6A8AiAkAAAAAAB1B5BdBgBBBBB"C)CICPCpCwCCCD&D4D@DqD}DDDDDDD EE$E0EGEQEfEpEEEEEEEEEEEF FF&F6F@FPFZFjFtFFFFFFFFFFFG GG'GlGpGGGGGGGGGGGHH(H2HBHLH\HfHvHHHHHHHHHHHHII(I0IBIJI\IdIvI~IIIIIIIIIJJ+J.JEJIJ_JbJzJ}JJJJJJJJJJJJJJJJJJJJ3333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333 KSpritzer5C:\Projects\FrontPageWebs\HAYS\ph4-summary-061909.docSg, ^`hH() ^`hH. pp^p`hH. @ @ ^@ `hH. ^`hH. ^`hH. ^`hH. ^`hH. PP^P`hH.SJ 0UWt0I@JJ@@GzTimes New Roman5Symbol3& zArial?5 zCourier New5& zTahoma"1h֦o֦!֦ 6 u=] @!!0dyK#LY1) put4 KSpritzer KSpritzerOh+'0  8 D P \hpx1) put4) p KSpritzer SprSpr Normal.dot KSpritzer12rMicrosoft Word 8.0@0@%:@t@ u=՜.+,D՜.+,8 hp  UCLA/yK1 1) put4 Title(RZ _PID_GUID _PID_HLINKSAN{94624692-2520-4F5B-8524-D6BA88D66E2E}A<wvhttps://www.coreprojects.org/PROMIS/analysis/default.aspx?RootFolder=%2fPROMIS%2fanalysis%2fPublic%20Documents%2fPsychometrics%20and%20Statistics%2fSoftware&FolderCTID=0x012000301B0BA5B921E949B4B96AC6C330D3E8&View=%7bE086E309%2d8A59%2d4B3F%2dB0AB%2d341CD00  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOQRSTUVWXYZ[\]^_`bcdefghijklmnopqstuvwxy{|}~Root Entry F06 4(Data P!1TableaH WordDocument"SummaryInformation(rDocumentSummaryInformation8zCompObjjObjectPool 4( 4(  FMicrosoft Word Document MSWordDocWord.Document.89q