1 Desempenhos de alunos de CFCs
1.1 Pacotes
library(lavaan)
library(semPlot)
library(semTools)
library(foreign)
library(tidyverse)
library(car)
library(dplyr)
library(rio)
library(ggplot2)
library(gridExtra)
library(ifaTools)
library(psych)
library(ggcorrplot)
library(polycor)2 Dataset
CFC_data = import(file = "CFCsFull.csv")
CFC_data$V1 = NULL
names(CFC_data)## [1] "Estado" "Q1A" "Q2A" "Q3A" "Q4A" "Q5A" "Q6A" "Q7A"
## [9] "Q8A" "Q9A"2.1 View
CFC_data = CFC_data %>% mutate_at(vars(starts_with("Q")), ordered)
head(CFC_data)2.2 Descrição
summary(CFC_data[, 2:10])## Q1A Q2A Q3A Q4A Q5A Q6A Q7A Q8A Q9A
## 1:259 1:260 1:285 1:158 1:124 1:214 1:107 1:111 1: 89
## 2:173 2:177 2:171 2:160 2:206 2:153 2:207 2:197 2:174
## 3:103 3: 99 3: 89 3:139 3:128 3:109 3:167 3:168 3:178
## 4:102 4:101 4: 92 4:180 4:179 4:161 4:156 4:161 4:1963 Recomendação
O IDEAL É TER 5 PESSOAS POR ITEM DA ESCALA
Obtendo dados por estado
CFC_data_RS = CFC_data %>% dplyr::filter(Estado=="RS")
CFC_data_SP = CFC_data %>% dplyr::filter(Estado=="SP")
CFC_data_PR = CFC_data %>% dplyr::filter(Estado=="PR")3.1 Dados com > 300 PESSOAS POR ESTADO
CFC_data_3UFs = rbind(CFC_data_PR, CFC_data_RS, CFC_data_SP)4 Filtro
CFC_data_3UFs_N = CFC_data_3UFs %>%
dplyr::select(Q1A:Q9A) %>%
dplyr::select(ends_with('A')) %>%
mutate_all(~as.character(.) %>%
as.numeric())4.1 Densidades dos itens
plot(density(CFC_data_3UFs_N$Q1A),
lwd = 2,
col = "#f72585")
lines(density(CFC_data_3UFs_N$Q2A),
lwd = 2,
col = "#b5179e")
lines(density(CFC_data_3UFs_N$Q3A),
lwd = 2,
col = "#7209b7")
lines(density(CFC_data_3UFs_N$Q4A),
lwd = 2,
col = "#560bad")
lines(density(CFC_data_3UFs_N$Q5A),
lwd = 2,
col = "#480ca8")
lines(density(CFC_data_3UFs_N$Q6A),
lwd = 2,
col = "#3f37c9")
lines(density(CFC_data_3UFs_N$Q7A),
lwd = 2,
col = "#4361ee")
lines(density(CFC_data_3UFs_N$Q8A),
lwd = 2,
col = "#4895ef")
lines(density(CFC_data_3UFs_N$Q9A),
lwd = 2,
col = "#4cc9f0")
5 CALCULO DE MATRIZ POLICORICA
CFC_EFA_MATRIZ = psych::mixedCor(CFC_data_3UFs_N, c=NULL, smooth = TRUE, p=1:9)
CFC_EFA_MATRIZ_POLI = CFC_EFA_MATRIZ$rho6 PLOTAGEM DE MATRIZ DE CORRELAÇÃO POLICORICA
ggcorrplot(CFC_EFA_MATRIZ_POLI) + annotate("text", x =3, y = 7, label = "BORAAAAAA",
color="BLACK", size=10, angle=45)+ annotate("text", x = 5, y = 3, label = "CARALEO",
color="BLACK", size=10, angle=45)
7 QUÃO FASTORÁVEL É ESTE BANCO
Critério 1: bartlett test p < 0.05
cortest.bartlett(CFC_EFA_MATRIZ_POLI)## $chisq
## [1] 560.6496
##
## $p.value
## [1] 2.201549e-95
##
## $df
## [1] 36Critério 2: Kaiser-Meyer-Olkin factor adequacy - MSA próximo de 1 (0 a 1)
KMO(CFC_EFA_MATRIZ_POLI)## Kaiser-Meyer-Olkin factor adequacy
## Call: KMO(r = CFC_EFA_MATRIZ_POLI)
## Overall MSA = 0.86
## MSA for each item =
## Q1A Q2A Q3A Q4A Q5A Q6A Q7A Q8A Q9A
## 0.88 0.84 0.89 0.79 0.92 0.82 0.87 0.89 0.868 EXPLORANDO O NÚMERO DE FATORES COM O SCREE PLOT plot
O “eigen values” indica o quanto de informação eu posso extrair do fator em questão
fa.parallel(CFC_EFA_MATRIZ_POLI, n.obs = nrow(CFC_data_3UFs_N), fa = "fa", fm = "ml")
## Parallel analysis suggests that the number of factors = 3 and the number of components = NA9 EFA
9.1 EFA BASEADO EM RESULTADOS DA ANALISE PARALELA
“oblimin” = Rotação Obliqua: (Permite que todos fatores carreguem informação sobre os itens, no caso de rotações ortogonais, teríamos apenas um fator por item).
rotate: “none”, “varimax”, “quartimax”, “bentlerT”, “equamax”, “varimin”, “geominT” and “bifactor” are orthogonal rotations. “Promax”, “promax”, “oblimin”, “simplimax”, “bentlerQ,”geominQ” and “biquartimin” and “cluster” are possible oblique transformations of the solution. The default is to do a oblimin transformation, although versions prior to 2009 defaulted to varimax. SPSS seems to do a Kaiser normalization before doing Promax, this is done here by the call to “promax” which does the normalization before calling Promax in GPArotation.
“ml” = maximum likelihood factor analysis:
Factoring method fm=“minres” will do a minimum residual as will fm=“uls”. Both of these use a first derivative. fm=“ols” differs very slightly from “minres” in that it minimizes the entire residual matrix using an OLS procedure but uses the empirical first derivative. This will be slower. fm=“wls” will do a weighted least squares (WLS) solution, fm=“gls” does a generalized weighted least squares (GLS), fm=“pa” will do the principal factor solution, fm=“ml” will do a maximum likelihood factor analysis. fm=“minchi” will minimize the sample size weighted chi square when treating pairwise correlations with different number of subjects per pair. fm =“minrank” will do a minimum rank factor analysis. “old.min” will do minimal residual the way it was done prior to April, 2017 (see discussion below). fm=“alpha” will do alpha factor analysis as described in Kaiser and Coffey (1965)
CFC_EFA_4F = psych::fa(CFC_EFA_MATRIZ_POLI, nfactors = 4, fm = "ml",
rotate = "oblimin", n.obs = nrow(CFC_data_3UFs_N))CFC_EFA_4F## Factor Analysis using method = ml
## Call: psych::fa(r = CFC_EFA_MATRIZ_POLI, nfactors = 4, n.obs = nrow(CFC_data_3UFs_N),
## rotate = "oblimin", fm = "ml")
## Standardized loadings (pattern matrix) based upon correlation matrix
## ML3 ML2 ML4 ML1 h2 u2 com
## Q1A 0.00 0.01 0.01 0.99 1.00 0.005 1.0
## Q2A -0.01 0.02 0.93 0.00 0.88 0.122 1.0
## Q3A 0.14 -0.02 0.43 0.19 0.44 0.564 1.6
## Q4A -0.04 0.97 -0.06 0.03 0.88 0.120 1.0
## Q5A 0.13 0.71 0.01 0.05 0.66 0.339 1.1
## Q6A 0.01 0.88 0.09 -0.04 0.82 0.177 1.0
## Q7A 0.86 0.01 0.04 -0.03 0.76 0.239 1.0
## Q8A 0.82 -0.04 0.01 0.04 0.67 0.332 1.0
## Q9A 0.89 0.04 -0.04 0.00 0.81 0.195 1.0
##
## ML3 ML2 ML4 ML1
## SS loadings 2.33 2.29 1.18 1.12
## Proportion Var 0.26 0.25 0.13 0.12
## Cumulative Var 0.26 0.51 0.64 0.77
## Proportion Explained 0.34 0.33 0.17 0.16
## Cumulative Proportion 0.34 0.67 0.84 1.00
##
## With factor correlations of
## ML3 ML2 ML4 ML1
## ML3 1.00 0.56 0.49 0.51
## ML2 0.56 1.00 0.43 0.46
## ML4 0.49 0.43 1.00 0.70
## ML1 0.51 0.46 0.70 1.00
##
## Mean item complexity = 1.1
## Test of the hypothesis that 4 factors are sufficient.
##
## df null model = 36 with the objective function = 5.89 with Chi Square = 3724.25
## df of the model are 6 and the objective function was 0.02
##
## The root mean square of the residuals (RMSR) is 0.01
## The df corrected root mean square of the residuals is 0.02
##
## The harmonic n.obs is 637 with the empirical chi square 2.3 with prob < 0.89
## The total n.obs was 637 with Likelihood Chi Square = 11.61 with prob < 0.071
##
## Tucker Lewis Index of factoring reliability = 0.991
## RMSEA index = 0.038 and the 90 % confidence intervals are 0 0.071
## BIC = -27.13
## Fit based upon off diagonal values = 1
## Measures of factor score adequacy
## ML3 ML2 ML4 ML1
## Correlation of (regression) scores with factors 0.95 0.97 0.95 1.00
## Multiple R square of scores with factors 0.91 0.93 0.90 0.99
## Minimum correlation of possible factor scores 0.82 0.87 0.79 0.99fa.diagram(CFC_EFA_4F)
Tucker Lewis Index of
factoring reliability = 1 | TLI > 0.9 = Bom
RMSEA index = 0 and the 90 % confidence intervals are 0 0.019 |
RMSEA < 0.08 = Bom
9.2 EFA BASEADO EM 3 FATORES
CFC_EFA_3F = psych::fa(CFC_EFA_MATRIZ_POLI, nfactors = 3, fm = "ml",
rotate = "oblimin", n.obs = nrow(CFC_data_3UFs_N))CFC_EFA_3F## Factor Analysis using method = ml
## Call: psych::fa(r = CFC_EFA_MATRIZ_POLI, nfactors = 3, n.obs = nrow(CFC_data_3UFs_N),
## rotate = "oblimin", fm = "ml")
## Standardized loadings (pattern matrix) based upon correlation matrix
## ML1 ML2 ML3 h2 u2 com
## Q1A 0.08 0.05 0.71 0.62 0.38 1.0
## Q2A 0.00 -0.05 0.88 0.72 0.28 1.0
## Q3A -0.05 0.06 0.69 0.50 0.50 1.0
## Q4A 0.98 -0.03 -0.04 0.88 0.12 1.0
## Q5A 0.71 0.13 0.04 0.66 0.34 1.1
## Q6A 0.87 0.00 0.06 0.81 0.19 1.0
## Q7A 0.00 0.86 0.02 0.76 0.24 1.0
## Q8A -0.04 0.82 0.04 0.67 0.33 1.0
## Q9A 0.04 0.89 -0.04 0.81 0.19 1.0
##
## ML1 ML2 ML3
## SS loadings 2.30 2.30 1.83
## Proportion Var 0.26 0.26 0.20
## Cumulative Var 0.26 0.51 0.71
## Proportion Explained 0.36 0.36 0.28
## Cumulative Proportion 0.36 0.72 1.00
##
## With factor correlations of
## ML1 ML2 ML3
## ML1 1.00 0.56 0.51
## ML2 0.56 1.00 0.59
## ML3 0.51 0.59 1.00
##
## Mean item complexity = 1
## Test of the hypothesis that 3 factors are sufficient.
##
## df null model = 36 with the objective function = 5.89 with Chi Square = 3724.25
## df of the model are 12 and the objective function was 0.04
##
## The root mean square of the residuals (RMSR) is 0.01
## The df corrected root mean square of the residuals is 0.02
##
## The harmonic n.obs is 637 with the empirical chi square 4.14 with prob < 0.98
## The total n.obs was 637 with Likelihood Chi Square = 24.32 with prob < 0.018
##
## Tucker Lewis Index of factoring reliability = 0.99
## RMSEA index = 0.04 and the 90 % confidence intervals are 0.016 0.063
## BIC = -53.16
## Fit based upon off diagonal values = 1
## Measures of factor score adequacy
## ML1 ML2 ML3
## Correlation of (regression) scores with factors 0.97 0.95 0.92
## Multiple R square of scores with factors 0.94 0.91 0.85
## Minimum correlation of possible factor scores 0.87 0.82 0.70 Tucker Lewis Index of
factoring reliability = 0.998 | TLI > 0.9 =
Bom
RMSEA index = 0.018 and the 90 % confidence intervals are 0.009 0.027
| RMSEA < 0.08 = Bom
9.3 Interpretação dos dados
Visualizando a matriz de cargas acima, o item Q4A é explicado majoritariamente pelo Fator ML1 \(0,96^2=92.16%\) enquanto o item Q9A é explicado majoritariamente pelo Fator ML2 0.88 \(0,88^2=77.44%\).
Standardized loadings (pattern matrix) based upon correlation
matrix
ML1 ML2 ML3 h2 u2 com
Q1A 0.03 0.05 0.73 0.61 0.39 1
Q2A 0.00 -0.05 0.86 0.70 0.30 1
Q3A -0.01 0.04 0.66 0.47 0.53 1
Q4A 0.96 -0.03 -0.02 0.86 0.14 1
Q5A 0.77 0.06 0.02 0.68 0.32 1
Q6A 0.87 0.01 0.03 0.79 0.21 1
Q7A 0.03 0.86 -0.01 0.75 0.25 1
Q8A -0.07 0.83 0.06 0.68 0.32 1
Q9A 0.05 0.88 -0.02 0.80 0.20
Os Fatores ML1,2,3…São correlacionados entre si (arcos) e eles explicam (carregam) como os itens (Q1A, Q2A…) são endossados. Nesse caso o Fator ML1 está carragando informação para explicar Q4A, Q6A, Q5A. Assim temos ML1 - 0.9 -> Q6A; ou seja, \(0.9^2 = 0.81\). Dessa maneira ML1 esplica 81% da variabilidade de Q6A.
fa.diagram(CFC_EFA_3F)
9.4 EFA DE 5 FATORES
Baseado na linha tracejada o scree plot sugere 5 fatores
CFC_EFA_5F = psych::fa(CFC_EFA_MATRIZ_POLI, nfactors = 5, fm = "ml",
rotate = "oblimin", n.obs = nrow(CFC_data_3UFs_N))
CFC_EFA_5F## Factor Analysis using method = ml
## Call: psych::fa(r = CFC_EFA_MATRIZ_POLI, nfactors = 5, n.obs = nrow(CFC_data_3UFs_N),
## rotate = "oblimin", fm = "ml")
## Standardized loadings (pattern matrix) based upon correlation matrix
## ML4 ML1 ML3 ML5 ML2 h2 u2 com
## Q1A 0.01 0.00 0.00 0.01 0.99 1.00 0.005 1.0
## Q2A 0.02 -0.02 0.01 0.87 0.01 0.78 0.218 1.0
## Q3A -0.02 0.11 -0.01 0.54 0.12 0.46 0.536 1.2
## Q4A 0.97 -0.02 -0.01 -0.07 0.04 0.88 0.117 1.0
## Q5A 0.71 0.08 0.06 0.00 0.05 0.66 0.339 1.1
## Q6A 0.88 0.01 0.00 0.11 -0.06 0.82 0.176 1.0
## Q7A 0.01 0.00 0.99 0.00 0.00 1.00 0.005 1.0
## Q8A 0.00 0.41 0.34 0.05 0.07 0.60 0.402 2.0
## Q9A 0.02 0.99 0.00 0.00 0.00 1.00 0.005 1.0
##
## ML4 ML1 ML3 ML5 ML2
## SS loadings 2.29 1.36 1.25 1.20 1.10
## Proportion Var 0.25 0.15 0.14 0.13 0.12
## Cumulative Var 0.25 0.41 0.54 0.68 0.80
## Proportion Explained 0.32 0.19 0.17 0.17 0.15
## Cumulative Proportion 0.32 0.51 0.68 0.85 1.00
##
## With factor correlations of
## ML4 ML1 ML3 ML5 ML2
## ML4 1.00 0.51 0.48 0.45 0.46
## ML1 0.51 1.00 0.78 0.45 0.45
## ML3 0.48 0.78 1.00 0.47 0.44
## ML5 0.45 0.45 0.47 1.00 0.74
## ML2 0.46 0.45 0.44 0.74 1.00
##
## Mean item complexity = 1.1
## Test of the hypothesis that 5 factors are sufficient.
##
## df null model = 36 with the objective function = 5.89 with Chi Square = 3724.25
## df of the model are 1 and the objective function was 0.01
##
## The root mean square of the residuals (RMSR) is 0.01
## The df corrected root mean square of the residuals is 0.03
##
## The harmonic n.obs is 637 with the empirical chi square 1.34 with prob < 0.25
## The total n.obs was 637 with Likelihood Chi Square = 5.25 with prob < 0.022
##
## Tucker Lewis Index of factoring reliability = 0.958
## RMSEA index = 0.082 and the 90 % confidence intervals are 0.025 0.156
## BIC = -1.2
## Fit based upon off diagonal values = 1
## Measures of factor score adequacy
## ML4 ML1 ML3 ML5 ML2
## Correlation of (regression) scores with factors 0.97 1.00 1.00 0.92 1.00
## Multiple R square of scores with factors 0.93 0.99 0.99 0.85 0.99
## Minimum correlation of possible factor scores 0.87 0.99 0.99 0.69 0.99fa.diagram(CFC_EFA_5F)
Tucker Lewis Index of
factoring reliability = 0.998 | TLI > 0.9 =
Bom
RMSEA index = 0.016 and the 90 % confidence intervals are 0 0.048
| RMSEA < 0.08 = Bom
9.5 EFA BI FATORIAL
Com fatores não correlacionados. Esta função entrega o valor de Ômega, que é como um Coeficiente Alfa de Cronbach (α ).
omega(CFC_EFA_MATRIZ_POLI, nfactors = 3, fm = "wls", rotate = "oblimin",
n.obs = nrow(CFC_data_3UFs_N), title = "EFA CFC BiFactor")
## EFA CFC BiFactor
## Call: omegah(m = m, nfactors = nfactors, fm = fm, key = key, flip = flip,
## digits = digits, title = title, sl = sl, labels = labels,
## plot = plot, n.obs = n.obs, rotate = rotate, Phi = Phi, option = option,
## covar = covar)
## Alpha: 0.89
## G.6: 0.92
## Omega Hierarchical: 0.75
## Omega H asymptotic: 0.79
## Omega Total 0.94
##
## Schmid Leiman Factor loadings greater than 0.2
## g F1* F2* F3* h2 h2 u2 p2 com
## Q1A 0.62 0.49 0.63 0.63 0.37 0.61 1.92
## Q2A 0.60 0.59 0.71 0.71 0.29 0.51 2.00
## Q3A 0.52 0.48 0.50 0.50 0.50 0.54 2.00
## Q4A 0.62 0.70 0.88 0.88 0.12 0.44 1.98
## Q5A 0.62 0.52 0.66 0.66 0.34 0.59 1.97
## Q6A 0.65 0.63 0.82 0.82 0.18 0.51 2.01
## Q7A 0.71 0.51 0.76 0.76 0.24 0.66 1.82
## Q8A 0.66 0.49 0.67 0.67 0.33 0.64 1.86
## Q9A 0.72 0.53 0.81 0.81 0.19 0.65 1.84
##
## With Sums of squares of:
## g F1* F2* F3* h2
## 3.67 1.16 0.78 0.82 4.70
##
## general/max 0.78 max/min = 5.99
## mean percent general = 0.57 with sd = 0.08 and cv of 0.13
## Explained Common Variance of the general factor = 0.57
##
## The degrees of freedom are 12 and the fit is 0.04
## The number of observations was 637 with Chi Square = 24.79 with prob < 0.016
## The root mean square of the residuals is 0.01
## The df corrected root mean square of the residuals is 0.02
## RMSEA index = 0.041 and the 10 % confidence intervals are 0.017 0.064
## BIC = -52.69
##
## Compare this with the adequacy of just a general factor and no group factors
## The degrees of freedom for just the general factor are 27 and the fit is 2.14
## The number of observations was 637 with Chi Square = 1353.99 with prob < 1.4e-268
## The root mean square of the residuals is 0.16
## The df corrected root mean square of the residuals is 0.18
##
## RMSEA index = 0.278 and the 10 % confidence intervals are 0.265 0.291
## BIC = 1179.66
##
## Measures of factor score adequacy
## g F1* F2* F3*
## Correlation of scores with factors 0.87 0.83 0.71 0.75
## Multiple R square of scores with factors 0.75 0.70 0.50 0.56
## Minimum correlation of factor score estimates 0.51 0.39 0.00 0.11
##
## Total, General and Subset omega for each subset
## g F1* F2* F3*
## Omega total for total scores and subscales 0.94 0.92 0.90 0.82
## Omega general for total scores and subscales 0.75 0.47 0.58 0.46
## Omega group for total scores and subscales 0.19 0.45 0.31 0.37Aqui podemos ler “With Sums of squares of” como “With eigenvalues of”. Dessa maneira o fator geral “g”, explica muito mais que os fatores F1, F2 e F3 que correspondem aos fatorews ML1, ML2 e ML3.
With Sums of squares of:
g F1* F2* F3*
3.80 1.15 0.70 0.69 4.56
Além disso, como os fatores são não correlacionados, podemos perceber que a matriz acima é uma matriz esparsa, com cada item possuindo a carga de um fator apenas.
10 DIRETO DOS DADOS
10.1 Scree ploty
fa.parallel(CFC_data_3UFs_N, fa="fa", fm = "ml")
## Parallel analysis suggests that the number of factors = 3 and the number of components = NACFC_EFA_3F = psych::fa(CFC_data_3UFs_N, nfactors = 3, fm = "ml", rotate = "oblimin")
CFC_EFA_3F## Factor Analysis using method = ml
## Call: psych::fa(r = CFC_data_3UFs_N, nfactors = 3, rotate = "oblimin",
## fm = "ml")
## Standardized loadings (pattern matrix) based upon correlation matrix
## ML1 ML2 ML3 h2 u2 com
## Q1A 0.07 0.06 0.66 0.53 0.47 1.0
## Q2A 0.00 -0.05 0.81 0.62 0.38 1.0
## Q3A -0.05 0.05 0.64 0.41 0.59 1.0
## Q4A 0.93 -0.03 -0.04 0.81 0.19 1.0
## Q5A 0.67 0.12 0.04 0.58 0.42 1.1
## Q6A 0.83 0.00 0.06 0.73 0.27 1.0
## Q7A 0.00 0.82 0.01 0.68 0.32 1.0
## Q8A -0.04 0.76 0.06 0.60 0.40 1.0
## Q9A 0.04 0.85 -0.04 0.72 0.28 1.0
##
## ML1 ML2 ML3
## SS loadings 2.07 2.06 1.56
## Proportion Var 0.23 0.23 0.17
## Cumulative Var 0.23 0.46 0.63
## Proportion Explained 0.36 0.36 0.27
## Cumulative Proportion 0.36 0.73 1.00
##
## With factor correlations of
## ML1 ML2 ML3
## ML1 1.00 0.54 0.48
## ML2 0.54 1.00 0.56
## ML3 0.48 0.56 1.00
##
## Mean item complexity = 1
## Test of the hypothesis that 3 factors are sufficient.
##
## df null model = 36 with the objective function = 4.31 with Chi Square = 2726.97
## df of the model are 12 and the objective function was 0.02
##
## The root mean square of the residuals (RMSR) is 0.01
## The df corrected root mean square of the residuals is 0.02
##
## The harmonic n.obs is 637 with the empirical chi square 3.61 with prob < 0.99
## The total n.obs was 637 with Likelihood Chi Square = 12.82 with prob < 0.38
##
## Tucker Lewis Index of factoring reliability = 0.999
## RMSEA index = 0.01 and the 90 % confidence intervals are 0 0.042
## BIC = -64.66
## Fit based upon off diagonal values = 1
## Measures of factor score adequacy
## ML1 ML2 ML3
## Correlation of (regression) scores with factors 0.95 0.93 0.89
## Multiple R square of scores with factors 0.90 0.87 0.79
## Minimum correlation of possible factor scores 0.79 0.74 0.59Utilizando a base direta, sem processamento o RMSEA melhora. No entanto, com os dados não tratados (sem a matriz) não é possível gerar informações como o ggcorrplot.
Tucker Lewis Index of
factoring reliability = 0.999 | TLI > 0.9 =
Bom
RMSEA index = 0.008 and the 90 % confidence intervals are 0 0.019
| RMSEA < 0.08 = Bom
fa.diagram(CFC_EFA_3F)
11 ANÁLISE DE REDE
library(EGAnet)
CFC_data_3UFs_NET = CFC_data_3UFs %>%
dplyr::select(Q1A:Q9A) %>%
dplyr::select(ends_with('A'))
CFC_data_3UFs_NET_OUT = EGA(CFC_data_3UFs_NET, model = "glasso", plot.EGA = FALSE)
CFC_data_3UFs_NET_OUT## Number of communities: 3
##
## Q1A Q2A Q3A Q4A Q5A Q6A Q7A Q8A Q9A
## 2 2 2 1 1 1 3 3 3
##
## Methods:
##
## Correlations = auto (from qgraph)
## Model = glasso
## Algorithm = walktrap
## Unidimensional Method = louvain with consensus clusteringpar(mfrow=c(1,1))
plot(CFC_data_3UFs_NET_OUT, label.cex=2)
11.1 CFA DA REDE
CFA_CFC_NET_OUT = CFA(ega.obj = CFC_data_3UFs_NET_OUT, data = CFC_data_3UFs_NET, estimator = "WLSMV",
plot.CFA = FALSE)## [1] "Q4A" "Q5A" "Q6A"
## [1] "Q1A" "Q2A" "Q3A"
## [1] "Q7A" "Q8A" "Q9A"CFA_CFC_NET_OUT = CFA(ega.obj = CFC_data_3UFs_NET_OUT, data = CFC_data_3UFs_NET, estimator = "WLSMV",
plot.CFA = TRUE)## [1] "Q4A" "Q5A" "Q6A"
## [1] "Q1A" "Q2A" "Q3A"
## [1] "Q7A" "Q8A" "Q9A"
summary(CFA_CFC_NET_OUT)## Length Class Mode
## fit 1 lavaan S4
## summary 3 summaryDefault character
## fit.measures 5 lavaan.vector numeric12 SEM
summary.factor(CFC_data$Q1A)## 1 2 3 4
## 259 173 103 10212.1 Modelo
Cada espectro ou constructo (Habilidade, Depressão e CNH) vai explicar cada item (cada “X” explica ois vários “Ys”). Dessa forma os constructos (variáveis latentes) são correlacionados com os itens para explicar a variancia dos itens associados ao fator. Portanto, a direcionalidade é do fator para os itens.
- WLSMV: para itens ordinais (não contínuos)
modelo = '
Habilidade =~ Q1A + Q2A + Q3A
Experiencia =~ Q4A + Q5A + Q6A
CNH =~ Q7A + Q8A + Q9A
'12.2 CFA
AJUSTE_CFC_COR = cfa(modelo, data = CFC_data, std.lv=T, estimator = "WLSMV", ordered=colnames(CFC_data), )
summary(AJUSTE_CFC_COR, fit.measures=T, standardized=TRUE)## lavaan 0.6.17 ended normally after 20 iterations
##
## Estimator DWLS
## Optimization method NLMINB
## Number of model parameters 39
##
## Number of observations 637
##
## Model Test User Model:
## Standard Scaled
## Test Statistic 26.472 55.569
## Degrees of freedom 24 24
## P-value (Chi-square) 0.330 0.000
## Scaling correction factor 0.515
## Shift parameter 4.185
## simple second-order correction
##
## Model Test Baseline Model:
##
## Test statistic 13564.460 6826.589
## Degrees of freedom 36 36
## P-value 0.000 0.000
## Scaling correction factor 1.992
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 1.000 0.995
## Tucker-Lewis Index (TLI) 1.000 0.993
##
## Robust Comparative Fit Index (CFI) 0.989
## Robust Tucker-Lewis Index (TLI) 0.984
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.013 0.045
## 90 Percent confidence interval - lower 0.000 0.030
## 90 Percent confidence interval - upper 0.036 0.061
## P-value H_0: RMSEA <= 0.050 0.999 0.660
## P-value H_0: RMSEA >= 0.080 0.000 0.000
##
## Robust RMSEA 0.051
## 90 Percent confidence interval - lower 0.025
## 90 Percent confidence interval - upper 0.074
## P-value H_0: Robust RMSEA <= 0.050 0.452
## P-value H_0: Robust RMSEA >= 0.080 0.019
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.027 0.027
##
## Parameter Estimates:
##
## Parameterization Delta
## Standard errors Robust.sem
## Information Expected
## Information saturated (h1) model Unstructured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Habilidade =~
## Q1A 0.841 0.026 32.428 0.000 0.841 0.841
## Q2A 0.794 0.027 29.226 0.000 0.794 0.794
## Q3A 0.696 0.032 21.727 0.000 0.696 0.696
## Experiencia =~
## Q4A 0.901 0.014 63.716 0.000 0.901 0.901
## Q5A 0.836 0.018 46.283 0.000 0.836 0.836
## Q6A 0.917 0.016 58.660 0.000 0.917 0.917
## CNH =~
## Q7A 0.874 0.017 52.189 0.000 0.874 0.874
## Q8A 0.811 0.020 40.460 0.000 0.811 0.811
## Q9A 0.898 0.016 56.899 0.000 0.898 0.898
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Habilidade ~~
## Experiencia 0.553 0.036 15.258 0.000 0.553 0.553
## CNH 0.614 0.035 17.297 0.000 0.614 0.614
## Experiencia ~~
## CNH 0.592 0.033 18.167 0.000 0.592 0.592
##
## Thresholds:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Q1A|t1 -0.236 0.050 -4.707 0.000 -0.236 -0.236
## Q1A|t2 0.463 0.052 8.952 0.000 0.463 0.463
## Q1A|t3 0.994 0.060 16.639 0.000 0.994 0.994
## Q2A|t1 -0.232 0.050 -4.628 0.000 -0.232 -0.232
## Q2A|t2 0.485 0.052 9.342 0.000 0.485 0.485
## Q2A|t3 1.000 0.060 16.707 0.000 1.000 1.000
## Q3A|t1 -0.132 0.050 -2.652 0.008 -0.132 -0.132
## Q3A|t2 0.571 0.053 10.816 0.000 0.571 0.571
## Q3A|t3 1.061 0.061 17.297 0.000 1.061 1.061
## Q4A|t1 -0.681 0.054 -12.578 0.000 -0.681 -0.681
## Q4A|t2 -0.002 0.050 -0.040 0.968 -0.002 -0.002
## Q4A|t3 0.575 0.053 10.893 0.000 0.575 0.575
## Q5A|t1 -0.861 0.057 -15.101 0.000 -0.861 -0.861
## Q5A|t2 0.045 0.050 0.911 0.363 0.045 0.045
## Q5A|t3 0.580 0.053 10.971 0.000 0.580 0.580
## Q6A|t1 -0.424 0.051 -8.248 0.000 -0.424 -0.424
## Q6A|t2 0.192 0.050 3.838 0.000 0.192 0.192
## Q6A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q7A|t1 -0.962 0.059 -16.299 0.000 -0.962 -0.962
## Q7A|t2 -0.018 0.050 -0.356 0.722 -0.018 -0.018
## Q7A|t3 0.691 0.054 12.730 0.000 0.691 0.691
## Q8A|t1 -0.937 0.059 -16.023 0.000 -0.937 -0.937
## Q8A|t2 -0.041 0.050 -0.831 0.406 -0.041 -0.041
## Q8A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q9A|t1 -1.082 0.062 -17.488 0.000 -1.082 -1.082
## Q9A|t2 -0.220 0.050 -4.391 0.000 -0.220 -0.220
## Q9A|t3 0.502 0.052 9.653 0.000 0.502 0.502
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .Q1A 0.292 0.292 0.292
## .Q2A 0.369 0.369 0.369
## .Q3A 0.516 0.516 0.516
## .Q4A 0.189 0.189 0.189
## .Q5A 0.302 0.302 0.302
## .Q6A 0.159 0.159 0.159
## .Q7A 0.236 0.236 0.236
## .Q8A 0.343 0.343 0.343
## .Q9A 0.194 0.194 0.194
## Habilidade 1.000 1.000 1.000
## Experiencia 1.000 1.000 1.000
## CNH 1.000 1.000 1.000lavaan::summary(AJUSTE_CFC_COR, rsquare=TRUE)## lavaan 0.6.17 ended normally after 20 iterations
##
## Estimator DWLS
## Optimization method NLMINB
## Number of model parameters 39
##
## Number of observations 637
##
## Model Test User Model:
## Standard Scaled
## Test Statistic 26.472 55.569
## Degrees of freedom 24 24
## P-value (Chi-square) 0.330 0.000
## Scaling correction factor 0.515
## Shift parameter 4.185
## simple second-order correction
##
## Parameter Estimates:
##
## Parameterization Delta
## Standard errors Robust.sem
## Information Expected
## Information saturated (h1) model Unstructured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|)
## Habilidade =~
## Q1A 0.841 0.026 32.428 0.000
## Q2A 0.794 0.027 29.226 0.000
## Q3A 0.696 0.032 21.727 0.000
## Experiencia =~
## Q4A 0.901 0.014 63.716 0.000
## Q5A 0.836 0.018 46.283 0.000
## Q6A 0.917 0.016 58.660 0.000
## CNH =~
## Q7A 0.874 0.017 52.189 0.000
## Q8A 0.811 0.020 40.460 0.000
## Q9A 0.898 0.016 56.899 0.000
##
## Covariances:
## Estimate Std.Err z-value P(>|z|)
## Habilidade ~~
## Experiencia 0.553 0.036 15.258 0.000
## CNH 0.614 0.035 17.297 0.000
## Experiencia ~~
## CNH 0.592 0.033 18.167 0.000
##
## Thresholds:
## Estimate Std.Err z-value P(>|z|)
## Q1A|t1 -0.236 0.050 -4.707 0.000
## Q1A|t2 0.463 0.052 8.952 0.000
## Q1A|t3 0.994 0.060 16.639 0.000
## Q2A|t1 -0.232 0.050 -4.628 0.000
## Q2A|t2 0.485 0.052 9.342 0.000
## Q2A|t3 1.000 0.060 16.707 0.000
## Q3A|t1 -0.132 0.050 -2.652 0.008
## Q3A|t2 0.571 0.053 10.816 0.000
## Q3A|t3 1.061 0.061 17.297 0.000
## Q4A|t1 -0.681 0.054 -12.578 0.000
## Q4A|t2 -0.002 0.050 -0.040 0.968
## Q4A|t3 0.575 0.053 10.893 0.000
## Q5A|t1 -0.861 0.057 -15.101 0.000
## Q5A|t2 0.045 0.050 0.911 0.363
## Q5A|t3 0.580 0.053 10.971 0.000
## Q6A|t1 -0.424 0.051 -8.248 0.000
## Q6A|t2 0.192 0.050 3.838 0.000
## Q6A|t3 0.666 0.054 12.350 0.000
## Q7A|t1 -0.962 0.059 -16.299 0.000
## Q7A|t2 -0.018 0.050 -0.356 0.722
## Q7A|t3 0.691 0.054 12.730 0.000
## Q8A|t1 -0.937 0.059 -16.023 0.000
## Q8A|t2 -0.041 0.050 -0.831 0.406
## Q8A|t3 0.666 0.054 12.350 0.000
## Q9A|t1 -1.082 0.062 -17.488 0.000
## Q9A|t2 -0.220 0.050 -4.391 0.000
## Q9A|t3 0.502 0.052 9.653 0.000
##
## Variances:
## Estimate Std.Err z-value P(>|z|)
## .Q1A 0.292
## .Q2A 0.369
## .Q3A 0.516
## .Q4A 0.189
## .Q5A 0.302
## .Q6A 0.159
## .Q7A 0.236
## .Q8A 0.343
## .Q9A 0.194
## Habilidade 1.000
## Experiencia 1.000
## CNH 1.000
##
## R-Square:
## Estimate
## Q1A 0.708
## Q2A 0.631
## Q3A 0.484
## Q4A 0.811
## Q5A 0.698
## Q6A 0.841
## Q7A 0.764
## Q8A 0.657
## Q9A 0.806# AJUSTE_CFC_COR = cfa(modelo, data = CFC_data, std.lv=T, estimator = "WLSMV")
# summary(AJUSTE_CFC_COR, fit.measures=T, standardized=TRUE)12.2.1 Verificação
User Model versus
Baseline Model: Comparative Fit Index (CFI) 0.999 0.997 |
Critério CFI: > 0.9
Tucker-Lewis Index (TLI) 0.999 0.996 | Critério TLI: >
0.9
RMSEA 0.022 0.0355 | Critério TLI: < 0.08
12.2.2 Carga Fatorial (fatcor loadings)
Para explicar o quanto
a Habilidade (fator latente) explica a variância dos itens, os
coeficientes deverão ser elevados ao quadrado para obter o
percentual.
Desse modo, o fator Habilidade explica 64% da variabilidade no item Q2A
\(0.801^2=64.16\).
12.2.3 Thresholds
Os thresholds indicam o ponto onde há 50% de chance para responder um valor ou outro (os valores 1 ou 2 são equiprováveis para o valor de fator latente. Ou seja, no exemplo de Q2A|t1 -0.172 é a quantidade de “habilidade” que indica que o indivíduo possui 50% de chance de responder 1 ou 2 para o item Q2A. Do mesmo modo, para Q2A|t2; o nível 0.545 indica o quando de “habilidade” ou fator latente é necessário possuir para ficar no limiar(50%) para responder 2 ou 3 no item Q2A. Por fim 1.049 é o quanto de fator latente o indivíduo precisa possuir para ficar no limiar entre 3 e 4, Q2A|t3.
É importante que cada fator latente atravesse cada espectro t1, t2 e
t3, pois isso indica que o item pega desde quem tem pouco fator latente
até muito.
No caso do item Q1A, ele exige muito fator latente para discriminar,
pois o valor positivo (Q1A|t3 = 1.133 ) acima de 1.0
(correspondendo a 1 escore Z) indica a presença de
muito fator latente/habilidade para discriminar o item.
Em resumo o factor loading informa o quanto o item é informado a partir do nível do traço/fator latente. O theshold por sua vez, informa a dificuldade (necessidade de habilidade) ou o quanto de fator latent é necessário para endossar os items (atitudes)
12.2.3.1 Covariâncias
Informa a associação entre os fatores latentes. Por exemplo, Habilidade está muito correlacionado com CNH (0.659).
12.3 Confiabilidade
Todos os fatores latentes apresental um Alfa de Cronbach (α ) > 0.8 indicando alta confiabilidade. O mesmo ocorre com ômega acima de 0.8.
semTools::reliability(AJUSTE_CFC_COR)## Habilidade Experiencia CNH
## alpha 0.7586045 0.8713943 0.8561321
## alpha.ord 0.8204260 0.9107757 0.8958506
## omega 0.7685805 0.8804376 0.8583869
## omega2 0.7685805 0.8804376 0.8583869
## omega3 0.7676358 0.8863171 0.8575766
## avevar 0.6076032 0.7835445 0.742284312.3.1 Plotagem
semPaths(AJUSTE_CFC_COR, "std", layout = "tree2", edge.label.cex = 0.9,
residuals = F, intercepts = F, posCol = c("skyblue3"))
semPaths(AJUSTE_CFC_COR, "std", edge.label.cex = 0.9,
residuals = T, intercepts = F, posCol = c("skyblue3"))
semPaths(AJUSTE_CFC_COR, residuals = T, intercepts = F, "std", sizeMan = 6,posCol = c("skyblue3"),
edge.label.cex = 1.2, layout = "circle")
12.4 Modelo de Segunda Ordem
modelo = '
Habilidade =~ Q1A + Q2A + Q3A
Experiencia =~ Q4A + Q5A + Q6A
CNH =~ Q7A + Q8A + Q9A
EfeitoPositivo =~ Habilidade + Experiencia + CNH
'
AJUSTE_CFC_SEGUNDA_ORDEM = cfa(modelo, data = CFC_data, std.lv=T, estimator = "WLSMV")Nesse caso, os factor loadings devem ser verificados na coluna “Std.all”.
summary(AJUSTE_CFC_SEGUNDA_ORDEM, fit.measures=T, standardized=TRUE)## lavaan 0.6.17 ended normally after 28 iterations
##
## Estimator DWLS
## Optimization method NLMINB
## Number of model parameters 39
##
## Number of observations 637
##
## Model Test User Model:
## Standard Scaled
## Test Statistic 26.472 55.569
## Degrees of freedom 24 24
## P-value (Chi-square) 0.330 0.000
## Scaling correction factor 0.515
## Shift parameter 4.185
## simple second-order correction
##
## Model Test Baseline Model:
##
## Test statistic 13564.460 6826.589
## Degrees of freedom 36 36
## P-value 0.000 0.000
## Scaling correction factor 1.992
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 1.000 0.995
## Tucker-Lewis Index (TLI) 1.000 0.993
##
## Robust Comparative Fit Index (CFI) 0.989
## Robust Tucker-Lewis Index (TLI) 0.984
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.013 0.045
## 90 Percent confidence interval - lower 0.000 0.030
## 90 Percent confidence interval - upper 0.036 0.061
## P-value H_0: RMSEA <= 0.050 0.999 0.660
## P-value H_0: RMSEA >= 0.080 0.000 0.000
##
## Robust RMSEA 0.051
## 90 Percent confidence interval - lower 0.025
## 90 Percent confidence interval - upper 0.074
## P-value H_0: Robust RMSEA <= 0.050 0.452
## P-value H_0: Robust RMSEA >= 0.080 0.019
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.027 0.027
##
## Parameter Estimates:
##
## Parameterization Delta
## Standard errors Robust.sem
## Information Expected
## Information saturated (h1) model Unstructured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Habilidade =~
## Q1A 0.550 0.040 13.737 0.000 0.841 0.841
## Q2A 0.519 0.036 14.357 0.000 0.794 0.794
## Q3A 0.455 0.034 13.552 0.000 0.696 0.696
## Experiencia =~
## Q4A 0.615 0.032 19.469 0.000 0.901 0.901
## Q5A 0.571 0.031 18.438 0.000 0.836 0.836
## Q6A 0.627 0.034 18.525 0.000 0.917 0.917
## CNH =~
## Q7A 0.512 0.043 11.988 0.000 0.874 0.874
## Q8A 0.474 0.039 12.017 0.000 0.811 0.811
## Q9A 0.525 0.043 12.120 0.000 0.898 0.898
## EfeitoPositivo =~
## Habilidade 1.159 0.129 8.992 0.000 0.757 0.757
## Experiencia 1.069 0.104 10.304 0.000 0.730 0.730
## CNH 1.385 0.172 8.077 0.000 0.811 0.811
##
## Thresholds:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Q1A|t1 -0.236 0.050 -4.707 0.000 -0.236 -0.236
## Q1A|t2 0.463 0.052 8.952 0.000 0.463 0.463
## Q1A|t3 0.994 0.060 16.639 0.000 0.994 0.994
## Q2A|t1 -0.232 0.050 -4.628 0.000 -0.232 -0.232
## Q2A|t2 0.485 0.052 9.342 0.000 0.485 0.485
## Q2A|t3 1.000 0.060 16.707 0.000 1.000 1.000
## Q3A|t1 -0.132 0.050 -2.652 0.008 -0.132 -0.132
## Q3A|t2 0.571 0.053 10.816 0.000 0.571 0.571
## Q3A|t3 1.061 0.061 17.297 0.000 1.061 1.061
## Q4A|t1 -0.681 0.054 -12.578 0.000 -0.681 -0.681
## Q4A|t2 -0.002 0.050 -0.040 0.968 -0.002 -0.002
## Q4A|t3 0.575 0.053 10.893 0.000 0.575 0.575
## Q5A|t1 -0.861 0.057 -15.101 0.000 -0.861 -0.861
## Q5A|t2 0.045 0.050 0.911 0.363 0.045 0.045
## Q5A|t3 0.580 0.053 10.971 0.000 0.580 0.580
## Q6A|t1 -0.424 0.051 -8.248 0.000 -0.424 -0.424
## Q6A|t2 0.192 0.050 3.838 0.000 0.192 0.192
## Q6A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q7A|t1 -0.962 0.059 -16.299 0.000 -0.962 -0.962
## Q7A|t2 -0.018 0.050 -0.356 0.722 -0.018 -0.018
## Q7A|t3 0.691 0.054 12.730 0.000 0.691 0.691
## Q8A|t1 -0.937 0.059 -16.023 0.000 -0.937 -0.937
## Q8A|t2 -0.041 0.050 -0.831 0.406 -0.041 -0.041
## Q8A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q9A|t1 -1.082 0.062 -17.488 0.000 -1.082 -1.082
## Q9A|t2 -0.220 0.050 -4.391 0.000 -0.220 -0.220
## Q9A|t3 0.502 0.052 9.653 0.000 0.502 0.502
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .Q1A 0.292 0.292 0.292
## .Q2A 0.369 0.369 0.369
## .Q3A 0.516 0.516 0.516
## .Q4A 0.189 0.189 0.189
## .Q5A 0.302 0.302 0.302
## .Q6A 0.159 0.159 0.159
## .Q7A 0.236 0.236 0.236
## .Q8A 0.343 0.343 0.343
## .Q9A 0.194 0.194 0.194
## .Habilidade 1.000 0.427 0.427
## .Experiencia 1.000 0.467 0.467
## .CNH 1.000 0.343 0.343
## EfeitoPositivo 1.000 1.000 1.000semTools::reliability(AJUSTE_CFC_SEGUNDA_ORDEM)## Habilidade Experiencia CNH
## alpha 0.7586045 0.8713943 0.8561321
## alpha.ord 0.8204260 0.9107757 0.8958506
## omega 0.7685806 0.8804376 0.8583870
## omega2 0.7685806 0.8804376 0.8583870
## omega3 0.7676360 0.8863171 0.8575767
## avevar 0.6076034 0.7835445 0.7422844semPaths(AJUSTE_CFC_SEGUNDA_ORDEM, "std", layout = "tree2", edge.label.cex = 0.9,
residuals = T, intercepts = F, posCol = c("skyblue3"))
semPaths(AJUSTE_CFC_SEGUNDA_ORDEM, residuals = T, intercepts = F, "std", sizeMan = 6,posCol = c("skyblue3"),
edge.label.cex = 1.2, layout = "circle")
12.5 Modelo com fatores não correlacionados (modelo ortogonal)
modelo = '
Habilidade =~ Q1A + Q2A + Q3A
Experiencia =~ Q4A + Q5A + Q6A
CNH =~ Q7A + Q8A + Q9A
Habilidade ~0* Experiencia
Habilidade ~0* CNH
Experiencia ~0* CNH
'
AJUSTE_CFC_NAO_COR = cfa(modelo, data = CFC_data, std.lv=T, estimator = "WLSMV")summary(AJUSTE_CFC_NAO_COR, fit.measures=T, standardized=TRUE)## lavaan 0.6.17 ended normally after 1 iteration
##
## Estimator DWLS
## Optimization method NLMINB
## Number of model parameters 36
##
## Number of observations 637
##
## Model Test User Model:
## Standard Scaled
## Test Statistic 3196.158 1604.472
## Degrees of freedom 27 27
## P-value (Chi-square) 0.000 0.000
## Scaling correction factor 2.009
## Shift parameter 13.561
## simple second-order correction
##
## Model Test Baseline Model:
##
## Test statistic 13564.460 6826.589
## Degrees of freedom 36 36
## P-value 0.000 0.000
## Scaling correction factor 1.992
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 0.766 0.768
## Tucker-Lewis Index (TLI) 0.688 0.690
##
## Robust Comparative Fit Index (CFI) 0.866
## Robust Tucker-Lewis Index (TLI) 0.821
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.430 0.303
## 90 Percent confidence interval - lower 0.417 0.291
## 90 Percent confidence interval - upper 0.442 0.316
## P-value H_0: RMSEA <= 0.050 0.000 0.000
## P-value H_0: RMSEA >= 0.080 1.000 1.000
##
## Robust RMSEA 0.170
## 90 Percent confidence interval - lower 0.154
## 90 Percent confidence interval - upper 0.187
## P-value H_0: Robust RMSEA <= 0.050 0.000
## P-value H_0: Robust RMSEA >= 0.080 1.000
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.322 0.322
##
## Parameter Estimates:
##
## Parameterization Delta
## Standard errors Robust.sem
## Information Expected
## Information saturated (h1) model Unstructured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Habilidade =~
## Q1A 0.791 0.030 26.346 0.000 0.791 0.791
## Q2A 0.834 0.029 28.792 0.000 0.834 0.834
## Q3A 0.708 0.030 23.241 0.000 0.708 0.708
## Experiencia =~
## Q4A 0.930 0.013 72.467 0.000 0.930 0.930
## Q5A 0.803 0.019 42.125 0.000 0.803 0.803
## Q6A 0.907 0.015 59.294 0.000 0.907 0.907
## CNH =~
## Q7A 0.873 0.017 50.777 0.000 0.873 0.873
## Q8A 0.819 0.021 39.854 0.000 0.819 0.819
## Q9A 0.892 0.018 50.864 0.000 0.892 0.892
##
## Regressions:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Habilidade ~
## Experiencia 0.000 0.000 0.000
## CNH 0.000 0.000 0.000
## Experiencia ~
## CNH 0.000 0.000 0.000
##
## Thresholds:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Q1A|t1 -0.236 0.050 -4.707 0.000 -0.236 -0.236
## Q1A|t2 0.463 0.052 8.952 0.000 0.463 0.463
## Q1A|t3 0.994 0.060 16.639 0.000 0.994 0.994
## Q2A|t1 -0.232 0.050 -4.628 0.000 -0.232 -0.232
## Q2A|t2 0.485 0.052 9.342 0.000 0.485 0.485
## Q2A|t3 1.000 0.060 16.707 0.000 1.000 1.000
## Q3A|t1 -0.132 0.050 -2.652 0.008 -0.132 -0.132
## Q3A|t2 0.571 0.053 10.816 0.000 0.571 0.571
## Q3A|t3 1.061 0.061 17.297 0.000 1.061 1.061
## Q4A|t1 -0.681 0.054 -12.578 0.000 -0.681 -0.681
## Q4A|t2 -0.002 0.050 -0.040 0.968 -0.002 -0.002
## Q4A|t3 0.575 0.053 10.893 0.000 0.575 0.575
## Q5A|t1 -0.861 0.057 -15.101 0.000 -0.861 -0.861
## Q5A|t2 0.045 0.050 0.911 0.363 0.045 0.045
## Q5A|t3 0.580 0.053 10.971 0.000 0.580 0.580
## Q6A|t1 -0.424 0.051 -8.248 0.000 -0.424 -0.424
## Q6A|t2 0.192 0.050 3.838 0.000 0.192 0.192
## Q6A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q7A|t1 -0.962 0.059 -16.299 0.000 -0.962 -0.962
## Q7A|t2 -0.018 0.050 -0.356 0.722 -0.018 -0.018
## Q7A|t3 0.691 0.054 12.730 0.000 0.691 0.691
## Q8A|t1 -0.937 0.059 -16.023 0.000 -0.937 -0.937
## Q8A|t2 -0.041 0.050 -0.831 0.406 -0.041 -0.041
## Q8A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q9A|t1 -1.082 0.062 -17.488 0.000 -1.082 -1.082
## Q9A|t2 -0.220 0.050 -4.391 0.000 -0.220 -0.220
## Q9A|t3 0.502 0.052 9.653 0.000 0.502 0.502
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .Q1A 0.374 0.374 0.374
## .Q2A 0.304 0.304 0.304
## .Q3A 0.499 0.499 0.499
## .Q4A 0.135 0.135 0.135
## .Q5A 0.356 0.356 0.356
## .Q6A 0.177 0.177 0.177
## .Q7A 0.237 0.237 0.237
## .Q8A 0.329 0.329 0.329
## .Q9A 0.205 0.205 0.205
## .Habilidade 1.000 1.000 1.000
## .Experiencia 1.000 1.000 1.000
## CNH 1.000 1.000 1.000semTools::reliability(AJUSTE_CFC_NAO_COR)## Habilidade Experiencia CNH
## alpha 0.7586045 0.8713943 0.8561321
## alpha.ord 0.8204260 0.9107757 0.8958506
## omega 0.7685326 0.8781469 0.8586901
## omega2 0.7685326 0.8781469 0.8586901
## omega3 0.7685326 0.8781469 0.8586901
## avevar 0.6077639 0.7774797 0.7428473semPaths(AJUSTE_CFC_NAO_COR, "std", layout = "tree2", edge.label.cex = 0.9,
residuals = T, intercepts = F, posCol = c("skyblue3"))
semPaths(AJUSTE_CFC_NAO_COR, residuals = T, intercepts = F, "std", sizeMan = 6,posCol = c("skyblue3"),
edge.label.cex = 1.2, layout = "circle")
12.6 Bifatorial
modelo = '
Habilidade =~ Q1A + Q2A + Q3A
Experiencia =~ Q4A + Q5A + Q6A
CNH =~ Q7A + Q8A + Q9A
EfeitoNegativo =~ Q1A + Q2A + Q3A + Q4A + Q5A + Q6A + Q7A + Q8A + Q9A
EfeitoNegativo ~0* Habilidade
EfeitoNegativo ~0* Experiencia
EfeitoNegativo ~0* CNH
Habilidade ~0* Experiencia
Habilidade ~0* CNH
Experiencia ~0* CNH
'
AJUSTE_CFC_BIFAT = cfa(modelo, data = CFC_data, std.lv=T, estimator = "WLSMV")summary(AJUSTE_CFC_BIFAT, fit.measures=T, standardized=TRUE)## lavaan 0.6.17 ended normally after 29 iterations
##
## Estimator DWLS
## Optimization method NLMINB
## Number of model parameters 45
##
## Number of observations 637
##
## Model Test User Model:
## Standard Scaled
## Test Statistic 7.606 22.693
## Degrees of freedom 18 18
## P-value (Chi-square) 0.984 0.203
## Scaling correction factor 0.390
## Shift parameter 3.208
## simple second-order correction
##
## Model Test Baseline Model:
##
## Test statistic 13564.460 6826.589
## Degrees of freedom 36 36
## P-value 0.000 0.000
## Scaling correction factor 1.992
##
## User Model versus Baseline Model:
##
## Comparative Fit Index (CFI) 1.000 0.999
## Tucker-Lewis Index (TLI) 1.002 0.999
##
## Robust Comparative Fit Index (CFI) 0.999
## Robust Tucker-Lewis Index (TLI) 0.998
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.000 0.020
## 90 Percent confidence interval - lower 0.000 0.000
## 90 Percent confidence interval - upper 0.000 0.043
## P-value H_0: RMSEA <= 0.050 1.000 0.989
## P-value H_0: RMSEA >= 0.080 0.000 0.000
##
## Robust RMSEA 0.018
## 90 Percent confidence interval - lower 0.000
## 90 Percent confidence interval - upper 0.055
## P-value H_0: Robust RMSEA <= 0.050 0.914
## P-value H_0: Robust RMSEA >= 0.080 0.001
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.016 0.016
##
## Parameter Estimates:
##
## Parameterization Delta
## Standard errors Robust.sem
## Information Expected
## Information saturated (h1) model Unstructured
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Habilidade =~
## Q1A 0.460 0.054 8.541 0.000 0.460 0.460
## Q2A 0.595 0.059 10.038 0.000 0.595 0.595
## Q3A 0.479 0.057 8.437 0.000 0.479 0.479
## Experiencia =~
## Q4A 0.727 0.039 18.685 0.000 0.727 0.727
## Q5A 0.469 0.040 11.590 0.000 0.469 0.469
## Q6A 0.610 0.041 15.010 0.000 0.610 0.610
## CNH =~
## Q7A 0.506 0.052 9.747 0.000 0.506 0.506
## Q8A 0.507 0.056 9.072 0.000 0.507 0.507
## Q9A 0.507 0.054 9.420 0.000 0.507 0.507
## EfeitoNegativo =~
## Q1A 0.655 0.037 17.796 0.000 0.655 0.655
## Q2A 0.589 0.040 14.629 0.000 0.589 0.589
## Q3A 0.518 0.044 11.727 0.000 0.518 0.518
## Q4A 0.606 0.038 15.789 0.000 0.606 0.606
## Q5A 0.669 0.034 19.801 0.000 0.669 0.669
## Q6A 0.660 0.036 18.225 0.000 0.660 0.660
## Q7A 0.711 0.036 19.855 0.000 0.711 0.711
## Q8A 0.644 0.039 16.378 0.000 0.644 0.644
## Q9A 0.734 0.036 20.354 0.000 0.734 0.734
##
## Regressions:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## EfeitoNegativo ~
## Habilidade 0.000 0.000 0.000
## Experiencia 0.000 0.000 0.000
## CNH 0.000 0.000 0.000
## Habilidade ~
## Experiencia 0.000 0.000 0.000
## CNH 0.000 0.000 0.000
## Experiencia ~
## CNH 0.000 0.000 0.000
##
## Thresholds:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## Q1A|t1 -0.236 0.050 -4.707 0.000 -0.236 -0.236
## Q1A|t2 0.463 0.052 8.952 0.000 0.463 0.463
## Q1A|t3 0.994 0.060 16.639 0.000 0.994 0.994
## Q2A|t1 -0.232 0.050 -4.628 0.000 -0.232 -0.232
## Q2A|t2 0.485 0.052 9.342 0.000 0.485 0.485
## Q2A|t3 1.000 0.060 16.707 0.000 1.000 1.000
## Q3A|t1 -0.132 0.050 -2.652 0.008 -0.132 -0.132
## Q3A|t2 0.571 0.053 10.816 0.000 0.571 0.571
## Q3A|t3 1.061 0.061 17.297 0.000 1.061 1.061
## Q4A|t1 -0.681 0.054 -12.578 0.000 -0.681 -0.681
## Q4A|t2 -0.002 0.050 -0.040 0.968 -0.002 -0.002
## Q4A|t3 0.575 0.053 10.893 0.000 0.575 0.575
## Q5A|t1 -0.861 0.057 -15.101 0.000 -0.861 -0.861
## Q5A|t2 0.045 0.050 0.911 0.363 0.045 0.045
## Q5A|t3 0.580 0.053 10.971 0.000 0.580 0.580
## Q6A|t1 -0.424 0.051 -8.248 0.000 -0.424 -0.424
## Q6A|t2 0.192 0.050 3.838 0.000 0.192 0.192
## Q6A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q7A|t1 -0.962 0.059 -16.299 0.000 -0.962 -0.962
## Q7A|t2 -0.018 0.050 -0.356 0.722 -0.018 -0.018
## Q7A|t3 0.691 0.054 12.730 0.000 0.691 0.691
## Q8A|t1 -0.937 0.059 -16.023 0.000 -0.937 -0.937
## Q8A|t2 -0.041 0.050 -0.831 0.406 -0.041 -0.041
## Q8A|t3 0.666 0.054 12.350 0.000 0.666 0.666
## Q9A|t1 -1.082 0.062 -17.488 0.000 -1.082 -1.082
## Q9A|t2 -0.220 0.050 -4.391 0.000 -0.220 -0.220
## Q9A|t3 0.502 0.052 9.653 0.000 0.502 0.502
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .Q1A 0.359 0.359 0.359
## .Q2A 0.298 0.298 0.298
## .Q3A 0.502 0.502 0.502
## .Q4A 0.104 0.104 0.104
## .Q5A 0.332 0.332 0.332
## .Q6A 0.192 0.192 0.192
## .Q7A 0.238 0.238 0.238
## .Q8A 0.327 0.327 0.327
## .Q9A 0.204 0.204 0.204
## .Habilidade 1.000 1.000 1.000
## .Experiencia 1.000 1.000 1.000
## CNH 1.000 1.000 1.000
## .EfeitoNegativo 1.000 1.000 1.000semTools::reliability(AJUSTE_CFC_BIFAT)## Habilidade Experiencia CNH EfeitoNegativo
## alpha 0.7586045 0.8713943 0.8561321 0.8650410
## alpha.ord 0.8204260 0.9107757 0.8958506 0.8946570
## omega 0.3875913 0.4506366 0.3554737 0.6148398
## omega2 0.3173885 0.3915203 0.2874596 0.7260001
## omega3 0.3173886 0.3915203 0.2874596 0.7264330
## avevar NA NA NA NAsemPaths(AJUSTE_CFC_BIFAT, "std", layout = "tree2", edge.label.cex = 0.9,
residuals = T, intercepts = F, posCol = c("skyblue3"))
semPaths(AJUSTE_CFC_NAO_COR, residuals = T, intercepts = F, "std", sizeMan = 6,posCol = c("skyblue3"),
edge.label.cex = 1.2, layout = "circle")