Here fit location-scale model to 23 schools from nels. The R belows will run the program mixreglsb.exe from R. The code makes use of the function R_mixregls.txt which is a function that I wrote to help you set up the input to mixreglsb.exe and than run the program.
The mixreglsb.txt use software described in the following paper. The R described in the paper no longer works; however, you can download programs and examples from the paper’s web-site. There are quite a few number of files, but you only need mixreglsb.exe.
Hedeker, D, & Nordgren, R. (2013), MIXREGLS: A Program for Mixed-Effects Location Scale Analysis. Journal of Statistical Software, 52. http://www.jstatsoft.org/
I suggest that you read the above paper to better understand what is going on.
The mixreglsb.exe only fits random intercept models, but an extended verision is described in the paper Nordrend, Hedeker, Donton & Yang (2019) in Statistics in Medicine. There is an appendix that contians information for fitting the extended version in SAS/NLMIXED and merlin in STATA. A R version is in development at the time the paper was written.
A few things if you run the R_mixregls function:
library(formula.tools)## Warning: package 'formula.tools' was built under R version 4.0.5library(stringr)This is where my data and the mixreglsb.exe files live:
setwd("D:/Dropbox/edps587/lectures/8 modelbuilding/MIXREGLS/school23")I will use as source the function
source("D:/Dropbox/edps587/lectures/8 modelbuilding/MIXREGLS/school23/R_mixregls.R")Most of the options are set to defaults but can easily be changed as will be illustrated below. Also all of these are described in Hekeke & Nordgren (2013)
R_mixedregls(fo, indata, idname, outdata, outresults, save_def,
title1=NULL, title2=NULL, PNINT=NULL, RNINT=NULL, SNINT=NULL,
CONV=NULL, NQ=NULL,AQUAD=NULL, MAXIT=NULL, STD=NULL, MISS=NULL,
NCOV=NULL) # The data
indata <- read.table("nels23_data.txt", header=TRUE)
# A formula
fo <- formula(math ~ homew + sex + schtype | schtype | schtype )
# Minimal input to run
R_mixregls(fo, indata, idname="school",
outdata="nels_example1.dat",
outresults="nels_example1.out",
save_def="nels_example1.def",
NCOV=1) ## [1] 0model_summaries<- readLines("nels_example1.out")
noquote(model_summaries)## [1] MIXREGLS: Mixed-effects Location Scale Model with BS and WS variance models
## [2]
## [3] ---------------------------
## [4] MIXREGLS.DEF specifications
## [5] ---------------------------
## [6]
## [7]
## [8]
## [9] data and output files:
## [10] nels_example1.dat
## [11] nels_example1.out
## [12]
## [13] CONVERGENCE CRITERION = 0.00001000
## [14] NQ = 11
## [15] QUADRATURE = 1 (0=non-adaptive, 1=adaptive)
## [16] MAXIT = 200
## [17]
## [18]
## [19] ------------
## [20] Descriptives
## [21] ------------
## [22]
## [23] Number of level-1 observations = 519
## [24]
## [25] Number of level-2 clusters = 23
## [26]
## [27] Number of level-1 observations for each level-2 cluster
## [28] 44 8 5 24 23 17 22 20 24 20 22 22 20
## [29] 16 23 20 19 67 21 21 27 20 14
## [30]
## [31] Dependent variable
## [32] mean min max std dev
## [33] --------------------------------------------------------
## [34] math 51.7225 30.0000 71.0000 10.7092
## [35]
## [36] Mean model covariates
## [37] mean min max std dev
## [38] --------------------------------------------------------
## [39] homew 1.9711 0.0000 7.0000 1.4839
## [40] sex 1.5202 1.0000 2.0000 0.5001
## [41] schtype 1.9769 1.0000 4.0000 1.3281
## [42]
## [43] BS variance model covariates
## [44] mean min max std dev
## [45] --------------------------------------------------------
## [46] schtype 1.9769 1.0000 4.0000 1.3281
## [47]
## [48] WS variance model covariates
## [49] mean min max std dev
## [50] --------------------------------------------------------
## [51] schtype 1.9769 1.0000 4.0000 1.3281
## [52]
## [53]
## [54] ---------------
## [55] Starting Values
## [56] ---------------
## [57]
## [58] BETA: mean model regression coefficients
## [59] 44.7379 2.3836 -0.8285 1.6463
## [60]
## [61] ALPHA: BS variance log-linear model regression coefficients
## [62] 2.8037 -0.0246
## [63]
## [64] TAU: WS variance log-linear model regression coefficients
## [65] 4.2624 -0.0246
## [66]
## [67] Random location (mean) effect on WS variance
## [68] 0.2004
## [69] Scale standard deviation
## [70] 0.4898
## [71]
## [72]
## [73] ------------------------------
## [74] Model without Scale Parameters
## [75] ------------------------------
## [76] Total Iterations = 6
## [77] Final Ridge value = 0.0
## [78]
## [79] Log Likelihood = -1861.041
## [80] Akaike's Information Criterion = -1868.041
## [81] Schwarz's Bayesian Criterion = -1872.016
## [82]
## [83] ==> multiplied by -2
## [84] Log Likelihood = 3722.083
## [85] Akaike's Information Criterion = 3736.083
## [86] Schwarz's Bayesian Criterion = 3744.031
## [87]
## [88]
## [89] Variable Estimate AsymStdError z-value p-value
## [90] -------- ------------ ------------ ------------ ------------
## [91] BETA (regression coefficients)
## [92] Intercpt 44.25150 2.02438 21.85926 0.00000
## [93] homew 2.34615 0.27790 8.44248 0.00000
## [94] sex -0.74888 0.76390 -0.98034 0.32692
## [95] schtype 1.93189 1.04046 1.85677 0.06334
## [96] ALPHA (BS variance parameters: log-linear model)
## [97] Intercpt 1.66763 0.72173 2.31060 0.02086
## [98] schtype 0.51767 0.29837 1.73502 0.08274
## [99] TAU (WS variance parameters: log-linear model)
## [100] Intercpt 4.26256 0.06353 67.09843 0.00000
## [101]
## [102]
## [103] ---------------------------
## [104] Model WITH Scale Parameters
## [105] ---------------------------
## [106] Total Iterations = 9
## [107] Final Ridge value = 0.0
## [108]
## [109] Log Likelihood = -1855.254
## [110] Akaike's Information Criterion = -1863.254
## [111] Schwarz's Bayesian Criterion = -1867.796
## [112]
## [113] ==> multiplied by -2
## [114] Log Likelihood = 3710.508
## [115] Akaike's Information Criterion = 3726.508
## [116] Schwarz's Bayesian Criterion = 3735.592
## [117]
## [118]
## [119] Variable Estimate AsymStdError z-value p-value
## [120] -------- ------------ ------------ ------------ ------------
## [121] BETA (regression coefficients)
## [122] Intercpt 44.29881 2.01211 22.01605 0.00000
## [123] homew 2.24645 0.26278 8.54883 0.00000
## [124] sex -0.71251 0.74706 -0.95375 0.34021
## [125] schtype 1.98936 1.04742 1.89930 0.05752
## [126] ALPHA (BS variance parameters: log-linear model)
## [127] Intercpt 1.54097 0.76640 2.01065 0.04436
## [128] schtype 0.56214 0.30219 1.86024 0.06285
## [129] TAU (WS variance parameters: log-linear model)
## [130] Intercpt 4.57824 0.11570 39.56894 0.00000
## [131] schtype -0.17045 0.04861 -3.50676 0.00045
## [132]
## [133]
## [134] -----------------------
## [135] Model WITH RANDOM Scale
## [136] -----------------------
## [137] Total Iterations = 13
## [138] Final Ridge value = 0.0
## [139]
## [140] Log Likelihood = -1854.457
## [141] Akaike's Information Criterion = -1864.457
## [142] Schwarz's Bayesian Criterion = -1870.134
## [143]
## [144] ==> multiplied by -2
## [145] Log Likelihood = 3708.913
## [146] Akaike's Information Criterion = 3728.913
## [147] Schwarz's Bayesian Criterion = 3740.268
## [148]
## [149]
## [150] Variable Estimate AsymStdError z-value p-value
## [151] -------- ------------ ------------ ------------ ------------
## [152] BETA (regression coefficients)
## [153] Intercpt 44.33596 2.01189 22.03697 0.00000
## [154] homew 2.22950 0.26997 8.25836 0.00000
## [155] sex -0.70867 0.74354 -0.95310 0.34054
## [156] schtype 1.97860 1.04404 1.89515 0.05807
## [157] ALPHA (BS variance parameters: log-linear model)
## [158] Intercpt 1.57915 0.76368 2.06781 0.03866
## [159] schtype 0.54966 0.29570 1.85884 0.06305
## [160] TAU (WS variance parameters: log-linear model)
## [161] Intercpt 4.58679 0.13137 34.91428 0.00000
## [162] schtype -0.17742 0.05765 -3.07738 0.00209
## [163] Random location (mean) effect on WS variance
## [164] Loc Eff 0.07342 0.08174 0.89821 0.36907
## [165] Random scale standard deviation
## [166] Std Dev 0.14119 0.09240 1.52798 0.12652readLines("MIXREGLS.EST")## [1] " 3722.082719 6 200"
## [2] " 44.25150073 2.34614536 -0.74887663 1.93188823"
## [3] " 1.66762733 0.51767392"
## [4] " 4.26255633"
## [5] " 2.02438210 0.27789758 0.76389761 1.04045609 0.72172967"
## [6] " 0.29836724 0.06352691"
## [7] " 3710.507994 9 200"
## [8] " 44.29880684 2.24645035 -0.71250645 1.98935823"
## [9] " 1.54096568 0.56214296"
## [10] " 4.57823614 -0.17044857"
## [11] " 2.01211389 0.26277875 0.74705607 1.04741550 0.76640313"
## [12] " 0.30218900 0.11570277 0.04860575"
## [13] " 3708.913495 13 200"
## [14] " 44.33596474 2.22949542 -0.70866999 1.97860291"
## [15] " 1.57915127 0.54965701"
## [16] " 4.58678581 -0.17741618"
## [17] " 0.07342047 0.14119008"
## [18] " 2.01189045 0.26996814 0.74354133 1.04403687 0.76368377"
## [19] " 0.29569915 0.13137276 0.05765173 0.08174101 0.09240311"readLines("MIXREGLS.ITS")## [1] " "
## [2] " ------------------------------"
## [3] " Model without Scale Parameters"
## [4] " ------------------------------"
## [5] " Newton-Raphson Iteration 1 with ridge 0.0000"
## [6] " maximum correction and derivative"
## [7] " 1.095834627451376 8.275792265556330"
## [8] " -2 Log-Likelihood = 3725.85087"
## [9] " Newton-Raphson Iteration 2 with ridge 0.0000"
## [10] " maximum correction and derivative"
## [11] " 7.693575178196128E-02 1.560921459705948"
## [12] " -2 Log-Likelihood = 3722.15739"
## [13] " Newton-Raphson Iteration 3 with ridge 0.0000"
## [14] " maximum correction and derivative"
## [15] " 1.510219549719200E-02 6.288897786075243E-02"
## [16] " -2 Log-Likelihood = 3722.08303"
## [17] " Newton-Raphson Iteration 4 with ridge 0.0000"
## [18] " maximum correction and derivative"
## [19] " 1.100064091041021E-04 3.813607024905075E-04"
## [20] " -2 Log-Likelihood = 3722.08272"
## [21] " Newton-Raphson Iteration 5 with ridge 0.0000"
## [22] " maximum correction and derivative"
## [23] " 5.488873141206965E-09 1.876020139235379E-08"
## [24] " -2 Log-Likelihood = 3722.08272"
## [25] " maximum correction and derivative"
## [26] " 3.555400562735551E-12 6.905809257773399E-12"
## [27] " -2 Log-Likelihood = 3722.08272"
## [28] " "
## [29] "---------------------------"
## [30] "Model WITH Scale Parameters"
## [31] "---------------------------"
## [32] " Newton-Raphson Iteration 1 with ridge 0.1000"
## [33] " maximum correction and derivative"
## [34] " 9.152074001000320E-02 37.82707410368207"
## [35] " -2 Log-Likelihood = 3719.52618"
## [36] " Newton-Raphson Iteration 2 with ridge 0.1000"
## [37] " maximum correction and derivative"
## [38] " 6.872585251269775E-02 27.60429747166064"
## [39] " -2 Log-Likelihood = 3715.01039"
## [40] " Newton-Raphson Iteration 3 with ridge 0.1000"
## [41] " maximum correction and derivative"
## [42] " 4.900772974775337E-02 19.24504975811226"
## [43] " -2 Log-Likelihood = 3712.64954"
## [44] " Newton-Raphson Iteration 4 with ridge 0.1000"
## [45] " maximum correction and derivative"
## [46] " 3.386780001410764E-02 13.23455334693403"
## [47] " -2 Log-Likelihood = 3711.50500"
## [48] " Newton-Raphson Iteration 5 with ridge 0.0000"
## [49] " maximum correction and derivative"
## [50] " 7.209622518538311E-02 9.066898920444936"
## [51] " -2 Log-Likelihood = 3710.96966"
## [52] " Newton-Raphson Iteration 6 with ridge 0.0000"
## [53] " maximum correction and derivative"
## [54] " 2.145642321368735E-03 0.5644640517330801"
## [55] " -2 Log-Likelihood = 3710.50825"
## [56] " Newton-Raphson Iteration 7 with ridge 0.0000"
## [57] " maximum correction and derivative"
## [58] " 1.942550391354983E-06 3.556557808477301E-04"
## [59] " -2 Log-Likelihood = 3710.50799"
## [60] " Newton-Raphson Iteration 8 with ridge 0.0000"
## [61] " maximum correction and derivative"
## [62] " 2.104218659652576E-11 2.586666436599216E-10"
## [63] " -2 Log-Likelihood = 3710.50799"
## [64] " maximum correction and derivative"
## [65] " 1.697030701641697E-13 5.917488721252084E-13"
## [66] " -2 Log-Likelihood = 3710.50799"
## [67] " "
## [68] "-----------------------"
## [69] "Model WITH RANDOM Scale"
## [70] "-----------------------"
## [71] " Newton-Raphson Iteration 1 with ridge 0.2000"
## [72] " maximum correction and derivative"
## [73] " 0.8030335520625563 20.23487348210596"
## [74] " -2 Log-Likelihood = 3719.80949"
## [75] " Newton-Raphson Iteration 2 with ridge 0.2000"
## [76] " maximum correction and derivative"
## [77] " 5.229800167115933E-02 6.559366352024711"
## [78] " -2 Log-Likelihood = 3709.80646"
## [79] " Newton-Raphson Iteration 3 with ridge 0.2000"
## [80] " maximum correction and derivative"
## [81] " 1.943389801601105E-02 4.236760491758237"
## [82] " -2 Log-Likelihood = 3709.09295"
## [83] " Newton-Raphson Iteration 4 with ridge 0.2000"
## [84] " maximum correction and derivative"
## [85] " 1.278885592177282E-02 2.950716334282731"
## [86] " -2 Log-Likelihood = 3708.99816"
## [87] " Newton-Raphson Iteration 5 with ridge 0.2000"
## [88] " maximum correction and derivative"
## [89] " 1.100376898095719E-02 2.117970339504217"
## [90] " -2 Log-Likelihood = 3708.95583"
## [91] " Newton-Raphson Iteration 6 with ridge 0.2000"
## [92] " maximum correction and derivative"
## [93] " 9.334912009394465E-03 1.568084308415922"
## [94] " -2 Log-Likelihood = 3708.93596"
## [95] " Newton-Raphson Iteration 7 with ridge 0.2000"
## [96] " maximum correction and derivative"
## [97] " 6.405304398932340E-03 0.8381090565351587"
## [98] " -2 Log-Likelihood = 3708.91981"
## [99] " Newton-Raphson Iteration 8 with ridge 0.2000"
## [100] " maximum correction and derivative"
## [101] " 4.347093843429422E-03 0.5463029925584184"
## [102] " -2 Log-Likelihood = 3708.91622"
## [103] " Newton-Raphson Iteration 9 with ridge 0.2000"
## [104] " maximum correction and derivative"
## [105] " 2.974613927236782E-03 0.3932119969612056"
## [106] " -2 Log-Likelihood = 3708.91490"
## [107] " Newton-Raphson Iteration 10 with ridge 0.0000"
## [108] " maximum correction and derivative"
## [109] " 6.899216427513330E-03 0.2944344631448630"
## [110] " -2 Log-Likelihood = 3708.91427"
## [111] " Newton-Raphson Iteration 11 with ridge 0.0000"
## [112] " maximum correction and derivative"
## [113] " 1.797344792833063E-05 1.058960327099956E-03"
## [114] " -2 Log-Likelihood = 3708.91350"
## [115] " Newton-Raphson Iteration 12 with ridge 0.0000"
## [116] " maximum correction and derivative"
## [117] " 3.392125295744964E-09 4.565051081595328E-08"
## [118] " -2 Log-Likelihood = 3708.91350"
## [119] " maximum correction and derivative"
## [120] " 7.338389002016553E-12 6.366963206660969E-11"
## [121] " -2 Log-Likelihood = 3708.91350"For the following, you can change the label on the columns and read them in as data frame or you can do something like what’s below.
std1 <- readLines("MIXREGLS.RE1")std2 <- readLines("MIXREGLS.RE2")In this example, I will set some of the options, one of which I recomend that is to give meaingful titles that will appead in the output.
R_mixregls(fo, indata, idname="school",
outdata="nels_example1.dat",
outresults="nels_example1.out",
save_def="nels_example1.def",
title1="NELS Example 1",
title2="Illusration of adding titles")## [1] 0# Table of output
model_summaries<- readLines("nels_example1.out")
noquote(model_summaries)## [1] MIXREGLS: Mixed-effects Location Scale Model with BS and WS variance models
## [2]
## [3] ---------------------------
## [4] MIXREGLS.DEF specifications
## [5] ---------------------------
## [6] NELS Example 1
## [7] Illusration of adding titles
## [8]
## [9] data and output files:
## [10] nels_example1.dat
## [11] nels_example1.out
## [12]
## [13] CONVERGENCE CRITERION = 0.00001000
## [14] NQ = 11
## [15] QUADRATURE = 1 (0=non-adaptive, 1=adaptive)
## [16] MAXIT = 200
## [17]
## [18]
## [19] ------------
## [20] Descriptives
## [21] ------------
## [22]
## [23] Number of level-1 observations = 519
## [24]
## [25] Number of level-2 clusters = 23
## [26]
## [27] Number of level-1 observations for each level-2 cluster
## [28] 44 8 5 24 23 17 22 20 24 20 22 22 20
## [29] 16 23 20 19 67 21 21 27 20 14
## [30]
## [31] Dependent variable
## [32] mean min max std dev
## [33] --------------------------------------------------------
## [34] math 51.7225 30.0000 71.0000 10.7092
## [35]
## [36] Mean model covariates
## [37] mean min max std dev
## [38] --------------------------------------------------------
## [39] homew 1.9711 0.0000 7.0000 1.4839
## [40] sex 1.5202 1.0000 2.0000 0.5001
## [41] schtype 1.9769 1.0000 4.0000 1.3281
## [42]
## [43] BS variance model covariates
## [44] mean min max std dev
## [45] --------------------------------------------------------
## [46] schtype 1.9769 1.0000 4.0000 1.3281
## [47]
## [48] WS variance model covariates
## [49] mean min max std dev
## [50] --------------------------------------------------------
## [51] schtype 1.9769 1.0000 4.0000 1.3281
## [52]
## [53]
## [54] ---------------
## [55] Starting Values
## [56] ---------------
## [57]
## [58] BETA: mean model regression coefficients
## [59] 44.7379 2.3836 -0.8285 1.6463
## [60]
## [61] ALPHA: BS variance log-linear model regression coefficients
## [62] 2.8037 -0.0246
## [63]
## [64] TAU: WS variance log-linear model regression coefficients
## [65] 4.2624 -0.0246
## [66]
## [67] Random location (mean) effect on WS variance
## [68] 0.2004
## [69] Scale standard deviation
## [70] 0.4898
## [71]
## [72]
## [73] ------------------------------
## [74] Model without Scale Parameters
## [75] ------------------------------
## [76] Total Iterations = 6
## [77] Final Ridge value = 0.0
## [78]
## [79] Log Likelihood = -1861.041
## [80] Akaike's Information Criterion = -1868.041
## [81] Schwarz's Bayesian Criterion = -1872.016
## [82]
## [83] ==> multiplied by -2
## [84] Log Likelihood = 3722.083
## [85] Akaike's Information Criterion = 3736.083
## [86] Schwarz's Bayesian Criterion = 3744.031
## [87]
## [88]
## [89] Variable Estimate AsymStdError z-value p-value
## [90] -------- ------------ ------------ ------------ ------------
## [91] BETA (regression coefficients)
## [92] Intercpt 44.25150 2.02438 21.85926 0.00000
## [93] homew 2.34615 0.27790 8.44248 0.00000
## [94] sex -0.74888 0.76390 -0.98034 0.32692
## [95] schtype 1.93189 1.04046 1.85677 0.06334
## [96] ALPHA (BS variance parameters: log-linear model)
## [97] Intercpt 1.66763 0.72173 2.31060 0.02086
## [98] schtype 0.51767 0.29837 1.73502 0.08274
## [99] TAU (WS variance parameters: log-linear model)
## [100] Intercpt 4.26256 0.06353 67.09843 0.00000
## [101]
## [102]
## [103] ---------------------------
## [104] Model WITH Scale Parameters
## [105] ---------------------------
## [106] Total Iterations = 9
## [107] Final Ridge value = 0.0
## [108]
## [109] Log Likelihood = -1855.254
## [110] Akaike's Information Criterion = -1863.254
## [111] Schwarz's Bayesian Criterion = -1867.796
## [112]
## [113] ==> multiplied by -2
## [114] Log Likelihood = 3710.508
## [115] Akaike's Information Criterion = 3726.508
## [116] Schwarz's Bayesian Criterion = 3735.592
## [117]
## [118]
## [119] Variable Estimate AsymStdError z-value p-value
## [120] -------- ------------ ------------ ------------ ------------
## [121] BETA (regression coefficients)
## [122] Intercpt 44.29881 2.01211 22.01605 0.00000
## [123] homew 2.24645 0.26278 8.54883 0.00000
## [124] sex -0.71251 0.74706 -0.95375 0.34021
## [125] schtype 1.98936 1.04742 1.89930 0.05752
## [126] ALPHA (BS variance parameters: log-linear model)
## [127] Intercpt 1.54097 0.76640 2.01065 0.04436
## [128] schtype 0.56214 0.30219 1.86024 0.06285
## [129] TAU (WS variance parameters: log-linear model)
## [130] Intercpt 4.57824 0.11570 39.56894 0.00000
## [131] schtype -0.17045 0.04861 -3.50676 0.00045
## [132]
## [133]
## [134] -----------------------
## [135] Model WITH RANDOM Scale
## [136] -----------------------
## [137] Total Iterations = 13
## [138] Final Ridge value = 0.0
## [139]
## [140] Log Likelihood = -1854.457
## [141] Akaike's Information Criterion = -1864.457
## [142] Schwarz's Bayesian Criterion = -1870.134
## [143]
## [144] ==> multiplied by -2
## [145] Log Likelihood = 3708.913
## [146] Akaike's Information Criterion = 3728.913
## [147] Schwarz's Bayesian Criterion = 3740.268
## [148]
## [149]
## [150] Variable Estimate AsymStdError z-value p-value
## [151] -------- ------------ ------------ ------------ ------------
## [152] BETA (regression coefficients)
## [153] Intercpt 44.33596 2.01189 22.03697 0.00000
## [154] homew 2.22950 0.26997 8.25836 0.00000
## [155] sex -0.70867 0.74354 -0.95310 0.34054
## [156] schtype 1.97860 1.04404 1.89515 0.05807
## [157] ALPHA (BS variance parameters: log-linear model)
## [158] Intercpt 1.57915 0.76368 2.06781 0.03866
## [159] schtype 0.54966 0.29570 1.85884 0.06305
## [160] TAU (WS variance parameters: log-linear model)
## [161] Intercpt 4.58679 0.13137 34.91428 0.00000
## [162] schtype -0.17742 0.05765 -3.07738 0.00209
## [163] Random location (mean) effect on WS variance
## [164] Loc Eff 0.07342 0.08174 0.89821 0.36907
## [165] Random scale standard deviation
## [166] Std Dev 0.14119 0.09240 1.52798 0.12652This is a random intercept version of Model 5 from lecture notes.
schMses <- as.data.frame(aggregate(ses~school, data=indata, FUN="mean"))
names(schMses) <- c('school', 'schMses')
indata <- merge(indata,schMses, by=c('school'))
indata$schCses <- indata$ses - indata$schMsesThis is similar to model 5 in lecture notes and Model_Building R example
fo <- formula(math ~ homew + sex + schCses + schMses + schtype | schtype | schtype )R_mixregls(fo, indata, idname="school",
outdata="nels_complex.dat",
outresults="nels_complex.out",
save_def="nels_complex.def",
title1="NELS Example 1",
title2="Illusration of adding titles")## [1] 0# Table of output
model_summaries<- readLines("nels_complex.out")
noquote(model_summaries)## [1] MIXREGLS: Mixed-effects Location Scale Model with BS and WS variance models
## [2]
## [3] ---------------------------
## [4] MIXREGLS.DEF specifications
## [5] ---------------------------
## [6] NELS Example 1
## [7] Illusration of adding titles
## [8]
## [9] data and output files:
## [10] nels_complex.dat
## [11] nels_complex.out
## [12]
## [13] CONVERGENCE CRITERION = 0.00001000
## [14] NQ = 11
## [15] QUADRATURE = 1 (0=non-adaptive, 1=adaptive)
## [16] MAXIT = 200
## [17]
## [18]
## [19] ------------
## [20] Descriptives
## [21] ------------
## [22]
## [23] Number of level-1 observations = 519
## [24]
## [25] Number of level-2 clusters = 23
## [26]
## [27] Number of level-1 observations for each level-2 cluster
## [28] 44 8 5 24 23 17 22 20 24 20 22 22 20
## [29] 16 23 20 19 67 21 21 27 20 14
## [30]
## [31] Dependent variable
## [32] mean min max std dev
## [33] --------------------------------------------------------
## [34] math 51.7225 30.0000 71.0000 10.7092
## [35]
## [36] Mean model covariates
## [37] mean min max std dev
## [38] --------------------------------------------------------
## [39] homew 1.9711 0.0000 7.0000 1.4839
## [40] sex 1.5202 1.0000 2.0000 0.5001
## [41] schCses 0.0000 -1.9971 1.7290 0.6250
## [42] schMs -0.0013 -1.0685 1.1763 0.6206
## [43] es sc 1.9769 1.0000 4.0000 1.3281
## [44]
## [45] BS variance model covariates
## [46] mean min max std dev
## [47] --------------------------------------------------------
## [48] schtype 1.9769 1.0000 4.0000 1.3281
## [49]
## [50] WS variance model covariates
## [51] mean min max std dev
## [52] --------------------------------------------------------
## [53] schtype 1.9769 1.0000 4.0000 1.3281
## [54]
## [55]
## [56] ---------------
## [57] Starting Values
## [58] ---------------
## [59]
## [60] BETA: mean model regression coefficients
## [61] 51.9841 2.1564 -0.6682 3.3185 9.1874 -1.8349
## [62]
## [63] ALPHA: BS variance log-linear model regression coefficients
## [64] 2.0175 -0.0642
## [65]
## [66] TAU: WS variance log-linear model regression coefficients
## [67] 4.2009 -0.0642
## [68]
## [69] Random location (mean) effect on WS variance
## [70] 0.2004
## [71] Scale standard deviation
## [72] 0.4898
## [73]
## [74]
## [75] ------------------------------
## [76] Model without Scale Parameters
## [77] ------------------------------
## [78] Total Iterations = 7
## [79] Final Ridge value = 0.0
## [80]
## [81] Log Likelihood = -1838.395
## [82] Akaike's Information Criterion = -1847.395
## [83] Schwarz's Bayesian Criterion = -1852.504
## [84]
## [85] ==> multiplied by -2
## [86] Log Likelihood = 3676.789
## [87] Akaike's Information Criterion = 3694.789
## [88] Schwarz's Bayesian Criterion = 3705.009
## [89]
## [90]
## [91] Variable Estimate AsymStdError z-value p-value
## [92] -------- ------------ ------------ ------------ ------------
## [93] BETA (regression coefficients)
## [94] Intercpt 50.31060 2.38227 21.11874 0.00000
## [95] homew 2.09594 0.27073 7.74180 0.00000
## [96] sex -0.50543 0.73996 -0.68305 0.49458
## [97] schCses 3.33994 0.57843 5.77410 0.00000
## [98] schMs 8.27633 1.94452 4.25624 0.00002
## [99] es sc -0.87835 1.22395 -0.71763 0.47298
## [100] ALPHA (BS variance parameters: log-linear model)
## [101] Intercpt 0.27658 1.08312 0.25536 0.79845
## [102] schtype 0.80865 0.43119 1.87540 0.06074
## [103] TAU (WS variance parameters: log-linear model)
## [104] Intercpt 4.19789 0.06344 66.16875 0.00000
## [105]
## [106]
## [107] ---------------------------
## [108] Model WITH Scale Parameters
## [109] ---------------------------
## [110] Total Iterations = 9
## [111] Final Ridge value = 0.0
## [112]
## [113] Log Likelihood = -1834.466
## [114] Akaike's Information Criterion = -1844.466
## [115] Schwarz's Bayesian Criterion = -1850.144
## [116]
## [117] ==> multiplied by -2
## [118] Log Likelihood = 3668.933
## [119] Akaike's Information Criterion = 3688.933
## [120] Schwarz's Bayesian Criterion = 3700.288
## [121]
## [122]
## [123] Variable Estimate AsymStdError z-value p-value
## [124] -------- ------------ ------------ ------------ ------------
## [125] BETA (regression coefficients)
## [126] Intercpt 50.17732 2.39892 20.91664 0.00000
## [127] homew 2.06480 0.25794 8.00509 0.00000
## [128] sex -0.47739 0.72904 -0.65482 0.51258
## [129] schCses 3.17400 0.58292 5.44502 0.00000
## [130] schMs 8.19304 1.96733 4.16456 0.00003
## [131] es sc -0.76985 1.24651 -0.61760 0.53684
## [132] ALPHA (BS variance parameters: log-linear model)
## [133] Intercpt 0.12962 1.15725 0.11201 0.91082
## [134] schtype 0.86771 0.44266 1.96022 0.04997
## [135] TAU (WS variance parameters: log-linear model)
## [136] Intercpt 4.45617 0.11453 38.90860 0.00000
## [137] schtype -0.13854 0.04825 -2.87140 0.00409
## [138]
## [139]
## [140] -----------------------
## [141] Model WITH RANDOM Scale
## [142] -----------------------
## [143] Total Iterations = 13
## [144] Final Ridge value = 0.0
## [145]
## [146] Log Likelihood = -1834.007
## [147] Akaike's Information Criterion = -1846.007
## [148] Schwarz's Bayesian Criterion = -1852.820
## [149]
## [150] ==> multiplied by -2
## [151] Log Likelihood = 3668.014
## [152] Akaike's Information Criterion = 3692.014
## [153] Schwarz's Bayesian Criterion = 3705.640
## [154]
## [155]
## [156] Variable Estimate AsymStdError z-value p-value
## [157] -------- ------------ ------------ ------------ ------------
## [158] BETA (regression coefficients)
## [159] Intercpt 50.10698 2.39753 20.89946 0.00000
## [160] homew 2.07714 0.26496 7.83943 0.00000
## [161] sex -0.49451 0.72641 -0.68076 0.49602
## [162] schCses 3.18663 0.58071 5.48751 0.00000
## [163] schMs 7.97654 2.00020 3.98786 0.00007
## [164] es sc -0.75606 1.23118 -0.61410 0.53915
## [165] ALPHA (BS variance parameters: log-linear model)
## [166] Intercpt 0.19162 1.15420 0.16602 0.86814
## [167] schtype 0.83273 0.43945 1.89496 0.05810
## [168] TAU (WS variance parameters: log-linear model)
## [169] Intercpt 4.46252 0.12808 34.84275 0.00000
## [170] schtype -0.14433 0.05635 -2.56152 0.01042
## [171] Random location (mean) effect on WS variance
## [172] Loc Eff 0.04180 0.09847 0.42451 0.67119
## [173] Random scale standard deviation
## [174] Std Dev 0.13677 0.09672 1.41412 0.15733A formula
fo <- formula(math ~ homew + sex + schtype | none | schtype )
R_mixregls(fo, indata, idname="school",
outdata="nels_no_bs.dat",
outresults="nels_no_bs.out",
save_def="nels_no_bs.def",
title1="Model without BS predictors of Variance",
title2="NELS data with 23 schools",
NCOV=0)## [1] 0# Table of output
model_no_bs<- readLines("nels_no_bs.out")
noquote(model_no_bs)## [1] MIXREGLS: Mixed-effects Location Scale Model with BS and WS variance models
## [2]
## [3] ---------------------------
## [4] MIXREGLS.DEF specifications
## [5] ---------------------------
## [6] Model without BS predictors of Variance
## [7] NELS data with 23 schools
## [8]
## [9] data and output files:
## [10] nels_no_bs.dat
## [11] nels_no_bs.out
## [12]
## [13] CONVERGENCE CRITERION = 0.00001000
## [14] NQ = 11
## [15] QUADRATURE = 1 (0=non-adaptive, 1=adaptive)
## [16] MAXIT = 200
## [17]
## [18]
## [19] ------------
## [20] Descriptives
## [21] ------------
## [22]
## [23] Number of level-1 observations = 519
## [24]
## [25] Number of level-2 clusters = 23
## [26]
## [27] Number of level-1 observations for each level-2 cluster
## [28] 44 8 5 24 23 17 22 20 24 20 22 22 20
## [29] 16 23 20 19 67 21 21 27 20 14
## [30]
## [31] Dependent variable
## [32] mean min max std dev
## [33] --------------------------------------------------------
## [34] math 51.7225 30.0000 71.0000 10.7092
## [35]
## [36] Mean model covariates
## [37] mean min max std dev
## [38] --------------------------------------------------------
## [39] homew 1.9711 0.0000 7.0000 1.4839
## [40] sex 1.5202 1.0000 2.0000 0.5001
## [41] schtype 1.9769 1.0000 4.0000 1.3281
## [42]
## [43] WS variance model covariates
## [44] mean min max std dev
## [45] --------------------------------------------------------
## [46] 1.9769 1.0000 4.0000 1.3281
## [47]
## [48]
## [49] ---------------
## [50] Starting Values
## [51] ---------------
## [52]
## [53] BETA: mean model regression coefficients
## [54] 44.7379 2.3836 -0.8285 1.6463
## [55]
## [56] ALPHA: BS variance log-linear model regression coefficients
## [57] 2.8037
## [58]
## [59] TAU: WS variance log-linear model regression coefficients
## [60] 4.2624 -0.0246
## [61]
## [62] Scale standard deviation
## [63] 0.4898
## [64]
## [65]
## [66] ------------------------------
## [67] Model without Scale Parameters
## [68] ------------------------------
## [69] Total Iterations = 4
## [70] Final Ridge value = 0.0
## [71]
## [72] Log Likelihood = -1862.738
## [73] Akaike's Information Criterion = -1868.738
## [74] Schwarz's Bayesian Criterion = -1872.145
## [75]
## [76] ==> multiplied by -2
## [77] Log Likelihood = 3725.477
## [78] Akaike's Information Criterion = 3737.477
## [79] Schwarz's Bayesian Criterion = 3744.290
## [80]
## [81]
## [82] Variable Estimate AsymStdError z-value p-value
## [83] -------- ------------ ------------ ------------ ------------
## [84] BETA (regression coefficients)
## [85] Intercpt 44.77335 2.06785 21.65210 0.00000
## [86] homew 2.38312 0.27738 8.59160 0.00000
## [87] sex -0.82827 0.76213 -1.08678 0.27713
## [88] schtype 1.61878 0.79299 2.04135 0.04122
## [89] ALPHA (BS variance parameters: log-linear model)
## [90] Intercpt 2.80354 0.36397 7.70277 0.00000
## [91] TAU (WS variance parameters: log-linear model)
## [92] Intercpt 4.26242 0.06351 67.11092 0.00000
## [93]
## [94]
## [95] ---------------------------
## [96] Model WITH Scale Parameters
## [97] ---------------------------
## [98] Total Iterations = 9
## [99] Final Ridge value = 0.0
## [100]
## [101] Log Likelihood = -1857.270
## [102] Akaike's Information Criterion = -1864.270
## [103] Schwarz's Bayesian Criterion = -1868.245
## [104]
## [105] ==> multiplied by -2
## [106] Log Likelihood = 3714.541
## [107] Akaike's Information Criterion = 3728.541
## [108] Schwarz's Bayesian Criterion = 3736.489
## [109]
## [110]
## [111] Variable Estimate AsymStdError z-value p-value
## [112] -------- ------------ ------------ ------------ ------------
## [113] BETA (regression coefficients)
## [114] Intercpt 44.90190 2.08913 21.49306 0.00000
## [115] homew 2.28393 0.26285 8.68924 0.00000
## [116] sex -0.79404 0.74560 -1.06497 0.28689
## [117] schtype 1.62101 0.79884 2.02920 0.04244
## [118] ALPHA (BS variance parameters: log-linear model)
## [119] Intercpt 2.84991 0.35481 8.03227 0.00000
## [120] TAU (WS variance parameters: log-linear model)
## [121] Intercpt 4.56816 0.11522 39.64722 0.00000
## [122] -0.16600 0.04869 -3.40907 0.00065
## [123]
## [124]
## [125] -----------------------
## [126] Model WITH RANDOM Scale
## [127] -----------------------
## [128] Total Iterations = 13
## [129] Final Ridge value = 0.0
## [130]
## [131] Log Likelihood = -1856.952
## [132] Akaike's Information Criterion = -1864.952
## [133] Schwarz's Bayesian Criterion = -1869.494
## [134]
## [135] ==> multiplied by -2
## [136] Log Likelihood = 3713.904
## [137] Akaike's Information Criterion = 3729.904
## [138] Schwarz's Bayesian Criterion = 3738.988
## [139]
## [140]
## [141] Variable Estimate AsymStdError z-value p-value
## [142] -------- ------------ ------------ ------------ ------------
## [143] BETA (regression coefficients)
## [144] Intercpt 44.86839 2.08928 21.47553 0.00000
## [145] homew 2.26090 0.26831 8.42648 0.00000
## [146] sex -0.74849 0.74368 -1.00646 0.31419
## [147] schtype 1.61813 0.80059 2.02118 0.04326
## [148] ALPHA (BS variance parameters: log-linear model)
## [149] Intercpt 2.85334 0.35418 8.05624 0.00000
## [150] TAU (WS variance parameters: log-linear model)
## [151] Intercpt 4.57387 0.12625 36.22758 0.00000
## [152] -0.16949 0.05528 -3.06595 0.00217
## [153] Random scale standard deviation
## [154] Std Dev 0.12806 0.09838 1.30168 0.19303Required input:
Options and their defaults are