Discussion post – Qualitative research methods

 

Primary Task Response: Within the Discussion Board area, write 400-600 words that respond to the following questions with your thoughts, ideas, and comments. This will be the foundation for future discussions by your classmates. Be substantive and clear, and use examples to reinforce your ideas.

Part 1

Top management of a large company has told you that they would like to be able to determine what the impact of years of service at their company has on workers’ productivity levels, and they would like to be able to predict potential productivity based upon years of service. The company has data on all of its employees and has been using a valid productivity measure that assesses each employee’s productivity.

What is a possible way to answer management’s question, and how does the analysis work? Reference the section on bi-variate linear regression from Green & Salkind.

Part 2

After reviewing the results of the analysis in Part I, the company found they are significantly related. However, a statistician reviewed the output and said that the results really do not explain much of what is happening in the total relationship.

What information is the statistician referring to from your output, and why is it important? Reference the section on bi-variate linear regression from Green & Salkind. 

Part 1

The data from this company is taking an independent variable (years of service) to try to predict a dependent variable (productivity level.)

 

A bi-variate linear regression can be used since it is one independent variable and a dependent variable per employee.  Since this dataset is similar to those procured from an experimental study, a Fixed-Effects model can be used to test for significance (Green & Salkind, 2017).  There are three assumptions that must be met in order to trust the p-value achieved from this model.  The first is that the dependent variable must be normally distributed for each level of the independent variable.  The large dataset will help minimize the effect of not having a normally-distributed variable should that be the case.  The second assumption is that the population variances for the dependent variable must be the same for each independent variable level.  This will require that all of the productivity levels vary the same amount for each level of years of service.  Finally, cases should represent a population sample and their scores are independent of each other.  An employee’s productivity level should have no effect on another’s productivity level.   This will ensure that any significance between years of service and productivity are true and not from another variable.

Part 2

            Looking at the analysis can tell us if years of service and productivity levels are significantly related.  It can also tell us in which way they are correlated.  Are they directly or inversely correlated?   The direct correlation will show that as years of service increase, so does productivity while an inverse correlation will show that productivity decreases as years of service increase.  If the 95% confidence interval of the slope goes from negative to positive, this correlation information will be null.  If this range was completely positive, it can be inferred that productivity is directly proportional to years of service and if the range was completely negative, the opposite can be said.

            Another finding could be that the standard deviation is large.   This will tell us that while years of service can predict productivity, it will be of little value.  For example, let us assume that productivity scores are from -3 to 6.  The model shows that the productivity scores can be predicted by years of service but with a standard deviation of 3.  This means if the standardized regression equation gives us a 1 for productivity, our actual score can be anywhere from -2 to 6.  This covers 90% of the range making it not a very useful prediction.

 

References

Green, S. B., & Salkind, N. J. (2017). Using SPSS. New York: Pearson.