Compare the old and new data frames, checking for accuracy of calculations in various ways.
keyDiagnostic(dfold, dfnew, keylist, max.values = 20, nametrunc = 8,
wide = 200, confidential = FALSE)Original data frame
The new recoded data frame
The imported variable key that was used to transform dfold into dfnew.
Show up to this number of values for the old variable
Truncate column and row names. Needed if there are long factor labels and we want to fit more information on table. Default = 6.
Number of characters per row in printed output. Suggest very wide screen, default = 200.
Should numbers in table be rounded to nearest "10" to comply with security standard enforced by some American research departments.
The simplest thing is to crosstabulate new variable versus old variable to see the mismatch. For tables of up to 10 values or so, that will be satisfactory.
For numeric variables, it appears there is no good thing to do except possibly to re-apply any transformations. Maybe a simple scatterplot will suffice.