ICPSR General Structural Equation Models Week 4 # 3 Panel Data (including Growth Curve Models) Causal models: 1 Eta-1 Ksi-1 ga2,1 1 ga1,2 Ksi-2 Cross-lagged panel coefficients [Reduced form of model on next slide] Eta-2 Causal models: 1 Ksi-1 Eta-1 1 Ksi-2 Eta-2 Reciprocal effects, using lagged values to achieve model identification Causal models: A variant Issue: what does ga(1,1) mean given concern over causal direction? TV Use gamma 1,1 Political Trust gamma2,1 Beta 2,1Pol Trust Time 2 Lagged and contemporaneous effects This model is underidentified 1 1 Lagged effects model Ksi-1 could be an “event” 1/0 dummy variable ksi-1 ksi-2 eta-1 eta-2 First order model for three wave data (univariate) 1 1 1 Time 1 1 1 1 1 1 1 1 1 Time 2 Time 3 1 First order model for three wave data (univariate) 1 1 1 1 1 1 1 1 b1 Tests: 1 1 b1 Equivalent of stability coefficients (b1) Mean differences (see earlier slide) 1 1 Second order model for three wave data (univariate) 1 1 1 1 1 1 1 1 1 b1 No longer comparable to b1 (t1 t2) b1 1 1 1 Second order model for three wave data (univariate) 1 1 1 1 1 1 1 1 1 b1 b1 1 1 Issue: adding appropriate error terms (2nd order) 1 Multivariate Model for Three-wave panel data: cross-lagged effects (first order) 1 1 1 1 Multivariate Model for Three-wave panel data: cross-lagged effects (first order) 1 1 1 1 Equivalence of parameters: T1 T2 T2 T3 Multivariate Model for Three-wave panel data: cross-lagged effects (second order) Multivariate Model for Four-wave panel data: cross-lagged effects (second order) Lagged and contemporaneous effects Three wave model with constraints: 1 1 a a d e d f c e f c b b 1 Under many circumstances, there will be an empirical under-ident. problem, though in theory this model is identified 1 Example: • Canada, Quality of Life data • In directory \Panel in Week4Examples Panel Data model Model for attitudes about labour unions, 1977-1979 Items: 5-pt. agree/disagree 199D QD6B Unions too much power Q156C QK16F Scabs (gov’t prohibit strikebreakers) Q156D QK16G Workers on Boards Q156B QK16E Teachers should not have right to strike Source: Cdn. Quality of life panel study, 1977-1979 waves 1 Union atts 1979 Union Atts 1977 1 1 1 1 1 1 1 1 1 1 Panel Data model LISREL Estimates (Maximum Likelihood) LAMBDA-Y Q199D LABOR77 -------1.000 LABOR79 -------- - Q156C -1.803 (0.141) -12.796 - - Q156D -1.148 (0.101) -11.350 - - Q156B 0.789 (0.098) 8.040 - - QD7B - - 1.000 QK16F - - -1.352 (0.109) -12.355 QK16G - - -0.755 (0.072) -10.479 QK16E - - 0.709 (0.084) 8.427 Panel Data model BETA PSI Note: This matrix is diagonal. LABOR77 LABOR79 -------- -------0.125 -0.066 (0.017) (0.018) 7.529 -3.611 LABOR77 LABOR77 -------- - LABOR79 Squared Multiple Correlations for Structural Equations LABOR77 LABOR79 -------- --------1.356 W_A_R_N_I_N_G: PSI is not positive definite 1.420 (0.138) 10.318 LABOR79 -------- - - Completely Standardized Solution LAMBDA-Y Q199D Q156C Q156D Q156B QD7B QK16F QK16G QK16E LABOR77 -------0.425 -0.559 -0.436 0.262 - - - - - LABOR79 -------- - - - 0.409 -0.524 -0.382 0.277 BETA LABOR77 LABOR79 LABOR77 -------- 1.165 LABOR79 -------- - - Panel Data model What is the problem here? Panel Data model Theta-epsilon was specified as diagonal Modification Indices for THETA-EPS Q199D Q156C Q156D Q156B QD7B QK16F QK16G QK16E Q199D -------- 2.845 3.439 17.009 83.881 10.361 19.366 0.158 Q156C -------- Q156D -------- Q156B -------- QD7B -------- QK16F -------- - 20.324 5.334 42.939 108.940 28.336 7.133 - 13.004 10.988 28.775 141.658 14.031 - 4.108 23.541 5.494 169.430 - 2.034 0.242 25.246 - 7.172 6.019 Panel Data model 1 Union atts 1979 Union Atts 1977 1 1 1 1 1 1 1 1 1 1 Panel Data model Added error covariances: FR TE 5 1 TE 6 2 TE 7 3 TE 8 4 BETA LABOR77 LABOR79 LABOR77 -------- - LABOR79 -------- - 1.094 (0.115) 9.547 - - Covariance Matrix of ETA LABOR77 LABOR79 LABOR77 -------0.116 0.127 LABOR79 -------0.199 Panel Data model Added error covariances: FR TE 5 1 TE 6 2 TE 7 3 TE 8 4 PSI Note: This matrix is diagonal. LABOR77 -------0.116 (0.020) 5.935 LABOR79 -------0.060 (0.016) 3.721 Squared Multiple Correlations for Structural Equations LABOR77 -------- - LABOR79 -------0.698 Panel data model Cdn. Quality of Life 1977-81 ! Model for mean differences SY='H:\QOL3WAVE\imputed_data.dsf' SE Q199D Q156C Q156D Q156B QD7B QK16F QK16G QK16E / MO NY=8 NE=2 LY=FU,FI PS=SY,FR TE=SY BE=FU,FI TY=FR AL=FI LE LABOR77 LABOR79 VA 1.0 LY 1 1 LY 5 2 FR LY 2 1 LY 3 1 LY 4 1 FR LY 6 2 LY 7 2 LY 8 2 FR TE 5 1 TE 6 2 TE 7 3 TE 8 4 EQ TY 5 TY 1 EQ TY 6 TY 2 EQ TY 7 TY 3 EQ TY 8 TY 4 EQ LY 2 1 LY 6 2 EQ LY 3 1 LY 7 2 EQ LY 4 1 LY 8 2 FR AL 2 OU ME=ML MI SC ND=3 Panel Data model Alternative specification with stability coefficient: PS=SY BE=SD [or BE=FU,FI then FR BE 2 1] Panel Data ALPHA LABOR77 -------- - LABOR79 -------0.043 (0.014) 3.051 Higher score = pro-union (ref. indicator: too much/too little power… too little=5 too much=1 Panel Data Panel data model Cdn. Quality of Life 1977-81 ! Impact of TV newspapers on labor union attitudes SY='H:\QOL3WAVE\imputed_data.dsf' SE Q258 Q260 Q261 Q199D Q156C Q156D Q156B QD7B QK16F QK16G QK16E / MO NY=11 NE=4 LY=FU,FI PS=SY TE=SY BE=FU,FI LE 1 NEWSP TV LABOR77 LABOR79 TV VA 1.0 LY 2 1 1 VA 1.0 LY 3 2 1 Newsp 1 FR LY 1 1 FI TE 3 3 VA 1.0 LY 4 3 LY 8 4 Union Atts FR LY 5 3 LY 6 3 LY 7 3 1977 FR LY 9 4 LY 10 4 LY 11 4 1 FR BE 4 3 FR BE 3 2 BE 3 1 FR BE 4 2 BE 4 1 1 1 1 1 1 FR PS 2 1 FR TE 11 7 TE 10 6 TE 9 5 TE 8 4 OU ME=ML MI SC ND=3 1 Union atts 1979 1 1 1 1 Panel Data LISREL Estimates (Maximum Likelihood) LAMBDA-Y Q258 NEWSP -------0.917 (0.176) 5.212 TV -------- - LABOR77 -------- - LABOR79 -------- - Q260 1.000 - - - - - - Q261 - - 1.000 - - - - Q199D - - - - 1.000 - - Q156C - - - - -1.891 (0.214) -8.819 - - Panel Data BETA NEWSP TV NEWSP -------- - - TV -------- - - LABOR77 0.061 (0.026) 2.325 -0.005 (0.011) -0.406 LABOR79 0.047 (0.030) 1.584 -0.017 (0.014) -1.216 LABOR77 -------- - LABOR79 -------- - - - - - - - - - 1.081 (0.113) 9.564 - - Panel Data Panel data model Cdn. Quality of Life 1977-81 ! Impact of TV newspapers on labor union attitudes ! Controls: education sex union membership SY='H:\QOL3WAVE\imputed_data.dsf' SE Q258 Q260 Q261 Q199D Q156C Q156D Q156B QD7B QK16F QK16G QK16E Q63 SEX Q201 RAGE Q157/ MO NY=11 NE=4 LY=FU,FI PS=SY TE=SY BE=FU,FI NX=5 NK=5 FIXEDX LE NEWSP TV LABOR77 LABOR79 LK MEMBER SEX EDUC AGE INCOME VA 1.0 LY 2 1 VA 1.0 LY 3 2 FR LY 1 1 FI TE 3 3 VA 1.0 LY 4 3 LY 8 4 FR LY 5 3 LY 6 3 LY 7 3 FR LY 9 4 LY 10 4 LY 11 4 FR BE 4 3 FR BE 3 2 BE 3 1 FR BE 4 2 BE 4 1 FR PS 2 1 FR TE 11 7 TE 10 6 TE 9 5 TE 8 4 OU ME=ML MI SC ND=3 Panel Data BETA NEWSP NEWSP -------- - TV - - TV -------- - - LABOR77 -------- - LABOR79 -------- - - - - - - - - - LABOR77 -0.025 (0.034) -0.738 -0.012 (0.011) -1.157 LABOR79 0.068 (0.042) 1.622 -0.010 (0.013) -0.751 1.033 (0.115) 8.970 MEMBER --------0.017 (0.039) -0.422 SEX -------0.011 (0.035) 0.311 EDUC --------0.097 (0.009) -11.303 AGE --------0.014 (0.001) -13.496 INCOME --------0.014 (0.005) -2.898 TV -0.013 (0.070) -0.182 -0.150 (0.062) -2.408 0.025 (0.015) 1.685 -0.017 (0.002) -9.807 0.001 (0.009) 0.113 LABOR77 0.286 (0.036) 7.880 -0.056 (0.026) -2.131 -0.039 (0.008) -5.158 -0.005 (0.001) -5.331 -0.010 (0.004) -2.557 LABOR79 0.045 (0.042) 1.082 0.114 (0.033) 3.487 0.001 (0.009) 0.069 0.001 (0.001) 0.966 -0.006 (0.004) -1.436 - - GAMMA NEWSP Another model (panel7) BETA INEQ77 LABOR77 INEQ77 -------- - - LABOR77 -------- - - INEQ79 -------- - LABOR79 -------- - - - - - INEQ79 0.704 (0.069) 10.214 0.012 (0.110) 0.105 - - - - LABOR79 -0.106 (0.044) -2.400 0.819 (0.124) 6.622 - - - - Re-expressing parameters: GROWTH CURVE MODELS Intercept & linear (& sometimes quadratic) terms • Suitable for panel models with >2 waves • Best for panel models with >3 waves Linear Growth Model Two Factor LGM LISREL: Parm1, Intercept Parm2, Slope 1 0 1 2 manifest variable, 2 latent variable model LY matrix 1 0 0 V1 - t1 V2 - t2 1 0, 0 1 0, 0 INT Slope V1 1 0 V2 1 1 TE matrix = elements equal TY zero PS matrix = SY,FR AL free (“parm1” and “parm2” above) (parm1 in model = variance of INT, parm2 = variance of Slope) Linear Growth Model Two Factor LGM Parm1, Intercept Parm2, Slope 1 0 1 1 0 0 V1 - t1 V2 - t2 1 0, 0 1 0, 0 Interpretation: • intercept factor represents initial status •Slope factor represents difference scores (V2-V1) With single indicators, cannot estimate error variances (as with any single indicator SEM model) Parm1 = mean intercept Parm2 = mean slope value Linear Growth Model Two Factor LGM E.g., TV use, adolescents, hours/day Parm1, Intercept Parm2, Slope 1 0 1 Parm1 = 2.5 Parm2 = 1.0 Increase of 1 hour/day from t1 to t2 1 0 0 V1 - t1 V2 - t2 1 0, 0 1 0, 0 We will also get variances for the Intercept and the Slope factors Parm1 = mean intercept Parm2 = mean slope value Some growth curve trajectories: • Parallel stability Some growth curve trajectories: • Strict stability Single-factor LGM •Actually nested within 2 factor model • take 2 factor model, intercept with 0 mean and 0 variance or strictly proportional to slope Curve 1 V1 B1 V2 B1 V3 (can estimate var(e1),(e2),(e3) if we impose constraint v(e1)=v(e2)=v(e3) ) Not generally the best model unless assumptions met: (cf. Duncan et al. p. 31: when rank ordering of individuals does not vary across time despite mean level changes) Linear Growth Model Two Factor LGM Parm1, Parm2, Intercept Slope 1 1 1 0 0 LV-t2 LV-t1 1 1 1 0, 0 A bit more complicated with latent variables instead of single manifest variables 1 0, 1 0, 1 1 1 0, 0, 0, … but the same basic principle. Linear Growth Model Two Factor Linear Growth Model Parm1, Parm2, Intercept LY matrix (LISREL) Slope 1 0 1 1 2 1 0 t1 1 0, 0 t2 1 0, 0 Int Slope V1 1 0 V2 1 1 V3 1 2 t3 1 0, *general test: vs. “unspecified growth model” Same principle would apply to k time points where k>3 More time points: test of linearity of “growth” (changes in mean)* Unspecified 2 factor Growth Curve Model Two Factor Unspecified Growth Model Parm1, Parm2, Intercept Slope 1 0 1 1 lambda 1 0 t1 1 0, 0 t2 1 0, 1 free lambda parameter in LY matrix 0 t3 1 0, In k time-point model, all but first 2 time points are represented by free parameters 3 factor Growth Curve Model Parm1, Intercept Parm2, Linear 1 0 1 1 Parm 3 0, Quadratic 0 2 1 1 0 t1 1 0, Non-linear growth 0 t2 1 0, 0 t3 1 0, 4 3 factor Growth Curve Model Parm1, Intercept Parm2, Linear 1 0 1 1 0, parm3 1 1 0, INT LIN Quad 1 0 t1 Quadratic 0 2 0 t2 1 0, LY matrix 4 V1 1 0 0 V2 1 1 2 V3 1 2 4 0 t3 1 0, TE is constrained to equality across t’s This is a “saturated” model (perfect fit by definition) PS is free AL is free (parm1-3) All TY elements 0 Examples: Z:\baer\Week4Examples\LatentGrowth Single variable models: LGMProg1.ls8 (output=.out) intercept model LGMProg2.ls8 - single factor curve model LGMProg3.ls8 - intercept + slope LGMProg4.ls8 – intercept + slope + quadratic Where do “growth factors” fit into models? • Examination of predictors (antecedents) and consequences of change Two Factor Linear Growth Model Parm1, Note: Intercept-slope covariance now disturbance covariance Parm2, Intercept Slope 1 0 1 1 2 1 0 t1 PROGRAM LGMProg5 1 0, 0 t2 1 0, 0 t3 1 0, Consequences Two Factor Linear Growth Model Parm1, Parm2, Intercept Slope 1 0 1 1 2 1 0 t1 1 0, 0 t2 1 0, 0 t3 1 0, Model LGMProg6.ls8 Dependent variable: job satisfaction, wave 8. Multiple indicators for the variable(s) involved in growth curves • “factor of curves” LGM • Intercept term and slope term (e.g.) constructed for each indicator • if there are 3 variables & 4 waves, we will have an intercept term based on 4 manifest variables representing time x 3 manifest variables per time (3 intercept terms) “common intercept” variable will have 3 indicators (intercept terms) “common slope” will have 3 indicators (slope terms) common intercept 1 x1-intercept 1 1 lambda2 lambda3 x2-Intercept x3-intercept 111 1 1 x1-t3 x2-t3 x3-t3 x1-t1 x2-t1 x3-t1 x1-t2 x2-t2 x3-t2 1 1 1 1 1 1 1 1 1 Error variances now estimated (not constrained to equality).. Could include corr. Errors too common intercept 1 lambda3 lambda2 1 1 x2-Intercept x1-intercept 1 1 x3-intercept 1 1 1 x1-t2 x2-t2 x3-t2 x1-t3 x2-t3 x3-t3 x1-t1 x2-t1 x3-t1 0 1 10 2 x2-slope x1-slope 2 1 1 1 1 x3-slope lambda3a lambda2a common slope 1 Interactions Easiest case: X1 is 0/1 X2 ix 0/1 Options: 1. Manually construct X3=X1*X2 outside SEM software, estimate model with X1,X2,X3 exogenous. Test for interaction: fix regression coefficient for X3 to 0. 2. Create two groups: X1=0 and X1=1. In each group, X2 as exogenous variable. Test for interaction would be H0: gamma[1] = gamma[2]. Extensions for X1, X2 >2 categories straightfoward (more groups/dummy variables) Interactions Option 3: Model as a 4-group problem. X1 1 0 X2 1 gr1 gr2 0 gr3 gr4 AL[1]=0 al[2], al[3],al[4] parameters to be estimated. Main effects model (no interaction) would allow for al[2]≠al[3] ≠al[4] but pattern of differences would be constrained such that….. Interactions Model as a 4-group problem. X1 1 X2 1 gr1 0 gr3 0 gr2 gr4 AL[1]=0 al[2], al[3],al[4] parameters to be estimated. Main effects model (no interaction) would allow for al[2]≠al[3] ≠al[4] but pattern of differences would be constrained such that….. The group1 vs. group 2 difference = group 3 vs. group 4 difference (or group 1 vs. 3 difference = group 2 vs. group 4). Programming in LISREL would be: Al[1] – Al[2] = al[3]- al[4] 0 – al[2] = al[3] – al[4] Al[2] = al[4]-al[3] LISREL: CO al 2 1 = al 4 1 – al 3 1 Test for interaction: run another model removing this constraint (all AL completely free except group 1) … more examples provided in text Interactions Interactions involving continuous variables. Case 1: One continuous (single or multiple indicator) and one categorical variable EASY: categorical variable becomes basis for grouping. Group 1 Eta = gamma[1] Ksi + zeta Group 2 Eta = gamma[2] Ksi + zeta Test for interaction: H0: gamma[1] = gamma[2] Case 2: Two continuous single indicator variables Also somewhat straightforward: Create single-indicator X3 = X2*X1 Case 3: Two continuous multiple indicator latent variables This is not so easy! Substantial literature on this question See course outline for extended list. (Schumacker and Mracoulides, eds., Interaction and Nonlinear Effects in Structural Equation Modeling). Case 3A, not talked about much: X1 single indicator Ksi1 (X2, X3,X4) Create: X1X2 , X1X3, X1,X4 Latent variable interactions Major approaches: • Kenny-Judd • Simplified variants of Kenny-Judd, modifications, etc. (Joreskog & Yang, 1996; Ping) • Two-stage least squares (get instrumental variables) • Use SEM to estimate 2 factor model, save latent variable “scores” (analogous to factor scores), then use these Latent variable interactions • Use SEM to estimate 2 factor model, save latent variable “scores” (analogous to factor scores), then use these In LISREL: Mo nx=6 nk=2 lx=fu,fi ph-sy,fr td=sy Va 1.0 lx 1 1 lx 4 2 Fr lx 2 1 lx 3 1 lx 5 2 lx 6 2 PS=Newfile.psf OU Latent variable interactions • Use SEM to estimate 2 factor model, save latent variable “scores” (analogous to factor scores), then use these In LISREL: Mo nx=6 nk=2 lx=fu,fi ph-sy,fr td=sy Va 1.0 lx 1 1 lx 4 2 Fr lx 2 1 lx 3 1 lx 5 2 lx 6 2 PS=Newfile.psf OU LISREL documentation suggests that a simple regression can be estimated in PRELIS: Sy=newfile.psf ne inter=ksi1*ksi2 rg y on ksi1 ksi2 ksi1 ksi2 ou Latent variable interactions LISREL documentation suggests that a simple regression can be estimated in PRELIS: Sy=newfile.psf ne inter=ksi1*ksi2 rg y on ksi1 ksi2 ksi1 ksi2 ou …. But it should also be possible to a) construct “inter” (=ksi1*ksi2) and read the 3 new “single indicator” variables back into LISREL for use with other variables (including those which form the basis of multiple-indicator endogenous variables. If all else fails, construct a LISREL model for Ksi1, Ksi2, and put FS (factor score regressions) on the OU line: OU ME=ML FS MI ND=4 .. And use factor score regressions to compute estimated factor scores in any stat package (incl. PRELIS) Example: INTERACTION MODEL WITH INTERACTION TERM CREATED EXTERNALLY SINGLE INDICATORS FOR EXOGENOUS LVS INVOLVED IN INTERACTION DA NO=1111 NI=10 MA=CM CM FI=G:\ICPSR\INTERACTIONS\INT5b.COV FU FO (10F10.7) LABELS lv1 lv2 interact sex race v217 v216 v125 v127 v130 se 8 9 10 1 2 3 4 5 6 7/ mo ny=3 ne=1 LY=FU,FI PS=SY,FR TE=SY c nx=7 nk=7 fixedx ga=fu,fr va 1.0 ly 1 1 fr ly 2 1 ly 3 1 ou me=ml se tv mi sc LISREL Estimates (Maximum Likelihood) LAMBDA-Y v125 ETA 1 -------1.00 v127 1.34 (0.24) 5.59 v130 0.65 (0.11) 5.74 Example: Dep var = inequality att’s (high score “more individual effort”) GAMMA ETA 1 lv1 --------0.04 (0.06) -0.65 lv2 --------0.21 (0.08) -2.57 interact -------0.85 (0.45) 1.89 sex -------0.22 (0.11) 2.10 race --------0.30 (0.13) -2.27 GAMMA ETA 1 v216 -------0.09 (0.03) 2.92 Lv1=relig. Lv2=econ. status v217 -------0.05 (0.03) 1.75 Kenny-Judd model Typically, literature (e.g., Kenny-Judd, 1984; Hayduk, 1987) starts with 2indicator example (2 LV’s each with 2 indicators). Ksi1 Ksi2 Ksi1*Ksi2 (interaction term) Indicators: Ksi1: x1 x2 Ksi2: x3 x4 Possible product terms: x1*x3 x1*x4 x2*X3 X2*x4 Kenny-Judd model use 4 product terms but Joreskog and Yang show that the model can be constructed with 1 product term. Kenny-Judd model Typically, literature (e.g., Kenny-Judd, 1984; Hayduk, 1987) starts with 2-indicator example (2 LV’s each with 2 indicators). Ksi1 Ksi2 Ksi1*Ksi2 (interaction term) Indicators: Ksi1: x1 x2 Ksi2: x3 x4 Possible product terms: x1*x3 x1*x4 x2*X3 X2*x4 Kenny-Judd model use 4 product terms but Joreskog and Yang show that the model can be constructed with 1 product term. Kenny-Judd do not include constant intercept terms (alpha, tau).. But even if dependent variable, Ksi1, Ksi2 and zeta have zero means, alpha will still be nonzero. - means of observed variables functions of other parameters in the model and therefore intercept terms have to be included. - Nonnormality even if x’s are normal (ADF estimation often recommended if sample size acceptable) Kenny-Judd model Kenny-Judd model alpha=1 term Kenny-Judd model, mod. INTERACTION MODEL KENNY JUDD MODIFICATION (JORESKOG AND YANG) ONE INTERACTION INDICATOR 3 INDICATORS PER L.V. DA NO=1111 NI=22 CM FI=G:\ICPSR2000\INTERACTIONS\INT5c.COV FU FO (22F20.11) ME FI=G:\ICPSR2000\INTERACTIONS\INT5C.MN FO (22F20.11) LABELS v181 v9 v190 v221 v226 v227 relinc1 relinc2 relinc3 relinc4 relinc5 relinc6 relinc7 relinc8 reling9 sex race v217 v216 v125 v127 v130 se 20 21 22 1 2 3 4 5 6 9 16 17 18 19/ mo ny=3 ne=1 NX=11 NK=7 LY=FU,FI PS=SY,FR C TE=SY TX=FR KA=FI C LX=FU,FI GA=FU,FR PH=SY,FR TD=SY AL=FI TY=FR va 1.0 ly 1 1 fr ly 2 1 ly 3 1 FI PH 3 1 PH 3 2 FR KA 3 VA 1.0 LX 1 1 LX 4 2 LX 7 3 LX 8 4 LX 9 5 LX 10 6 LX 11 7 FR TD 1 1 TD 2 2 TD 3 3 TD 4 4 TD 5 5 TD 6 6 TD 7 7 FR LX 2 1 LX 3 1 LX 5 2 LX 6 2 LX 7 1 LX 7 2 CO LX(7,1)=TX(1) CO LX(7,2)=TX(4) CO KA(3) = PH(2,1) FI PH 3 1 PH 3 2 CO PH(3,3) = PH(1,1)*PH(2,2) + PH(2,1)**2 CO TX(6) = TX(1)*TX(4) FI TD(8,8) TD(9,9) TD(10,10) TD(11,11) CO TD(7,7) = TX(1)**2*TD(3,3) + TX(4)**2*TD(1,1) + PH(1,1)*TX(4) + C PH(2,2)*TX(1) + TD(1,1)*TD(4,4) OU ME=ML SE TV ND=3 AD=off Kenny-Judd model, modified Joreskog/Yang Parameter Specifications LAMBDA-Y ETA 1 -------0 1 2 v125 v127 v130 LAMBDA-X v181 v9 v190 v221 v226 v227 relinc3 sex race v217 v216 KSI 1 -------0 3 4 0 0 0 Constr'd 0 0 0 0 KSI 2 -------0 0 0 0 5 6 Constr'd 0 0 0 0 KSI 3 -------0 0 0 0 0 0 0 0 0 0 0 KSI 4 -------0 0 0 0 0 0 0 0 0 0 0 KSI 5 -------0 0 0 0 0 0 0 0 0 0 0 KSI 6 -------0 0 0 0 0 0 0 0 0 0 0 Kenny-Judd model, modified Joreskog/Yang GAMMA ETA 1 KSI 1 --------0.023 (0.009) -2.557 GAMMA ETA 1 KSI 7 -------0.080 (0.029) 2.735 KSI 2 --------0.003 (0.015) -0.198 KSI 3 --------0.008 (0.004) -1.984 KSI 4 -------0.209 (0.098) 2.130 KSI 5 --------0.324 (0.125) -2.593 KSI 6 -------0.051 (0.024) 2.094
© Copyright 2026 Paperzz