Intro to Marketing Research: Hypothesis Testing - Nonparametric Alternatives, MannWhitney U Wilcoxon KruskalWallis Friedman

What It Is

Non-parametric alternatives are statistical methods used when the data does not meet the assumptions of parametric tests, such as normality or equal variances. A famous example is the use of the Mann-Whitney U test in a study by David H. Kaye and George C. Thornton Jr. (1999) to analyze the effectiveness of a new police lineup procedure. The study found that the new procedure significantly reduced the number of false identifications, demonstrating the importance of non-parametric methods in real-world applications.

Key Terms & Concepts

• Mann-Whitney U test: A non-parametric test used to compare two independent samples. It is used to determine if there is a significant difference between the medians of two groups. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not.)
• Wilcoxon rank-sum test: A non-parametric test used to compare two related samples. It is used to determine if there is a significant difference between the medians of two groups. (Example: A company wants to compare the sales performance of two different sales teams.)
• Kruskal-Wallis H test: A non-parametric test used to compare more than two independent samples. It is used to determine if there is a significant difference between the medians of three or more groups. (Example: A company wants to compare the satisfaction levels of customers who received different levels of customer service.)
• Friedman test: A non-parametric test used to compare more than two related samples. It is used to determine if there is a significant difference between the medians of three or more groups. (Example: A company wants to compare the sales performance of three different sales teams.)
• Non-parametric test: A statistical test that does not assume a specific distribution of the data. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not, but the data is not normally distributed.)
• Median: The middle value of a dataset when it is ordered from smallest to largest. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not, and the median satisfaction level is 4 out of 5.)
• Rank: A numerical value assigned to each data point based on its position in the ordered dataset. (Example: A company wants to compare the sales performance of two different sales teams, and the sales data is ranked from highest to lowest.)
• Hypothesis testing: A statistical method used to determine if there is a significant difference between two or more groups. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not, and the hypothesis is that the new feature will increase satisfaction.)
• P-value: A measure of the probability of observing a result at least as extreme as the one observed, assuming that the null hypothesis is true. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not, and the p-value is 0.01.)
• Null hypothesis: A statement that there is no significant difference between two or more groups. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not, and the null hypothesis is that there is no difference in satisfaction levels.)
• Alternative hypothesis: A statement that there is a significant difference between two or more groups. (Example: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not, and the alternative hypothesis is that the new feature will increase satisfaction.)

Common Misunderstandings

• Misunderstanding: The Mann-Whitney U test is only used for comparing two independent samples.
• Correction: The Mann-Whitney U test can also be used for comparing two related samples, but the Wilcoxon rank-sum test is more commonly used for this purpose.
• Misunderstanding: The Kruskal-Wallis H test is only used for comparing more than two independent samples.
• Correction: The Kruskal-Wallis H test can also be used for comparing more than two related samples, but the Friedman test is more commonly used for this purpose.
• Misunderstanding: Non-parametric tests are only used when the data is not normally distributed.
• Correction: Non-parametric tests can also be used when the data is normally distributed, but the sample size is small or the data is not normally distributed.

Quick Application / Identification

Scenario: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not. The data is not normally distributed, and the company wants to use a non-parametric test to compare the medians of the two groups.

• What type of test should the company use?
• Answer: The Mann-Whitney U test.
• Explanation: The Mann-Whitney U test is a non-parametric test used to compare two independent samples, and it is suitable for this scenario because the data is not normally distributed.

Scenario: A company wants to compare the sales performance of three different sales teams. The data is normally distributed, but the company wants to use a non-parametric test to compare the medians of the three groups.

• What type of test should the company use?
• Answer: The Kruskal-Wallis H test.
• Explanation: The Kruskal-Wallis H test is a non-parametric test used to compare more than two independent samples, and it is suitable for this scenario because the data is normally distributed, but the company wants to use a non-parametric test.

Scenario: A company wants to compare the satisfaction levels of customers who received a new product feature versus those who did not. The data is normally distributed, and the company wants to use a non-parametric test to compare the medians of the two groups.

• What type of test should the company use?
• Answer: The Wilcoxon rank-sum test.
• Explanation: The Wilcoxon rank-sum test is a non-parametric test used to compare two related samples, and it is suitable for this scenario because the data is normally distributed, and the company wants to compare the medians of two related groups.

Last-Minute Revision

• The Mann-Whitney U test is used to compare two independent samples, and it is based on the ranks of the data.
• The Wilcoxon rank-sum test is used to compare two related samples, and it is based on the ranks of the data.
• The Kruskal-Wallis H test is used to compare more than two independent samples, and it is based on the ranks of the data.
• The Friedman test is used to compare more than two related samples, and it is based on the ranks of the data.
• Non-parametric tests are used when the data is not normally distributed or when the sample size is small.
• The median is the middle value of a dataset when it is ordered from smallest to largest.
• The p-value is a measure of the probability of observing a result at least as extreme as the one observed, assuming that the null hypothesis is true.
• The null hypothesis is a statement that there is no significant difference between two or more groups.
• The alternative hypothesis is a statement that there is a significant difference between two or more groups.
• The Wilcoxon signed-rank test is used to compare two related samples, and it is based on the ranks of the data.
• The Spearman rank correlation coefficient is used to measure the strength and direction of the relationship between two variables.
• The Kendall's tau coefficient is used to measure the strength and direction of the relationship between two variables.
• The Mann-Whitney U test is sensitive to outliers, and it is not suitable for datasets with extreme values.
• The Kruskal-Wallis H test is sensitive to outliers, and it is not suitable for datasets with extreme values.
• The Friedman test is sensitive to outliers, and it is not suitable for datasets with extreme values.
• The p-value is not the probability that the null hypothesis is true.
• The p-value is not the probability that the alternative hypothesis is true.
• The null hypothesis is not the same as the alternative hypothesis.