acs (version 2.1.4)

sum-methods: acs Methods for Function sum

Description

Returns the sum of all the estimates present in its arguments, along with proper treatment of standard errors.

Usage

# S4 method for acs
sum(x, agg.term=c("aggregate", "aggregate"),
one.zero=FALSE, ..., na.rm=FALSE)

Arguments

x

the acs object to be summed

agg.term

a character vector (length 1 or 2) of labels to use for the geography or acs.colnames of the new object

one.zero

a logical flag indicating whether to include standard errors for only one zero-value estimates or all (the default); see details.

...

reserved for other arguments to pass

na.rm

whether to remove NAs from the values before summing; defaults to FALSE.

Methods

signature(object = "acs")

When passed an acs object (possibly involving subsetting), sum will return a new acs object created by aggregating (adding) all estimates in the object, and adding the corresponding standard errors in a statistically appropriate way. (Aggregate standard errors are computed by taking the square root of the sum of the squared standard errors of the terms to be aggregated.)

If the original object contains a single row, the geographic metadata and row name is preserved; if not, the geographic metadata is replaced with the term "aggregate" (or the contents of the first item of the (vector) option agg.term).

If the original object contains a single column, the column names and acs.units data are preserved; if not, the column names are replaced with the term "aggregate" or the contents of the second item of the (vector) option agg.term; note: if agg.term is only one item in length, it will be repeated here if needed.

All other acs-class metadata is preserved, except for the modified flag, which is set to TRUE.

Details

Note: when aggregating ACS data, users may want to sum many fields with "0" values for estimates, especially when working with small geographies or detailed tables that split the population into many categories. In these cases, some analysts have suggested that the traditional summation procedure for standard errors (taking the square-root of the sum of the squares of the errors) may over-inflate the associated margins of error; instead, they recommend an alternative method, which ignores all but the single largest of the standard errors for any "zero-estimate" fields. Although this is somewhat unconventional, it is provided as an additional user-specified option here, through the "one.zero" argument.

Examples

# NOT RUN {
# load ACS data
data(kansas09)

# aggregate the third column, all rows
sum(kansas09[,3])

# aggregate the fifth row, all column
sum(kansas09[5,])

# aggregate all rows, columns 3 through 25, rename rows "Kansas" and columns "Total Males"
sum(kansas09[, 3:25], agg.term=c("Kansas","Total Males"))
# }