/ Opening

The method and the sample jointly determine what you can know

A perfect survey of the wrong people is still weak evidence. Ten rich interviews are not a market estimate.

QUESTION

What do we need to know?

Explore why, estimate how much, compare, or test.

METHOD

What evidence can answer it?

Qualitative, quantitative, behavioral, experimental.

SAMPLE

Who or what is included?

Frame, size, selection, missing groups.

TAKEAWAYChoose the method and sample before deciding how strong the claim can be.

/ Quick Test

Which design can answer the question?

A company wants to know why trial users do not renew and how common each reason is.

ONE-METHOD PLAN

Survey 10,000 users
List possible reasons
Report percentages

SEQUENCED PLAN

Interview and observe first
Build answer categories
Survey a representative sample
Compare behavior by group

TAKEAWAYExploration and estimation are different jobs that often require different methods.

/ Model

Research design is a chain of fit

Each choice limits the next.

01QuestionDefine the decision and unknown.

02MethodChoose evidence suited to the unknown.

03SampleInclude the people or events that matter.

04ClaimState only what the design supports.

TAKEAWAYA research result cannot travel farther than its method and sample allow.

/ Distinctions

Common evidence families answer different questions

No method sees the whole field.

QUALITATIVE

Why, how, language, context

Depth and mechanism; not prevalence.

QUANTITATIVE

How many, how often, difference

Scale and comparison; depends on measurement quality.

BEHAVIORAL

What happened under conditions

Action evidence; motive may remain unknown.

EXPERIMENTAL

What changes when a condition changes

Stronger causal inference; requires control and clean design.

TAKEAWAYMethod choice should follow the decision—not organizational habit.

/ Team Challenge

Design the research and the sample together

Create a credible plan for one business question.

Why applicants abandon an MBA applicationWhat prevents software renewalWhich coffee offer could grow morningsWhy a campaign earned attention but not sales

OUTPUT: decision · research question · method sequence · sample frame · excluded groups · claim limit

TAKEAWAYA complete plan names who is missing and what the evidence will not prove.

/ Pressure Test

How far can the finding travel?

Representativeness is a property of the claim, not a compliment about sample size.

1.Who had a chance to be included?

2.Which groups are over- or underrepresented?

3.What response or selection bias may exist?

4.What population and timeframe can the result legitimately describe?

TAKEAWAYThe size of the sample cannot repair a biased frame.

/ Example

One question, two methods, one honest claim

The methods work together without pretending to be interchangeable.

STAGEDESIGNWHAT IT SUPPORTS

Explore20 interviews + session observationMechanisms and language

EstimateSurvey representative non-renewersPrevalence by segment

ValidateBehavioral cohort comparisonWhether stated reasons match action

Claim“Among sampled non-renewers…”Bounded, auditable conclusion

TAKEAWAYMixed methods are strongest when each method has a distinct job and claim boundary.

/ Close

Question first. Sample before claim.

Research rigor is the fit among the decision, evidence, and population.

ASKDefine the unknown

DESIGNChoose method and sample

BOUNDLimit the claim honestly

TAKEAWAYDo not call evidence representative until you can name what it represents.