CREATE MULTIPLE DICHOTOMY has 3 variants - each producing a multiple response output.
The following table describes the key differences between them so you can choose which command is appropriate for your task. Each command also includes a link to the help article for that command.
The CREATE MULTIPLE SELECTION command is also considered in this table because it also outputs a multiple response variable:
|What it does
|CREATE MULTIPLE DICHOTOMY
|Creates a multiple response variable from a list of categorical variables.
- The imported data lacked metadata to define variable groupings, so you need to define them after import.
- Grouping the “Yes” response(s) from different questions to create a summary.
|CREATE MULTIPLE DICHOTOMY FROM
|Creates a new derived Multiple Response variable using an existing categorical array variable as input, by indicating which categories are “selected” in the condensed output.
- You want to summarize an existing array for a simplified analysis of “top two” categories of a scale.
|CREATE MULTIPLE DICHOTOMY WITH RECODE
|Operates on an input set of disparately-coded categorical variables and indicate which category to use on each of them to create a multiple dichotomy array.
- You want to group variables but they all have different values/labels (so you need to recode them).
- Only selected responses are recorded and each column is otherwise missing.
- Selected responses take a different value or code and label in each input variable (cf. SPSS CATEGORYLABELS=VARLABELS).
- Inputs contain missing data that should belong to a valid, non-selected category. You can Recode using expressions representing your survey routing logic if needed.
|CREATE MULTIPLE SELECTION
- Converts a fixed series of categorical inputs containing “selections” into dichotomies. Each of the resulting subvariables takes for its selected value each of the categories in any of the input variables.
- For example, if you have 50 unique categories across the input variables (typically far fewer than the number of categories), the result will have 50 subvariables in the new multiple response variable.
- Multiple selection is “lossy” encoding in that the items all must be valid or missing for an entire row.
- You have multiple response data that has been stored in multiple selections format with many categories, and you need to transform them into dichotomous subvariables that make a multiple response variable. (e.g., inputs are “Selection 1”, “Selection 2”; each categorical indicating “Item B chosen as Selection 1”. Instead, you want to know how many respondents chose Item B.)
- You import categorical data that represents ‘coded’ text, with a variable for matched strings in order of appearance (a row may match up to the number of input columns). The output of interest is the selection of each category in any of the input variables.