Document Lo8RBvObvonwKk41m2YDZO5Yg

rtOHMfc, UNITED STATES ENVIRONMENTAL PROTECTION AGENCY ZHAO 20?. C. &" nCT 2 1985 Dr, David Weil Environmental Criteria and Assessment Office U.S. EPA Research Triangle Park, N.C. Dear David: I am Enclosing ray latest memorandum in response to Richard Royall's questions about the consistency of the blood lead blood pressure relationship in the NHANES II data. In addition to these results, which show that lead is significant as a preditor of blood pressure even when dummy variables for all of the PSUs are used, and that the coefficient is less affected when alternative dummy variables that represent location with less collinearity with lead are used instead of PSU, ny August 21st memo used a proportionate hazard model with and without blocking. The blocking case calculates the regression separately in each of the 64 PSUs and then averages the results, computing the significance of that lead coefficient using a Mantel Haenzsel approach to summing the log liklihoods. It showed that lead was significant as a predictor of blood pressure with or without the blocking, and that the result using the 64 PSUs gave a coefficient only 20% lower than the coefficient obtained treating the sample as a whole. If Dr. Royall does not have this memo I would appreciate your sending it to him. Sincerely, 0 Joel Schwartz TEH 0412421 OCT 2 1985 Subject: Response to Richard Royall'a questions on the blood lead -blood pressure relationship in NHANES IX From: Joel Schwartz Office of Policy Analyaia To: David Weil r Environmental Criteria and Assessment Office This memo is in response to Dr. Royall'a comments to you regarding the NHANES II analysis. Dr. Royal1 basic question was whether the blood lead-blood pressure relationship in NHANES II showed internal consistency, and he proposed a number of different models and wished the analysis to be discussed with resgect to those models. I believe that I did use his Second'model, as well as his first in my previous memo, but I will start from scratch to avoid confusion. Dr. Royall has phrased his questions formally, in terms of a series of models, and 1 will respond in a similar fashion. My first response is that I do not believe that any of those model-are exactly the correct one, since they omit an important factor that I discussed at some length in my previous memo, and which has considerable bearing on the outcome. This factor is measurement error. To deal with it more formally than I did previously consider the following model, which is similar to the first model Dr. Royall discussed, except for incorporating the effect of measurement error. I will first discuss the impact of measurement error in the framework of this model and them go on to the more complex models he raised. MODEL TYPE 1 Let Y represent the outcome, X the independent variable of interest, 2 the covariatea. Let X = x u, where x ia the true TEH 0412422 DUP050453474 N 30991.01 value of blood lead, and u the measurement error. Kuaeroua authors have shown that in this case the regression coefficient is biased downward, aa ia ita t-statiatic <i). Thus the aodel is E(Y ) * A BCX > ii C(Z > i <1> and the regression coefficient B ia biased, with, in the simple bivariate case B = b(l - Var(u) > <21 Var<x)+Var<u> where b is the coefficient that would result if x rather than X were used in model 1. Moreover, if there ia also measurement error in the dependent variable, then the error term variance is also inflated. Since Var(B) is proportional to Var(e), this measurement error will not bias the regression coefficient, but will bias the h-ststiatic. The implications of this are that measurement error in Y and X leads to bias in B and its significance level in model 1. The multivariate equations are more complicated in form, but not in substance than <2>, however, errors in measuring the Z's can produce both positive and negative effects on B. Fortunately, th most important covariates, body mass index and age, are very accurately determined, and the impacts of the other variables can be determined empirically. When we looked at the coefficient of lead using only age, race, and body mass index as variables, and compared it to the results using the selected nutritional and other variables signi ficant at the 0.05 level, or the ones significant at the 0.15 level, or the intermediate results, the coefficient of blood lead changed by less than 10k among all of the regressions (1st seated systolic,. 1st seated diastolic, 2nd seated systolic, 2nd seated diastolic, recumbent systolic, recumbent diastolic), indicating TEH 0412423 DUP050453475 little aenaitivity to those variables Part of the bias due to the measurement error can be quanti fied, because the 1100 blind quality control samples allow the computation o Var(u). Supercarp is a program that will do regression analysis on survey data accounting for the stratified nature of the data, and will also take into account the measure ment error if the variance ia known. In my August 21 memo to David Weil, I enclosed the results of using that program, and the variance of the 0C data. As equation (2> suggests, the coeffi cient of blood lead increases by about 20k after accounting for this error, and the t-statistic also increases. The results are shown in the appendix in the back. This only accounts for part of the measurement error bias since blood lead is an imprecise measure of the toxicologically active soft tissue lead, however I have no good estimates for this factor. Supercarp will not allow estimates of the effect of measurement error in the dependent variable, however. To indicate the magnitude of this effect, however, I regressed the first seated blood^prassure measurement on the second seated measurement. The R of this regression is only 0.6432, indicating that there is a significant further downward bias in the t-statistic that needs to be at least recognized. This low value of explained variance somewhat overstates the noisyness of blood pressure measurements because the first measurement was taken upon first entering the medical examination center, and seeing all the white coats and equipment. Medical experience suggests that this would result in a slight increase in both the mean blood pressure and its 2 variance, which would reduce the R of the regression. Neverthe less, the results show that between the actual' measurement error in blood pressure, and the real fluctuation in blood pressure form hour to hour, (which is clearly measurement error in this context) blood pressure is a quite noisy variable. 1 am going through this somewhat lengthy discussion because TEH 0412424 DUP050453476 of the simple fact that bias in the regression coefficient and significance level in this first stage of the analysis has important implications for the more complicated models. In particular, introducing variables col linear with blood lead wil_l_ Increase the bias In the regression. Or. Royall has already indicated that our search of many alternative variables and model specifications within the context of model <1) shows considerable internal consistency within this specification. Additional examination of internal consistency within this model was performed by looking at age subgroups, as described in my memo to CASAC, by using principle component analysis on the Z variables, and repeating the stepwise analysis described in our paper, by excluding persons on hypertensive medication, etc. I believe that the consistency within this model is well established, and next turn to Dr. Royall'a second model. MODEL TYPE 2 ` The second set of models to be considered are : g :y ) * A * BX * CZ 13 1 iJ iJ where X * x u (3) and again 8 is biased downward. In this scenario, the A^ represent groups formed in various ways to control for factors by class variables rather than continuous variables. These class variables can be of two types, either they themselves represent a factor known or suspected of being related to blood pressure that is better modeled categorically than continuously , or they may be proxies for some known or unknown unmeasured factors. I have also used this model to examine the internal consistency of the evidence, using several different classes. I have described in previous memorandum the results of using TEH 0412425 DUP050453477 age aa a classification variable. Inserting dummy variables for age subgroups has no impact on the lead coefficient. Inserting dummy variables for the time period in the survey has no impact either. The same results obtained for seasonal dummies. Dupont has stated that the results are different when the classification variable represents location. The use of location n aa a classification variable creates some difficulty, however, since location is collinear with lead. This will, particularly in the context of e model with substantial measurement error < the instrument measurement error variance accounts for l&a of the variance of lead, and the inexactitude of blood lead accounting for toxicologically active soft tissue lead probably accounts for a similar fraction of the variance of lead} create a substantial increase in the bias of B. There is one further problem. The NHANES II survey was stratified by picking several health outcomes, and , using NHANES I data, finding within each region the demographic factors that best predicted that outcome. All candidate PSU's were then classified into superstrabd^based oif"si-milarities in their predic- ted mean levels of tho^'e medical outcomes. The medical outcomes chosen were prevalence bf^yperte'SSXdnr'Tnfant mortality rate, prevalence of high serum cholesterol levels, prevalence of heart disease, and prevalence of cancer. These are discussed in the Plan And Design of the NHANES II Survey. Thus the PSU's were chosen so that dummy variables for them represent more of the X variation in blood pressure than would be the case in a random survey. This increases the collinearity of the PSUa with the outcome variable. It is possible (although unlikely, given the broad range of potential variables investigated), that location represents some risk factor that is omitted and is correlated with blood lead. It is possible to investigate this possibility by finding classification variables that explain location whil minimizing, TEH 0412426 DUP050453478 -to the extent possible, their collinearity with blood lead. I will report the results of a number of such classification schemes, including using the 64 PSU's in the NHANES survey, and empirically demonstrate that the blood lead -blood pressure relationship is not due to a confounding with location. I will also show that the impact of a classification scheme on the blood lead coefficient is proportional to its collinearity with lead, and not to its accuracy in explaining location, thereby indicating that the impact on lead is due to insufficient within site variation and not omitted variable bias. I will use the results for all men over 20 years old in this analysis. (Since supercarp only allows 50 Independent variables it cannot be used in this analysis.) One factor that makes it easier to separate out the two possibilities is that, as I mentioned in my previous memo, PSU is not the same as location. WHile a stand was set up in a given location, the sample'persons were drawn from a considerable distance around that location, sometimes from 100 miles away. The variable size on the NHANES II tape codes each examined person with a degree of urbanization levc? from 1 to 8, where 1 means that they lived in a city of over 3 million, 7 that they live in a town of 2500-10,000 people, and 6 that they live in a rural area, and not even in a town. A univariate scan of the NHANES data shows that each PSU contains people with a range of urbanization levels, in fact PSUs in cities of over 1 million people have sampled individuals who live in rural areas. As I also noted previously, some PSU's contain more than one urban location, (i.e. New Haven and Hartford), and, because the sample was designed to be representative, large metropolitan areas had more than one stand (There were 5 in the New York metropolitan area). Because of this we can consider various alternative forms of model 2 to represent the locational factor and see what impact they have on the blood lead coefficient. All regressions were run in weighted least squares, since SURREGR cannot be used when the stratifica- TEH 0412427 DUP050453479 tion ia entered as variables My irat choice of the A 'a to explore the consistency of the blood pressure relationship is to use the Strata in the NHANES II survey. These strata represent the same basic geographic areas# the same size of county# of demographic profile etc., and are therefore a good measure of locational and demographic factors. The two PSU's in each strata are supposed to be similar on these characteristics, and the use of strata minimizes the collinearity with lead. The results are shown in table 2 in the appendix. For systolic blood pressure the coefficient was 4.84 and the tatatisbic was 4.732 compared to baseline values of 5.232 and 5.505 respectively. For diastolic blood pressure lead has a coefficient of 2.262, and t-statistic of 3.498, compared to a baseline value of 2.956 and a t-statistic of 4.724. Thus accounting for location and stratification in this manner has very little impact on the blood lead-blood pressure relationship. The next consistency check was made by further refining the locational dummies. The NHANES data contains a variable for whether or not each examined person lived inside the central city of the SMSA where they resided, if they resided in an 5KSA. I split each strata into two substrata, one containing all persons living inside the central city, the other for persons not living inside the central city of an SMSA. This gives-64 dummy variables to use as the A in equation <3>. However# 11 of the substrata were empty# because there was no central city in the strata. Since this approach does not lump people residing in rural areas and living in cities together# as the Dupont approach does# it would appear to be a more precise subcategorization of strata to account for location then their use of PSU. It is also less collinear with lead. The results are shown in table 3 of the appendix. For systolic blood pressure the coefficient is 4.7 and the t-statistic is 4.753, for diastolic blood pressure the coefficient is 2.44 and the t-statistic is 3.735. Again# these are little changed from the results in the basecase (equation ItH 0412428 DUP050453480 A different approach to forming the subgroupa to uae aa the A in equation 3 is to subdivide each of the 4 regions of the country into the 8 different degree of urbanization categories. This is less precise than the Dupont categories in identifying specific geographical locations, but more precise in identifying the urbanization level in which the subject resides. The results of this analysis, shown in table 4, gives a coefficient of lead of 5.22 for systolic, with a t-statistic of 5.34, and a coefficient of lead for diatolic blood pressure of 3.18, with a t-statistic of 4.89. This indicates that the reduction in the blood lead coefficient from controlling for PSU'a is not due to their representing urbanization. The remaining tests of the consistency focus on representing /precise geographical locations with less collinearity with lead \]then using the 64 psu's. If they also show little impact on lead, then we can conclude that the reduction in the lead coefficient when using the 64 PSU's is due to its collinearity with lead interacting with the measurement error, and not due to its representing a genuine ommitted factor. .. Two approaches were taken to this consistency test. First, since 48 of the PSU's represented cities or counties with populations over 100,000, and the others represented more difuse rural areas without any concentration of people in one central location, and with populations drawn from up to 100 miles away from the site of the examination stand, dummy variables were created for just those 48 more localized pau's. The remaining rural areas were represented by 4 regional dummy variables. Thus represent 52 subgroups. The results of this analysis are shown in table 5. For systolic blood pressure lead had a coefficient of 4.15 and a t-statistic of 4.19. For diastolic blood pressure lead had a coefficient of 2.16 and a t-statistic of 3.28. Using 48 PSU's represents exactly the same level of control for those urban TEH 0412429 DUP050453481 locations aa the 64 PSU approach, and for the rural areas the difference is small. However the difference in collinearity with lead is significant, because all of the time and seasonal variation in lead is not removed. A final consistency check was done using all of the 64 PSU'a. However, to obtain more precision on exact location, use of the same water supply, etc., the PSU's were split. Those persons coded as residing in completely rural areas, or in towns of between 2500 and 10,000 in population, who clearly do not reside in the cities where the stands were located were excluded from the categories. They were all subdivided into four regional rural locations. This categorization provides a finer measure of location than using the 64 PSU'a as representative of location, but again has lower collinearity with lead. 11 psu's were not independent of the remaining variables presumably because they consisted primarily of rural locations. Thus the total number of categories was 57. The results ar.e shown in table 6. For systolic the coefficient of lead was 5.28 with a t-statistic of 5.32, for diastolic blood pressure, the coefficient of lead was 2.65, with a t-statistic of 4.02. Table 7 shows the results of using the 64 PSU'a aa categorization variables. For systolic blood pressure the coefficient of lead is 3.30 with a t-statistic of 3.33, for diastolic it is 1.41 with a t-statistic of 2.15. This represents a reduction of 37X for the systolic coefficient and a reduction of 52* for diastolic blood pressure. Both coefficients remain statistically aignif CONCLUSIONS Comparing all of these results suggests that: -lead is still significant as a predictor of blood pressure when model 2 is used and the A i represent PSU, although the magnitude is reduced by TEH 0412430 DUP050453482 1/3 to 1/2 When alternative subcategorizations of the sample are used which are more specific for urbanization or location, but less collinear with lead, the impact of using model 2 on the lead coefficient is substantially reduced or eliminated. -Even when almost all of the PSU'a are used as variables but the collinearity is reduced by dealing differently with the rural areas, the impact of model 2 is minimal. -The relationship between blood lead and blood pressure in the NHANES 11 data is consistent, and robust under both a wide range of alternative model specifications and a wide range of alternative classification schemes. It is also consistent with the result of other general population surveys. -The reduction in regression coefficient from using the classification scheme that is most collinear with lead seems to be due to the collinearity, rather then the representation of location and urbanization. TEH 0412431 DUP050453483 REFERENCES 1. W.G. Chochran Technometrics 10,637 1968 TEH 0412432 ; DUP050453484 TABLE 1 RESULTS OF BASELINE ANALYSIS AFTER CORRECTING FOR MEASUREMENT ERROR TEH 0412433 DUP050453485 N 30991.02 -CRESSION COEFFICIENTS - DEPENDENT VARIABLE IS BP2DIA * RIABLE 3TERCPT ;e y r s ;e s q 31 JG2INC )GVITC -BSQ ;c h o l 10KNUM iLCIUM 1ETVITC :n c :mo g l o b UCEPR VWHYPER OFFEE EIGHT /PERMED KCE COEFFICIENT 8.9215230D+01 B.5179156D-01 -8.0111895D-03 7.2B499I8D-01 -3.6633675D+01 1.1480018D+00 4.3278731D-03 3.5014183D+00 --5.6466130D-02 1.6846338D-06 --9.9106581D-03 3.3071074D-01 1.2372121D-01 1.2142064D-02 2.2039796D+00 -2.3803678D-01 1.4564105D+0I 5.4619252D+00 .2-'62'ra951D+00^ STANDARD ERROR 3.1480411D+01 1.0894348D-01 1.1681387D-03 9.2730583D-02 8.8970814D+00 3.15897290-01 7.2320342D-04 1.5948060D+00 2.3875944D-02 8.3832945D-07 3.2752592D-03 9.4888104D--02 2.8875438D-02 5.3527929D-03 3.9217988D-01 9.4315350D-02 3.2050318D+00 8.4050559D-01 8.6263038D-01 T-STATISTIC 2.8339919D+00 7.8186558D+00 -6.85808080+00 7.85608330+00 -4.11749360+00 3.6340982D+00 5.9843094D+00 2.1955136D+00 -2.3649801D+00 2.0095128D+00 -3.0259156D+00 3.4852708D+00 4.2846521D+00 2.2683604D+00 5.6198182D+00 -2.5238392D+00 4.54413740+00 6.4983806D+00 3.0384915D+00 a CHWARTZCARPRUN1--Tr`50PEirC RELEASE 10-80 IOWA STATE UNIVERSITY TEH 0412434 DUP050453486 GRESSION COEFFICIENTS - DEPENDENT VARIABLE IS BP2SYS r RIABLE TERCPT EVRS ESQ 2 GZINC BSQ LCIUM FFEE WHYPER :ig h t PERMED - COEFFICIENT 6.3811359D+01 -4.7275936D-01 9.1906898D-03 1.2797492D+00 -S.3776212D+00 8.4716657D--03 2.6327408D-06 -3.88U590D-01 3.33^03470+00 I.848I922D+0I jr""6.0841298D+00/ STANDARD ERROR 1.7701916D+01 1.6341380D-01 1.7899688D-03 7.8843626D-02 2.5976941D+00 1.3558794D-03 1.2264S31D-06 9.5377448D-02 6.4388941D-01 6.25B7973D+00 1.4178121D+00 1.7029155D+00 T-STATISTIC 3.6047712D+00 -2.8930198D+0D 5.134553OD+OO 1.6231486D+D1 -3.2250222D+00 6.2480968D+00 2.1465598D+00 -4.0692628D+00 5.1810678D+00 2.9529510D+00 6.1149462D+00 35727726D+00 i TEH 0412435 DUP050453487 TABLE 2 REGRESSION RESULTS SUBCATEG0RI2ING ON STRATA # TEH 0412436 DUP050453488 N 30991.03 3 AS OFF VARIABLE! 6P2SY3 SnytJCE OF SUM OF SQUARES MODEL 42 4816026404 FKfl 1216 12570140704 C TOTAL 2258 173*6167138 ROOT *s e OFF 'NfAN CV| 2381,690 128,270 1856,773 me a n SQUARE 114667295 5072446 R-8QUARE a d j k-sn F VALUE 20.215 0,2770 0,2633 PR03>F 0.0001 v ar iabl e o f p a r a me t e r e s t ima t e STANDARD T FOR HQ 1 ERROR PARAMETERS X'iTEHCEP i 69,669024 13.517554 AlErk S t 0,467125 0,154801 AfifcSu i 0,009049477 0,001700945 0*1 l 1,273142 0,087264 l c s ?ih c l 8,983279 2.136954 At HSU l 0,008925774 0.401299349 CALCIUM i ,00000222671 ,00000118297 COFFEE i 0,340072 0.104939 F *mh y E9 i 3.253691 0,689177 XOr(T i 17,670201 4,964062 HfPNUED i 8,423290 1,272749 Ll l 4,643092 0.963153 LOCI i -1,867091 2.462489 Lf,C2 ! 1.754628 2,513575 LOCI 1 7,628028 2.566929 LOCa 1 4.401316 2.305211 Lot5 l 0,995434 2.603146 LnC6 l 2.999698 2.763960 LOCT - 1 6,479525 2.694921 LOCA 1 1.286B06 2.383631 LOCR 1 4.557120 2,687180 Lnctti i 2,358493 2,594702 LOCli I 0,703654 2,414058 L0C12 1 3,523361 2,454088 L0C15 LC1 1 7.128818 1 -4,445573 2.772353 2,477844 LOCI 5 l 0.839914 2,620637 Lncib 1 -1,782119 2.715365 LOCI 7 1 1,0300C 2,337656 L0CI8 ** *r i* aT 1 7,043275 2.617896 1 3.650958 2.69950 LJ.C20 * 0.755070 2.529495 tffcai 1 3.711074 2.679642 L0C22 1 -1,213803 2,467882 LOC23 5 2.087329 2.512583 L0C20 1 3.954312 3.195376 LPC?5 1 -0.517636 2,466198 LOC?o 1 6.993977 2.622417 Lf`C27 1 -5.234867 2.559611 5,154 3.018 5,320 14,586 4,204 6,669 1.8*2 3.241 4.721 3,560 6,618 4.723 0,758 0,698 2,972 1.909 0,362 -1,085 2,404 0,540 1.748 0,909 0.291 1,436 2.571 -1,794 0,320 -0,656 0,446 2.690 1,352 0,299 1,3*5 -0,442 -0,831 1,236 0,210 2.667 2,045 PROB > IT 1 0,0001 0.0026 0,0001 0,0001 0,0001 0,0001 0,0599 0,0012 0,0001 0.0004 o.oooi 0,0001 0,4484 0.4852 0.0030 0.0564 0.7022 0,2779 0,0163 0,5894 0.0*06 0,3635 0,7707 0,1512 0,0102 0,0729 0.7486 0,5116 0.6554 0.0072 0,1764 0.7653 0.1662 0.6229 0,4062 0.2160 0.8339 0.0077 0,0410 TEH 0412437 DUP050453489 VAKUSu E LOC28 LnC?9 L^CSO L0C31 DF J 1 1 1 PARAMETER ESTI-ATE -2.201571 3,R16tt76 !,391270 2.13*735 STANOa RD T FOR H0| ERROR PARARETERaO 2.315075 2,59t>570 2.627099 2.576270 0.951 1,70 3,955 0,830 PROS > ITI 0,301* 0,101* 0,0001 0,0085 TEH 0412438 DUP050453490 ZP VARIABLE* BP20IA QuRCE DF SU* OF SQUARES ODEL 50 0 i2202 TOTAL i2 252 2425512775 <*936109o09 7361622342 ROUT mSE OEP iMEAN c.v. 1497,214 82,334660 1016,444 MEAN SQUARE 48510255 2241648 R-SQUARE ADJ R-SD F VALUE 21.640 0.3295 0.3143 PR0B5F o.oooi /a k ia s l e OF PARAMETER ESTIMATE st andar d T FOR HOl ERROR p a r a me t e r s I`*TECEP a*;e v k s tQEStf i" J .nuZJHC .'GVITC tiasu .SCHOL i'JtrNIIM :-`LCI>.'R iTETVTTC IC lEMUGLO* f^ICEPR *AMMTPR rFfEE if.XGHT lYPER^EO <*CE .oci .OC2 .*'C3 ."C .ncs .OCA .ncT ,,uc .OC* .nC10 -'CV1 .oci 2 ."C13 _ 'Ll** ,i?C!5 .OCU .rci 7 .OCtB -OCi 9 -OC20 1 04,780959 31,907511 l 0,436602 0,101692 1 -9,00790779 0,001103976 1 0,763602 0,073688 1 -33,081732 8.846927 1 1,194129 0.264748 1 9,004716*155 0,0004499976 1 2.723858 1,096896 t 0,046725 0.016463 1 ,00040143821 7.452J0E-07 l y.0099379 0.002651983 1 0,292206 0.095258 i 0,115033 0.021609 1 0,010384 0,004035374 1 2,051442 0.434356 1 -0,21**94 4* 13.157620 0,068643 3,156060 1 5,453748 0,604722 1 2,696115 0,639400 J 0,044370 1 2,605264 1,562574 1,596267 1 6,208475 1.627223 1 5,852086 1,461687 1 1,006459 1,656593 1 0,43e394 1.747412 1 2.262472 1.699021 i 0.485113 1.505122 * 5,396629 1,647645 1 4,241301 1,643598 1 2,621625 1.513940 1 0,088071 1,555594 1 6,399910 1.759493 1 -3,251645 1.566737 1 5.211291 1,650436 1 1,995103 1,715198 1 5,291024 1.462253 1 4.270204 1.651800 1 2.401707 1.722482 1 0,601268 1,598171 2.657 8.230 -7.163 10.432 -3.739 4.510 5,582 2.483 -2,835 1,930 -3.747 3,068 5.323 2.573 4.723 -3.125 4,169 6,777 3.214 0,028 1.757 3,8)5 4,064 0,608 -0,250 1.332 -0.322 3.27S 2,58 0 1,732 0,057 3.637 2,075 3,158 1,163 3,618 2,585 1,394 0.501 PRUB > ITI 0.0079 0.0001 0,0001 0.0001 0.0002 0.0001 0,0001 0,0131 0,0046 0.0537 0,0002 0,0022 0.0001 0,0101 0,0001 0,0018 0,0001 0,0001 0,0013 0,9776 0,0790 0,0001 0,001 0.5436 0,8028 0,1831 0.7472 0.0011 0.0099 0.0835 0,959 0,0003 0.0381 0.0016 0.2449 0,0003 0,0098 0 #4634 0,6162 VARIABLE DF -OC21 -DC22 =OC 23 .OC2* .1C25 .nL26 .fC27 .rC2B .UC29 .OC30 .0C31 .L l 1 1 l 1 1 1 1 1 1 1 1 PAkAHgTER ESTIMATE B,*23672 0.799969 0,661668 -2,095327 -2.333977 5,794529 0.59729 2.625975 2.659320 1.943221 1.139151 2.262086 st andar d T FOR HOI ERROR PARAMETERS 1.701812 l.5725a 1.599907 2.016668 1.555697 1.655660 1.613468 1.470188 1,639291 1.657719 1.628297 0.646757 5.185 0.509 0.414 -1.039 -1,500 3.469 0.295 1.786 1.622 1.172 0,700 3,498 PROS > IT 1 0.0001 0.6110 0.6792 0.2990 0.1338 0,0005 0,7757 0,0742 0.1049 0.2412 0,4843 0,0005 TEH 0412440 DUP050453492 TABLE 3 REGRESSION RESULTS SUBCATEGORIZING ON STRATA AND INSIDE CENTER CITY TEH 0412441 DUP050453493 N 30991.04 DEP VARIABLE! S2Sr5 SPUCE OF SUN OF SQUARES me a n SQUARE F VALUE PR08>F mmo e l NnR 63 2195 C TOTAL 2258 SiayB9fc69S 12285270413 17366167108 81601535 5576711 18.627 0.0001 ROOT "SE DEP MEAN C.V. 2361.929 128.270 1681,366 RSQUARE ADJ RSQ 0.2957 0.2755 NOTE! HOOEL 13 NQT PULL RANK, LEAST SOUAHES SOLUTIONS FOR TmE PARAMETERS ar e NOT UNIQUE, SQNE STATISTICS KILL be MISLEADING, A REPORTED OF OF 0 OR B KEANS TMAT THE ESTIMATE IS BIASED. Th e FDLLOmINC PARAMETERS h a v e BFEN SET TO 0 * SINCE TE v a r ia b l e s a r e A l in e a r c o 6TNATION OP OTHER VARIABLES AS SHQ*N, L30C7 L3UC8 L3UC19 L30C15 UOC17 L3QC16 L30C2! L3UC28 L3UC29 L30C32 L3UC51 a a a a . v a r ia b l e OF p a r a me t e r ESTIMATE STANDARD t f u r h o * ERROR p a r a me t e r s In TERCEP AGEvRS AGE SO 8MI tnGZINC AL6SVI CALCIUM COFFEE Fa mh v p ER HEIGHT MVPERmEO LL L30C1 L3UC3 L3UC4 UDC5 L3UC6 L3UC7 L3UC8 68,866410 13.891881 1 0,879896 0,159607 1 0,009119988 0.001701388 1 1.266633 0,087376 1 6.967072 2,181638 1 0.006951855 0.001303386 l . 00000216782 .00000117783 1 0.361922 0.100866 1 3,223187 0.688180 1 18.871226 8.959886 1 B.59S105 1.272764 1 8.700876 0.968896 l 2.260686 2.984936 1 2.373382 3.336112 1 1.653332 3.590935 1 0.316657 4.135512 1 1.012355 3.095904 1 0.101965 3.550881 A0 00 5.076 -3.097 5.360 14.519 4,187 6.868 1.657 3.451 4.684 3.724 6.753 4,753 0.768 0.711 0.516 0.077 0.327 0.029 PROS > 171 o.ooot 0.0020 0,0001 0.0001 0.0001 0.0001 0.0634 0.0006 0.0001 0.0002 0.0001 o.ocoi 0.8826 0,9769 0.6058 0,9389 0.7837 0 H,9771 0412442 DUP050453494 Va HIa UE OF L3UCY L3UC1J 13UCH 13UCI2 130C13 UJOCia L3UC15 13UC16 U1UC17 S.30C16 L3UC19 L3UC20 L3UC21 L30C22 13UC23 LJVC24 L3UC25 L3t'C26 L3UC27 UUC2* L3UC29 UUC3* L?UC3l t3UC32 L3UC33 U3'JC3U U0C35 LH>r.lh L3UC37 LlVCl* 13u C3< L3U.C4 0 13UC1 L3UC42 L3UC3 L3UC44 13UC45 L3Uefe 13VC7 U30C8 LJ0C49 IUUtS#ivv.CUfCS5>O14 Llocia L3VC4S L30C5 h> Juv.55 USt'CSfe L30C57 ^3uC53 ,,3t;CS9 .3OC60 .3UC62 1 1 i 1 1 0 0 1 0 0 1 1 0 1 1 0 1 1 i t 0 1 i 0 l i 1 l t l i 1 1 l 'l l l l 1 l i QlA l * i i i i l l i i i p a r a me t e r e s t ima t e u.ia^teR }.406004 7.36*333 5.726328 6,131682 0 0 -0.05SI18 0 0 3.704683 0.294975 0 -0.952005 8,802907 0 3.163652 1.361097 -2.32*236 3.706169- O 1.583750 0.455276 0 1.261529a.501399 13.9M922 5.36*302 3,644485 5.474004 6.544478 1.338989 2.046265 3.646332 4.256766 2.132916 8.461001 4.3S5443 0.923967 2,148416 1.117551 7.101540 A V 0.91A305 3.744371 1.3U84S 0,351050 4.064654 0.097044 8.685578 6.764829 0,86*613 3.895643 13.175531 s t a n d a r d T FOR HOI e r r o r p a r a e t e r o 4.237276 2.917754 3.515234 3.268072 3,376025 5.017810 2.677407 5.648897 3.194103 4,486539 4,530297 .52397 3.607616 2,851969 4.424912 2.B26234 3,089559 2.920231 2,920326 2.302717 3.430882 3.413408 2.672*78 2.364101 2.826447 3.772082 2,647042 2.823109 3,656456 2.4S7719 2,599005 2.8*8266 2.288853 2,59t>i37 2.592103 2.658077 2,879526 2.651909 3.169238 2.578218 2.807633 2.905920 2,708784 2,575244 2,834553 2.867 0,6*4 2,096 1,752 1.816 0.011 1.3*4 0.050 0.298 1,961 * 0,698 0,305 0,645 1 ,300 0,358 -0,161 0,408 -1.541 4.473 2.244 1.062 1.604 2.448 0.566 0,724 0,967 1.609 0,756 2,314 -1,772 0,356 0,744 0.4*6 2,735 0*354 1.4C9 -0.456 -0,132 1.2*3 0.038 3.094 -2.328 0.321 1.513 4,646 348 P90B > IT 1 0.0042 0.5135 0.0362 0,0799 0.0695 0.9912 0.1666 0.959* 0.7657 0.0500 0,4850 0.7602 0,518* 0.1939 0,7204 0,8721 0.6831 0.1234 0.0001 0.u250 0.2682 0,1069 0.0144 0.5712 0,4692 0.333* 0.1078 0.4500 0.0208 0,0765 0,7222 0.4571 0.6254 0.0063 0^7232 0.1591 0.6467 0.8947 0.199* 0.9700 . 0.0020 0.0200 c.faea 0.1305 0,0001 TEH 0412443 DUP050453495 V4KI AfcUE OF LSUCoS 1 PARAhFTER ESTIMATE P.757046 STANDARD T FQR HOi ERROR PARAMETER*!? J,974039 2,2o<| SAS P*OB > IT! 0,0277 $ TEH 0412444 DUP050453496 DEP VARIABLE* 0P2PIA SOURCE DP SUM OP SQUARES KEAN SQUARE P VALUE PROP>P MODEL 71 er r o r 2iai C TOTAL 2252 2513232262 4848390100 7361622382 3539763* 2223012 15*923 0*0001 HOOT *5E DEP MEAN C.V. 1090.977 62*334*60 1*10.S7 R-SQUARE ADJ R-SQ 0.3414 0.3200 NOTE! MODEL IS NOT PULL An K, LEAST SQUARES SOLUTIONS FOR TE PARAMETERS ARE NOT u n iq u e . s o me s t a t is t ic s w il l BE MISLEADING. A REPORTED DP OP 0 OR B MEANS THAT t h e ESTIMATE IS BIASED. THE POLLUTING PARAMETERS h a v e BEEN SET TO 0, SINCE THE VARIABLES ARE A LINEAR COMBINATION OP OTHER VARIABLES AS SHOWN, L0C7 4.0CS L0C14 LOC15 LOCI 7 LOCI* LDC21 L0C24 L0C29 L0C32 LOC51 a a a a a a a a a a v a r ia b l e or PARAMETER ESTIMATE STANDARD T POR H0 ERROR PARAMETERaO JNTERCEP AGETRS AGESQ MI LQ6Z2MC LOGVJTC ALBSU LSCHQL S"QKNUM CALCIUM OIETVITC 2 INC HEMOGLOB TflCEPR f'AMHYPEP COPPEE HEIGHT h y p e r me d RACE LOCI 1 31,585182 31.900054 -1 0.636176 0.102391 1 >0.00791101 9.001109707 1 0.787347 0,073987 1 31.470660 8*658652 1 1.120961 0.265699 1 0*004594211 0.0008501269 1 2*623909 1,100*44 1 0,044498 0,016595 1 ,00000141999 7.45091E-07 1 0,00936215 0.00266636 1 0.275308 0,095292 1 0,107609 0,021645 1 0,009602622 6.004031703 1 1.997015 0,435643 1 0.242506 0,068658 1 12.545104 3*166506 1 5.53*864 0.806625 t 2.67*836 0.860587 1 0,204698 1.922R52 2,557 8.167 7.129 10,642 -3,553 4.236 5.404 2*384 2.6*1 1,906 3,511 2.8*9 4.935 2.431 4,584 S5?2 3,959 6.650 3,113 0.107 PRC8 > ITI 6,0106 0.0001 0,0001 0,0001 0.0004 0.0001 O.OOOi 0,0172 0,0074 0,0568 0,0005 0.0039 0,0001 0,0151 0.0001 0.0004 0,0001 0,0001 0.9152 TEH 0412445 DUP050453497 VARIABLE LCC2 LUC3 LPC 4 L0C5 LOG 6 LCC7 LOCB L0C9 L0C10 LOCH LOCI 2 LOCI 3 LOCH LOC15 LOCI 6 L0CI7 LOCH LOCI 9 LOCZQ LOC21 LOC22 LOC23 LOC2 LOC25 L0C26 LOC27 L0C2A LOC29 LOCSO LOC31 LOC32 L0C3J LOCSO LOCS5 LOCife LCC37 L0C38 LOC19 LOC0 LOCl LOCO 2 L06 3 LOCoo LOCOS LOCOO LOC07 LOCO 8 A LOCSO LQC51 LOCS2 LOC53 LOCSO L PC'S5 DF 1 1 1 i 2 0 0 1 1 1 1 1 0 0 1 0 0 1 * 0 1 1 0 1 1 i 1 0 1 1 0 1 1 1 1 ,,t 2 1 1 1 J I s 1 1 t 1 1 1 0 2 2 2 2 p a r a me t e r ESTIMATE 3.520081 2.307012 0.2 07600 1.332936 -0.338516 0 0 0,325107 3.000002 3.121902 1.906322 5.326420 0 0 2,536859 0 0 2.302772 -0,565566 0 2.38*635 -6.210597 0 -3.009819 3.873407 1.522771 1.794906 0 1 .9*6350 -1.199736 0 ft.075261 2.315901 8.005352 6,233191 0,059329 -0.102771 2.217988 -0.060230 5,752304 6,236168 2,386374 1,046632 7.619927 3.216073 5.187175 3.09S045 5,313633 0.229185 0 0.918367 8.790658 0.391392 2.394237 st andar d T FOR HO* ERROR PARAMETER**) 2.112252 2,286035 2.617271 1.986312 2.252151 2,679770 1.855309 2.220206 2,078390 2.162627 3.180028 a 1.716400 3.699300 2.080821 2.604026 2.820922 2,860592 2.283420 1.622460 2.607666 1,795005 1.959833 1,660456 1.865470 1.523227 2.179607 2.166807 1.692283 1,099151 1,794521 2,393249 1.662283 1.800639 2.313230 1,560670 1.643671 1.831365 1.456496 1,645195 1.645127 1,695461 1,627330 1.696038 1.670 1,026 1,569 0.671 0,372 1,614 1.833 1.400 0.918 2.060 0.798 l3t>5 0.158 1.1*8 -2.187 1.208 1.350 0,667 0,9*7 a ft.700 0,668 . 0,243 1,2*2 0,506 0.092 1.662 0,047 1.311 -0,310 3.205 2.606 1.436 0,603 3.2*0 -2.061 3.156 -1,690 3,6*8 2.571 0.558 5.185 0.214 1.412 PR08 > IT1 0.0950 0.3051 0.1167 0.S023 0.7097 0,1067 0,0670 0.1606 0.3586 0.0138 a 0,4251 0,1720 0,8742 0,2509 0,0289 0,2270 0.1759 0.5049 0,3236 . 0,4838 0,5040 o.afteo 0,2143 0.0001 0,0001 0.0627 0.962? 0,1901 0.7568 0,0014 0,0092 0.1513 0.4216 0,0010 0.0394 0,0016 0.0912 0,0003 0,0102 0.5767 0.0001 0.8304 0.1582 VARU*LI 0 F L0C56 L0C5? 10C56 L0C59 LOC60 L0C61 L0C62 LOCoJ LL 1 i l 1 1 l 1 1 l PARAMETER ESTIMATE -2,050622 -2.130704 6.218680 0.132016 3.298981 2.703266 1.859202 6.850650 2.04319J STANDARD T FOR H0| ERROR p a r a me t e r s 2,008956 1,633938 1.782799 1.800211 1,719522 1.632646 1.794386 2.513787 0.650193 -1.021 -1.304 3.486 -0.072 1.919 1,656 1.036 2.725 3,735 PROS > IT( 0,3075 0,1924 0.0005 0.9428 0.0552 0.0979 0.3003 0.0065 0,0002 TEH 0412447 DUP050453499 TABLE 4 REGRESSION RESULTS SUBCATEGORIZING ON REGION AND DEGREE OF URBANIZATION si TEH 0412448 DUP050453500 N 30991.05 SAS DtP VAR!ALf 8P2SYS SOURCE OF SU- OF SOUAKC5 me a n SQUARE F VALUE PRDB>F MODEL 4! ERROR 221? C ?or*L 2258 <1597799915 12768367195 17386167108 112141461 5766321 19.441 0,0001 BOOT *$E DP MEAN c.v. 2401.733 126,270 1872.396 R-SOUARE ADJ ft-SQ 0.2665 0.2508 MOTE I MODEL IS NOT FULL RANK. LEAST SQUARES SOLUTIONS FOR THE PARAMETERS ARE SOT UslLUE, SI'mE STATISTICS *ILL BE MISLEADING. A REPORTED OF OF 0 OR 8 MEANS THAT TmC ESTI -ATE IS 6IASED. THE FOLLOWING PARAMETERS h av e BEEN SET TO 0 t SINCE Th E VARIABLES ARE A LINEAR COMBINATION OF *1Th ER VARIABLES AS SHOWN. me 17 VARIABLE In TERCEP AGEVk S ag es* 8*1 Lu GZIn C AlttSU C a l CI'j m c o f f ee FAMHYaR HEIGHT HfPERMED U LOCI tOC2 mes LOC tOCS L0C6 LOC7 me a LUC 9 t^eio Locn me 12 toe 13 LOCI * LPC15 LOCI 6 me 17 toe *6 p a r a me t e r OF ESTIMATE st ano ar d er r o r 1 66.193932 13,474574 1 0.453678 0,156504 1 0,009095806 0.001718458 1 1,269571 0,087963 1 8.565174 2.136266 1 0,009311152 0.00130619 1 00000236951 .00000119309 1 0.353227 0,105660 1 3.237436 0.694207 1 17.447637 4.977566 1 9.130530 1,230989 1 5.215524 0,976702 1 1.301681 1.634366 1 -2,054040 2.303587 1 5.640242 3,592573 & 2.242364 2.044459 1 0.040604 4.064251 1 10,320259 7,066484 1 7,752211 3.660378 1 1,281436 1.328629 1 ,896643 2,190028 1 3.574425 2.132136 1 2,229195 1.915307 1 3,127611 2.378921 1 3.594051 4.235027 1 3.660663 3.015830 1 8.200051 3.078787 1 4,213463 1.662965 00 1 3.639667 1.936825 T FOR HO 1 PARA1ETER*0 4.913 2.9C0 5.293 14.431 3.972 7,128 1.986 -3,343 4.634 3,505 7,131 5,340 0,746 0,892 -1,570 -1.097 0,010 1.460 2.196 0.701 2,237 1.676 1.164 1.315 0,849 -1,214 2.663 2.504 1.879 PROP > 1T t 0,000! 0.0036 0.0001 0.0001 0,0001 0,0001 0,0072 0,0006 0,0001 0,0005 0.0001 0,0001 0,4259 0,3726 0.1166 0.2728 0,9920 0.1443 0.0353 0.4835 0.0254 0.0938 0.2446 0.1687 0,3962 0.2250 0.0078 0,0124 O 0,0603 0412449 DUP050453501 VkHlAHlt UF L5C19 L0C20 LHC21 LOC22 I.PC23 L0C24 1.0025 Ln-C?t> LOC27 LtiC28 1.CC29 LOC30 L0C31 1 i l 1 t 1 i 1 ! 1 1 1 1 p a r a me t e r ESTIMATE 0.349734 3.680524 2.096323 1.408897 0,759819 0.775258 7.154788 3.190030 0.798380 0.760701 1.960534 5.245297 -1.418137 s t a n o ar d T FOR HOl ERROR PARA-ETERs O 2.916223 2,779290 2,826463 2,855259 2,701628 1,468037 2.316479 1,994376 1,896115 2,659208 4,202921 2.454497 2,874465 -0,120 1.324 0,742 0.493 0.2R1 0.528 3,089 1.600 0.421 0.286 0,466 2,137 0,493 pr o s > m 0,9946 0.1856 0.4584 0.6216 0.7765 0.5975 0.0020 0.1098 0.6738 0.7749 0.6409 0.0327 0.6218 TEH 0412450 DUP050453502 V SiS OEP VARIABLE* BP20IA SOURCE of SUM OF SQUARES MEAN SQUARE F VALI'E PR0B>F MODEL 49 koR 2203 C TOTAL 2252 SOOT i'SE DEP MEAN c.v. 80644.145 2S3&53 334497 10.734552 6?,494452 13,01245 1645.799 115.231 R-S9UARE ADJ K-S<3 14,283 0.24U 0.2242 0.0001 NOTE* mo d el is VQT FuLL h a n k . LEAST SQUARES SOLUTIONS FOR THE PARAMETERS ARE NQT UNIQUE. SOE STATISTICS RILL BE MISLEADING. A REPORTED DF OF 0 OR 6 NEA*S THAT ThE ESTIMATE 1$ fiUSEO. THE FOLLOh ING PARAMETERS HAVE BEEN SET TO 0, SINCE THE VARIAFLE5 ARE A LINEAR COMBINATION OF OTHER VARIABLES as Sh o w n . LOCI 7 v a r ia b l e DF In 1RCP 1 AGEYRS 1 AGESQ 1 *I 1 LOU7INC 1 LOUVITC 1 ALtSSU 1 LSCHOL 1 SMu h n u m J CALCIHH i OtETVITC I ZINC 'l HEh o g LOS s TRICEPR FAMHYPR i i c o f f ee l HEIGHT l h y p e r me d l RACE t LOCI i LOC2 t LOC3 i LOC l LOC5 i LOCfe i LOC7 i LOGS t L0C9 i LOCIO l LOCI 1 i p a r a me t e r ESTIMATE st andar d ERROR 97.923307 34.627925 0.769753 0.109816 0,00742647 0.001153988 0.640662 0.077107 *34.3ol386 9.679975 0.561096 0.284318 0.00336726 0.0009042316 3.744377 1-. 1918 3 0 -0,045076 0.018467 a .25416E-07 B.Q9161E-07 -0,0076974 0.003040004 0.310774 0,065646 0.105302 0,022177 0.011741 0.004359689 2,051327 0,471363 -0,226337 0.079159 12.796897 3.403347 4.889610 0.758129 2.357143 0.863127 2.343059 1.151455 -0.748307 1,585534 -3.781676 2.268292 0,791438 1.396501 2,314672 2,596493 13.436675 5,450635 3.807950 2.323768 -0.339619 1.244300 3,199818 1,598269 0.968530 1.556813 -0.780973 1.269551 T FOR H0| P4RAME TER0 2.628 7.192 *6.435 8,309 -3.550 1.973 3.724 3,142 -2,441 1,020 -2,598 2.951 3,671 2.693 4.352 -2.859 3.760 6.450 2.731 2.035 -0.472 -1.653 -0.567 0.891 2.465 1.639 -0,273 2.002 0.634 0.615 P908 > III 0.0047 0.0001 C.0001 0.0001 0.0004 0.0486 0.0002 0.0017 0,0147 0.3078 0,0094 0,0032 0,0001 0.0071 0.0001 0.0043 0.0002 0.0001 0.0064 0.0420 0.6370 0,0986 0,5710 0.372 0.0138 0,1014 0,7849 0,0454 0,5260 0,5385 TEH 0412451 DUP050453503 VARIABLE OF LOC12 LOCI 3 LOCI LOC15 LOCI ft LOCI 7 LOCI 3 L0C19 L0C20 LOC21 LOC22 LnC23 L0C24 L0C25 LOC2ft LOC27 LOC2B LnC29 LOC30 LCC3I LL 1 1 l 1 l 0 1 1 1 1 1 I 1 1 1 1 1 1 1 l 1 PARAMETER ESTIMATE !.3o919 3.269541 -4.055265 2.637949 0,532309 0 1.054463 3,409606 1,632151 2,072950 2,179375 2,454317 1,315596 1,576354 0,4*2661 0,463355 2.293939 4,149099 4.901924 2.498873 3,013737 STANDARD T FOR HO 1 ERROR PARAMETERS 1.612615 2.541900 1.905113 1.926464 1.059855 1.371457 2.339372 2.097935 1,994433 2.079842 1.609199 0.965733 1.671195 1.530028 1.347999 2.025452 3.197372 1.780905 1.737091 0.685731 0.891 1.2R3 2.129 1,369 0.508 0.791 1.457 0.781 1.039 1.043 1,525 1.362 -0.943 -0,315 0,359 1,133 1.298 2,752 -1.439 4.395 SAS PPOB > ITI 0.3730 0.1997 0.0334 0,1710 0.6116 0,4292 0.1451 0.4350 0.2988 0,2948 0,1274 0,1732 0.3457 0.7524 0,7199 0.2575 0,1945 0,0060 0,1504 0.0001 TEH 0412452 DUP050453504 TABLE 5 REGRESSION RESULTS SUBCATEGORIZING ON THE 48 URBAN PSUa AND REGION TEH 0412453 DUP050453505 N 30991.06 DEP VARIABLE I BP2SYS SOURCE DF SU* OF SQUARES MODEL 6! 5198170504 ERROR 2189 12094235109 C TOTAL 2250 17292405613 ROOT MS E DEP MEAN C ,V, 2350.533 126,265 1632.554 MEAN SQUARE 85215910 5525005 R-SQUARE AOJ R-SQ F VALUE 15,424 0,3006 0.2811 PR08>F 0.0001 NOTE! MODEL ZS w q T FULL RANK LEAST SQUARES SOLUTIONS FOR THE PARAMETERS ARE s o t UNIQUE, SOME STATISTICS KILL be MISLEADING, a r e p o r t e d o f o f o o r b MEANS t h a t THE ESTIMATE is BIASED. THE FOLLOWING PARAMETERS h a v e BEEN SET TO 0# SINCE THE VARIABLES ARE A UNEAk COMBINATION OF OTHER VARIABLES AS Sh o r n , L0C14 VARIABLE JNTERCEP AGEVRS AGESQ auj l d g zih c ALt'SU CALCIUM COFFEE F*MHTBR HEIGHT hVPER-ED LL LOCI LOC2 L0C3 LOCR L0C5 LOGS L0C7 LOC LOCiO uOCIl -0CS2 ,,GU3 .0C16 .0C19 .0C20 .OC21 -OC22 .QC23 PARAMETER DF e s t ima t e 1 72.35S550 t 0,534019 1 0.0G9853689 i 1.2309(13 i 7,93*313 1 0.006733535 1 .00000207336 1 0.3)2364 1 2.957627 1 17,306958 1 8.590758 4.153181 1 4.087719 I 5.779067 I 0.723204 1 0.960444 I -0.921357 l 0.043432 1 0.871o55 I 4,667432 I -0.649751 1 -2.*48136 1 -8.571792 I -3,343950 1 -1.524402 I -6.479085 1 -10.437877 1 4,604720 1 6,965148 1 4,606820 STANDARO ERROR 13.518903 0.154293 0.001694417 0.087266 2.168657 0.001306967 00000117447 0.104305 0.669364 4.949642 1.259848 0.992303 2,4*5963 2.558405 3,180629 2.372720 2.646283 3.668946 3.411119 3.700078 2.578581 3.264356 3.391798 3.378777 2.784546 3.658658 3.009551 3.046181 2.607904 3.225554 T FOR HOI PARAMETERS 5.352 -3,481 5.815 14,105 3,660 6,6*2 1 7&5 -3.762 4.290 3,497 6,819 4,185 -1.644 -2,259 -0,227 0.413 0,323 -0,012 -0,256 -1,261 -0.252 -0.872 -2.527 -0,990 -0.547 -1,771 -3.468 1.512 -2.461 -1,428 PPOH > 1TI 0,0001 0.U0G5 0.0001 0.0001 0.0003 0,0001 0,0776 0,0002 0,0901 0,0005 0.0001 0.0001 0,1003 0,0240 0,8202 0,6795 0.7464 0.9906 0,7983 0,2073 0,8011 0.3*30 0.0116 0.3224 0.5641 0.0767 0.0005 0.$308 0.0132 0.1534 TEH 0412454 DUP050453506 v ar iabl e tOC24 LOC25 tOC?7 tOC28 LOC 30 L0C31 U0C32 L0C33 LOC34 IOC35 L0C36 UC 3 7 toe 3 a Local toc2 toe tt3 t0C4 tOC5 tOC8 L0C49 LOC51 L0C52 tOC53 L0C54 tocss tOC57 L0C59 LOC60 L0C63 L0C64 Ri R2 R3 OF 0 1 1 1 1 i i i i l l l l i l i i i i i l l l i i i l ! 1 1 1 1 1 PARAMETER ESTIMATE 0 -3.430385 8.498893 1.*39084 8.8a4433 15.124843 9.528033 3.128230 0.751404 10.643567 3.488299 4.812966 8.238452 8.149014 2.100280 19.024361 6.771305 6.668082 .7.440588 4.079035 0.098962 0.303510 12.406882 8.371903 10.788197 -8.787723 -6.331115 -10.547497 2.667094 -13.783164 3.635769 8.956461 0,690675 st andar d T FOR 601 ERROR PARAMETERS 3.102062 2.830421 2.266217 3.514942 3.4B12B4 3.213581 2.958639 3.466808 2,645312 2.753578 3.092389 2,996408 3.451426 4,013163 3.327071 3.368256 4.904991 3.662072 2.343488 3.086352 3.111017 3.546940 3.164630 2,905704 2,823270 3.110599 3.024758 3.629691 3.350348 2.014020 2.585666 1.604291 -1.106 3,Of3 0.723 2.516 4,35 2.965 -1.057 -0.217 4.024 1,267 -1.556 2.749 -2.361 0.523 -5,718 -2,010 -1.359 2.032 -1.741 0.032 0,098 3.498 2.629 3,713 3,113 -2.035 -3.467 -0.735 -4.114 1.805 3,464 -0.431 PR08 > 1T1 0.2689 0.0027 0.4696 0.0119 0,0001 0,0031 0.2905 0,8284 0.0001 0,2054 0,1198 0,0960 0.0183 0.6008 0,0001 0,0445 0.1741 0.0423 0,0819 0.9744 0.9223 0.0005 0,008* 0.0002 0,0019 0.0419 0.0005 0.4625 0.0001 0.0712 0.OC05 0.6669 TEH 0412455 DUP0504535Q7 V S*S DEP VARIABLES fiPCIA SOURCE DF SUM OF SQUARES MEAN SQUARE F VALUE P06>F MODEL 69 ERROR 2181 C TOTAL 2250 2509*53211 4850093003 7359926814 36374394 2223793 16.357 O.OCOl ROOT ms e DEP MEAN C V a 1491,239 82.340136 1811,072 R-SQUARE ADJ R*SQ 0.34t0 0.3202 NOTES MODEL IS NOT FULL RANK, LEAST SQUARES SOLUTIONS FOR THE PARAMETERS ARE n o t UNIQUE# SOME STATISTICS WILL BE MISLEADING. A REPORTED OF OF 0 OR 8 MEANS THAT THE ESTIMATE IS b ia s e d . THE FOLLOWING PARAMETERS HAVE BEEN SET TO 0 , SINCE THE v a r ia b l e s a r e A LINEAR COMBINATION OF OTHER VARIABLES a s SHOWN, LDC24 v a r ia b l e DF PARAMETER ESTIMATE st andar d T FOR h o ERROR PARAMETEH0O INTERCEP ag eyr s AGfcSQ 0MJ LOGZI`=C LUGVITC AIBSQ LSCHOL SMUKNIIH CALCIUM OTETVITC 2 IRC MEMOGLOfl TRICEPR Fa mh y p e R COFFEE WEIGHT HYPERw eo RACE LL LOCI L0C2 L0C3 LOCO LCC5 L0C6 LOCT LOCP LCCIO LDC1I 1 90,110799 32.039433 t 0,827253 0.102018 l 0.00776924 0,001105339 l 0,752460 0.07537 1 33.5291*6 8.894478 1 1.053905 0.265399 1 0,004504224 0,0008538185 1 2.727746 1.104813 I 0,049479 0.01654J 1 00000124755 7.46150E-07 t 0.30962747 0,00267727 1 0.308196 0.095657 1 0.107621 0.022130 I 0.011475 0,004067018 I 2,311404 0,437681 1 -0.232094 0.068974 1 12.683363 3.175072 1 5.476364 0.600810 i 2.406955 0,865635 1 2.164772 0.659640 1 3.803491 1.567160 l 2.410796 1.636063 1 0.051390 2.033760 I 5.155247 1.513016 1 4.616265 1.825042 1 -2.359595 2.338970 t 2.95*403 2.169954 1 3.132462 2.351937 1 2.378140 1.644982 1 1.288447 2.073994 2.812 8.109 7.029 10.095 3.770 3.971 5.275 2.469 2.991 1.672 -3.596 3.222 4.863 2.821 4.596 -3,365 3.995 6.839 2.781 3,2*2 2,396 1,474 -0,025 3.407 -2,529 -1.008 1.363 1.332 1.446 0.621 PROS > ITI 0,0*50 0.0001 0,0001 - 0,0001 0.0002 0,0001 0,0001 0,0136 0.0028 0.0947 0.0003 0.0013 0,0001 0,004* 0.0001 0.0008 0.0001 0.0001 0.0055 0.0010 0,0166 0.1408 0.9798 0.0007 0.0115 0.313 0,1729 0.1830 0,1464 TEH 0412456 0,5345 DUP050453508 VARIABLE OF L0C12 L0C13 LOClb (.CC19 IDC20 LOCHi L0C22 U0C23 10C24 1.0C25 . (.0C27 L0C28 LOC50 L0C31 L0C32 LDC53 toe 3 a L0CI5 LDC36 L0C37 L0C36 LOCtti i,OC a 2 l L0C43 i.OC (.OCAS l tocae L0C49 LCCSl 1 10CS2 iOCSJ L0C5a 1 L0C55 IDC57 tOC59 - f IOC 60 tOC 6 3 IDC64 t Rl ft? l< R3 1 1 I 2 1 1 1 1 0 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 i l l l 2 2 2 2 2 2 2 2 2 2 2 2 2 p a r a me t e r e s t ima t e 4,084241 6.230369 6,702734 1.339098 6,257740 8.715519 -1.174121 0.222882 0 -3.558478 -1.270224 5,079406 0.304463 4.567965 2.444392 2.236172 1.089232 5.317464 4,604197 1.087214 0.664086 5.20307b 6,213957 1,679657 1.463585 2.338762 2.219954 1.217420 C.119289 2.135376 1.428221 3.692559 3.986783 6.959078 0.551986 5,299429 4.966930 0.100974 2.166997 0.335719 2.238645 c c STANDARD T FOR HO* ERROR PARA*ETR*0 2.160339 2,155754 2,772034 2.350473 1.916880 1.943633 1.791030 2.061716 1.971832 1,602606 2.477977 2,247313 2.223667 2.040891 1,964217 2,207641 1,667560 1,760865 1.977653 1,917515 2.196839 2.5So70l 2,ll607 2.142136 3.144179 2.330605 1.4*8952 1.975348 1.989576 2.270144 2,023713 1,848118 1.606496 1.976910 1.923476 2.3094B6 2.129802 1,280904 1,644273 1,024385 -1.691 2.690 3.782 0.570 3.261 4.484 -0,656 -0.103 -1.805 0.649 3.437 -0.135 2,063 1,198 1,086 0,493 3,151 2,615 0.550 0,346 2.366 2.436 0.794 0.6*3 0.74 -0,953 -0,616 0,060 1.073 0,6?9 -1.825 2.156 -3.652 0.279 -2.755 2.151 0.047 1.692 0,204 2.185 S*S PROfi > IT 0.0586 0,0039 0.0002 0,5669 o.oon 0,0001 0.5122 0.9178 0.0713 0.51b3 0,0006 0.8922 0.0392 0,2312 0.2769 0,6218 0.001b 0,0090 0,5*25 0,7291 0.0160 0,014 0.4275 0,4945 0.4571 0.3409 0.4137 0.9519 0,2e33 0.5293 0,0682 0.0310 0.0001 0.7801 0.0059 0.0316 0,9622 0.090* 0.8382 0.0290 * TEH 0412457 DUP050453509 TABLE 6 REGRESSION RESULTS SUBCATEGORIZING ON 64 PSUa, REGION, AND RURAL TEH 0412458 DUP050453510 N 30991.07 DEP VARIABLE* SF2SYS SOURCE OF SUM OF sq uar es me a n SQUARE F VALUE PROB>F M^DEL 67 ERROR 2183 C Tu TAL 2250 09929709*6 12299430627 17292005613 70521955 5630169 13.227 0.0001 MOOT **SE OFF EAN C.V, 2373.645 128.265 1850,573 R5LUAR AOJ K-SG 0.2887 0.2669 NC-TEl MODEL IS NOT FULL RANK. LEAST SQUARES SOLUTIONS FOR THE PA*.TKS aw e NOT UNIUUE. SUE STATISTICS WJLL BE mis l e a d in g , a wEPuwreo o f c f o o r 8 MEANS THAT THE ESTIMATE IS BIASED. THE FOLLPMIMC PARAMETERS MAVE SEEN SET TO 0 , SINCE Th E VARIABLES ARE A LINEAR COMBINATION Qf OTHER VARIABLES AS S''0*N, L0C7 L0C1 L0C20 LOC29 LPC39 IOC0 LPC7 tOC50 LQC56 L0C61 toeHi a a a a a a a a VAKXaatg OF PARAMETER ESTIMATE STANDARD T FOP HO: ERROR PARa *ETERQ IMER CEP" AGEVRS AGESU 8** I U>tiZXC ALOSU CLCI c o f f ee KJ R2 3 FAr.rtYP3 ........ .. HYWR"EO LI LOCI Lr1CZ LnC3 LPC L.CS 1 65.370153 13.580557 1 0.061025 0.156217 1 0,009205219 0.001713552 1 1.202552 0.088476 t R.0262C2 2,173828 1 0.00915O2S2 0.001321977 1 , 00000231063 0,0000011692 l p .370831 0.105570 ! 2.000295 1.673263 1 5,010009 1.533798 I 0,999802 1.346131 I 3.051307 0,690157 J 17,094703 0.975816 l 8.807996 1,277676 t 5.278603 0,992308 1 0.433160 2,066832 l 3.060360 2,978472 l 3,924669 3.312823 1 3.901661 2,436408 I 1.306356 2.839380 4.812 -2.950 5,372 10,040 3,692 6,925 t.9#7 -3.551 1.034 3.269 0.743 4,396 1,436 6.894 5.319 1.797 1.036 1.185 1.60 0 0.060 PROM > IT1 0,0001 0,0032 0.0001 0,0001 0.0002 0.0001 6,0517 0,0004 0,1516 0,0011 0.0577 0.0001 0.0006 0,0001 0,0001 0.0725 0,3005 0.2363 1,1097 0,6455 TEH 0412459 DUP050453511 VARIABLE or ICC6 L0C7 toC*' LOCQ l o c io LOCI 2 LOC13 L0C14 LOC15 LOCI 6 LPCI? LOC IB ICC *9 LOC20 Lr-C21 LOC22 Lf'C2J L0C24 LOC25 Lf>C2o LOC27 L'iCPB LrC29 LOC30 LCCM Lnesa LPC33 LOC3 Lnt,5 Lf'CSo LUC37 LOC 3 8 inC39 Lf'C^O LOCl LOC2 LOC a 3 LuC a a LOC as Locuo Lu C"7 LSC"8 LOCaO LC-CSO L=CSi LOC52 LCC53 LC-C54 LOC55 LOCSe LOCS7 LOCSB LOC59 l o 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 t 1 1 1 i 1 1 _1 0 0 1 1 1 1 1 1 0 1 1 0 1 1 1 1 1 0 1 1 1 Pa k a h ETER ESTIMATE I.9l24 0 7.83*022 1.886110 1.042938 2.216595 2,271771 1.852487 0 -0.778497 -0,179709 -7,613019 -1,007999 -3.638382 -7.042641 6.625518 >4,(>63508 --5,0682*3 0 9.964691 22.502340 7.440641 -0.aib829 0 1.170762 -9.575931 7,564178 -3.254559 0.765196 *>.563207 3.327703 -6.562144 -6.623244 0 0 3.217319 4.U1935 -13.491988 -1,529830 -1.335794 6.405042 0 5,639656 -6.379495 0 2.778660 2.934462 -10.054094 -7,697902 -8.337394 0 -7.537152 1.944363 6,219761 STANDARD T FOR HOt ERROR Pa Kime t ERb O 3.652679 4.912750 3.354133 2.594115 2.958188 3.217297 2.763384 10.553056 4,722347 8.370221 6,704026 3,586385 3,503316 2,954145 2.767758 6,200153 3.03u200 11.343460 2.84*732 2.580871 3.779271 2.894487 3,143454 2.951402 3.415660 3,410091 3.062170 3,44259 3.3*3090 3.543578 M,762e65 2,957867 2.765769 4.547289 6,486205 4,203918 3.566426 2,991757 3,018698 3,477534 3.1S61S8 3,534064 2,995358 3.690216 3.240003 0.389 1.595 0.556 0.402 -0,749 0.706 0.670 0,074 -0.038 0.910 0,150 1.014 -2,010 2.3)0 1.612 i2o7 -0.319 1.984 -2,614 -0.If2 0,310 -3.308 -2.438 -1.1C3 0.224 2,810 1.087 -1,906 1,969 -0,908 0,863 4,561 -0.553 -9,294 0.987 1.334 -1.769 0.929 0,939 -2.891 -2.439 -2,359 -2.516 -0.2*8 -1,920 S AS PROP > ITI 0.6974 0.1108 0,5780 0.6877 0,4538 0.4802 0.5027 0.9412 0,9696 0.3632 0.8805 0.3105 0,0445 0,0210 0.1072 0.2277 0,7496 0,0474 0,0099 0.6717 0,7568 o.oom 0,0148 0.2703 0,8227 0,0050 0.2773 0,0566 0.0490 0,3640 0.3680 o.ooot 0.5802 0.7690 0,3235 0.1823 0.0738 0.3531 0.3479 0,0039 0.0148 0.0164 ''0 0.0550 ltM <M1Z46Q DUP050453512 \. VARIABLE OP LOC60 LOCfcl LHC62 LHC63 LOCfeM 1 0 0 1 1 PARAMETER e s t ima t e -11.259665 0 0 2.666219 1.702663 STANDARD T FOR HOi e r r o r PARAMETERS 3.290249 3.061715 4.420435 3.013 oI*36 0.395 SAS PROP > tTI 0.0066 0*3093 0.7001 TEH 0412461 DUP050453513 DgP VARIABLES BP2DJA SCARCE DF SUM OF SQUARES MEAN SQUARE F VALUE PRQ9>F 0DEL 75 E*KQH 2175 C TOTAL 2250 2447B28728 491209908b 735992b81 32637716 2258436 14.451 0,0001 KCOT SE OEP MIAN C.v. 1502.809 82.34013b 1825.124 R-SUUARE ADJ R-SQ 0,3326 0.3096 MOTES MODEL IS NOT PULL RAM* . LEAST SQUARES SOLUTIONS FOR THE Pi*4ffTHS ARE SQT Un Iu ME. SOmE STATISTICS *ILL 8E mis l e a d in g , a r e p o r t e d OF OF 0 OR B MEANS THAT t h e e s t ima t e IS BIASED, TH FOLLOIHG p a r a me t e r s h a v e b e e n SET TO 0 , SI*CE THE VARIABLES ARE A LINEAR COMBINATION F OTHER v a r ia b l e s a s Sh o w n , L0C7 L0C14 LOC2a LDC29 LOG59 Lncao tr>L'<7 L0L50 LnCSb L'CM LOCa2 a a * a a * a a a - VARIABLE DF PARAMETER ESTIMATE STANDARD T FQ HO I ERROR PARAMETERaO X**TRC ACEYRS AC*S ftM-J LOG7JMC LlilV|TC ALbStt LSC^OL SyCtNiJM CALCI'tn DIETVITC ZtNC h imo g i yo twicep m k? as f A>4yP;? C'*fPEE HEIGmT 1 95.76S610 32.159641 1 0.867135 0.103272 1 0.0081418b 0 .001117208 1 0.7392*5 0.U75043 1 35.899095 8.928973 1 1.081755 0.267497 1 0,004328288 0, 0OC8617418 1 2.329990 1.114731 1 0.0431O5 0,016691 * 0000014R894 7 .53849E-07 1 0.010122 0 .002697159 1 0.332334 0,095999 1 0.112025 0,022033 1 0,012494 0,00407175 1 0,330441 1.065643 1 0.008171034 0.977594 1 0.747141 0.860097 1 2.016104 0.439866 I 0,223805 0,069604 1 13.083942 3,186144 2.978 8,397 7.288 9,851 -4.021 4,044 5,020 2.539 2.583 1.988 -3,753 3,462 5.084 3.069 0,310 0,008 0,869 4.583 -3,215 4,107 PRO*1 > t T1 0.0029 0.0001 0.0001 o.ocoi 0,0001 0,0001 0,0001 0,0112 0,0099 0,0469 0.0002 0,0005 0,0001 0.0022 0,7565 0,933 0,J51 0,0001 0,0013 0.0001 TEH 0412462 DUP050453514 \ VAHIAPLE Df p a r a me t e r ESTIMATE MY R*CE LL L$L1 1.9C2 LOCI tec* L0C5 IOC 6 LCC7 LOC.9 LPC9 L-'CIO LOCU LCl2 L?C13 LPC14 LCC15 L^C 17 LoC18 LOCI 9 LHC20 L0C21 LOC22 L"C?3 LrC24 LnC25 LUZ LOC27 LiK? 8 LnC?9 Lr|C3u L^CSl LrtC32 LOC3J L0C34 LOC35 L<*C36 LOG 37 LOC39 LOG 59 LOCttW LOCttj tnc2 LOC43 L"C45 LOG 6 LOG *7 tncua L0C9 L-USO LOCSt 1 5.R77315 1 2.381505 1 2.6543S 1 3,356863 1 0.574513 1 0.240291 1 fl.776787 1 4.192329 1 3,666911 00 1 2,434456 1 2.906957 1 1,467349 1 0.499996 1 4,681147 1 5.383479 00 1 2.36435 1 3.105707 1 6,460677 i 1.035347 j 0.231697 i 2.P49I05 i 10.064613 1 0.294026 1 0.009906102 00 1 2,057100 1 23.014455 1 2.706265 t 3.72C434 00 1 1.60x663 1 5,223668 -1 S,409674 1 -1.625269 i 0.005*43911 i 3.654107 l 2.9495*6 j 2,666316 l 1.383630 00 00 1 3.430536 1 6.3673*8 1 1.053155 1 0,714494 I 1.690730 1 5,227510 00 1 0,4722*5 1 -1.2W9290 00 1 1.412911 st andar d ERROR " 0.810305 0.673612 0.660066 1,572869 1.669662 2,115490 1.550275 1.816242 2.449688 a 3.115066 2.127667 1,650909 1.874*63 2.049621 1,762746 6.689447 2,991905 5,3*3025 4.256952 2.302636 2,22346.6 1.876449 1.761979 2.667064 1.955411 7,199935 1,807997 1,673669 2,404821 1.846540 1.992452 1.956613 2.170516 2,168076 1.953181 2.191323 2.140374 2.255394 3.017909 1,880436 1.756426 2.913145 4,116614 a 2.669741 2.259735 a 1,912230 T FOR HO* PARAMETERS 6.760 2.726 4.021 2.134 0.304 -0.114 3.0*1 -2.308 -1,506 0.7*2 1.366 0.8*9 0.266 2.2*4 3.054 a 0,353 1.038 1.220 0,243 0,1*1 1.281 5,364 0,1*7 0,0*4 -1,052 3.197 -1,497 2,223 0,667 -2,829 2.715 -0.831 0,003 1.6*5 1.510 1.217 0,646 a a 1.525 2.110 0,560 0,407 0.5*0 1.270 a -0.177 -0.535 a 0,739 SAS PROP > 1T o.ooot 0.0065 0.0001 0,0329 0,7611 0.9396 0,0021 0,0211 0.1322 0.4346 0.1720 0.3742 0.7901 0.0225 0.0023 a 0.7238 0.2994 0,2227 0.8060 0.919* 0,2002 0,0001 0.8675 0.9970 a 0,2*29 0,0014 0,1346 0,0263 0.5046 0.0047 0.0067 0.4063 0.9979 0.0921 0.1312 0,223* 0,5181 0.1275 0.0350 0.5755 0,6*42 0,5617 0.2043 0 0.8591, 0*5926 0.4601 TEH 0412463 DUP050453515 \ VAKJALE UF t*C52 LCC53 tnC55 L0CS6 LftCS7 iucsa LOC59 L``CftO tCCfcl i.nc<>2 1.0C63 1 1 1 1 0 1 1 t 1 0 0 1 1 PARAMETER ESTIMATE 3.576612 2.8oS592 2.2925*6 5.086244 0 7,238970 1,792074 1,816415 4,975886 0 0 tt,240625 2,904017 STANDARO T FOR HO* ERROR PARAMETERS 1,924065 2,213778 2.001040 2,240254 1.910150 2,471952 2,056325 2,089323 1.944610 2.606842 1.899 1.267 1.146 2.270 3.790 0,705 0,683 2.392 2,100 1.035 SAS PROS > ITl 0.0632 0,2052 0.2520 0,0233 0.0002 0,4819 0,3772 0.0173 a 0.0293 0.3010 TEH 0412464 DUP050453516 TABLE 7 REGRESSION RESULTS SUBCATEG0RI2ING ON 64 PSUa Jt TEH 0412465 DUP050453517 N 30991.08 d e p v a r ia b l e t 8P2SVS SOURCE OF SUM OP SQUARES MEAN SQUARE F VALUE PROB>F mo d e l ERROR 73 2185 C TOTAL 2258 5700213835 11685^53872 17326167108 78633061 5329956 18.753 0.0001 ROOT MSB DEP MEAN C.V. 2308,670 128.270 1799,6 R-SGUANC ADJ R-S9 0.3302 0.3078 NOTE* MODEL IS NOT PULL RANK, LEAST SQUARES SOLUTIONS FOR THE PARAMETERS ARE MOT UNIQUE. SOME STATISTICS WILL BE mis l e a d in g , a r e p o r t e d OP OF 0 OR B MEANS THAT THE ESTIMATE IS BUSED. t h e p o l l o -in g p a r a me t e r s h a v e b e e n SET TO 0 * SIn CE Tug VARIABLES ARE A LINEAR COM8INATION OP OTHER v a r ia b l e s iIS SHOWN. Lnc? VARIAPLE OF Pa HamETER e s t ima t e STANDARD T FOR h o i e r r o r p a r a me t e r s IMTERCEP Ar,VH3 AOESQ 8*1 LOG2I-JC ALBSU CALCIUM / COFFEE Fa mh v PER h eig h t HYPERmED LL LOCI LUC2 LOCI LOCO L0C5 LnC6 LnCT LOC9 LOC LOCI 0 LOCli LOCI 2 LOCI 3 LOCt LOCI5 LOUP LOC17 Loue 1 67,801299 13.859680 I 0.565090 0.151336 1 0.010194 0.001662603 i 1.261290 0,065567 l -7,601723 2.135768 1 0,009a28836 0.001296598 1 0 .0000020816 .00000115875 l 0,318386 0.103200 1 2.812202 0,677257 1 16,612281 4.863655 l Rj Jl z t s a a l _ 1.239833 1 /''ftjoioss 0.990175 1 ^ 3.280839 l -0,938327 3.270483 1 3,801003 3.628726 1 5,589078 3.113671 l 3,358593 3.528196 l 8.535891 8,175169 i 13.068950 3,885059 \ 8.335066 3.208918 l 9, 1 82392 3,759337 l 3.720626 3,276576 l 11.015632 3,387383 l 5.035230 3,820641 i 10.655592 3,882878 l -7,626950 3.328632 i 7.882165 3,676317 I 2.109233 3.506463 i 7.58*225 3,288663 1 9.611263 4.003717 5,037 3,738 6,131 18,70 3.559 7.269 1.803 3,0*5 4,152 3.816 6.880 3.334 0.013 0,286 0.889 1.795 0,952 1.0*6 3.749 2.601 2.832 1,135 3,291 1.472 3,095 2,354 2.0P4 0,602 2,307 2,801 PROB > IT1 0,0001 0.0002 0.0001 0,0001 0,0008 0,0001 0,0716 0,0021 0,0001 0,0006 0,0009 0 0,7751 0.3740 0.0728 0.3412 0.2775 0.0002 0.0098 0.0151 0,2566 0.0010 0,1012 0.0020 0.0186 0,0031 0.5076 0,0211 0,0165 VARIABLE DF LPC19 tc>C20 L0C21 L0C22 L0C?3 10C24 LOC25 U?C26 Lnt27 LOC2S LPC29 LCC30 LOC31 LOC32 L0C33 L0C3<* L0C3S LDC3 LDC3? IDC38 L0C39 L0C40 LHCI L0C2 trc3 L0C4 L0Ca5 L0t6 unc7 L0C48 LOC9 LOC50 S.OC51 1.PC52 L0C53 LOC5 UC5S LOCSs LPC57 10C58 L0C59 L0C60 LOCbl LPC62 L0C6J 1 1 i l i 0 s t i i i l i i t i i i i i j i i s l i i i i i i i i i i -s i i i t i i i i l PARAMETER ESTIMATE 1.966037 -2.183103 13.145301 I.9fe2099 3.896323 0 9.915158 15.936734 3.715187 3.16"922 8,235630 4.646054 -1.369620 4.170954 0.930469 3.927950 14.6lTft9a 6.129764 0.698877 -3.244779 1.230625 -3.517102 5.768743 6.923611 -5,399054 6.896920 6.951604 6.060262 0.694005 -3.683191 --0.471809 9.58B839 8.533111 8.873071 3.753241 0.120315 2.527665 6.204024 *0.345002 4.660617 1,621616 -5.959308 3.103652 22,252666 11.040994 s t an d a r d T FOR HOI ERROR p a r a me t e r s 4.036879 3.481594 3.536705 3.297655 3.648046 3.502077 3.742411 3.462446 3.044671 3,49749? 3.565097 3,524119 3,27742} 3,654106 3,966418 3.447168 3.396033 3.748942 3.616096 4,208452 3.512601 3.525316 4,452663 3.39V578 3.448343 4.914404 3,663690 3,630037 4.215238 3.171166 3.419646 3.554428 3.576940 3.959169 3.642583 3,397001 3.56B551 3.331026 3.565541 3.707S42 3.633195 3.761297 3.772603 3.659777 0.487 0,627 3.717 0.595 1,068 1.372 4.125 -1,073 1,038 2,355 1.303 -0,389 1.273 0.25S 0.990 4,280 2.394 0.186 *0.897 0.285 -1,001 1.676 1.555 1.588 2.001 1,415 1.660 0,191 -0.874 *0.149 2.804 2.401 2.441 0,948 0,033 -0.744 1.739 0,104 1.307 0,491 1.640 0.825 5.898 3.017 Sts PROP > IT 1 0,6263 0,5307 0,0002 0.5519 0.2856 0,1702 0.0001 0.2834 0.2993 0.0186 0.1926 0,6975 0,2033 0,7990 0,3221 0,0001 0,0168 0.852! 0.3696 0.7684 0.3168 C.1019 0.1201 0.112a 0.0455 0.1573 0,0971 0,8464 0.3823 0,8817 0.0051 0,0164 0,0132 0.3432 0,9737 0,4569 0.0823 0,9175 0.1913 0.6232 0.1011 0,4097 0.0001 0.0026 TEH 0412467 DUP050453519 SAS DEP VARIABLE! 8P2MA SOURCE OF SUM OF SQUARES MEAN SQUARE F VALUE PRflR>F MODEL 61 ERROR 2171 C TOTAL 22 52 2767030997 *594591305 7361622362 30160877 2116346 16.161 0.0001 ROOT MSE DEP MEAN C.V. 1656.767 62.336660 1766.691 R-SQUARE AOJ *SQ 0.3759 0.3526 NOTE 1 MODEL IS NOT FULL RANK, LEAST SQUARES SOLUTIONS FOR THE PARAMETERS ARE NOT UNIQUE. SOME STATISTICS MLL BE MISLEADING. A REPORTED UP OP 0 OR B MEANS THAT THE ESTIMATE IS BIASED. THE FOLLOWING p a r a me t e r s h a v e BEEN SET 10 0 SINCE THE VARIABLES ar e A LINEAR COMBINATION OF o t h e r v a r ia b l e s a s s h o r n . L0C24 VARIABLE OF p a r a me t e r ESTIMATE STANDARD T FOR HOI ERROR PARAMETERS Jn TEKCEP AGEVR3 1 1 91.571616 0.795972 31.322787 0,09*042 A5SU 1 0.007*5637 0,001079794 8*1 1 6,79*919 0.072013 L"SZXNC 1 -33.7*9507 6.6809*2 L0GV1TC 1 1.166078 0,259028 al bsq 1 0,005160703 0,000*422091 LSCHOL 1 3.06620S 1.002216 SNDKNtJM 1 0.045059 0,016179 CALCIUM DIETVITC * 0000012*473 7.29U23E-07 1 0,010104 0.00261529 ZIn C - | 0.307324 0.093361 HEMOGLOB 1 0.103736 0.02ib72 TRICEPR 1 0.O1G531 0.003983413 fAMH^PgR 1 1.926219 0.427949 COFFEE 1 -0,176898 0,067543 h eig h t 1 12.012304 3,101684 HVPERHEO 1 5.447130 0.785475 RACE 1 2.313058 0.6*7916 LOCI 1 1.*65459 2.003514 LOC2 1 1.996755 2.089065 LC3 1 1.8712*2 2.43*629 LOCa 1 4,619544 1.982V62 LnC5 1 2.522967 2.255*00 LCC6 1 2.935037 2.652586 LOCT 1 3.087947 2.209733 LnC0 1 -0.472729 2,032064 L0C9 1 3.115892 2.350512 L0C10 1 1.634609 2.087562 LOC1I 1 1.199201 2.104089 2,923 7,9*8 6,904 10,7*0 -3.808 4,*06 6,126 2,833 -2.7*5 1,706 3,863 3.292 4.787 2.644 4,5*6 2,619 3,873 6.935 2.726 -0.895 0,956 0,769 2.331 -1.119 -1.107 1.397 -0,233 1.309 0,783 0,570 PROB > tT 1 0,0035 0,0001 0.0001 0.0001 0,0001 0,0001 0,0001 0.0047 0.005* 0.0881 0.0001 0.0010 0,0001 0.0063 0,0001 0,0009 0,0001 0,0001 0,006* 0.3707 0.3393 0.4*22 0.0199 0.263* 0.2685 0.1624 0.8161 0,1907 Pt.4337 0.5688 TEH 0412468 DUP050453520 v a r ia b l e OF PARAMETER ESTIMATE ST ABOARD T FOR HO 1 ERROR PARAMETERs O LPC12 toe 13 tOC14 t^ClS L0C16 ICC17 Lt'cie U&C19 LCC20 tnC21 L0C22 tOC23 L0C2O tOC 25 tOC26 tOC27 tOC 2 8 LOC29 tf-C30 tOC31 LOC52 tOC3 3 tnCI tocss toes* toe 37 tocse tOC19 t^coo tnci l o c 2 toe3 U'C04 tocos tncod tOC7 tocoa ttC09 tOC50 toes 1 toesi L0C53 Lnc5 tOC55 toes* tOCS 7 i (T* U'C59 tOCbO tOCbl toc*2 L0Ct>3 tl 1 -0.207915 2.180058 1 6.370782 2.186118 1 -9.065587 1 0.721035 2.110091 2,329536 1 -5.207493 2.220797 1 7.865067 2,063057 1 2.775200 2,530989 1 0.506580 2.571004 1 -0.578305 2.200171 1 10.575113 2.248004 i 0.08160b 2.121308 i 1.561203 2.335039 00 * l -1.686605 2.262252 l 8.387190 2.372910 l -1.570016 2.193282 i 0.b99S7b 1,904870 i 3.7b90l9 2.210999 l 0.6b?597 2.271761 l -0.607921 2.200998 l -2.523908 2,076083 l 0.297yuo 2.376168 l 0,635b77 2.512079 i 7.106006 2.172020 i 0.112652 2.162004 i 2,992610 2.378770 i -0.837167 2.301053 l 2.938789 2,6*1993 l -3.*35082 2,225110 i 5.279755 2.235608 l 5,963930 2.819057 l 1.579630 2.152129 l 1,366813 2.185779 l 2.262867 3.124520 l 2.326378 2.321068 - i 3.069200 2.299006 i -0.716505 2.677985 i 0.399083 2.025009 l 3.145093 2,169989 i 1.6706*6 2.259760 ! 3,983018 2.282290 1 3.360091 2.520301 1 1.9017*2 2.312267 1 2,006000 2,161631 1 3,065830 2,262999 1 -5.137598 2.110033 1 1.278072 2,262350 1 0,068179 2,300557 1 -5.855321 2.303563 1 -2.052673 2,386026 1 1.865107 2.366003 1 _iUA8A3-9L6_ v 2.345-7.5.1. 1 ' 1.007711 X 0.655606 ^ -1,927 2.914 0.077 2.027 -2.359 3.775 1.095 0.197 -2.077 0.700 0.227 0.669 0.706 3.535 0.716 2.017 1.7 02 -0.292 2,056 -1.215 -0,125 0.253 3,290 1,902 1,256 -0.360 -1.10 -1.720 2.362 2.123 0.710 0.625 0.720 1.002 1.300 0.268 0.197 1.050 0,739 1,705 1,335 0,622 1.132 -1,532 2.030 0,565 -0,029 -2,542 -0,860 0,782 2.111 2,1*7 3i$ PROB > 1T1 0,05020,0036 0.0001 0.0028 0.0180 0.0002 0,2737 0.8038 0.0379 0.0001 0.8204 0.5038 0,0560 0.0000 0.0701 0,0156 0.0289 0.7706 0,0399 0.2243 0,9005 0.8003 0,0010 0.0573 0,2085 0,7161 0.2*97 0,0609 0,0183 0,0339 0,0630 0.5318 0.0690 0.3163 0.1792 0.7891 0.803* 0.1073 0.4598 0.0611 0,1821 0,4109 0.2579 0,1258 0.0152 0.5722 0,9768 0,0111 0,3698 0,0305 0,0309 0.0319 TEH 0412469 DUP050453521