Documentation

randomEffects

Class: LinearMixedModel

Estimates of random effects and related statistics

Description

example

B = randomEffects(lme) returns the estimates of the best linear unbiased predictors (BLUPs) of random effects in the linear mixed-effects model lme.

example

[B,Bnames] = randomEffects(lme) also returns the names of the coefficients in Bnames. Each name corresponds to a coefficient in B.

example

[B,Bnames,stats] = randomEffects(lme) also returns the estimated BLUPs of random effects in the linear mixed-effects model lme and related statistics.

example

[B,Bnames,stats] = randomEffects(lme,Name,Value) also returns the BLUPs of random effects in the linear mixed-effects model lme and related statistics with additional options specified by one or more Name,Value pair arguments.

Input Arguments

expand all

Linear mixed-effects model, specified as a LinearMixedModel object constructed using fitlme or fitlmematrix.

Name-Value Pair Arguments

Specify optional comma-separated pairs of Name,Value arguments. Name is the argument name and Value is the corresponding value. Name must appear inside quotes. You can specify several name and value pair arguments in any order as Name1,Value1,...,NameN,ValueN.

Significance level, specified as the comma-separated pair consisting of 'Alpha' and a scalar value in the range 0 to 1. For a value α, the confidence level is 100*(1–α)%.

For example, for 99% confidence intervals, you can specify the confidence level as follows.

Example: 'Alpha',0.01

Data Types: single | double

Method for computing approximate degrees of freedom for the t-statistics that test the random-effects coefficients against 0, specified as the comma-separated pair consisting of 'DFMethod' and one of the following.

 'residual' Default. The degrees of freedom are assumed to be constant and equal to n – p, where n is the number of observations and p is the number of fixed effects. 'satterthwaite' Satterthwaite approximation. 'none' All degrees of freedom are set to infinity.

For example, you can specify the Satterthwaite approximation as follows.

Example: 'DFMethod','satterthwaite'

Output Arguments

expand all

Estimated best linear unbiased predictors of random effects of linear mixed-effects model lme, returned as a column vector.

Suppose lme has R grouping variables g1, g2, ..., gR, with levels m1, m2, ..., mR, respectively. Also suppose q1, q2, ..., qR are the lengths of the random-effects vectors that are associated with g1, g2, ..., gR, respectively. Then, B is a column vector of length q1*m1 + q2*m2 + ... + qR*mR.

randomEffects creates B by concatenating the best linear unbiased predictors of random-effects vectors corresponding to each level of each grouping variable as [g1level1; g1level2; ...; g1levelm1; g2level1; g2level2; ...; g2levelm2; ...; gRlevel1; gRlevel2; ...; gRlevelmR]'.

Names of random-effects coefficients in B, returned as a table.

Estimates of random effects BLUPs and related statistics, returned as a dataset array that has one row for each of the fixed effects and one column for each of the following statistics.

 Group Grouping variable associated with the random effect Level Level within the grouping variable corresponding to the random effect Name Name of the random-effect coefficient Estimate Best linear unbiased predictor (BLUP) of random effect SEPred Standard error of the estimate (BLUP minus random effect) tStat t-statistic for a test that the random effect is zero DF Estimated degrees of freedom for the t-statistic pValue p-value for the t-statistic Lower Lower limit of a 95% confidence interval for the random effect Upper Upper limit of a 95% confidence interval for the random effect

Examples

expand all

Load the sample data.

Fit a linear mixed-effects model for miles per gallon (MPG), with fixed effects for acceleration and horsepower, and potentially correlated random effects for intercept and acceleration, grouped by the model year. First, store the data in a table.

tbl = table(Acceleration,Horsepower,Model_Year,MPG);

Fit the model.

lme = fitlme(tbl, 'MPG ~ Acceleration + Horsepower + (Acceleration|Model_Year)');

Compute the BLUPs of the random-effects coefficients and display the names of the corresponding random effects.

[B,Bnames] = randomEffects(lme)
B = 26×1

3.1270
-0.2426
-1.6532
-0.0086
1.2075
-0.2179
4.4107
-0.4887
-1.3103
-0.0208
⋮

Bnames=26×3 table
Group         Level           Name
______________    ______    ________________

{'Model_Year'}    {'70'}    {'(Intercept)' }
{'Model_Year'}    {'70'}    {'Acceleration'}
{'Model_Year'}    {'71'}    {'(Intercept)' }
{'Model_Year'}    {'71'}    {'Acceleration'}
{'Model_Year'}    {'72'}    {'(Intercept)' }
{'Model_Year'}    {'72'}    {'Acceleration'}
{'Model_Year'}    {'73'}    {'(Intercept)' }
{'Model_Year'}    {'73'}    {'Acceleration'}
{'Model_Year'}    {'74'}    {'(Intercept)' }
{'Model_Year'}    {'74'}    {'Acceleration'}
{'Model_Year'}    {'75'}    {'(Intercept)' }
{'Model_Year'}    {'75'}    {'Acceleration'}
{'Model_Year'}    {'76'}    {'(Intercept)' }
{'Model_Year'}    {'76'}    {'Acceleration'}
{'Model_Year'}    {'77'}    {'(Intercept)' }
{'Model_Year'}    {'77'}    {'Acceleration'}
⋮

Since intercept and acceleration have potentially correlated random effects, grouped by model year of the cars, randomEffects creates a separate row for intercept and acceleration at each level of the grouping variable.

Compute the covariance parameters of the random effects.

[~,~,stats] = covarianceParameters(lme)
stats=2×1 cell
{3x7 classreg.regr.lmeutils.titleddataset}
{1x5 classreg.regr.lmeutils.titleddataset}

stats{1}
ans =
Covariance Type: FullCholesky

Group         Name1                   Name2                   Type
Model_Year    {'(Intercept)' }        {'(Intercept)' }        {'std' }
Model_Year    {'Acceleration'}        {'(Intercept)' }        {'corr'}
Model_Year    {'Acceleration'}        {'Acceleration'}        {'std' }

Estimate    Lower       Upper
3.3475      1.2862      8.7119
-0.87971    -0.98501    -0.29674
0.33789      0.1825     0.62558

The correlation value suggests that random effects seem negatively correlated. Plot the random effects for intercept versus acceleration to confirm this.

plot(B(1:2:end),B(2:2:end),'r*') Load the sample data.

The dataset array includes data from a split-plot experiment, where soil is divided into three blocks based on the soil type: sandy, silty, and loamy. Each block is divided into five plots, where five different types of tomato plants (cherry, heirloom, grape, vine, and plum) are randomly assigned to these plots. The tomato plants in the plots are then divided into subplots, where each subplot is treated by one of four fertilizers. This is simulated data.

Store the data in a dataset array called ds, for practical purposes, and define Tomato, Soil, and Fertilizer as categorical variables.

ds = fertilizer;
ds.Tomato = nominal(ds.Tomato);
ds.Soil = nominal(ds.Soil);
ds.Fertilizer = nominal(ds.Fertilizer);

Fit a linear mixed-effects model, where Fertilizer and Tomato are the fixed-effects variables, and the mean yield varies by the block (soil type), and the plots within blocks (tomato types within soil types) independently.

lme = fitlme(ds,'Yield ~ Fertilizer * Tomato + (1|Soil) + (1|Soil:Tomato)');

Compute the BLUPs and related statistics for random effects.

[~,~,stats] = randomEffects(lme)
stats =
Random effect coefficients: DFMethod = 'Residual', Alpha = 0.05

Group                  Level                     Name
{'Soil'       }        {'Loamy'         }        {'(Intercept)'}
{'Soil'       }        {'Sandy'         }        {'(Intercept)'}
{'Soil'       }        {'Silty'         }        {'(Intercept)'}
{'Soil:Tomato'}        {'Loamy Cherry'  }        {'(Intercept)'}
{'Soil:Tomato'}        {'Loamy Grape'   }        {'(Intercept)'}
{'Soil:Tomato'}        {'Loamy Heirloom'}        {'(Intercept)'}
{'Soil:Tomato'}        {'Loamy Plum'    }        {'(Intercept)'}
{'Soil:Tomato'}        {'Loamy Vine'    }        {'(Intercept)'}
{'Soil:Tomato'}        {'Sandy Cherry'  }        {'(Intercept)'}
{'Soil:Tomato'}        {'Sandy Grape'   }        {'(Intercept)'}
{'Soil:Tomato'}        {'Sandy Heirloom'}        {'(Intercept)'}
{'Soil:Tomato'}        {'Sandy Plum'    }        {'(Intercept)'}
{'Soil:Tomato'}        {'Sandy Vine'    }        {'(Intercept)'}
{'Soil:Tomato'}        {'Silty Cherry'  }        {'(Intercept)'}
{'Soil:Tomato'}        {'Silty Grape'   }        {'(Intercept)'}
{'Soil:Tomato'}        {'Silty Heirloom'}        {'(Intercept)'}
{'Soil:Tomato'}        {'Silty Plum'    }        {'(Intercept)'}
{'Soil:Tomato'}        {'Silty Vine'    }        {'(Intercept)'}

Estimate    SEPred    tStat       DF    pValue      Lower      Upper
1.0061     2.3374     0.43044    40     0.66918     -3.718     5.7303
-1.5236     2.3374    -0.65181    40     0.51825    -6.2477     3.2006
0.51744     2.3374     0.22137    40     0.82593    -4.2067     5.2416
12.46     7.1765      1.7362    40    0.090224    -2.0443     26.964
-2.6429     7.1765    -0.36827    40     0.71461    -17.147     11.861
16.681     7.1765      2.3244    40    0.025269     2.1766     31.185
-5.0172     7.1765    -0.69911    40     0.48853    -19.522     9.4872
-4.6874     7.1765    -0.65316    40     0.51739    -19.192     9.8169
-17.393     7.1765     -2.4235    40    0.019987    -31.897    -2.8882
-7.3679     7.1765     -1.0267    40     0.31075    -21.872     7.1364
-8.621     7.1765     -1.2013    40     0.23671    -23.125     5.8833
7.669     7.1765      1.0686    40     0.29165    -6.8353     22.173
0.28246     7.1765    0.039359    40      0.9688    -14.222     14.787
4.9326     7.1765     0.68732    40     0.49585    -9.5718     19.437
10.011     7.1765      1.3949    40     0.17073    -4.4935     24.515
-8.0599     7.1765     -1.1231    40      0.2681    -22.564     6.4444
-2.6519     7.1765    -0.36952    40     0.71369    -17.156     11.852
4.405     7.1765      0.6138    40     0.54282    -10.099     18.909

The first three rows contain the random-effects estimates and the statistics for the three levels, Loamy, Sandy, and Silty of the grouping variable Soil. The corresponding $p$-values 0.66918, 0.51825, and 0.82593 indicate that these random-effects are not significantly different from 0. The following 15 rows include the BLUPS of random-effects estimates for the intercept, grouped by the variable Tomato nested in Soil, i.e. interaction of Tomato and Soil.

Load the sample data.

Fit a linear mixed-effects model with a random intercept grouped by operator, to assess if there is a significant difference in the performance according to the time of the shift. Use the restricted maximum likelihood method.

lme = fitlme(shift,'QCDev ~ Shift + (1|Operator)');

Compute the 99% confidence intervals for random effects using the residuals option to compute the degrees of freedom. This is the default method.

[~,~,stats] = randomEffects(lme,'Alpha',0.01)
stats =
Random effect coefficients: DFMethod = 'Residual', Alpha = 0.01

Group               Level        Name                   Estimate    SEPred
{'Operator'}        {'1'}        {'(Intercept)'}        0.57753     0.90378
{'Operator'}        {'2'}        {'(Intercept)'}         1.1757     0.90378
{'Operator'}        {'3'}        {'(Intercept)'}        -2.1715     0.90378
{'Operator'}        {'4'}        {'(Intercept)'}         2.3655     0.90378
{'Operator'}        {'5'}        {'(Intercept)'}        -1.9472     0.90378

tStat      DF    pValue      Lower       Upper
0.63902    12     0.53482     -2.1831     3.3382
1.3009    12     0.21772     -1.5849     3.9364
-2.4027    12    0.033352     -4.9322    0.58909
2.6174    12    0.022494    -0.39511     5.1261
-2.1546    12    0.052216     -4.7079    0.81337

Compute the 99% confidence intervals for random effects using the Satterthwaite approximation to compute the degrees of freedom.

[~,~,stats] = randomEffects(lme,'DFMethod','satterthwaite','Alpha',0.01)
stats =
Random effect coefficients: DFMethod = 'Satterthwaite', Alpha = 0.01

Group               Level        Name                   Estimate    SEPred
{'Operator'}        {'1'}        {'(Intercept)'}        0.57753     0.90378
{'Operator'}        {'2'}        {'(Intercept)'}         1.1757     0.90378
{'Operator'}        {'3'}        {'(Intercept)'}        -2.1715     0.90378
{'Operator'}        {'4'}        {'(Intercept)'}         2.3655     0.90378
{'Operator'}        {'5'}        {'(Intercept)'}        -1.9472     0.90378

tStat      DF        pValue      Lower       Upper
0.63902    6.4253      0.5449      -2.684     3.839
1.3009    6.4253     0.23799     -2.0858    4.4372
-2.4027    6.4253    0.050386      -5.433      1.09
2.6174    6.4253    0.037302    -0.89598     5.627
-2.1546    6.4253    0.071626     -5.2087    1.3142

The Satterthwaite method usually produces smaller values for the degrees of freedom (DF), which results in larger p-values (pValue) and larger confidence intervals (Lower and Upper) for the random-effects estimates.