How to stop Text Analysis for Surveys from grouping terms within the extraction phase?
I would like to know who to stop "Text Analysis for Surveys" grouping terms within the extraction phase, i.e. I only want single words within the results.
I am looking at search strings so the frequency and combination of words is of interest. When the software groups words it makes the patterns harder to see.
There isn't a way to do this, as it defeats the entire purpose of the software.
During the extraction phase the product is extracting concepts, not individual words. There are settings to allow you to extract 'uniterms' -- single words to be treated as concepts -- but nothing to go the other way around. There is an underlying presumption with this product that the corpus to be analyzed is comprised of short answer narrative. This tool is largely for extracting the meaning of the content rather than simply counting the occurrences and co-occurrences of words.
The only surefire way to do this, would be to take the extracted multi-word concepts and alias them as synonyms for uniterms. So basically, you set the product up to extract uniterms, run the extraction, then individually set your multi-word concepts as synonymous with existing uniterms and run the extraction again. This time your concepts list will contain only uniterms. Then use the rules processing to check for patterns.
A possible work around solution would be to parse the response strings in IBM SPSS Statistics and then collect the patterns from the individually co-occurring words.