Mplus VERSION 6.1 MUTHEN & MUTHEN 10/16/2010 5:07 AM INPUT INSTRUCTIONS Data: File is all_wide.dat ; Variable: Names are patno age count pritreat tx race gender stage icda mets lnode pmryt survcns surv opfscns opfs pfscns pfs ttopdcns ttopd ttpdcns ttpd tttfcns tttf ttprcns ttpr kps_0 kps_1 kps_2 kps_3 kps_4 kps_5 kps_6 kps_7 kps_8 kps_9 kps_10 kps_11 kps_12 anrx_0 anrx_1 anrx_2 anrx_3 anrx_4 anrx_5 anrx_6 anrx_7 anrx_8 anrx_9 anrx_10 anrx_11 anrx_12 ftg_0 ftg_1 ftg_2 ftg_3 ftg_4 ftg_5 ftg_6 ftg_7 ftg_8 ftg_9 ftg_10 ftg_11 ftg_12 cgh_0 cgh_1 cgh_2 cgh_3 cgh_4 cgh_5 cgh_6 cgh_7 cgh_8 cgh_9 cgh_10 cgh_11 cgh_12 dysp_0 dysp_1 dysp_2 dysp_3 dysp_4 dysp_5 dysp_6 dysp_7 dysp_8 dysp_9 dysp_10 dysp_11 dysp_12 hmpty_0 hmpty_1 hmpty_2 hmpty_3 hmpty_4 hmpty_5 hmpty_6 hmpty_7 hmpty_8 hmpty_9 hmpty_10 hmpty_11 hmpty_12 pain_0 pain_1 pain_2 pain_3 pain_4 pain_5 pain_6 pain_7 pain_8 pain_9 pain_10 pain_11 pain_12 sx_0 sx_1 sx_2 sx_3 sx_4 sx_5 sx_6 sx_7 sx_8 sx_9 sx_10 sx_11 sx_12 intfr_0 intfr_1 intfr_2 intfr_3 intfr_4 intfr_5 intfr_6 intfr_7 intfr_8 intfr_9 intfr_10 intfr_11 intfr_12 qol_0 qol_1 qol_2 qol_3 qol_4 qol_5 qol_6 qol_7 qol_8 qol_9 qol_10 qol_11 qol_12; Missing are all (-9999) ; usev = pfs1-pfs9 c1-c9 group x; survival = pfs1(all) pfs2(all) pfs3(all) pfs4(all) pfs5(all) pfs6(all) pfs7(all) pfs8(all) pfs9(all); timecensored = c1 c2 c3 c4 c5 c6 c7 c8 c9 (0 = not 1 = right); categorical = group; classes = cg(2); ! use the "known class" approach to capture the treatment and control groups Define: if (pfs>=0.7) then pfs1=0.7; ! Visit 1 if (pfs>=0.7) then c1=1; if (pfs<0.7) then pfs1=pfs; if (pfs<0.7) then c1=pfscns; if (pfs>=1.4) then pfs2=0.7; ! Visit 2 if (pfs>=1.4) then c2=1; if (pfs<1.4) then pfs2=pfs-0.7; if (pfs<1.4) then c2=pfscns; if (pfs<0.7) then pfs2=_missing; if (pfs<0.7) then c2=_missing; if (pfs>=2.1) then pfs3=0.7; if (pfs>=2.1) then c3=1; if (pfs<2.1) then pfs3=pfs-1.4; if (pfs<2.1) then c3=pfscns; if (pfs<1.4) then pfs3=_missing; if (pfs<1.4) then c3=_missing; if (pfs>=2.8) then pfs4=0.7; if (pfs>=2.8) then c4=1; if (pfs<2.8) then pfs4=pfs-2.1; if (pfs<2.8) then c4=pfscns; if (pfs<2.1) then pfs4=_missing; if (pfs<2.1) then c4=_missing; if (pfs>=3.5) then pfs5=0.7; if (pfs>=3.5) then c5=1; if (pfs<3.5) then pfs5=pfs-2.8; if (pfs<3.5) then c5=pfscns; if (pfs<2.8) then pfs5=_missing; if (pfs<2.8) then c5=_missing; if (pfs>=4.2) then pfs6=0.7; ! Visit 6 if (pfs>=4.2) then c6=1; if (pfs<4.2) then pfs6=pfs-3.5; if (pfs<4.2) then c6=pfscns; if (pfs<3.5) then pfs6=_missing; if (pfs<3.5) then c6=_missing; if (pfs>=4.9) then pfs7=0.7; if (pfs>=4.9) then c7=1; if (pfs<4.9) then pfs7=pfs-4.2; if (pfs<4.9) then c7=pfscns; if (pfs<4.2) then pfs7=_missing; if (pfs<4.2) then c7=_missing; if (pfs>=5.6) then pfs8=0.7; if (pfs>=5.6) then c8=1; if (pfs<5.6) then pfs8=pfs-4.9; if (pfs<5.6) then c8=pfscns; if (pfs<4.9) then pfs8=_missing; if (pfs<4.9) then c8=_missing; pfs9=pfs-5.6; c9=pfscns; if (pfs<5.6) then pfs9=_missing; if (pfs<5.6) then c9=_missing; ! define the group indicator group = 0; ! control group if(tx eq 2) then group = 1; ! treatment group ! make the qol item have a smaller scale x = qol_0/10; Analysis: type = mixture; starts = 0; ! classes are known and we don't need multiple starts process = 4(starts); Model: %overall% pfs1-pfs9 on x; %cg#1% ! control group [group$1@15]; ! this gives P(group = 0)=1, where group=0 is controls pfs1 on x (d1); pfs2 on x (d2); pfs3 on x (d3); pfs4 on x (d4); pfs5 on x (d5); pfs6 on x (d6); pfs7 on x (d7); pfs8 on x (d8); pfs9 on x (d9); ! the below uses the Mplus v6 approach of holding baseline hazards equal across class ! and letting the [pfs] intercept reflect the effect of the treatment covariate in ! the cg#2 group - see parameters p11-p19 below [pfs1@0]; [pfs2@0]; [pfs3@0]; [pfs4@0]; [pfs5@0]; [pfs6@0]; [pfs7@0]; [pfs8@0]; [pfs9@0]; %cg#2% ! treatment group [group$1@-15]; ! this gives P(group = 1)=1 pfs1 on x (d11); pfs2 on x (d12); pfs3 on x (d13); pfs4 on x (d14); pfs5 on x (d15); pfs6 on x (d16); pfs7 on x (d17); pfs8 on x (d18); pfs9 on x (d19); [pfs1*0] (p11); [pfs2*0] (p12); [pfs3*0] (p13); [pfs4*0] (p14); [pfs5*0] (p15); [pfs6*0] (p16); [pfs7*0] (p17); [pfs8*0] (p18); [pfs9*0] (p19); Model constraint: new(b2*0 g1*0 g2*0 gdiff*0 ddiff1*0 ddiff2*0 ddiff3*0 ddiff4*0 ddiff5*0 ddiff6*0 ddiff7*0 ddiff8*0 ddiff9*0); ! the below adds a linear constraint on the hazards for the tx covariate p12 = p11+b2*1; p13 = p11+b2*2; p14 = p11+b2*3; p15 = p11+b2*4; p16 = p11+b2*5; p17 = p11+b2*6; p18 = p11+b2*7; p19 = p11+b2*8; ! the 2 sections below add linear constraints on the hazards for the x (qol) covariate d2 = d1+g1*1; d3 = d1+g1*2; d4 = d1+g1*3; d5 = d1+g1*4; d6 = d1+g1*5; d7 = d1+g1*6; d8 = d1+g1*7; d9 = d1+g1*8; d12 = d11+g2*1; d13 = d11+g2*2; d14 = d11+g2*3; d15 = d11+g2*4; d16 = d11+g2*5; d17 = d11+g2*6; d18 = d11+g2*7; d19 = d11+g2*8; ! the below describes effect estimates of interest gdiff = g2 - g1; ddiff1 = d11-d1; ddiff2 = d12-d2; ddiff3 = d13-d3; ddiff4 = d14-d4; ddiff5 = d15-d5; ddiff6 = d16-d6; ddiff7 = d17-d7; ddiff8 = d18-d8; ddiff9 = d19-d9; Output: tech1 patterns sampstat standardized; *** WARNING Data set contains cases with missing on x-variables. These cases were not included in the analysis. Number of cases with missing on x-variables: 29 1 WARNING(S) FOUND IN THE INPUT INSTRUCTIONS SUMMARY OF ANALYSIS Number of groups 1 Number of observations 214 Number of dependent variables 10 Number of independent variables 1 Number of continuous latent variables 0 Number of categorical latent variables 1 Observed dependent variables Binary and ordered categorical (ordinal) GROUP Time-to-event (survival) PFS1 PFS2 PFS3 PFS4 PFS5 PFS6 PFS7 PFS8 PFS9 Observed independent variables X Categorical latent variables CG Variables with special functions Time-censoring variables C1 C2 C3 C4 C5 C6 C7 C8 C9 Estimator MLR Information matrix OBSERVED Optimization Specifications for the Quasi-Newton Algorithm for Continuous Outcomes Maximum number of iterations 100 Convergence criterion 0.100D-05 Optimization Specifications for the EM Algorithm Maximum number of iterations 500 Convergence criteria Loglikelihood change 0.100D-06 Relative loglikelihood change 0.100D-06 Derivative 0.100D-05 Optimization Specifications for the M step of the EM Algorithm for Categorical Latent variables Number of M step iterations 1 M step convergence criterion 0.100D-05 Basis for M step termination ITERATION Optimization Specifications for the M step of the EM Algorithm for Censored, Binary or Ordered Categorical (Ordinal), Unordered Categorical (Nominal) and Count Outcomes Number of M step iterations 1 M step convergence criterion 0.100D-05 Basis for M step termination ITERATION Maximum value for logit thresholds 15 Minimum value for logit thresholds -15 Minimum expected cell size for chi-square 0.100D-01 Maximum number of iterations for H1 2000 Convergence criterion for H1 0.100D-03 Optimization algorithm EMA Link LOGIT Base Hazard EQUAL ACROSS CLASSES Input data file(s) all_wide.dat Input data format FREE SUMMARY OF DATA Number of missing data patterns 1 Number of y missing data patterns 0 Number of u missing data patterns 1 SUMMARY OF MISSING DATA PATTERNS COVARIANCE COVERAGE OF DATA Minimum covariance coverage value 0.100 UNIVARIATE PROPORTIONS AND COUNTS FOR CATEGORICAL VARIABLES GROUP Category 1 0.481 103.000 Category 2 0.519 111.000 SAMPLE STATISTICS SAMPLE STATISTICS Means X ________ 1 4.642 Covariances X ________ X 6.830 Correlations X ________ X 1.000 THE MODEL ESTIMATION TERMINATED NORMALLY MODEL FIT INFORMATION Number of Free Parameters 7 Loglikelihood H0 Value -522.001 H0 Scaling Correction Factor 0.956 for MLR Information Criteria Akaike (AIC) 1058.001 Bayesian (BIC) 1081.563 Sample-Size Adjusted BIC 1059.382 (n* = (n + 2) / 24) Chi-Square Test of Model Fit for the Binary and Ordered Categorical (Ordinal) Outcomes Pearson Chi-Square Value 0.000 Degrees of Freedom 0 P-Value 1.0000 Likelihood Ratio Chi-Square Value 0.000 Degrees of Freedom 0 P-Value 1.0000 FINAL CLASS COUNTS AND PROPORTIONS FOR THE LATENT CLASSES BASED ON THE ESTIMATED MODEL Latent Classes 1 103.00000 0.48131 2 111.00000 0.51869 FINAL CLASS COUNTS AND PROPORTIONS FOR THE LATENT CLASS PATTERNS BASED ON ESTIMATED POSTERIOR PROBABILITIES Latent Classes 1 103.00000 0.48131 2 111.00000 0.51869 CLASSIFICATION QUALITY Entropy 1.000 CLASSIFICATION OF INDIVIDUALS BASED ON THEIR MOST LIKELY LATENT CLASS MEMBERSHIP Class Counts and Proportions Latent Classes 1 103 0.48131 2 111 0.51869 Average Latent Class Probabilities for Most Likely Latent Class Membership (Row) by Latent Class (Column) 1 2 1 1.000 0.000 2 0.000 1.000 MODEL RESULTS Two-Tailed Estimate S.E. Est./S.E. P-Value Latent Class 1 PFS1 ON X 0.109 0.060 1.810 0.070 PFS2 ON X 0.082 0.050 1.641 0.101 PFS3 ON X 0.055 0.043 1.285 0.199 PFS4 ON X 0.028 0.040 0.692 0.489 PFS5 ON X 0.001 0.043 0.018 0.986 PFS6 ON X -0.026 0.051 -0.517 0.605 PFS7 ON X -0.053 0.061 -0.868 0.385 PFS8 ON X -0.080 0.073 -1.092 0.275 PFS9 ON X -0.107 0.086 -1.239 0.215 Intercepts PFS1 0.000 0.000 999.000 999.000 PFS2 0.000 0.000 999.000 999.000 PFS3 0.000 0.000 999.000 999.000 PFS4 0.000 0.000 999.000 999.000 PFS5 0.000 0.000 999.000 999.000 PFS6 0.000 0.000 999.000 999.000 PFS7 0.000 0.000 999.000 999.000 PFS8 0.000 0.000 999.000 999.000 PFS9 0.000 0.000 999.000 999.000 Thresholds GROUP$1 15.000 0.000 999.000 999.000 Latent Class 2 PFS1 ON X 0.176 0.082 2.140 0.032 PFS2 ON X 0.164 0.067 2.455 0.014 PFS3 ON X 0.152 0.052 2.890 0.004 PFS4 ON X 0.139 0.041 3.419 0.001 PFS5 ON X 0.127 0.034 3.697 0.000 PFS6 ON X 0.115 0.036 3.173 0.002 PFS7 ON X 0.103 0.045 2.271 0.023 PFS8 ON X 0.091 0.058 1.557 0.119 PFS9 ON X 0.079 0.073 1.075 0.282 Intercepts PFS1 -1.729 0.559 -3.093 0.002 PFS2 -1.507 0.469 -3.210 0.001 PFS3 -1.284 0.393 -3.269 0.001 PFS4 -1.062 0.339 -3.137 0.002 PFS5 -0.840 0.318 -2.641 0.008 PFS6 -0.617 0.337 -1.831 0.067 PFS7 -0.395 0.390 -1.011 0.312 PFS8 -0.172 0.466 -0.369 0.712 PFS9 0.050 0.555 0.090 0.928 Thresholds GROUP$1 -15.000 0.000 999.000 999.000 Categorical Latent Variables Means CG#1 -0.075 0.137 -0.547 0.585 New/Additional Parameters B2 0.222 0.114 1.944 0.052 G1 -0.027 0.015 -1.778 0.075 G2 -0.012 0.017 -0.695 0.487 GDIFF 0.015 0.023 0.649 0.517 DDIFF1 0.067 0.101 0.666 0.506 DDIFF2 0.082 0.083 0.992 0.321 DDIFF3 0.097 0.067 1.440 0.150 DDIFF4 0.112 0.057 1.955 0.051 DDIFF5 0.126 0.055 2.292 0.022 DDIFF6 0.141 0.062 2.271 0.023 DDIFF7 0.156 0.076 2.060 0.039 DDIFF8 0.171 0.093 1.837 0.066 DDIFF9 0.186 0.112 1.653 0.098 RESULTS IN PROBABILITY SCALE Latent Class 1 GROUP Category 1 1.000 0.000 0.000 1.000 Category 2 0.000 0.000 0.000 1.000 Latent Class 2 GROUP Category 1 0.000 0.000 0.000 1.000 Category 2 1.000 0.000 0.000 1.000 LATENT CLASS ODDS RATIO RESULTS Latent Class 1 Compared to Latent Class 2 GROUP Category > 1 0.000 0.000 999.000 999.000 STANDARDIZED MODEL RESULTS STDYX Standardization Two-Tailed Estimate S.E. Est./S.E. P-Value STDY Standardization Two-Tailed Estimate S.E. Est./S.E. P-Value STD Standardization Two-Tailed Estimate S.E. Est./S.E. P-Value R-SQUARE Class 1 Class 2 QUALITY OF NUMERICAL RESULTS Condition Number for the Information Matrix 0.570E-07 (ratio of smallest to largest eigenvalue) TECHNICAL 1 OUTPUT PARAMETER SPECIFICATION FOR LATENT CLASS 1 NU X ________ 1 0 LAMBDA X ________ X 0 THETA X ________ X 0 ALPHA X ________ 1 0 BETA X ________ X 0 PSI X ________ X 0 PARAMETER SPECIFICATION FOR LATENT CLASS 2 NU X ________ 1 0 LAMBDA X ________ X 0 THETA X ________ X 0 ALPHA X ________ 1 0 BETA X ________ X 0 PSI X ________ X 0 PARAMETER SPECIFICATION FOR LATENT CLASS INDICATOR MODEL PART TAU(U) FOR LATENT CLASS 1 GROUP$1 ________ 1 0 TAU(U) FOR LATENT CLASS 2 GROUP$1 ________ 1 0 PARAMETER SPECIFICATION FOR LATENT CLASS REGRESSION MODEL PART ALPHA(C) CG#1 CG#2 ________ ________ 1 1 0 GAMMA(C) X ________ CG#1 0 CG#2 0 PARAMETER SPECIFICATION FOR THE CENSORED/NOMINAL/COUNT MODEL PART NU(P) FOR LATENT CLASS 1 PFS1#1 PFS1 PFS2#1 PFS2 PFS3#1 ________ ________ ________ ________ ________ 1 0 0 0 0 0 NU(P) FOR LATENT CLASS 1 PFS3 PFS4#1 PFS4 PFS5#1 PFS5 ________ ________ ________ ________ ________ 1 0 0 0 0 0 NU(P) FOR LATENT CLASS 1 PFS6#1 PFS6 PFS7#1 PFS7 PFS8#1 ________ ________ ________ ________ ________ 1 0 0 0 0 0 NU(P) FOR LATENT CLASS 1 PFS8 PFS9#1 PFS9 ________ ________ ________ 1 0 0 0 KAPPA(P) FOR LATENT CLASS 1 X ________ PFS1#1 0 PFS1 2 PFS2#1 0 PFS2 3 PFS3#1 0 PFS3 4 PFS4#1 0 PFS4 5 PFS5#1 0 PFS5 6 PFS6#1 0 PFS6 7 PFS7#1 0 PFS7 8 PFS8#1 0 PFS8 9 PFS9#1 0 PFS9 10 NU(P) FOR LATENT CLASS 2 PFS1#1 PFS1 PFS2#1 PFS2 PFS3#1 ________ ________ ________ ________ ________ 1 0 11 0 12 0 NU(P) FOR LATENT CLASS 2 PFS3 PFS4#1 PFS4 PFS5#1 PFS5 ________ ________ ________ ________ ________ 1 13 0 14 0 15 NU(P) FOR LATENT CLASS 2 PFS6#1 PFS6 PFS7#1 PFS7 PFS8#1 ________ ________ ________ ________ ________ 1 0 16 0 17 0 NU(P) FOR LATENT CLASS 2 PFS8 PFS9#1 PFS9 ________ ________ ________ 1 18 0 19 KAPPA(P) FOR LATENT CLASS 2 X ________ PFS1#1 0 PFS1 20 PFS2#1 0 PFS2 21 PFS3#1 0 PFS3 22 PFS4#1 0 PFS4 23 PFS5#1 0 PFS5 24 PFS6#1 0 PFS6 25 PFS7#1 0 PFS7 26 PFS8#1 0 PFS8 27 PFS9#1 0 PFS9 28 PARAMETER SPECIFICATION FOR THE ADDITIONAL PARAMETERS NEW/ADDITIONAL PARAMETERS B2 G1 G2 GDIFF DDIFF1 ________ ________ ________ ________ ________ 1 29 30 31 32 33 NEW/ADDITIONAL PARAMETERS DDIFF2 DDIFF3 DDIFF4 DDIFF5 DDIFF6 ________ ________ ________ ________ ________ 1 34 35 36 37 38 NEW/ADDITIONAL PARAMETERS DDIFF7 DDIFF8 DDIFF9 ________ ________ ________ 1 39 40 41 STARTING VALUES FOR LATENT CLASS 1 NU X ________ 1 0.000 LAMBDA X ________ X 1.000 THETA X ________ X 0.000 ALPHA X ________ 1 0.000 BETA X ________ X 0.000 PSI X ________ X 3.415 STARTING VALUES FOR LATENT CLASS 2 NU X ________ 1 0.000 LAMBDA X ________ X 1.000 THETA X ________ X 0.000 ALPHA X ________ 1 0.000 BETA X ________ X 0.000 PSI X ________ X 3.415 STARTING VALUES FOR LATENT CLASS INDICATOR MODEL PART TAU(U) FOR LATENT CLASS 1 GROUP$1 ________ 1 15.000 TAU(U) FOR LATENT CLASS 2 GROUP$1 ________ 1 -15.000 STARTING VALUES FOR LATENT CLASS REGRESSION MODEL PART ALPHA(C) CG#1 CG#2 ________ ________ 1 0.000 0.000 GAMMA(C) X ________ CG#1 0.000 CG#2 0.000 STARTING VALUES FOR THE CENSORED/NOMINAL/COUNT MODEL PART NU(P) FOR LATENT CLASS 1 PFS1#1 PFS1 PFS2#1 PFS2 PFS3#1 ________ ________ ________ ________ ________ 1 -20.000 0.000 -20.000 0.000 -20.000 NU(P) FOR LATENT CLASS 1 PFS3 PFS4#1 PFS4 PFS5#1 PFS5 ________ ________ ________ ________ ________ 1 0.000 -20.000 0.000 -20.000 0.000 NU(P) FOR LATENT CLASS 1 PFS6#1 PFS6 PFS7#1 PFS7 PFS8#1 ________ ________ ________ ________ ________ 1 -20.000 0.000 -20.000 0.000 -20.000 NU(P) FOR LATENT CLASS 1 PFS8 PFS9#1 PFS9 ________ ________ ________ 1 0.000 -20.000 0.000 KAPPA(P) FOR LATENT CLASS 1 X ________ PFS1#1 0.000 PFS1 0.000 PFS2#1 0.000 PFS2 0.000 PFS3#1 0.000 PFS3 0.000 PFS4#1 0.000 PFS4 0.000 PFS5#1 0.000 PFS5 0.000 PFS6#1 0.000 PFS6 0.000 PFS7#1 0.000 PFS7 0.000 PFS8#1 0.000 PFS8 0.000 PFS9#1 0.000 PFS9 0.000 NU(P) FOR LATENT CLASS 2 PFS1#1 PFS1 PFS2#1 PFS2 PFS3#1 ________ ________ ________ ________ ________ 1 -20.000 0.000 -20.000 0.000 -20.000 NU(P) FOR LATENT CLASS 2 PFS3 PFS4#1 PFS4 PFS5#1 PFS5 ________ ________ ________ ________ ________ 1 0.000 -20.000 0.000 -20.000 0.000 NU(P) FOR LATENT CLASS 2 PFS6#1 PFS6 PFS7#1 PFS7 PFS8#1 ________ ________ ________ ________ ________ 1 -20.000 0.000 -20.000 0.000 -20.000 NU(P) FOR LATENT CLASS 2 PFS8 PFS9#1 PFS9 ________ ________ ________ 1 0.000 -20.000 0.000 KAPPA(P) FOR LATENT CLASS 2 X ________ PFS1#1 0.000 PFS1 0.000 PFS2#1 0.000 PFS2 0.000 PFS3#1 0.000 PFS3 0.000 PFS4#1 0.000 PFS4 0.000 PFS5#1 0.000 PFS5 0.000 PFS6#1 0.000 PFS6 0.000 PFS7#1 0.000 PFS7 0.000 PFS8#1 0.000 PFS8 0.000 PFS9#1 0.000 PFS9 0.000 STARTING VALUES FOR THE ADDITIONAL PARAMETERS NEW/ADDITIONAL PARAMETERS B2 G1 G2 GDIFF DDIFF1 ________ ________ ________ ________ ________ 1 0.000 0.000 0.000 0.000 0.000 NEW/ADDITIONAL PARAMETERS DDIFF2 DDIFF3 DDIFF4 DDIFF5 DDIFF6 ________ ________ ________ ________ ________ 1 0.000 0.000 0.000 0.000 0.000 NEW/ADDITIONAL PARAMETERS DDIFF7 DDIFF8 DDIFF9 ________ ________ ________ 1 0.000 0.000 0.000 Beginning Time: 05:07:56 Ending Time: 05:07:57 Elapsed Time: 00:00:01 MUTHEN & MUTHEN 3463 Stoner Ave. Los Angeles, CA 90066 Tel: (310) 391-9971 Fax: (310) 391-8971 Web: www.StatModel.com Support: Support@StatModel.com Copyright (c) 1998-2010 Muthen & Muthen