| Course | HRM-7640 Technology, Data, and Analytics as Change Agents |
|---|---|
| Module | Module 3 |
| Paper type | Module 3 written assignment: predictive statistics for competitive advantage |
| Length | About 773 words, 5 pages |
| Format | APA 7 student paper |
| School | Thomas Edison State University |
| Program | MBA in Human Resource Management |
| Updated | October 2026 |
Free sample paper for HRM 7640 Module 3
Who Leaves in Ninety Days? A Logistic Regression of Early Turnover Among Home Health Aides
Student Name
MBA in Human Resource Management, Thomas Edison State University
HRM-7640: Technology, Data, and Analytics as Change Agents
Mentor Name
Month Day, Year
Who Leaves in Ninety Days? A Logistic Regression of Early Turnover Among Home Health Aides
Introduction
The question set in Module 1 was simple: which new aides will probably quit inside ninety days? In Module 2, the agency began cleaning its data. With three years of records on 1,480 aides hired, 38 percent of whom left within ninety days, the agency can now build a model. Fitz-enz and Mattox (2014) see the advantage of prediction in acting sooner and more precisely than rivals on whatever drives results. In home health, where agencies compete for the same limited pool of aides, keeping new aides longer is a direct advantage. Marler and Boudreau (2017) found that the value of HR analytics depends on connecting people data to business outcomes, which early turnover in home care does directly.
Method
The outcome is binary: an aide either left within ninety days or did not. Logistic regression is the standard method for such outcomes, because it estimates the probability of an event from several predictors at once (Hosmer et al., 2013). The predictors were chosen from supervisors' suggestions and the data available: a commute of more than 30 minutes to assigned patients, a weekend-only schedule, a starting wage below $18 an hour, assignment of a peer mentor and prior home care experience. The model was built on 70 percent of the hires and tested on the remaining 30 percent to check that it predicts new cases, not only the data it was built on.
Results
An odds ratio above one means the factor increases the odds of leaving, holding the other factors constant, and below one means it decreases them. All five predictors were statistically significant. On the test sample, the model's area under the ROC curve was 0.71, meaning that for a randomly chosen leaver and stayer, the model ranks the leaver as higher risk about 71 percent of the time. That is useful but far from perfect.
| Predictor | Coefficient | Odds ratio | Interpretation |
|---|---|---|---|
| Commute over 30 minutes | 0.62 | 1.86 | Odds of leaving 86 percent higher |
| Weekend-only schedule | 0.48 | 1.62 | Odds 62 percent higher |
| Starting wage below $18 | 0.41 | 1.51 | Odds 51 percent higher |
| Assigned a peer mentor | -0.71 | 0.49 | Odds roughly halved |
| Prior home care experience | -0.35 | 0.70 | Odds 30 percent lower |
Predicted Probabilities
The model's value is clearest in predicted probabilities for individual profiles. An aide with a long commute, a weekend-only schedule and a starting wage below $18, with no mentor or prior experience, has a predicted probability of leaving within ninety days of about 72 percent. An aide with a short commute, a weekday schedule, a mentor and prior experience has a predicted probability of about 17 percent. Assigning a mentor to the first aide alone lowers the predicted probability to about 56 percent.
Turning Predictions Into Advantage
The results suggest three actions. First, assign peer mentors to every new aide, starting with those at highest predicted risk, since mentoring is the strongest protective factor and is within the agency's control. Second, assign patients to new aides within a 30-minute commute wherever possible, a scheduling change that costs little. Third, review starting pay for weekend-only roles, where the combination of schedule and pay raises risk most. These actions are prescriptive uses of the model: they turn predictions into decisions. Because they are built on Shoreline's own data and processes, competitors cannot simply copy them.
Limits
The model shows associations, not proof of cause. Aides with mentors may differ in other ways, for example if supervisors assigned mentors to aides they thought likely to succeed. A small pilot that assigns mentors to a randomly selected half of new aides for three months would test the effect directly. The model also omits factors the agency does not record, such as family circumstances, and its accuracy will drift as the workforce changes, so it should be retrained each year.
Fairness Checks
Before using predictions in any decision about individuals, Shoreline must check that the model does not disadvantage protected groups. Commute distance, for example, may correlate with neighborhood and therefore with race or ethnicity. The agency should not use the model to screen applicants; it should use it to direct support to new hires, and it should compare predicted and actual turnover by group to confirm the model is equally accurate for all.
Conclusion
A logistic regression of ninety-day turnover gives Shoreline Home Health a practical, interpretable model. Mentoring, commute and schedule emerge as levers the agency controls. Used to target support rather than to screen people, tested with a pilot and checked for fairness, the model can help the agency keep aides longer than its competitors.
References
Fitz-enz, J., & Mattox, J. R., II. (2014). Predictive analytics for human resources. John Wiley & Sons.
Hosmer, D. W., Lemeshow, S., & Sturdivant, R. X. (2013). Applied logistic regression (3rd ed.). John Wiley & Sons.
Marler, J. H., & Boudreau, J. W. (2017). An evidence-based review of HR analytics. The International Journal of Human Resource Management, 28(1), 3-26. https://doi.org/10.1080/09585192.2016.1244699
HRM 7640 Module 3 instructions, in plain terms
The second HRM-7640 written assignment covers the use of predictive statistics to create competitive advantage. Expect to build or interpret a predictive model of an HR outcome, such as turnover, performance or engagement, explain its results in plain terms and show how they would change decisions. You may be handed data, or you may be asked to design the analysis yourself. Explain the method, interpret coefficients or other results carefully, discuss accuracy and limits and address fairness. Write for HR leaders who are not statisticians, and use APA style for tables and citations. If you design the analysis rather than run it, explain the data you would need and how you would interpret each result.
Inside the HRM 7640 Module 3 example
The paper restates the prediction question and the data. A method section explains why logistic regression fits a yes-or-no outcome and how the model was validated. A results table gives coefficients, odds ratios and plain-language interpretations, followed by a measure of accuracy. A section on predicted probabilities shows the model's meaning for individual profiles. Three actions turn predictions into advantage. Sections on limits and fairness explain what the model cannot show and how to use it responsibly. The order moves from method to results to decisions. A section on limits proposes a pilot to test whether the strongest factor actually causes better retention.
Where the marks sit in the HRM 7640 Module 3 rubric
The modeling paper is marked on sound methods, careful reading of the output, clear links from results to decisions and honest treatment of limits and fairness. Answers that explain odds ratios and accuracy in plain language, and that show how predictions would change specific practices, demonstrate applied skill. Recognizing that associations are not causes, and proposing a test, shows statistical maturity. Fairness checks before using predictions about people reflect professional responsibility. A readable results table and proper citations round out the mark. Accuracy reported on a holdout sample is valued, because it shows the model works on cases it has not seen. Using predicted probabilities for example profiles makes results meaningful to leaders. Fairness checks before using predictions demonstrate responsible practice.
HRM 7640 Module 3 help from the desk
Students lose points by misreading coefficients, treating associations as causes or presenting accuracy figures without explaining them. Others build models and stop short of a recommendation. Interpret each result in a sentence a manager would understand. Hold back part of the data to test the model. Propose actions within the organization's control. Check fairness before using predictions about individuals. A free first sample using your own HRM-7640 Module 3 data is available on request. Explain odds ratios in plain language. Report accuracy on a holdout sample. Recommend actions the organization can control. State what the model cannot show and how you would test it.
Write yours, or have the desk draft it
This paper is an original model document written by our desk, not a submitted student paper and not an official Thomas Edison State University document. Read it for the moves, then write your own to the instructions in your classroom. If you want one built to your exact prompt and rubric, the first custom sample is free and arrives in 24 to 48 hours.
More HRM 7640 and MBA in Human Resource Management sample papers
- HRM 7640 Module 1: Discussion: From Reporting to Prediction in HR Analytics
- HRM 7640 Module 2: Written Assignment: Building an Analytics Culture in HR
- HRM 7640 Module 4: Final Project: An HR Analytics Plan That Links Activity to Outcomes
- HRM 7630 Module 4: Final Project: The Perfect Relationship Handbook for HR and Counsel
- HRM 7610 Module 1: Discussion: Culture and the Limits of a Global HR Policy
- HRM 7620 Module 5: Discussion: The Talent Infrastructure Behind Inclusion
HRM 7640 Module 3 questions, answered
Where can I find a free HRM 7640 Module 3 sample paper?
The complete Module 3 paper with a logistic regression of ninety-day turnover at Shoreline Home Health is above, with odds ratios and actions.
What is logistic regression?
A regression method for outcomes that either happen or do not, such as quitting in the first ninety days, which turns several predictors into a probability.
What does an odds ratio mean?
The factor by which a predictor multiplies the odds of the outcome, holding other predictors constant; above one raises the odds.
What does an area under the ROC curve of 0.71 mean?
That the model ranks a random leaver above a random stayer about 71 percent of the time, which is useful but imperfect.
How can predictive models in HR be used fairly?
By using them to target support rather than screen people, and by checking accuracy and outcomes across groups.