GB891 Unit 6 Seminar Reflection Example

When the Person Holding the Door Leaves: A Unit 6 Seminar Reflection on Site Dependence

[Student Name]

Purdue University Global

GB891: Doctoral Project II: Project Design and Implementation

Unit 6 Seminar Reflection

[Instructor Name]

[Date]

The classmate's study, Corliss Baking Company and its people are composite; the sources cited are real.

The Classmate's Account

Teresa's study tested a new handoff checklist on three medical units of a regional hospital, with two other units as comparisons. Her sponsor was the director of nursing, who had approved the study, introduced Teresa to the unit managers and signed the letter of cooperation. In week five, the director accepted a position at another hospital. Her interim replacement, cautious about commitments he had not made, asked Teresa to pause until he could review the study. The review took six weeks. By the time it ended, the comparison units had begun using a version of the checklist on their own, and Teresa's comparison had lost its meaning.

The seminar did not dwell on what Teresa could have done differently. It asked what she could still use: four weeks of clean data before the pause, staff interviews about why the comparison units adopted the checklist, and the adoption itself, which said something about how good practices spread within a hospital.

What I Saw in My Own Study

As the discussion went on, I realized I had a version of Teresa's problem. The Dunmore plant, one of my two comparison plants, is managed by someone who has been there nineteen years and who agreed to the study because he trusts me. He has announced that he will retire in late March, which falls in week eight of the study. His successor has not been named.

I had treated the comparison plants as the safe part of the design. Nothing is asked of them except to carry on as usual while pulse questions are asked and observations made. But a new manager could easily decide that outside observers at his stations, or a research assistant interviewing his crews at break, were a distraction he did not need in his first month. Or, worse for the design, he could arrive with his own safety program, which would turn my comparison plant into a second intervention plant, as Teresa's comparison units had become.

I had protected the plants that were receiving something and left unguarded the plant that was only asked to stay the same.

Changes I Made

I made three changes the following week.

First, I asked Corliss's chief operating officer to tell the incoming Dunmore manager, as part of his onboarding, that the plant is a comparison site for a company-approved study through April, and that no new safety program should start there before May. The letter of cooperation is signed by the company rather than by individual managers, so this request is consistent with what the company has already agreed.

Second, I added a weekly check to the collection log: a line asking whether any safety initiative, campaign or training has begun at any study plant that week. If one begins at a comparison plant, the log will show it, and the analysis can account for it. Shadish et al. (2002) describe this kind of contamination, when comparison groups receive some version of the treatment, as a threat that can only be addressed if it is detected.

Third, I wrote down in advance what I would do if Dunmore were lost. The design would continue with one comparison plant, Calder, and the analysis would compare each intervention plant with Calder alone. The loss of statistical power would be real, but the study would still answer its question for the crews that remained.

What Remains Exposed

Some exposure cannot be removed. If the new manager declines to continue, I cannot compel him; a letter of cooperation is a commitment from the company, not a command to its managers. And if he introduces changes quietly, through how he talks to supervisors, the weekly check may not catch them. Carroll et al. (2007) note that fidelity concerns not only whether an intervention is delivered as planned but whether conditions in comparison groups remain as planned, and that is exactly the part I cannot fully control.

What the Seminar Changed

The session also changed how I talk about the study with plant managers. I had described the comparison plants to their managers as places where nothing would change, which invited them to think nothing was being asked of them. I now describe them as half of the study, whose value depends on staying as they are for twelve weeks, and I ask each manager to tell me before any new safety effort begins.

Before the session, I thought about site risk as the risk of losing access to the plants receiving the intervention. Teresa's story showed me that comparison sites carry their own risk, and that it is easier to miss because nothing seems to depend on them. The design depends on them entirely. Even the intervention that inspired this study, run by Zohar (2002) with comparison groups inside the same plant, depended on those groups staying untouched, and his groups at least shared one manager who knew the study.

References

Carroll, C., Patterson, M., Wood, S., Booth, A., Rick, J., & Balain, S. (2007). A conceptual framework for implementation fidelity. Implementation Science, 2, Article 40. https://doi.org/10.1186/1748-5908-2-40

Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Houghton Mifflin.

Zohar, D. (2002). Modifying supervisory practices to improve subunit safety: A leadership-based intervention model. Journal of Applied Psychology, 87(1), 156-163. https://doi.org/10.1037/0021-9010.87.1.156