The purpose of this empirical study was to determine the extent to which three different objective analytical methods--sequence analysis, surface cohesion analysis, and lexical cohesion analysis--can most accurately identify specific characteristics of online interaction. Statistically significant differences were found in all points of comparison: duration of discussions, rates of participation and extent of interaction. Sequence analysis over-reports the duration of discussions but consistently under-reports both participation rates and the extent of participant interaction. Surface cohesion analysis accurately reports discussion duration and participation rates but under-reports the extent and nature of interaction between participants. Lexical cohesion analysis accurately reports all three characteristics and also reflects the integrated nature of participant interaction. The findings support the recommendation that at least a surface cohesion analysis and preferably a lexical cohesion analysis be conducted prior to engaging in a thematic analysis of discussion content. Surface and lexical cohesion analyses reveal the central role of the instructor in online discussions and have implications for course design and delivery. (Contains 3 figures and 1 table.)