Dependent variable and Hypothesis

¶ … consumer preference for longer sleeves is unrelated to the threat of West Nile. In this study, the independent variable will be the threat of West Nile in a given region. The dependent variable will be the consumer preference for longer sleeves in a given region.

The best research design is a survey design. The hypothesis can be tested with simple answers, so no substantial qualitative responses are needed. Observational design will not allow us to determine levels of fear about West Nile, nor will it leave us with evidence of causation. The research conducted is not an experiment. The survey design allows us to gauge the fear of West Nile in a region and the preference for longer sleeves. The latter can be corroborated with sales figures.

The best type of sampling for this survey is simple random probabilistic. This assumes that ABC does not have a customer database -- that would allow for a more targeted form of sampling. In lieu of that, we can assume that even if the person being surveyed has a different taste in style to our product, they will still have the same feelings with regards to West Nile and long sleeves. Non-probabilistic methods are not random enough to gauge the feelings of a general population, and none of the more systematic types of probabilistic sampling will yield better results given our target market and our ability to segment out the general population.

The data collected will be ordinal, largely strongly agree, agree, disagree, strongly disagree type of answers. This is practical because we are trying to gauge not only whether the sentiment exists, but the intensity of the sentiment on the part of the consumer. The decision rule that will confirm or deny the hypothesis is the correlation of responses indicating fear of West Nile and responses indicating that this drives a purchase of longer-sleeved shirts. A statistically significant correlation must exist in order to prove the null hypothesis. The two co-workers who disagree will need to both be happy with the decision. In this case, the respondents should indicate a level of concern about West Nile and a purchasing decision with greater intensity than "agree" in order to demonstrate intent. This is a more conservative approach, but the cost of action in this case is likely higher than the cost of inaction, so Ms Szekely should have to demonstrate clearly that her view is backed up by research in order to justify the additional expense and risk.