T1. Taller encuestas. Descriptivos3
VIII Jornadas R Albacete 20162
Álvaro Hernández Vicente, Elvira Ferre Jaén, Antonio José Perán Orcajada, Ana Belén Marín Valverde, Antonio Maurandi López1
17 de noviembre de 2016
Lectura de datos
Se leen los datos saeraq.Rdata, que provienen del fichero raq.dat utilizado en (A. Field, Miles, and Field 2012) (disponible en la web del libro).
Y
load( "saeraq.RData" )Descriptivos con paquete tables
Se realizan descriptivos de cada variable con el paquete tables (Murdoch (2016)).
# library( tables )
op <- booktabs()
# Función para imprimir tablas en latex
tablaLatex <- function( tabla, caption = NULL ){
cat( '\\begin{table} \\centering\n' )
if( !is.null( caption ) ) cat( paste0( '\\caption{', caption, '}\n' ) )
latex( tt )
cat( '\\end{table}' )
}Variable sexo
tt <- tabular( ~ ( Sexo = sexo ) + ( Total = 1 ), data = df )
# tablaLatex( tt, caption = "Variable \\texttt{sexo}." )
html( tt, options = htmloptions( HTMLcaption = "Variable sexo", pad = TRUE ) )| Sexo | |||
|---|---|---|---|
| Mujer | Hombre | Total | |
| 1927 | 644 | 2571 | |
Variable actividadS
tt <- tabular( ~ ( `Actividad S` = actividadS ) + ( Total = 1 ), data = df )
# tablaLatex( tt, "Variable \\texttt{actividadS}." )
html( tt, options = htmloptions( HTMLcaption = "Variable actividadS", pad = TRUE ) )| Actividad S | |||||
|---|---|---|---|---|---|
| Nada | Poco | Mucho | Muchísimo | Total | |
| 613 | 658 | 669 | 631 | 2571 | |
Variable ingresos
ic1 <- function(x){
mean( x ) - qt( 0.975, df = length( x ) - 1 ) * sd( x ) / sqrt( length( x ) )
}
ic2 <- function(x){
mean( x ) + qt( 0.975, df = length( x ) - 1 ) * sd( x ) / sqrt( length( x ) )
}
tt <- tabular( ~ ( Ingresos = ingresos ) * ( ic1 + mean + sd + ic2 + median ) +
( Total = 1 ), data = df )
# tablaLatex( tt, "Variable \\texttt{ingresos}." )
html( tt, options = htmloptions( HTMLcaption = "Variable ingresos", pad = TRUE ) )| Ingresos | ||||||
|---|---|---|---|---|---|---|
| ic1 | mean | sd | ic2 | median | Total | |
| 29320 | 29672 | 9095 | 30024 | 27771 | 2571 | |
Variable origen
tt <- tabular( ~ ( Origen = origen ) + ( Total = 1 ), data = df )
# tablaLatex( tt, "Variable \\texttt{origen}." )
html( tt, options = htmloptions( HTMLcaption = "Variable origen", pad = TRUE ) )| Origen | ||||
|---|---|---|---|---|
| Albacete | Murcia | Helsinki | Total | |
| 1178 | 1112 | 281 | 2571 | |
Variable nivelIngles
tt <- tabular( ( `Nivel de inglés` = nivelIngles ) + ( Total = 1 ) ~ ( Frecuencia = 1 ),
data = df )
# tablaLatex( tt, "Variable \\texttt{nivelIngles}." )
html( tt, options = htmloptions( HTMLcaption = "Variable nivelIngles", pad = TRUE ) )| Nivel de inglés | Frecuencia |
|---|---|
| Nulo | 297 |
| CasiNulo | 289 |
| A1 | 451 |
| A2 | 455 |
| B1 | 473 |
| B2 | 377 |
| C1 | 145 |
| C2 | 79 |
| IsabelII | 3 |
| Shakespeare | 2 |
| Total | 2571 |
Descriptivos con paquete likert
Se realizan gráficos descriptivos de los datos con el paquete likert (Bryer and Speerschneider (2015)) que, al estar basados en ggplot2 (Wickham (2009)), permite modificarlos de forma sencilla.
# library( likert )
dfLikert <- df[ , grep( "^Q", colnames( df ) ) ]
colnames( dfLikert ) <- dicc[ grep( "^Q", dicc$item ), "spanish" ]
bloque1 <- 1:8
bloque2 <- 9:17
bloque3 <- 18:23
items1 <- likert( items = dfLikert[ , bloque1 ] )
items2 <- likert( items = dfLikert[ , bloque2 ] )
items3 <- likert( items = dfLikert[ , bloque3 ] )Gráficos de barras
plot( items1, centered = TRUE, group.order = colnames( items1$items ),
legend.position = "right" ) +
theme( axis.text.x = element_text( size = 10 ),
axis.text.y = element_text( size = 13, hjust = 0 ),
legend.text = element_text( size = 10 ),
legend.title = element_text( size = 10 ) )
Gráfico de barras de los ítems 1 a 8.
plot( items2, centered = TRUE, group.order = colnames( items2$items ),
legend.position = "right" ) +
theme( axis.text.x = element_text( size = 10 ),
axis.text.y = element_text( size = 13, hjust = 0 ),
legend.text = element_text( size = 10 ),
legend.title = element_text( size = 10 ) )
Gráfico de barras de los items 9 a 17.
plot( items3, centered = TRUE, group.order = colnames( items3$items ),
legend.position = "right" ) +
theme( axis.text.x = element_text( size = 10 ),
axis.text.y = element_text( size = 13, hjust = 0 ),
legend.text = element_text( size = 10 ),
legend.title = element_text( size = 10 ) )
Gráfico de barras de los items 18 a 23.
Gráficos de densidad
plot( items1, type = "density" )
Gráfico de densidad de los items 1 a 8.
plot( items2, type = "density" )
Gráfico de densidad de los items 9 a 17.
plot( items3, type = "density" )
Gráfico de densidad de los items 18 a 23.
Gráficos de calor
plot( items1, type = "heat", group.order = colnames( items1$items ) ) +
theme( axis.text.x = element_text( size = 8 ),
axis.text.y = element_text( size = 13, hjust = 0 ),
legend.text = element_text( size = 10 ),
legend.title = element_text( size = 10 ) )
Gráfico de calor de los items 1 a 8.
plot( items2, type = "heat", group.order = colnames( items2$items ) ) +
theme( axis.text.x = element_text( size = 8 ),
axis.text.y = element_text( size = 13, hjust = 0 ),
legend.text = element_text( size = 10 ),
legend.title = element_text( size = 10 ) )
Gráfico de calor de los items 9 a 17.
plot( items3, type = "heat", group.order = colnames( items3$items ) ) +
theme( axis.text.x = element_text( size = 8 ),
axis.text.y = element_text( size = 13, hjust = 0 ),
legend.text = element_text( size = 10 ),
legend.title = element_text( size = 10 ) )
Gráfico de calor de los items 18 a 23.
Correlaciones
Se estudian correlaciones con los paquetes corrr (Jackson (2016)) y corrplot (Wei and Simko (2016)).
# library( corrr )
# library( corrplot )
dfCor <- df[ , grep( "^Q", colnames( df ) ) ]Matriz de correlaciones
Se calcula la correlación con la función cor de la librería stats.
dfCor <- lapply( dfCor, as.numeric ) # devuelve una lista
dfCor <- as.data.frame( dfCor ) # coarcionamos a data.frame
corr <- cor( dfCor )
kable( corr[ 1:8, 1:8 ],
caption = "Tabla de correlaciones (solo se muestran las ocho primeras)" )| Q01 | Q02 | Q03 | Q04 | Q05 | Q06 | Q07 | Q08 | |
|---|---|---|---|---|---|---|---|---|
| Q01 | 1.0000000 | -0.0987240 | -0.3366489 | 0.4358602 | 0.4024399 | 0.2167340 | 0.3053651 | 0.3307376 |
| Q02 | -0.0987240 | 1.0000000 | 0.3183902 | -0.1118597 | -0.1193466 | -0.0742097 | -0.1591745 | -0.0496226 |
| Q03 | -0.3366489 | 0.3183902 | 1.0000000 | -0.3804602 | -0.3103088 | -0.2267405 | -0.3819533 | -0.2586342 |
| Q04 | 0.4358602 | -0.1118597 | -0.3804602 | 1.0000000 | 0.4006722 | 0.2782015 | 0.4086150 | 0.3494294 |
| Q05 | 0.4024399 | -0.1193466 | -0.3103088 | 0.4006722 | 1.0000000 | 0.2574601 | 0.3393918 | 0.2686270 |
| Q06 | 0.2167340 | -0.0742097 | -0.2267405 | 0.2782015 | 0.2574601 | 1.0000000 | 0.5135805 | 0.2228318 |
| Q07 | 0.3053651 | -0.1591745 | -0.3819533 | 0.4086150 | 0.3393918 | 0.5135805 | 1.0000000 | 0.2974970 |
| Q08 | 0.3307376 | -0.0496226 | -0.2586342 | 0.3494294 | 0.2686270 | 0.2228318 | 0.2974970 | 1.0000000 |
Otra opción es utilizar las funciones correlate y fashion de librería corrr (Jackson 2016).
corr2 <- correlate( dfCor )
kable( fashion( corr2[ 1:8, 1:9 ] ),
caption = "Tabla de correlaciones (solo se muestran las ocho primeras)" )| rowname | Q01 | Q02 | Q03 | Q04 | Q05 | Q06 | Q07 | Q08 |
|---|---|---|---|---|---|---|---|---|
| Q01 | -.10 | -.34 | .44 | .40 | .22 | .31 | .33 | |
| Q02 | -.10 | .32 | -.11 | -.12 | -.07 | -.16 | -.05 | |
| Q03 | -.34 | .32 | -.38 | -.31 | -.23 | -.38 | -.26 | |
| Q04 | .44 | -.11 | -.38 | .40 | .28 | .41 | .35 | |
| Q05 | .40 | -.12 | -.31 | .40 | .26 | .34 | .27 | |
| Q06 | .22 | -.07 | -.23 | .28 | .26 | .51 | .22 | |
| Q07 | .31 | -.16 | -.38 | .41 | .34 | .51 | .30 | |
| Q08 | .33 | -.05 | -.26 | .35 | .27 | .22 | .30 |
Gráficos de correlaciones
network_plot( corr2, min_cor = 0.4 )
Red de correlaciones. Valor correlación mínima
corrplot.mixed( corr, tl.pos = "lt", diag = 'n', upper = "ellipse",
number.cex = 0.4, tl.cex = 0.8,
order = "hclust" )
Matriz de correlaciones mixta
Referencias y bibliografía
Bryer, Jason, and Kimberly Speerschneider. 2015. Likert: Functions to Analyze and Visualize Likert Type Items. http://CRAN.R-project.org/package=likert.
Field, Andy, Jeremy Miles, and Zoe Field. 2012. Discovering Statistics Using R. 1st edition. Sage Publications Ltd.
Jackson, Simon. 2016. Corrr: Correlations in R. https://CRAN.R-project.org/package=corrr.
Murdoch, Duncan. 2016. Tables: Formula-Driven Table Generation. http://CRAN.R-project.org/package=tables.
Wei, Taiyun, and Viliam Simko. 2016. Corrplot: Visualization of a Correlation Matrix. http://CRAN.R-project.org/package=corrplot.
Wickham, Hadley. 2009. Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. http://ggplot2.org.
Xie, Yihui. 2015. Dynamic Documents with R and Knitr. Vol. 29. CRC Press.
Servicio de Apoyo Estadístico; alvarohv@um.es, elvira@um.es, antoniojose.peran@um.es, anabelen.marin4@um.es, amaurandi@um.es↩
doc:T1_descriptivos.Rmd↩