tests/testthat/test-50-pisa.R

In EdSurvey: Analysis of NCES Education Survey and Assessment Data

skip_on_cran()
require(testthat)
context("PISA data reads in correctly")
require(WeMix)
options(width = 500)
options(useFancyQuotes = FALSE)
source("REF-4-pisa.R") # has REF output in it


if (!dir.exists(edsurveyHome)) {
  dir.create(edsurveyHome)
}

# able to toggle 'forceReread' for recaching the data if necessary
if (!exists("forceCacheUpdate")) {
  forceCacheUpdate <- FALSE
}

# Test =================

# 1. Check data read-in
test_that("PISA data reads in correctly", {
  expect_silent(downloadPISA(root = edsurveyHome, year = c(2000, 2003, 2006, 2009, 2012, 2015, 2018), cache = FALSE, verbose = FALSE))
  expect_silent(downloadPISA(root = edsurveyHome, year = 2012, database = "CBA", cache = FALSE, verbose = FALSE))
  expect_silent(usaINT2015 <<- readPISA(file.path(edsurveyHome, "PISA/2015"), countries = "usa", verbose = FALSE, forceReread = forceCacheUpdate))
  expect_silent(usaINT2012 <<- readPISA(file.path(edsurveyHome, "PISA/2012"), countries = "usa", verbose = FALSE, forceReread = forceCacheUpdate))
  expect_silent(qcnCBA2012 <<- readPISA(file.path(edsurveyHome, "PISA/2012"), database = "CBA", countries = "qcn", verbose = FALSE, forceReread = forceCacheUpdate)) # Shanghai
  expect_silent(jpn2009 <<- readPISA(file.path(edsurveyHome, "PISA/2009"), countries = "jpn", verbose = FALSE, forceReread = forceCacheUpdate))
  expect_silent(aus2006 <<- readPISA(file.path(edsurveyHome, "PISA/2006"), countries = "aus", verbose = FALSE, forceReread = forceCacheUpdate))
  expect_silent(aus2003 <<- readPISA(file.path(edsurveyHome, "PISA/2003"), countries = "aus", verbose = FALSE, forceReread = forceCacheUpdate))
  expect_warning(
    usa2000 <<- readPISA(file.path(edsurveyHome, "PISA/2000"), countries = "usa", verbose = FALSE, forceReread = forceCacheUpdate),
    "Cannot find both PSU and Stratum variables on data."
  )
  # 2000 complains about the PSU variable not being present

  expect_is(usaINT2015, "edsurvey.data.frame")
  expect_is(usaINT2012, "edsurvey.data.frame")
  expect_is(qcnCBA2012, "edsurvey.data.frame")
  expect_is(jpn2009, "edsurvey.data.frame")
  expect_is(aus2006, "edsurvey.data.frame")
  expect_is(usa2000, "edsurvey.data.frame")
  expect_equal(dim(usaINT2015), c(5712, 3715))
  expect_equal(dim(usaINT2012), c(4978, 1263))
  expect_equal(dim(qcnCBA2012), c(5177, 1346))
  expect_equal(dim(jpn2009), c(6088, 982))
  expect_equal(dim(aus2006), c(14170, 1022))
})

# Check multiple-path read-in
test_that("PISA multiple path read-in", {
  multiESDFL <- readPISA(paste0(edsurveyHome, c("PISA/2009", "PISA/2012", "PISA/2015")), countries = "usa", verbose = FALSE)
  expect_is(multiESDFL, "edsurvey.data.frame.list")
  expect_equal(colnames(multiESDFL$covs), c("subject", "year", "country"))
  expect_equal(length(multiESDFL$datalist), 3)
})

context("PISA showPlausibleValues and showWeights verbose output agrees")
test_that("PISA showPlausibleValues and showWeights verbose output agrees", {
  co <- capture.output(showPlausibleValues(usaINT2012, verbose = TRUE))
  expect_equal(co, pvREF)

  co <- capture.output(showWeights(usaINT2012, verbose = TRUE))
  expect_equal(co, swREF)
})

context("PISA getData")
test_that("PISA getData", {
  expect_known_value(head(gd0 <- EdSurvey::getData(usaINT2015, c("st004d01t", "st001d01t"))), file = "PISAgd0.rds", update = FALSE)
  expect_known_value(head(gd1 <- EdSurvey::getData(usaINT2012, c("st04q01", "st20q01"))), file = "PISAgd1.rds", update = FALSE)
  expect_known_value(head(gd2 <- EdSurvey::getData(qcnCBA2012, c("st04q01", "st20q01"))), file = "PISAgd2.rds", update = FALSE)
  expect_known_value(head(gd3 <- EdSurvey::getData(jpn2009, c("s514q03", "bookid"))), file = "PISAgd3.rds", update = FALSE)
  expect_known_value(head(gd4 <- EdSurvey::getData(aus2006, c("isi"))), file = "PISAgd4.rds", update = FALSE)
})

context("PISA subset data")
test_that("PISA subset data", {
  usa12_female <- subset(usaINT2012, st04q01 == "Female", verbose = FALSE)
  expect_equal(dim(usa12_female), c(2453, 1263))
})

context("PISA showCutPoints")
test_that("PISA showCutPoints", {
  co <- capture.output(showCutPoints(qcnCBA2012))
  expect_equal(co, scREF)
})

# 2. Check analytical functions
context("PISA edsurveyTable")
test_that("PISA edsurveyTable", {
  edTable1 <- edsurveyTable(math ~ st04q01 + st20q01, data = usaINT2012)
  withr::with_options(
    list(digits = 7),
    edTable1c <- capture.output(edTable1)
  )
  expect_equal(edTable1c, pisaedTable1REF)
})


context("PISA lm.sdf")
test_that("PISA lm.sdf", {
  plm1 <- lm.sdf(macq ~ st29q06 + sc01q01, data = usaINT2012)
  withr::with_options(
    list(digits = 7),
    plm1c <- capture.output(plm1)
  )
  expect_equal(plm1c, plm1REF)
})

context("PISA gap")
test_that("PISA gap", {
  pgap1 <- gap(
    variable = "math", data = usaINT2012,
    groupA = st04q01 == "Male", groupB = st04q01 == "Female", weightVar = "w_fstuwt"
  )
  withr::with_options(
    list(digits = 7),
    pgap1c <- capture.output(pgap1)
  )
  expect_equal(pgap1c, pgap1REF)

  pgap2 <- gap(
    variable = "math", data = usaINT2012,
    groupA = st04q01 == "Male", groupB = st04q01 == "Female", weightVar = "w_fstuwt", percentiles = c(50, 90), pctMethod = "symmetric"
  )
  withr::with_options(list(digits = 7), pgap2c <- capture.output(pgap2))
  expect_equal(pgap2c, pgap2REF)
})

context("PISA achievementLevels")
test_that("PISA achievementLevels", {
  al1 <- achievementLevels(achievementVars = "cpro", aggregateBy = "st04q01", data = qcnCBA2012)
  withr::with_options(list(digits = 7), al1c <- capture.output(al1))
  expect_equal(al1c, al1REF)
})

context("PISA glm")
test_that("PISA glm", {
  logit1 <- logit.sdf(I(st04q01 == "Male") ~ st20q01 + st48q01 + st87q01, data = qcnCBA2012)
  coefREF <- c(
    "(Intercept)" = 0.898510256928794,
    "st20q01Other country" = -0.637039679070128,
    "st48q01Courses after school Test Language" = 0.0570384010137383,
    "st87q01Agree" = -1.02420699129252,
    "st87q01Disagree" = -0.984759761693152,
    "st87q01Strongly disagree" = -0.896205635267306
  )
  expect_equal(coef(logit1), coefREF)
  se <- c(
    "(Intercept)" = 0.383392563610423,
    "st20q01Other country" = 0.481575743890542,
    "st48q01Courses after school Test Language" = 0.1112482333064,
    "st87q01Agree" = 0.418721001158844,
    "st87q01Disagree" = 0.38157294795411,
    "st87q01Strongly disagree" = 0.400877207648506
  )
  expect_equal(logit1$se, se)
})

context("PISA cor")
test_that("PISA cor", {
  usa2000$omittedLevels <- c(usa2000$omittedLevels, "Mis") # add 'Mis' here to match
  cor_drop <- cor.sdf("read", "st21q02", usa2000, method = "Pearson", weightVar = "w_fstuwt_read", jrrIMax = Inf, dropOmittedLevels = TRUE)
  cor_nodrop <- cor.sdf("read", "st21q02", usa2000, method = "Pearson", weightVar = "w_fstuwt_read", jrrIMax = Inf, dropOmittedLevels = FALSE)
  withr::with_options(list(digits = 7), cor_dropc <- capture.output(cor_drop))
  withr::with_options(list(digits = 7), cor_nodropc <- capture.output(cor_nodrop))
  cor1REF <- c(
    "Method: Pearson",
    "full data n: 3846",
    "n used: 3634",
    "",
    "Correlation: -0.1345117",
    "Standard Error: 0.01830193",
    "Confidence Interval: [-0.170717, -0.09794362]",
    "",
    "Correlation Levels:",
    "  Levels for Variable 'st21q02' (Lowest level first):",
    "    1. Yes",
    "    2. No"
  )
  expect_equal(cor_dropc, cor1REF)
  cor2REF <- c(
    "Method: Pearson",
    "full data n: 3846",
    "n used: 3846",
    "", 
    "Correlation: -0.125838",
    "Standard Error: 0.03567593",
    "Confidence Interval: [-0.1958661, -0.054533]",
    "",
    "Correlation Levels:",
    "  Levels for Variable 'st21q02' (Lowest level first):", 
    "    1. Yes", "    2. No", "    3. Mis")
  expect_equal(cor_nodropc, cor2REF)
})