Drawing a conclusion, and evaluating the research

Drawing the conclusion

• The conclusion answers the aim, in one or two sentences, using the data.
• Then it says what the business should do, because research exists to inform a decision.

A conclusion has four parts:

1. The answer to the aim — "Around three quarters of students would buy a $4 hot lunch at least occasionally, and about a fifth would buy it most days."
2. The evidence it rests on — the specific figures, not a general impression.
3. The business knowledge that explains it — why the result is what it is.
4. The recommendation — what the business should do next.

Explaining the conclusion with business knowledge

• The standard requires business knowledge relevant to the conclusion, and at Merit that knowledge must support the conclusion rather than sit beside it.
• Useful concepts to explain a market research result:
  • Target market — who the product is for, and whether the research shows that group exists in sufficient numbers
  • Price and value — willingness to pay depends on what the customer compares the product to
  • Competition — what else the customer could buy instead, including doing nothing
  • Capacity — whether the business can actually produce the volume the demand implies
  • Unit cost and margin — whether the price customers accept covers the cost of producing it
• Include a Māori business concept where it is genuinely relevant. For a food product sold in a school or community setting, manaakitanga — providing for people properly — and whanaungatanga — the relationships the product supports — often are.

Evaluating: strengths and weaknesses

• Achieved requires you to state the strengths and/or weaknesses of the research.
• Merit requires reasoned explanations of them and how this impacts on the validity of the findings or conclusions.

So every weakness needs three things:

• What it was — "the sample over-represented students who already buy lunch".
• Why it happened — "because collection took place near the canteen at lunchtime".
• What it does to validity — "so the stated willingness to buy is likely to be an over-estimate, and the conclusion should be treated as an upper bound".

Where the weaknesses usually are

Area	Typical weakness	Effect on validity
Sample	Too small, or unrepresentative of the population in the aim	Findings describe the sample, not the population
Method	One method only, so no explanation for the numbers	Conclusion can state size but not cause
Questions	Leading, ambiguous or unbalanced options	Measures the wording rather than the view
Timing and place	Collected at one time or in one location	Findings reflect that moment, not general behaviour
Response rate	Many forms not returned	Those who responded may differ systematically from those who did not
Data type	Stated intention rather than observed behaviour	Overstates demand

• Strengths matter too, and should be as specific: "the stratified sample matched the roll proportions, so each year level is represented as it occurs in the school", not "my research was quite good".

Excellence: discussing ways to improve

• The Excellence criterion is discussing ways to improve the market research process — the process, not the product.
• An improvement is only worth writing if it is:
  • Specific — "pilot the questionnaire on ten students first", not "be more careful"
  • Tied to a weakness you identified — each improvement should answer one of them
  • Realistic — within the time and resources actually available
  • Explained — say why it would improve the research, not just that it would
• Strong improvements that apply to most projects:
  • Pilot the questionnaire to find ambiguous or leading questions before the sample is spent
  • Collect at several times and locations to remove time-and-place bias
  • Add a second method aimed at whatever the first cannot explain
  • Use secondary data to set sample proportions, so the sample is defensible
  • Replace an intention question with an observation or a test, moving from what people say to what they do
  • Increase the response rate by shortening the questionnaire, rather than by distributing more copies