Engaging the Senses: Triangulating Qualitative Data Analysis Using Transcripts, Audio, and Video

Background and Purpose: Transcript analysis of qualitative interview data is widely used for its ease and presumed rigour, but it is not without limitations. Multifaceted dimensions of the interview (e.g., emotion, context) are lost once converted to text, compromising a ‘thick’ analysis and data fidelity. As audio-video recording technologies have become increasingly accessible, opportunities have emerged to innovate. Recent developments in computer-assisted qualitative data analysis (CAQDAS) programs support multi-modal coding systems (i.e., synchronization of text, audio, video data coding). Yet, a notable literature gap exists examining this approach.

The purpose of this study was to explore the utility of a multi-modal coding framework for qualitative analysis through a grounded theory study investigating sexual and gender minority (SGM) young adults’ engagement with offline and online media.

Methods: A grounded theory study was conducted with SGM young adults (n=19). Interviews with individual participants (age 18–22) ranged from 45–90 minutes and were simultaneously audio and video recorded. Recruitment occurred throughout data collection, until theoretical saturation was achieved. The data was prepared for analysis in three formats: via (1) transcripts (with embedded timecodes), (2) audio files, and (3) video files. Six independent, heterogeneous coders participated in the preliminary analysis process. Each coder was assigned to code data in a single data format (e.g., only transcripts, only audio). Each data format was used in the analysis of all interviews by two coders (e.g., two coders coded each interview in video format). Analysis for all data formats was undertaken using CAQDAS program ATLAS.ti.

Following independent open coding, the research team conducted four three-hour meetings to compare preliminary analyses across formats. This comparison was undertaken in two minute increments of the interviews, where codes for particular segments were paralleled across data formats. To record this process, a code table was employed and collaborative coding strategies were utilized during triangulation. Intercoder agreement ranged from 60-90%.

Results: Notable similarities and differences were found in the coding of the same interview segments based on the data format used in analysis. Codes related to finding identities and community online, and issues of offline and online safety, were not only similar but strengthened with the multi-modal approach. Significant coding differences attributable to the data format during analysis included the importance of participant affect, perceived contradictions, discrepancies between verbal statements and body language, the level of comfort and engagement, and distress when discussing traumatic experiences (e.g., violence). Of the three data formats, video provided the most comprehensive data for analysis and facilitated coder attunement to participants, yet simultaneously inhibited attention to narrative details. 

Conclusions and Implications: A multi-modal coding approach – using a combination of text, audio, and video – expands the analytical lens for an enhanced holistic analysis, generating a more accurate and nuanced conceptualization of the phenomena being investigated. Social work researchers should leverage advancements in technology for analysis as multi-modal approaches may more fully capture and illuminate the experiences of marginalized populations. The presentation will illustrate the results using multi-modal codes and share strategies for coding and integrating data types.