What is your process for reconciling significant coding differences?

WHAT THIS TESTS: This question evaluates whether you understand that qualitative coding is a systematic team practice, not a purely interpretive solo activity. The interviewer wants to see if you know how to measure agreement, facilitate structured reconciliation, and iterate on a codebook to achieve intercoder reliability. At the senior level, they are also listening for awareness of bias management, the value of multiple perspectives, and how to balance rigor with pragmatism in a research timeline.

A GOOD ANSWER COVERS: A strong response moves through four phases in order. First, quantify the divergence by calculating an intercoder agreement metric such as percent agreement or Cohen's kappa so the team is not relying on vague impressions of how far apart the codes are. Second, schedule a consensus session where both coders review every discrepant passage together, discuss the reasoning behind each code, and identify whether the disagreement stems from ambiguous definitions, overlapping codes, or genuine interpretive differences. Third, revise the codebook based on that discussion by tightening definitions, adding inclusion and exclusion examples, and creating decision rules for edge cases so the same passage would trigger the same code regardless of who reads it. Fourth, pilot the revised codebook on a new sample of transcripts, recalculate agreement, and repeat the cycle until reliability reaches an acceptable threshold before coding the full dataset.

COMMON WRONG ANSWERS: Red flags include treating coding differences as unresolvable subjective opinion, suggesting that the senior researcher simply overrules the junior researcher, or proposing to merge the two codebooks by averaging codes without reconciliation. Another weak pattern is jumping straight to re-coding everything without first diagnosing why the codes diverged, which wastes time and leaves the root cause unaddressed. Saying you would work alone next time to avoid conflict also signals poor collaborative research maturity.

LIKELY FOLLOW-UPS: An interviewer may ask how you would handle persistent disagreement after multiple consensus rounds, how you decide between consensus coding and split coding given timeline constraints, or what you do when a passage legitimately supports two different codes. They might also probe whether you have ever used software to track coding decisions or how you document the rationale for codebook changes for an audit trail.

ONE CONCRETE EXAMPLE: Imagine you and a colleague are coding interviews about remote work experiences. You coded a passage about unstable internet as technical barrier while your colleague coded it as stress. During reconciliation you discover your codebook defines technical barrier too broadly and stress too vaguely. You split the difference by rewriting technical barrier to cover only infrastructure and tool failures, rewriting stress to require explicit emotional language, and adding a new code logistical friction for gray-area cases. You then re-code ten transcripts and see agreement jump from sixty-two percent to ninety-one percent, at which point you lock the codebook and divide the remaining transcripts for split coding.

Source: delvetool.com