Stats 2 Final Project

Seth Sauer, Madhur Mittal, Chris Driessen, Dan Song, Xiaoxu Mu


A statistical analysis of employee satisfaction using Confidence Intervals, ANOVA, Paired Hypothesis Testing, and Multiple Regression.


Executive Summary: Overview of the dataset, the analysis, and results.
In our project, we first took the dataset and saw if there was a significant difference in employee satisfactions among the three teams. We calculated the confidence interval for team one, two and three at the 95% confidence interval to see if there is a mean difference among the three teams. In testing for the confidence intervals, we found that the confidence interval for team 1 ranged from 19.88 to 26.59. Additionally, for Team 2 it was 21.84 to 24.69, and 19.62 to 25.19 for Team 3. From looking at the confidence interval data, we are able to see that they clearly overlap, and conclude that there is not any significant difference in the means of the three teams. Next, we tested our dataset to see if there was a significant difference between the support an employee receives from the organization as a wholeand the manager specifically. We tested this with an alpha level of 0.05. We said the null hypothesis was that the support received by an employee is the same from organization, as well as the manager individually. For our alternative hypothesis, we said that the support received by an employee is not the same from the organization compared to the manager. In doing our testing, we found that 2.79E08<0.05, so we reject the null hypothesis, and were able to say that that support received by an employee is not the same from the organization as the manager; along with also being able to safely say that they are significantly different from each other. Next, we tested to see if there was a significant mean difference in employee satisfaction between the three teams. We used an ANOVA test to do this, and compared team 1, team 2 and team 3 to find whether there was a significant difference between these three groups. Our null hypothesis was that there is no difference amongst the teams (Team1=Team2=Team3); with our alternative hypothesis being that not all teams are equal, or at least two of the teams are different. From running the ANOVA test, we found that the pvalue is 0.8208greater than the alpha value of 0.05. Given this, we cannot reject the null hypothesis. Furthermore, we can now say that there is no significant difference amongst the three teams. In this situation, there is no need for the “post –hoc” test, as there was no significant difference found between the teams in the ANOVA test performed. We would perform a posthoc test only if there was a difference found, and do so to determine which groups were different. Next, we saw if there was there a correlation between three independent variables and a dependent variable. With the dependent variable being employee satisfaction; the independent variables we used were organizational support, managerial support, and peer support. Using the variable inflation factors (VIF’s), we first tested to see if there was multicollinearity between our three independent variables. We found that all of the VIF’s for the three variables were below two. This means and shows us that there is no concern with multicollinearity. Our next step was to test for outliers utilizing the Cook’s D Test; with the Cook’s D cutoff value being 0.04705. We ran the test and were able to remove 5 pieces of data. After doing this, we next ran a multiple regression statistical analysis. Additionally, we made scatter plots for each of the three independent variables. We found that organizational support and managerial support were significant predictors of employee satisfaction. Our predictor variables can account for about 38% of employee satisfaction. This means there are other factors that attribute to employee satisfaction. Description of Sample/ Background:
The...
Please join StudyMode to read the full document