Discussion:
In a specific company, there are three levels of workers: skilled, unskilled, and manager. The owner of this company wants to know if the level of job satisfaction differs amongst the three levels of workers.
Data: 30 workers in each level are randomly selected to take the job satisfaction survey. The survey asks each worker on a scale of 1-10 how satisfied he or she is with his or her job. The data will consist of 90 workers total (30 belonging to skilled, 30 belonging to unskilled, and 30 belonging to manager) and their satisfaction rating.
Null Hypothesis: The population means of satisfaction ratings of each of level of worker are equal to each other.
Alternative Hypothesis: The population means of satisfaction ratings of each of level of worker are not all equal to each other.
Type 1 Error: We incorrectly reject the null hypothesis that the population means of satisfaction ratings of each of level of worker are equal to each other.
Type 2 Error: We fail to reject the null hypothesis that the population means of satisfaction ratings of each level of worker are equal to each other even though the null hypothesis if false.
If we obtained a p-value a .2 from a test statistic, that means we have a 20 percent probability of obtaining a test statistic at least as extreme as our test statistic assuming the null hypothesis is true. At a .05 level of significance, we would fail to reject the null hypothesis.
If we obtained a p-value a .01 from a test statistic, that means we have a 1 percent probability of obtaining a test statistic at least as extreme as our test statistic assuming the null hypothesis is true. At a .05 level of significance, we would reject the null hypothesis.
a. Reformulate your hypothesis test to incorporate a 2-sample hypothesis test. What would be your data? What is your null hypothesis? What is your alternate hypothesis? What would be your Type 1 and Type 2 errors relative to your decision? Suppose you have a p-value of 0.01, what does this mean relative to your problem and decision? Suppose the p-value is 0.20, what does this mean relative to your problem and decision?
b.If you reformulated your design for 3 or more samples, what would be the implications of interaction? When would you use Tukey-Kramer test?