Atividade 1 - Alpha Cronbach

2005_uma análise utilizando o coeficiente alfa de Cronbach_XII SIMPEP

Atividade parcial do artigo:

LUÍS, A.; FREITAS, P. uma análise utilizando o coeficiente alfa de Cronbach. In: XII SIMPEP, Bauru. Anais… Bauru: 2005.

library(readxl)
# IMPORTÂNCIA DO ITENS
GI <- read_excel("2005_SIMPEP.xlsx", sheet = "GI")
# EXCLUIR A PRIMEIRA COLUNA
GI <- GI[,-1]
# DESEMPENHO DA IES
GD <- read_excel("2005_SIMPEP.xlsx", sheet = "GD")
# EXCLUIR A PRIMEIRA COLUNA
GD <- GD[,-1]

head(GI)

head(GD)

um primeiro exemplo passo a passo

adiante vamos usar o comando alpha do package psych

calcular o alpha de Cronbach da dimensão organização Institucional (I1 a I14) usando a base de dados Desempenho da IES armazenada em GD

library(psych)
#length(GD)
# str(GD)
I1aI14 <-GD[1:14,1:36]
I1aI14

# substituir em cada coluna os valores NA pelos valores médios das linhas
I1aI14$P5 [is.na(I1aI14$P5 )==T] <- mean(t(I1aI14$P5 ), na.rm = T)
I1aI14$P10[is.na(I1aI14$P10)==T] <- mean(t(I1aI14$P10), na.rm = T)
I1aI14$P16[is.na(I1aI14$P16)==T] <- mean(t(I1aI14$P16), na.rm = T)
I1aI14$P21[is.na(I1aI14$P21)==T] <- mean(t(I1aI14$P21), na.rm = T)
I1aI14$P25[is.na(I1aI14$P25)==T] <- mean(t(I1aI14$P25), na.rm = T)
I1aI14$P31[is.na(I1aI14$P31)==T] <- mean(t(I1aI14$P31), na.rm = T)
I1aI14$P33[is.na(I1aI14$P33)==T] <- mean(t(I1aI14$P33), na.rm = T)
I1aI14$P36[is.na(I1aI14$P36)==T] <- mean(t(I1aI14$P36), na.rm = T)
# variância por coluna (qi)
somavar <- sum(apply(I1aI14,2,var))
#variância total
# soma todas as linhas e obtem a var da coluna resultante
vartot <-var(apply(I1aI14,1,sum))
k <- length(I1aI14)
alfa <- (k/(k-1))*(1-(somavar)/vartot)
alfa

[1] 0.8874479

Existe uma forma mais simples para calcular o Alfa com o comando alpha usando o package psych

psych::alpha()

o nome antes do comando alpha é usado para não gerar conflito com outros pacotes que usam o mesmo nome (por ex: o ggplot2 também tem um comando denominado alpha)

# vamos usar os dados I1aI14 com valores médios já imputados quando há obs ausente (NA)
# Uma outra opção seria usar a base de dados original que o procedimento a seguir já excluiria as linhas com NA.
alfa_cronb <- psych::alpha(as.matrix(I1aI14))

Matrix was not positive definite, smoothing was doneSome items were negatively correlated with the total scale and probably 
should be reversed.  
To do this, run the function again with the 'check.keys=TRUE' option

Some items ( P16 P17 P18 P21 P28 ) were negatively correlated with the total scale and 
probably should be reversed.  
To do this, run the function again with the 'check.keys=TRUE' option

Matrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was doneMatrix was not positive definite, smoothing was done

alfa_cronb$total

library(ggplot2)
library(ggcorrplot)
# Correlation matrix
corr <- round(cor(I1aI14), 1)
#corr
# Plot
ggcorrplot(corr, hc.order = TRUE, 
           type = "lower", 
           lab = TRUE, 
           lab_size = 3, 
           method="circle", 
           colors = c("tomato2", "white", "springgreen3"), 
           title="Correlograma", 
           ggtheme=theme_bw)