Recoding text variables allows you to create a new set of variables who have their values based on existing variables with a transformation applied. To start recoding open your data file, goto the "Editing" tab, select one or more text variables to which you want to apply the same transformation. Then click on "Recode" in the top menu bar, this will show you the following screen:
You now have the following options:
- Specify the destination variable names;
You can use the asterisks('*') in the destination name which will be replaced by the source variable name, in the example above this would mean that the destination variable would be named "Q02c". This feature makes it easy to determine destination names when multiple source variables have been selected. Another option is to write a comma delimited list of detination names, in which case you must make sure that each source variable has a destination name.
- The open answer recode list is by default created using the "case insensitive match" algorithm, you can change this by selecting a different algorithm and clicking on "Regenerate coding list".
BE CAREFULL: Any manually made changes to the coding list will be lost when regenerating the coding list.
IMPORTANT: All of the algorithms available work best on short texts only (one or two words), trying to run these against sentences will give unpredictable results.
You have the following algorithm options available:
- Exact match - Generates a dstinct coding list, mainly usefull only when you want to apply recode changes manually.
- Case insensitive match - Generates a distinct list of values not taken into account case sensitivity.
- Soundex match - Matches entries based on how they sound using a standard Soundex function. if you wish to know more about soundex please look here
- Double metaphone match - Matches entries based on how they sound using a standard double metaphone function. If you wish to know more about "Double metaphone" please have a look here.
- Levenshtein match - Matches entries based on a cost calculation of adding, removing and changing letters to get from one word to the other. If you would like to know more on the levenshtein agorithm please have a look here
- Once you have a code list that is a good starting point you can start making manual changes by:
- combining multiple values by selecting them and clicking on "Combine values" on te right.
- Split already combined values, by selecting the entry and clicking on "Split values" on the right
- Renaming the new categories by clicking on the third column in the grid and editing the category name.
- Once the recode list is to your satisfaction you can click on "Apply coding" on the right to create the new variables with their transformed values as defined in the recode list.