Reliability and validity in dissertation research

Reliability and validity in quantitative work

Reliability in quantitative research concerns consistency: would the same instrument, applied under the same conditions, produce the same results again? Internal consistency (often reported via Cronbach's alpha for multi-item scales) and test-retest reliability are the most common checks at student level.

Validity is broader. Construct validity asks whether your instrument measures the underlying concept it claims to; content validity asks whether it covers the full range of that concept; criterion validity compares your measure against an established benchmark.

• Internal validity: are you confident the relationship you found is not explained by something else?
• External validity: how far can findings generalise beyond your sample?
• Construct validity: does your measure actually capture the concept it names?

Trustworthiness in qualitative work

Because qualitative research does not aim for replicable measurement, reliability and validity are usually replaced with trustworthiness criteria. Credibility is the qualitative equivalent of internal validity — are your interpretations a believable account of participants' experience?

Transferability replaces external validity: rather than claiming generalisability, you provide enough contextual detail (thick description) for a reader to judge whether findings might apply elsewhere. Dependability replaces reliability, showing your process was logical and documented; confirmability replaces objectivity, showing your interpretations are grounded in the data rather than your own assumptions.

Practical strategies to strengthen either

For quantitative designs, pilot your instrument, report reliability statistics, and be explicit about sampling limitations that constrain external validity.

For qualitative designs, keep a reflexive journal, use member checking or peer debriefing where feasible, and maintain an audit trail of coding decisions so a supervisor could retrace your reasoning.

• Pilot the instrument or interview schedule before full data collection.
• Triangulate data sources or methods where your design allows it.
• Keep an audit trail: raw data, codes, memos, and decision points.
• Be explicit in your limitations section about what you cannot claim.

Writing this into your methodology chapter

Do not simply list the four qualitative terms or the three validity types as a glossary. Show how each applies to your specific design: name the strategy you used and the limitation that remains despite it.

A sentence such as 'dependability was supported by maintaining a coding log, though a single researcher coding all transcripts limits confirmability' does more work than a definition copied from a textbook.

Students also ask

What is the difference between reliability and validity?

Reliability is about consistency — would the same method produce the same result again? Validity is about accuracy — are you actually measuring or capturing the thing you claim to be studying? A measure can be reliable without being valid, but not usefully valid without some reliability.

What are the four criteria for qualitative trustworthiness?

Credibility (believable interpretation of the data), transferability (enough context for a reader to judge relevance elsewhere), dependability (a logical, documented process) and confirmability (interpretations grounded in data rather than researcher bias). They replace validity and reliability in most qualitative designs.

Do I need to discuss reliability and validity in a qualitative dissertation?

You need to discuss rigour, but using the correct terminology matters: use trustworthiness criteria rather than quantitative reliability/validity language unless your design specifically calls for both (as in some mixed-methods work).