edsurveyTable {EdSurvey}

EdSurvey Tables With Conditional Means

Description

Returns a summary table (as a data.frame) that shows the number of students, the percentage of students, and the mean value of the outcome (or left-hand side) variable by the predictor (or right-hand side) variable(s).

Usage

edsurveyTable(
  formula,
  data,
  weightVar = NULL,
  jrrIMax = 1,
  pctAggregationLevel = NULL,
  returnMeans = TRUE,
  returnSepct = TRUE,
  varMethod = c("jackknife", "Taylor"),
  drop = FALSE,
  dropOmittedLevels = TRUE,
  defaultConditions = TRUE,
  recode = NULL,
  returnVarEstInputs = FALSE,
  omittedLevels = deprecated()
)

Arguments

formula

object of class formula, potentially with a subject scale or subscale on the left-hand side and variables to tabulate on the right-hand side. When the left-hand side of the formula is omitted and returnMeans is TRUE, then the default subject scale or subscale is used. You can find the default composite scale and all subscales using the function showPlausibleValues. Note that the order of the right-hand side variables affects the output.

data

object of class edsurvey.data.frame. See readNAEP for how to generate an edsurvey.data.frame.

weightVar

character string indicating the weight variable to use. Note that only the name of the weight variable needs to be included here, and any replicate weights will be automatically included. When this argument is NULL, the function uses the default. Use showWeights to find the default.

jrrIMax

a numeric value; when using the jackknife variance estimation method, the default estimation option, jrrIMax=1, uses the sampling variance from the first plausible value as the component for sampling variance estimation. The V_{jrr} term (see the Details section of lm.sdf to see the definition of V_{jrr}) can be estimated with any number of plausible values, and values larger than the number of plausible values on the survey (including Inf) will result in all of the plausible values being used. Higher values of jrrIMax lead to longer computing times and more accurate variance estimates.

pctAggregationLevel

the percentage variable sums up to 100 for the first pctAggregationLevel columns. So, when set to 0, the PCT column adds up to 1 across the entire sample. When set to 1, the PCT column adds up to 1 within each level of the first variable on the right-hand side of the formula; when set to 2, then the percentage adds up to 100 within the interaction of the first and second variable, and so on. Default is NULL, which will result in the lowest feasible aggregation level. See Examples section.

returnMeans

a logical value; set to TRUE (the default) to get the MEAN and SE(MEAN) columns in the returned table described in the Value section.

returnSepct

set to TRUE (the default) to get the SEPCT column in the returned table described in the Value section.

varMethod

a character set to jackknife or Taylor that indicates the variance estimation method to be used.

drop

a logical value. When set to the default value of FALSE, when a single column is returned, it is still represented as a data.frame and is not converted to a vector.

dropOmittedLevels

a logical value. When set to the default value of TRUE, drops those levels of all factor variables that are specified in an edsurvey.data.frame. Use print on an edsurvey.data.frame to see the omitted levels.

defaultConditions

a logical value. When set to the default value of TRUE, uses the default conditions stored in an edsurvey.data.frame to subset the data. Use print on an edsurvey.data.frame to see the default conditions.

recode

a list of lists to recode variables. Defaults to NULL. Can be set as recode = list(var1 = list(from = c("a", "b", "c"), to = "c")).

returnVarEstInputs

a logical value set to TRUE to return the inputs to the jackknife and imputation variance estimates, which allows for the computation of covariances between estimates.

omittedLevels

this argument is deprecated. Use dropOmittedLevels.

Details

This method can be used to generate a simple one-way, two-way, or n-way table with unweighted and weighted n values and percentages. It also can calculate the average of the subject scale or subscale for students at each level of the cross-tabulation table.

A detailed description of all statistics is given in the vignette titled Statistical Methods Used in EdSurvey.

Value

A table with the following columns:

RHS levels

one column for each right-hand side variable. Each row regards students who are at the levels shown in that row.

N

count of the number of students in the survey in the RHS levels

WTD_N

the weighted N count of students in the survey in RHS levels

PCT

the percentage of students at the aggregation level specified by pctAggregationLevel (see Arguments). See the vignette titled Statistical Methods Used in EdSurvey in the section “Estimation of Weighted Percentages” and its first subsection “Estimation of Weighted Percentages When Plausible Values Are Not Present.”

SE(PCT)

the standard error of the percentage, accounting for the survey sampling methodology. When varMethod is the jackknife, the calculation of this column is described in the vignette titled Statistical Methods Used in EdSurvey in the section “Estimation of the Standard Error of Weighted Percentages When Plausible Values Are Not Present, Using the Jackknife Method.” When varMethod is set to Taylor, the calculation of this column is described in “Estimation of the Standard Error of Weighted Percentages When Plausible Values Are Not Present, Using the Taylor Series Method.”

MEAN

the mean assessment score for units in the RHS levels, calculated according to the vignette titled Statistical Methods Used in EdSurvey in the section “Estimation of Weighted Means When Plausible Values Are Present.”

SE(MEAN)

the standard error of the MEAN column (the mean assessment score for units in the RHS levels), calculated according to the vignette titled Statistical Methods Used in EdSurvey in the sections “Estimation of Standard Errors of Weighted Means When Plausible Values Are Present, Using the Jackknife Method” or “Estimation of Standard Errors of Weighted Means When Plausible Values Are Present, Using the Taylor Series Method,” depending on the value of varMethod.

When returnVarEstInputs is TRUE, two additional elements are returned. These are meanVarEstInputs and pctVarEstInputs and regard the MEAN and PCT columns, respectively. These two objects can be used for calculating covariances with varEstToCov.

Author(s)

Paul Bailey and Ahmad Emad

References

Binder, D. A. (1983). On the variances of asymptotically normal estimators from complex surveys. International Statistical Review, 51(3), 279–292.

Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. New York, NY: Wiley.

Examples

## Not run: 
# read in the example data (generated, not real student data)

sdf <- readNAEP(path=system.file("extdata/data", "M36NT2PM.dat", package = "NAEPprimer"))

# create a table that shows only the breakdown of dsex
edsurveyTable(formula=composite ~ dsex, data=sdf, returnMeans=FALSE, returnSepct=FALSE)

# create a table with composite scores by dsex
edsurveyTable(formula=composite ~ dsex, data=sdf)

# add a second variable
edsurveyTable(formula=composite ~ dsex + b017451, data=sdf)

# add a second variable, do not omit any levels
edsurveyTable(formula=composite ~ dsex + b017451 + b003501, data=sdf, omittedLevels=FALSE)

# add a second variable, do not omit any levels, change aggregation level
edsurveyTable(formula=composite ~ dsex + b017451 + b003501, data=sdf, omittedLevels=FALSE,
	            pctAggregationLevel=0)

edsurveyTable(formula=composite ~ dsex + b017451 + b003501, data=sdf, omittedLevels=FALSE,
	            pctAggregationLevel=1)

edsurveyTable(formula=composite ~ dsex + b017451 + b003501, data=sdf, omittedLevels=FALSE,
	            pctAggregationLevel=2)

# variance estimation using the Taylor series 
edsurveyTable(formula=composite ~ dsex + b017451 + b003501, data=sdf, varMethod="Taylor")

## End(Not run)