Content Coding Info

For a description of the content coding output variable names, please see the Content Coding Output page.

Years ago, I began including a content coding system as an option in RIOT Scan. These days, this is the primary purpose of the software. In its current form, it codes the following dictionary schemes (full citation information for all coding schemes is generated in your program output directory upon execution):

Content coding may be somewhat slow as it works through multiple coding systems for each file and can take a considerable amount of time to complete for large files. Please be patient while your text is processing. The RIOT Scan system has been tested across data from over 100 studies / corpora and converges with other content coding packages very nicely. For example, each of the LIWC 2007 categories that have been tested typically correlate between .98 and 1.00 with the output delivered by the LIWC 2007 software package (with the exception of the “Numbers” and “Filler” categories, which we score somewhat differently).

Any minor variations between the two packages are due to differences in our particular text parsing systems. RIOT Scan does some text preparation, which LIWC does not do, in order to ensure that minor text/composition errors do not exclude words from analysis. I do not know what the LIWC 2007 text parsing system is, but I prefer the system of RIOT Scan as its decisions as to what qualifies as a word appear to align more closely with other mainstream software packages (e.g., Microsoft Office).

If you publish using any of the coding systems performed by RIOT Scan, the burden is on YOU, the user, to properly cite the individuals responsible for those particular systems in addition to RIOT Scan. A lot of very dedicated people put in some excellent work in developing these systems. Please ensure that they receive proper credit for their efforts.