| title | output |
|---|---|
ECLS-K-2011 |
html_document |
knitr::opts_chunk$set(echo = TRUE)
Here we are setting the WD to google drive. See earlier versions for getting the original data source.
#setwd("~/Google Drive/PARCS/Projects/ECLSK2011/Data")
#data = read.csv("ELCS-K-2011.csv", header = TRUE)
# Need to grab the second year versions of these. If no second version, create a second version with the "2 and 4" title so we can aggregate the data later.
data1 = cbind(X1PRNCON = data$X1PRNCON, X2PRNCON = data$X2PRNCON, X4PRNCON = data$X4PRNCON, X1PRNSOC = data$X1PRNSOC, X2PRNSOC = data$X2PRNSOC, X4PRNSOC = data$X4PRNSOC, X1BMI = data$X1BMI, X2BMI = data$X2BMI, X4BMI = data$X4BMI, X1PAR1AGE = data$X1PAR1AGE, X2PAR1AGE = data$X2PAR1AGE, X4PAR1AGE = X4PAR1AGE_R, X1PAR1EMP = data$X1PAR1EMP, X2PAR1EMP = data$X1PAR1EMP, X4PAR1EMP = data$X4PAR1EMP_I, X1HTOTAL = data$X1HTOTAL, X2HTOTAL = data$X2HTOTAL, X4HTOTAL = data$X4HTOTAL, X1NUMSIB = data$X1NUMSIB, X2NUMSIB = data$X2NUMSIB, X4NUMSIB = data$X4NUMSIB, X1POVTY = data$X2POVTY, X2POVTY = data$X2POVTY, X4POVTY_I = data$X4POVTY_I, X1SESL = data$X12SESL, X2SESL = data$X12SESL, X4SESL_I = data$X4SESL_I, X1PAR1ED_I = data$X12PAR1ED_I, X2PAR1ED_I = data$X12PAR1ED_I, X4PAR1ED_I = data$X4PAR1ED_I, X1LANGST = data$X12LANGST, X2LANGST = data$X12LANGST, X4LANGST = data$X4LANGST, X1_CHSEX_R = data$X_CHSEX_R, X2_CHSEX_R = data$X_CHSEX_R, X4_CHSEX_R = data$X_CHSEX_R, X1HPARNT = data$X1HPARNT, X2HPARNT = data$X2HPARNT, X4HPARNT = data$X4HPARNT, S1REGSKL = data$S2REGSKL, S2REGSKL = data$S2REGSKL, S4REGSKL = data$S2REGSKL, W1P0 = data$W1P0, data[,10894:10973])
head(data1)
# These are variables that will need to be changed to binary at a later date.
data2 = cbind(X12LANGST = data$X12LANGST, X_CHSEX_R = data$X_CHSEX_R, X1RESREL = data$X1RESREL, X1HPARNT = data$X1HPARNT, X1PRIMNW = data$X1PRIMNW)
head(data2)
# We will need to create new variables for each race with this variable.
data3 = cbind(X_RACETHP_R = data$X_RACETHP_R)
data4 = cbind( X1PAR1RAC = data$X1PAR1RAC)
data4 = as.data.frame(data4)
#Combining all four of the datas. Keeping them seperate for data transformations below
data1 = cbind(data1, data2, data3, data4)
head(data1)
# Change the -9 to NAs
data1 = apply(data1, 2, function(x){ifelse(x == -9, NA, x)})
data1 = as.data.frame(data1)
head(data1)
Now we need to alter the variables to be binary in necessary. First we create get all the variables where 1 is the interest and get those as 1 and rest as zero. Then for parent ethnicty we change the ones to zero and everything else to one to have a non-white be one. Then we need to grab a seperate subset of the all the remaining variables, so we don't double up on those variables when we cbind them togehter at the end. I then needed to get the variables in the same order and rename with meaningful names. Need to grab the correct data from data sets two and three, because those have the binary transformations.
data2 = cbind(X12LANGST = data1$X12LANGST, X_CHSEX_R = data1$X_CHSEX_R, X1RESREL = data1$X1RESREL, X1HPARNT = data1$X1HPARNT, X1PRIMNW = data1$X1PRIMNW)
data2 = ifelse(is.na(data2), NA, ifelse(data2 == 1, 1,0))
data2 = as.data.frame(data2)
head(data2)
# Here is the ethnicity variable that needs to be transformed into the original variables that you used. Remember that original variable were incorrect and just keeping the names the same here for consistency.
X_HISP_R = ifelse(is.na(data3), NA, ifelse(data3 == 3 | data3 == 4, 1, 0))
X_HISP_R = as.data.frame(X_HISP_R)
names(X_HISP_R) = c("X_HISP_R")
#Instead of changing all of the names, just replaced the other ethnicity with white
X_WHITE_R = ifelse(is.na(data3), NA, ifelse(data3 == 1, 1, 0))
X_WHITE_R = as.data.frame(X_WHITE_R)
names(X_WHITE_R) = c("X_WHITE_R")
sum(X_WHITE_R, na.rm =TRUE)
X_BLACK_R = ifelse(is.na(data3), NA, ifelse(data3 == 2, 1, 0))
X_BLACK_R = as.data.frame(X_BLACK_R)
names(X_BLACK_R) = c("X_BLACK_R")
X_ASIAN_R = ifelse(is.na(data3), NA, ifelse(data3 == 5, 1, 0))
X_ASIAN_R = as.data.frame(X_ASIAN_R)
names(X_ASIAN_R) = c("X_ASIAN_R")
X_AMINAN_R = ifelse(is.na(data3), NA, ifelse(data3 == 7, 1, 0))
X_AMINAN_R = as.data.frame(X_AMINAN_R)
names(X_AMINAN_R) = c("X_AMINAN_R")
X_HAWPI_R = ifelse(is.na(data3), NA, ifelse(data3 == 6, 1, 0))
X_HAWPI_R = as.data.frame(X_HAWPI_R)
names(X_HAWPI_R) = c("X_HAWPI_R")
X_MULTR_R = ifelse(is.na(data3), NA, ifelse(data3 == 8, 1, 0))
X_MULTR_R = as.data.frame(X_MULTR_R)
names(X_MULTR_R) = c("X_MULTR_R")
data4 = cbind(X_HISP_R, X_WHITE_R, X_BLACK_R, X_ASIAN_R, X_AMINAN_R, X_AMINAN_R, X_HAWPI_R, X_MULTR_R)
data4 = as.data.frame(data4)
# Need to grab the public (only regular public classrooms) versus private and magnet and charter: S2REGSKL. Need to reverse code, because it has 1 as public and one 1 as public.
data7 =
# Need to change 2 through 8 to be 1 and 1 to be zero and therefore 1 is all non-white parents
data6 = cbind(X1PAR1RAC = data1$X1PAR1RAC)
data6 = ifelse(is.na(data6), NA, ifelse(data6 == 1, 0,1))
data6 = as.data.frame(data6)
Reording the variables to be in the correct order. Grab each variable from the correct data set from above.
# Rearrange and then rename variables to get them in the correct order. This includes getting data2 and data3 into the correct order as well, because you need to rename all of these variables.
data5 = cbind(X1PRNCON = data1$X1PRNCON, X1PRNSOC = data1$X1PRNSOC, X1PRNSAD = data1$X1PRNSAD, X1PRNIMP = data1$X1PRNIMP, X1PRNAPP = data1$X1PRNAPP,X_HISP_R = data4$X_HISP_R, X_WHITE_R = data4$X_WHITE_R, X_BLACK_R = data4$X_BLACK_R, X_ASIAN_R = data4$X_ASIAN_R, X_AMINAN_R = data4$X_AMINAN_R, X_HAWPI_R = data4$X_HAWPI_R, X_MULTR_R = data4$X_MULTR_R, X_CHSEX_R = data2$X_CHSEX_R, X1BMI = data1$X1BMI,X1RESREL = data2$X1RESREL,X1HPARNT = data2$X1HPARNT, X1PAR1AGE = data1$X1PAR1AGE, X1PAR1RAC = data6$X1PAR1RAC,X12PAR1ED_I = data1$X12PAR1ED_I, X1PAR1EMP = data1$X1PAR1EMP, X1HTOTAL = data1$X1HTOTAL, X1NUMSIB = data1$X1NUMSIB, X1PRIMNW = data2$X1PRIMNW, X2POVTY = data1$X2POVTY, X12SESL = data1$X12SESL)
head(data5)
data1 = cbind(data5, W1P0 = data1[,19], data1[,20:99])
data1 = as.data.frame(data1)
head(data1)
Here we will use Amelia. Need to set m as five for five imputed data sets. Then we place each of the variables into their appropriate categories.
library(Amelia)
library(mitools)
library(survey)
m = 5
a.out = amelia(x = data1, m=m, ords = c("X1PAR1EMP", "X2POVTY", "X12PAR1ED_I"), logs = c("X1HTOTAL", "X1NUMSIB"), noms = c("X_HISP_R", "X_WHITE_R", "X_BLACK_R", "X_ASIAN_R", "X_AMINAN_R", "X_HAWPI_R", "X_MULTR_R", "X_CHSEX_R", "X1RESREL", "X1HPARNT", "X1PRIMNW", "X1PAR1RAC"))
# Now we can creat seperate data set and then analyze them seperately and combine them later with the mi.meld function in Ameila
write.amelia(obj = a.out, file.stem = "ECLSK")
head(a.out.imp[c(1:6)])
test = a.out.imp$imputations$imp1
Here we are analyzing data from the first imputted data set ECLSK1
setwd("~/Google Drive/PARCS/Projects/ECLSK2011/Data/ImputedFirst")
ECLSK1 = read.csv("ECLSK1.csv", header = TRUE)
ECLSK1 = ECLSK1[c(-1)]
ECLSK1 = as.data.frame(ECLSK1)
scdrep1 = svrepdesign(data = ECLSK1, type="JKn", repweights = ECLSK1[,26:106], weights = ECLSK1[,26], combined.weights = TRUE, rscales = 1, scale = 1, nest = TRUE)
# Grab descriptives
svyMean1 = svymean(ECLSK1, scdrep1)
modelSI1 = svyglm(log(X1PRNSOC) ~X_HISP_R + X_BLACK_R + X_ASIAN_R+ X_AMINAN_R+ X_HAWPI_R+ X_MULTR_R+ X_CHSEX_R+ + X1BMI + X1HPARNT + X1PAR1AGE + X1PAR1RAC + X12PAR1ED_I + X1PAR1EMP + X1HTOTAL + X1NUMSIB + X1PRIMNW + X2POVTY + X12SESL, scdrep1)
modelSI1Coef= summary(modelSI1)$coefficients[,1:2]
modelSI1Coef = as.data.frame(t(modelSI1Coef))
Here we are analyzing data from the second imputted data set ECLSK2
setwd("~/Google Drive/PARCS/Projects/ECLSK2011/Data/ImputedFirst")
ECLSK2 = read.csv("ECLSK2.csv", header = TRUE)
ECLSK2 = ECLSK2[c(-1)]
ECLSK2 = as.data.frame(ECLSK2)
scdrep2 = svrepdesign(data = ECLSK2, type="JKn", repweights = ECLSK2[,26:106], weights = ECLSK2[,26], combined.weights = TRUE, rscales = 1, scale = 1, nest = TRUE)
# Grab descriptives
svyMean2= svymean(ECLSK2, scdrep2)
modelSI2 = svyglm(log(X1PRNSOC) ~X_HISP_R + X_BLACK_R + X_ASIAN_R+ X_AMINAN_R+ X_HAWPI_R+ X_MULTR_R+ X_CHSEX_R+ + X1BMI + X1HPARNT + X1PAR1AGE + X1PAR1RAC + X12PAR1ED_I + X1PAR1EMP + X1HTOTAL + X1NUMSIB + X1PRIMNW + X2POVTY + X12SESL, scdrep2)
modelSI2Coef= summary(modelSI2)$coefficients[,1:2]
modelSI2Coef = as.data.frame(t(modelSI2Coef))
Now with imputed data set three
setwd("~/Google Drive/PARCS/Projects/ECLSK2011/Data/ImputedFirst")
ECLSK3 = read.csv("ECLSK3.csv", header = TRUE)
ECLSK3 = ECLSK3[c(-1)]
ECLSK3 = as.data.frame(ECLSK3)
scdrep3 = svrepdesign(data = ECLSK3, type="JKn", repweights = ECLSK3[,26:106], weights = ECLSK3[,26], combined.weights = TRUE, rscales = 1, scale = 1, nest = TRUE)
# Grab descriptives
svyMean3= svymean(ECLSK3, scdrep3)
modelSI3 = svyglm(log(X1PRNSOC) ~X_HISP_R + X_BLACK_R + X_ASIAN_R+ X_AMINAN_R+ X_HAWPI_R+ X_MULTR_R+ X_CHSEX_R+ + X1BMI + X1HPARNT + X1PAR1AGE + X1PAR1RAC + X12PAR1ED_I + X1PAR1EMP + X1HTOTAL + X1NUMSIB + X1PRIMNW + X2POVTY + X12SESL, scdrep3)
modelSI3Coef= summary(modelSI3)$coefficients[,1:2]
modelSI3Coef = as.data.frame(t(modelSI3Coef))
How with imputed data four
setwd("~/Google Drive/PARCS/Projects/ECLSK2011/Data/ImputedFirst")
ECLSK4 = read.csv("ECLSK4.csv", header = TRUE)
ECLSK4 = ECLSK4[c(-1)]
ECLSK4 = as.data.frame(ECLSK4)
scdrep4 = svrepdesign(data = ECLSK4, type="JKn", repweights = ECLSK4[,26:106], weights = ECLSK4[,26], combined.weights = TRUE, rscales = 1, scale = 1, nest = TRUE)
# Grab descriptives
svyMean4= svymean(ECLSK4, scdrep4)
modelSI4 = svyglm(log(X1PRNSOC) ~X_HISP_R + X_BLACK_R + X_ASIAN_R+ X_AMINAN_R+ X_HAWPI_R+ X_MULTR_R+ X_CHSEX_R+ + X1BMI + X1HPARNT + X1PAR1AGE + X1PAR1RAC + X12PAR1ED_I + X1PAR1EMP + X1HTOTAL + X1NUMSIB + X1PRIMNW + X2POVTY + X12SESL, scdrep4)
modelSI4Coef= summary(modelSI4)$coefficients[,1:2]
modelSI4Coef = as.data.frame(t(modelSI4Coef))
Now with the fifth imputed data set
setwd("~/Google Drive/PARCS/Projects/ECLSK2011/Data/ImputedFirst")
ECLSK5 = read.csv("ECLSK5.csv", header = TRUE)
ECLSK5 = ECLSK5[c(-1)]
ECLSK5 = as.data.frame(ECLSK5)
scdrep5 = svrepdesign(data = ECLSK5, type="JKn", repweights = ECLSK5[,26:106], weights = ECLSK5[,26], combined.weights = TRUE, rscales = 1, scale = 1, nest = TRUE)
# Grab descriptives
svyMean5= svymean(ECLSK5, scdrep5)
modelSI5 = svyglm(log(X1PRNSOC) ~X_HISP_R + X_BLACK_R + X_ASIAN_R+ X_AMINAN_R+ X_HAWPI_R+ X_MULTR_R+ X_CHSEX_R+ + X1BMI + X1HPARNT + X1PAR1AGE + X1PAR1RAC + X12PAR1ED_I + X1PAR1EMP + X1HTOTAL + X1NUMSIB + X1PRIMNW + X2POVTY + X12SESL, scdrep5)
modelSI5Coef= summary(modelSI5)$coefficients[,1:2]
modelSI5Coef = as.data.frame(t(modelSI5Coef))
# SEtting the degrees of freedom here.
df = modelSI5$df.null
Now we need to rbind all of the parameter and sd estimates into two columns. However, we need to grab the first row of these data sets, because the second row is the standard error. We need seperate data set for the se's, because we need to stack the different sets on top of each other.
Here we have the SI model
ParsSI = rbind(modelSI1Coef[1,], modelSI2Coef[1,], modelSI3Coef[1,], modelSI4Coef[1,], modelSI5Coef[1,])
ParsSI = as.matrix(ParsSI)
ParsSI
SeSI = rbind(modelSI1Coef[2,], modelSI2Coef[2,], modelSI3Coef[2,], modelSI4Coef[2,], modelSI5Coef[2,])
SeSI = as.matrix(SeSI)
SeSI
SIModelSIombine = mi.meld(q = ParsSI, se = SeSI)
SIModelSIombine = as.data.frame(cbind(Estimate = t(SIModelSIombine$q.mi),SE = t(SIModelSIombine$se.mi)))
names(SIModelSIombine) = c("Estimate", "SE")
SIModelSIombine
SIModelSIombine$T_Stat = SIModelSIombine$Estimate / SIModelSIombine$SE
SIModelSIombine$P_Value = 2*pt(-abs(SIModelSIombine$T_Stat), df = df)
SIModelSIombine = round(SIModelSIombine,3)
SIModelSIombine$Sig = ifelse(SIModelSIombine$P_Value <= .000, "***", ifelse( SIModelSIombine$P_Value <= .01, "**", ifelse(SIModelSIombine$P_Value <= .05, "*","NS")))
SIModelSIombine
Now we need to grab the means and standard deviations and combine them. Need to transpose them, because Amelia takes the five different version and combines them.
allSvyMeans = t(as.matrix(cbind(svyMean1, svyMean2, svyMean3, svyMean4, svyMean5)))
# For Se's we need to create dataframes so that we can grab the se's them combine them
SvySes1 = as.data.frame(svyMean1)
SvySes2 = as.data.frame(svyMean2)
SvySes3 = as.data.frame(svyMean3)
SvySes4 = as.data.frame(svyMean4)
SvySes5 = as.data.frame(svyMean5)
SvySes1 = SvySes1$SE
SvySes2 = SvySes2$SE
SvySes3 = SvySes3$SE
SvySes4 = SvySes4$SE
SvySes5 = SvySes5$SE
allSvySes = t(as.matrix(cbind(SvySes1, SvySes2, SvySes3, SvySes4, SvySes5)))
allSvyMeansSes = mi.meld(q = allSvyMeans, se = allSvySes)
allSvyMeansComb = t(as.data.frame(allSvyMeansSes$q.mi))
allSvySesComb = t(as.data.frame(allSvyMeansSes$se.mi))
allSvyMeansSesComb = data.frame(Mean = allSvyMeansComb, SE = allSvySesComb)
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent