Co-occurrence And Correlation: Challenges Using ATLAS.ti
ATLAS.ti is a software package used for pooling and analyzing qualitative data. Whereas statistical software packages have limited qualitative functionality, programs like ATLAS.ti offer the ability to store and code interview transcripts, field notes and multimedia data. In doing that, the program tries to emulate some established statistical tools by creating parallel qualitative analysis tools. Hence, the C-index or the co-occurrence index designed to produce a normalized measure of how frequently one code overlaps with another code. The tool is fully incorporated in the regular functionality of recent versions of ATLAS.ti: "Analysis / Code cooccurrence table." As any similar normalized coefficient, values close to 0 indicate weak relationship and values close to 1 indicate a strong relationship.
For instance, in the screenshot above, the light-green cell with the 0.63 C-coefficient indicates a higher intensity than the 0.15 C-coefficient in the darker green cell on the top right. Within the research analyzing "We Are The 99 Percent" and "We Are The 53%" blogs above, these coefficients seem to indicate that the anti-Occupy blog is more invested in denying the legitimacy of the pro-Occupy blog than the opposite. However, explanations of the coefficient's use and how it should be interpreted are limited. For instance, it is unclear whether the 0.63 value above indicates a very strong relationship or a moderate relationship. By contrast, a Pearson statistical correlation coefficient of that value would indicate a moderately strong association between the two variables.
Although limited, there are some resources outside of the official manual and guide that offer insights into the nature of the C-coefficient:
- Overview, formula and key issues: "Some Cooc Table Explorer (CTE) mysteries unveiled" by ATLAS.ti's Thomas Muhr
- Short article with examples and screenshots: "Examining the Context in Qualitative Analysis: The Role of the Co-Occurrence Tool in ATLAS.ti" by Ricardo B. Contreras.