LESSON 27 - RELIABILITY: METHODS OF DETERMINING RELIABILITY / DEPENDABILITY IN QUALITATIVE RESEARCH

Qualitative reliability is reframed as **dependability** because identical results across different researchers and times can’t be guaranteed in an interpretive setting. Instead of asking whether the same findings will emerge under the same conditions, dependability asks whether the research process is transparent and auditable—so outsiders can judge whether the conclusions genuinely follow from the raw data. In practice, dependability measures how well qualitative findings remain consistent with the data collected in the field, and whether other researchers could reach similar interpretations if they reviewed the same evidence.

This shift matters because qualitative research treats the **researcher as the main research instrument**, meaning bias, decisions, and interpretation are inseparable from data collection and analysis. That’s why reliability can’t be handled with statistical repeatability in the way it often is in quantitative studies. Dependability becomes the mechanism for demonstrating that nothing essential was missed, that procedures were reasonable, and that the report isn’t misleading relative to what participants said, what was observed, and what documents contained.

The lesson lays out **five practical methods** for establishing dependability in qualitative research. The central method is an **audit trail**, which lets an external auditor trace the study’s procedures, verify that the data exist as reported, and evaluate whether analytic decisions match the collected evidence. Audit trail work includes checking how data were gathered, how analysis was conducted, and how findings were presented, with the “trail” functioning as a record that supports the “audit.”

A second method is ensuring **transcripts are error-free**. Because qualitative data often come from interviews and observations, transcription must preserve participants’ meaning; otherwise, the analysis may reflect the researcher’s distortions rather than the participant’s account. Third is **interrelator reliability** (also called **interrelator agreement**), where an independent researcher cross-checks coding decisions—testing whether multiple coders categorize themes from the same text in comparable ways.

When more than one researcher is involved, the lesson adds two related safeguards: **coordination among researchers** during coding so theme development stays aligned, and **crosstaking the codes**, where researchers compare independently derived results to see whether their coding converges.

Coding itself is treated as a core reliability concern. Qualitative coding means organizing and labeling data to identify patterns and relationships, often using participants’ own wording through **in vivo** codes. To support consistent coding, researchers need a **code book** and must ensure that all interviews, observations, and documents are fully transcribed and recorded.

Finally, the lesson highlights threats to validity and reliability—framed as sources of error. It cites Brink (1993) and lists four main threats: the **researcher** (behavior, attitude, and bias), **participants** (artificial responses, withholding or distortion), the **social context** of data collection, and the **methods of data collection and analysis**. Inadequate **triangulation** can lead to thin data, weaker codes, and less dependable conclusions. The takeaway is that dependability is built through multiple, overlapping strategies embedded in the research proposal, not through a single check at the end.