Within the data dictionary, the user can define multiple value sets for each data item. For use in Tabulation applications, these sets may have value names for both individual values and value ranges. If one table requires five-year age groups but another uses single years of age, both requirements can be met by having two value set definitions for age.
If an item in the dictionary has more than one value set, you can choose which of the value sets to use when adding that item to a table. If you drag the item itself from the dictionary, CSPro will add the first value set for that item to the table. However, if you drag one of the value sets underneath the item in the dictionary tree window, CSPro will use the value set that you dragged.
When creating a value set, you should take into account the following:
Value labels must be given for the value set since these appear as text (row or column) in the tables. If no label is given, the corresponding row/column label will be blank.
Counts associated with the value labels are based on the item values defined for the label. This might be a single value, e.g., Male (sex=1); a range of values, e.g., Under 5 years old (age from 0 to 4); disjoint values, e.g., Teacher (occupation=23, 42, or 67); or some combination of these. That is, each value in the data file that falls into a particular range/value in the value set will cause the count to be incremented for the corresponding cell in the table.
Counts are ONLY made in the table for item values defined in the value set. Values in the data file that are not defined in the value set are not counted. In the example below, the 'Marital Status' value set is missing codes 2 and 3 from its value set and the total reflects only the values listed in the value set. The exception to this rule is when Special Values are included in the Tally Attributes for the table or subtable.
In the current version of the software, alphanumeric items CANNOT be tallied.
By default, certain summary statistics (median, n-tiles and mode) are calculated using the counts in the values/ranges in the value set. The accuracy of these summary statistics is highly dependent on the number and size of values/ranges in the value set. If the range for a particular variable is large and there are a small number of categories in the value set, these statistics will be inaccurate. For best results, when calculating median, mode and n-tiles, use either single values for discrete variables or small, uniformly sized ranges for continuous variables. Note that by setting the appropriate options for n-tiles and median, you can use different categories for the n-tile/median calculation than those in the value set. See Tally Attributes for a Variable for details.
See Also: Include/Exclude Special Values in a Variable, Tally Attributes for a Table, Value Sets Description, Create a Table, Add Summary Statistics to a Table, Tabulate categories with disjoint values, Debug Table Totals