Hypothesis Testing Report
A hypothesis is similar to a problem statement in that it helps the researcher to develop “A statement about a population parameter… for the purpose of testing.” (Lind, et al, 2004, p. 317). Hypothesis testing is a five-step procedure using “sample evidence and probability theory to determine whether the hypothesis is a reasonable statement.” (Lind, et al, 2004, p. 318). The five steps used are: “State null and alternative hypothesis,” “Select a level of significance,” “Identify the test statistic,” “Formulate a decision rule,” and “Take a sample, arrive at a decision” (Lind, et al, 2004, p. 318). Many items are factored into the hypothesis testing such as the null hypothesis, alternative hypothesis, Type I error, Type II error, test statistics, critical value, and more. Understanding the method for testing hypothesis is essential for success.
Max Bazerman, a professor at Harvard University once said “People are ‘erroneously confident’ in their knowledge and underestimate the odds that their information or beliefs will be proved wrong. They tend to seek additional information in ways that confirm what they already believed.” (Cooper & Schindler, 2006, p. 536). This leads us to understand the importance of objective research and the careful consideration of a hypothesis so that it includes all possibilities while still being both “inclusive and mutually exclusive” (Hypothesis Testing, n.d., para. 3). The null hypothesis and the alternative hypothesis are nearly exact opposites of each other. For instance, a researcher might state that most people prefer Pepsi over Coke then develop a hypothesis that states, “People prefer the taste of Pepsi products over Coke” and the alternative hypothesis would be: “People prefer the taste of Coke products over Pepsi.”
The statement may be developed based on secondary data, which demonstrates that more Pepsi products are sold than Coke products. This information may be developed to determine which beverage to carry in a restaurant, and could be based on community preferences and local store sales. The next step would be to determine the “level of significance” which determines the risk factors involved in research information not containing the proper parameters and developing the wrong analysis. A Type I error is “rejecting the null hypothesis when it is true” and a Type II error is “accepting the null hypothesis when it is false” (Lind, et al, 2004). In the scenario of Pepsi versus Coke, benchmarking results may determine that beverage preference does not greatly influence the number of consumers interested in a restaurant; therefore, either error may not greatly affect the success of the restaurant. However, if the situation were more detrimental to consumers it would need a higher confidence level such as 99% or better, as is the case with a hypothesis, which states “Product A, shampoo, is both hypoallergenic and does not irritate sensitive skin.” A Type I could cost the consumers and the company (legal ramifications as well as loss of brand reliability), while a Type II error may cost the company research and development funding. Understanding these possible errors will allow the researcher to develop possible scenarios of what might occur.
Step 3 is the “Test Statistic – A value, determined from sample information, used to determine whether to reject the null hypothesis.” (Lind, et al, 2004, p. 321). This step involves using a formula to enable us to develop information gathered and go on to the final step. Parametric tests are also known as the “Z test” or the “t-test” and examine the “statistical significance between a sample distribution mean and a population parameter” (Cooper & Schindler, 2006, p. 550). As a census, or study of the whole population, is rarely compatible for a study, these means enable the population to be studied as a sample when the standard deviation is not known or if the sample size is large. For instance, if we use a sample of ten loaves of bread, per shift, it would only be 30 loaves of bread per day and be a sample standard deviation. As a quality hypothesis, which will demonstrate the quality of product produced by that bread manufacturer, a 99% confidence level would be important; and we would use the t-test due to the smaller sample size.
Selecting the correct tests is essential to the success of the hypothesis testing. It is important to know how many samples are involved, if the samples are “independent or related,” or if the “measurement scale [is] nominal, ordinal, interval, or ratio” (Cooper & Schindler, 2006, p. 548). These factors determine if we will use a One-Sample, Two-Independent-Samples, Two-Related-Samples, k-Independent-Samples, k-Related-Samples Tests (Cooper & Schindler, 2006). For example, a researcher could use the Two-Independent-Samples tests to evaluate the results of two different diet programs on similar target groups. The results could be tailored to evaluate the reaction to program, impact on lifestyle, and success rates.
In the next step, we develop a decision rule: “A decision rule is a statement of the specific conditions under which the null hypothesis is rejected and the conditions under which it is not rejected.” (Lind, et al, 2004, p. 321). This step includes a critical value which is “[t]he dividing point between the region where the null hypothesis is rejected and the region where it is not rejected.” (Lind, et al, 2004, p. 321). Using the appropriate tests enables the conditions to be analyzed and the critical value to be determined. From the example above, the diet program, we could set the null hypothesis, as “There is no difference between the two diet programs.” The alternative hypothesis might state that one diet program is better than the other. A t-test would be used because samples are independent and the data is interval. Next, we would calculate the value using a one-tailed test for the significance level of 0.05. Finally, we would go to the last step – interpreting the results.
The final step is “computing the test statistic, comparing it to the critical value, and making a decision to reject or not to reject the null hypothesis.” (Lind, et al, 2004, p. 322). This occurs when all the results are evaluated, hypothesis is rejected or accepted, and the results are reported. A focus group, using a taste-testing stand as a One-Sample Test, may determine that the local community does prefer Pepsi over Coke. After developing the information, it is important to evaluate it with an objective mind to ensure that the results are accurate and avoid the error as described by Professor Bazerman, in which information is gathered to show biased results.
“We can never know for sure if something is true, but we can make probability statements based on the Null Hypothesis Procedures, and this is almost as good.” (Hypothesis Testing, n.d., para. 2). Hypothesis testing is used to develop accurate statements and enable companies to make educated decisions on project management. Collecting, analyzing, and interpreting data requires a solid foundation and understanding of the questions that need answered. While it is impossible to develop statistics using entire populations, such as the quality of every loaf of bread produced, or the exact number of people who will purchase a new car in 2009, it is possible to use a sample of the population to estimate both of these scenarios.
References:
Cooper, D., & Schindler, P. (2006). Marketing Research. New York, NY: McGraw-Hill Irwin.
Hypothesis Testing. (n.d.). University of Phoenix. Retrieved on March 15, 2008, from University
of Phoenix, eResource.
Lind, D., Marchal, W., & Wathen, S. (2004). Statistical Techniques in Business and Economics,
12e. Retrieved on March 2, 2008, from University of Phoenix, eResource.
