January 10th, 2018
Type I and II Errors
To assess aspects such as customer satisfaction with an entity’s products, an entity uses information from a sample rather than the entire population of customers who use its products. This arises since it might be impossible for an entity to correct information (feedback) from all its customers especially when it serves a large population spanning different geographical locations. Such decisions made about a population based on information derived from sample data are referred to as statistical decisions (Spiegel and Stephens 245). To make decisions on a population based on sample data, one has to make assumptions (which could be true or not true) about such a population. These assumptions – referred to as statistical hypotheses – help in assessing the extent to which the sample is representative of the population, hence the relevance of the data obtained from such a sample to the population (Spiegel and Stephens 245). The hypotheses formulated may either be of two kinds – a null hypothesis (a hypothesis that the researcher seeks to disprove, usually by showing that differences exist among observations made) and an alternative hypothesis (any hypothesis that is different from a stated hypothesis) (Spiegel and Stephens 245). In evaluating whether such hypotheses are true, two types of errors – type I and type II – may occur. This section discusses these errors in the context of marketing research.
In type I error, a researcher rejects and hypothesis that should have been accepted thus resulting into an erroneous conclusion (Spiegel and Stephens 246; FAO 1). In type II error, the researcher accepts an hypothesis that should have been rejected, thus also leading to an erroneous conclusion (Spiegel and Stephens 246; FAO 1). The two errors are mutually exclusive (since one cannot accept and reject the hypothesis) and as one tries to avoid one of the errors the chances of committing the other type increase (Spiegel and Stephens 246). Reducing the chances of both errors only occurs with increase in sample size, since such an increase enhances the likelihood that the sample will be representative of the population under survey (Spiegel and Stephens 246). Such kind of errors may affect accuracy in marketing decisions to a varying extent as exemplified in the subsequent paragraph.
An example of a marketing research would be to evaluate whether the introduction of a new product or service improves customer satisfaction with an entity’s products. In this respect a null hypothesis would be that customer satisfaction levels have not changed following introduction of the new products or services. To test such a null hypothesis, the entity would collect data from customers prior to the introduction and following introduction of the service or product. Using such data, the company would evaluate any changes in customer satisfaction following the introduction to conclude whether customer satisfaction levels have changed significantly. In the case of type I error the company would reject the null hypothesis that customer satisfaction has not changed following introduction of the product and conclude that the new product has enhanced customer satisfaction, whereas in reality such customer satisfaction has not improved even with the introduction of the new product. Such an error would make the company complacent, perceiving the introduction of the product to have been effective in enhancing customer service, while, in reality, the product does not offer any benefits with respect to enhancing customer satisfaction.
In type II errors, the company would accept the hypothesis that should have been rejected. In this case, the entity by accepting the hypothesis that no changes in customer satisfaction have occurred following introduction of the new product, may decide to eliminate a product that, in reality, has improved customer satisfaction. Type I and II errors may thus affect the accuracy of decisions made by an entity. To reduce such risks, the sampling procedures ought to ensure that a sufficient sample is obtained and comparing results of various samples instead of relying on data from a single sample.
Cravens, David W., Nigel F. Piercy and Ashley Prentice. “Developing Market-driven Product Strategies.” Journal of Product & Brand Management 9.6 (2000): 369-388. Print.
Food and Agricultural Organization of the United Nations – FAO. “Sampling in Marketing Research.” Marketing Research and Information Systems. Web. 19 October 2011.
Kotler, Philip. Marketing Management Millenium Edition. 10th ed. Upper Saddle River, NJ: Prentice-Hall, Inc., 2000. Print.
Muffatto, Moreno. “Platform Strategies in International New Product Development.” International Journal of Operations & Production Management 19.5/6 (1999): 449-459. Print.
Norling, Parry M. “Lessons in New Product Development from Dupont.” Engineering Management Journal 10.3 (1998): 5-10. Print.
Robinson, William T. and Jeongwen Chiang. “Product Development Strategies for Established Market Pioneers, Early Followers, and Late Entrants.” Strategic Management Journal 23 (2002): 855-866. Print.
Rogers, Dale S., Douglas M. Lambert and A. Michael Knemeyer. “The Product Development and Commercialization Process.” The International Journal of Logistics Management 15.1 (2004): 43-56. Print.
Smith, Alan D. “Competitive Approaches to New Product Development: A Comparison of Successful Organizations in an Unstable Economic Environment.” Team Performance Management 17.3/4 (2011): 124-145. Print.
Spiegel, Murray R. and Larry J. Stephens. Schaum’s Outline of Theory and Problems of Statistics. 4th ed. New York: McGraw Hill, 2008. Web. 19 October 2011.