Applying the Vertically Integrated Project Model to Youth-Inclusive Social Media Coding of Zyn-Related YouTube...

The rapid evolution of the nicotine and tobacco product (NTP) marketplace continues to outpace existing research. Although adolescents are active participants on social media who both consume and generate content, this age group has rarely been included in analyses of NTP content, raising a methodological challenge for accurate interpretation and coding of youth-oriented NTP content. Most research examining oral nicotine product content online has focused on TikTok, despite YouTube's widespread adolescent use. One promising framework is the vertically integrated project (VIP) model. Situating content analysis within the VIP model suggests the methodological potential of age-diverse teams to capture nuances in social media data coding, reveal areas of interpretive divergence, and ultimately advance research methodologies in public health.

This study aimed to understand how perspectives across different age groups, academic levels, and lived experiences shape the interpretation of Zyn-related YouTube shorts by engaging high school (HS) students, undergraduate public health students, and faculty researchers in parallel coding.

We manually collected 300 publicly available YouTube shorts containing #Zyn on September 18, 2024, as a methodological case example. Three coder pairs, including 2 HS students (GC and NC), 2 undergraduate students (PN and TGL), and 2 public health faculty members (BLH and JES), independently coded the same 50 videos per round across 6 iterative rounds (N=300). For each round, interrater reliability was assessed using Cohen κ and percent agreement as complementary indicators. After all coding rounds were completed, the full coding team participated in a debriefing session. The transcript was analyzed using a combined deductive-inductive thematic approach to contextualize reliability patterns and identify challenges and practical insights related to age-diverse collaborative coding within the VIP-informed model.

Interrater reliability varied across 6 iterative rounds, with the undergraduate team showing the highest κ values in the final coding round (κ range: 0.40-0.63). Several constructs showed consistently high percent agreement but low and variable κ values, particularly within the HS team. For individual codes, mean κ values for male stereotype were higher for the undergraduate and faculty teams than for the HS team (0.56, 0.47, and 0.10, respectively). Debriefing findings indicated that disagreement reflected factors such as evolving code definitions, interpretive drift across rounds, and differences in cultural and experiential lenses. Coders noted that the codebook struggled to capture ambiguity in short-form content and that teams developed distinct heuristics for resolving uncertainty. Suggested improvements included rotating coding pairs and expanding codebook examples.

This methodological case study illustrates how a VIP-informed, age-diverse coding structure can be used to examine interpretive challenges in coding youth-oriented nicotine-related short-form video content. For public health studies examining how different groups interpret online health messages, the VIP model may offer a practical structure for incorporating those perspectives into the research process.