| Course | GB892 Doctoral Project III: Data Analysis |
|---|---|
| Unit | Unit 9 |
| Paper type | Departure log |
| Length | About 710 words, 5 pages double-spaced |
| Format | APA 7 student paper |
| School | Purdue University Global |
| Program | MBA |
| Updated | October 2026 |
Free sample paper for GB892 Unit 9
Seven Departures From the Plan: A Unit 9 Departure Log for a Dock Appointment Analysis
[Student Name]
Purdue University Global
GB892: Doctoral Project III: Data Analysis
Unit 9 Departure Log
[Instructor Name]
[Date]
Haverly Logistics and every estimate are composite and illustrative; the approver cells are left for the chair, and the methods sources cited are real.
Why a Departure Log
A proposal commits to an analysis before the data are seen. When the data force changes, the credibility of the results depends on showing what changed, why and what difference it made. Simmons et al. (2011) demonstrated that undisclosed choices made after seeing data can make almost any result look significant; a log that records each choice with its reason, and reports results both ways where possible, is the remedy. Each row below has seven parts: a number, a date, what was planned, what was done instead, the reason, an approver line left for the dissertation chair and the effect on results.
Row 1: Outcome Scale
Date. Week three of analysis, during data preparation. Planned. Dwell in minutes as the primary outcome. Done instead. Dwell in log minutes. Reason. Skewness of 1.77 in raw minutes, reported in the Unit 4 assumption check, meant the longest waits would dominate a model of means. Approver. Left blank for the chair. Effect. The planned model put the saving at 6.2 minutes, between 2.3 and 10.0 at 95 percent confidence; the adopted model gives roughly 4 percent, or about 5 minutes at the pre-adoption median. Both appear in the results, the planned one in an appendix.
Row 2: Inference Method
Date. Week four. Planned. Conventional clustered standard errors. Done instead. A wild cluster bootstrap. Reason. With a dozen centers, conventional cluster-robust errors claim more precision than the data hold (Cameron et al., 2008). Effect. Intervals are wider than conventional errors would have given; the primary interval of 1 to 7 percent would have been about 2 to 6 percent conventionally.
Row 3: Secondary Outcome
Date. Week four. Planned. Detention charges per visit. Done instead. The two-hour share, meaning the proportion of visits running past two hours. Reason. Detention is invoiced monthly by carrier with no visit number, so charges cannot be linked to visits. Effect. A different secondary outcome, reported as a proxy for exposure to detention rather than as dollars.
Row 4: Excluded Center-Months
Date. Week five. Planned. All 216 center-months. Done instead. 215 center-months, excluding DC-03's migration month in the primary model. Reason. The yard system migration filled missing departures with false times. Effect. Small; the sensitivity model including the month leaves the estimate essentially where it was.
Row 5: An Added Sensitivity Model
Date. Week five. Planned. No center-specific sensitivity models. Done instead. A model excluding DC-09. Reason. DC-09 stepped down by about 11 percent in month five after adding a receiving shift, as the Unit 4 check found. Effect. The estimate without DC-09 is about 5 percent rather than 4.
Row 6: Site Ranking Removed
Date. Week six, after the seminar. Planned. Site-by-site estimates ranked from largest to smallest. Done instead. One pooled estimate, with site medians shown descriptively. Reason. Each site estimate would rest on one treated center, too few for a meaningful interval. Effect. No ranking appears in the results.
Row 7: Fewer Interviews
Date. Week eight. Planned. Twenty-four interviews, twelve supervisors and twelve dispatchers. Done instead. Seventeen interviews, eleven supervisors and six dispatchers. Reason. Carrier dispatchers were harder to reach than expected, and two carriers declined to allow staff to participate. Effect. Dispatchers' accounts are thinner, and the thematic analysis notes this where themes depend on them.
Every departure appears with the estimate it moved, so a reader can see that the headline result does not depend on any single change.
What the Log Shows Overall
Taken together, the departures moved the primary estimate within a narrow range: about 4 percent in the adopted model, about 5 percent without DC-09 and 6.2 minutes in the planned raw-minute model, which corresponds to roughly the same proportional change. None reversed the direction of the result. Angrist and Pischke (2009) describe robustness to reasonable alternative specifications as a core test of a credible estimate, and the log provides the material for that test.
References
Angrist, J. D., & Pischke, J.-S. (2009). Mostly harmless econometrics: An empiricist's companion. Princeton University Press.
Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. The Review of Economics and Statistics, 90(3), 414-427. https://doi.org/10.1162/rest.90.3.414
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359-1366. https://doi.org/10.1177/0956797611417632
What the GB892 Unit 9 assignment asks for
The ninth GB892 assignment asks you to log every departure between your proposed and actual analysis. Explain why such a log matters, citing research on undisclosed analytic flexibility. For each departure, record a number, the date, what was planned, what was done instead, the reason, an approver line for your chair and the effect on results, reporting both estimates where possible. Distinguish changes forced by the data from changes chosen for other reasons. Include changes to qualitative data collection as well as models. Then summarize what the departures do together to the main result, including whether any reversed its direction. Leave approver cells blank for the chair rather than filling them yourself. Cite methods sources that justify individual changes. About five pages in a single table with one row per departure and seven columns is a common format for this log.
How this GB892 Unit 9 example is built
The log explains, with Simmons and colleagues, why disclosure protects credibility. Row one switches to log minutes after skewness of 1.77, with both the 6.2-minute and roughly 4 percent estimates printed, as a margin note stresses. Row two adopts the wild bootstrap, citing Cameron and colleagues, and reports how intervals widened. Row three replaces detention charges with the two-hour share, and a second note explains why the original was impossible. Row four drops DC-03's migration month, row five adds a model without DC-09, row six removes the site ranking and row seven records seventeen interviews against twenty-four. The highlighted sentence states the purpose, and a summary with Angrist and Pischke shows the direction never reversed.
GB892 Unit 9 rubric: what earns full points
Points in the GB892 departure log follow completeness and candor. Instructors expect every departure, including qualitative ones, with a date, plan, change, reason and effect. Reporting both planned and adopted estimates is weighted heavily because it lets readers judge the change. Leaving approver cells for the chair shows respect for oversight. Distinguishing changes forced by data from discretionary ones shows understanding of analytic flexibility. A summary of what the departures do together, especially whether any reversed the result, demonstrates judgment. Methods sources should justify individual changes. Clear tabular format with consistent columns makes the log usable, and the dates should match the analysis record so the sequence of decisions can be followed.
Common GB892 Unit 9 mistakes, and how to avoid them
Many GB892 departure logs list changes without their effect on results, which hides whether any mattered. Others omit changes to interviews or secondary outcomes. Students sometimes fill in approver cells themselves. The planned estimate is dropped once the adopted one exists. Logs rarely say whether changes were forced by the data or chosen. A summary of the changes' combined effect is usually missing. Log every departure with its date, plan, change, reason and effect, print both estimates, include qualitative changes, leave approval to the chair, separate forced from chosen changes, cite justifying sources and summarize the overall effect. Share your proposal and final analysis, and our team can help draft the log with you.
Get a GB892 Unit 9 example written to your instructions
Put your approved analysis plan beside what you actually ran and send both, with the Unit 9 prompt and rubric. Returned to you: a departure log with every change dated, sorted into forced and chosen, and measured against the estimate it moved. First custom sample, no charge; 24-48h. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
More GB892 papers and MBA samples
- GB892 Unit 1: Discussion Board Post
- GB892 Unit 2: Data Preparation Record
- GB892 Unit 3: Codebook
- GB892 Unit 4: Assumption Check Summary
- GB892 Unit 5: Descriptive Results Table
- GB892 Unit 6: Seminar Reflection
- GB892 Unit 7: Thematic Analysis
- GB892 Unit 8: Inferential Results Write-Up
- GB892 Unit 10: Results Chapter Draft
- GB891 Unit 9: Deviation Report
- GB500 Unit 9: Professional Development Plan
- GB890 Unit 9: Source Appraisal
- GB512 Unit 9: Business Presentation Deck
GB892 Unit 9 questions, answered
Must a departure log include the results of the original planned analysis?
Where the planned analysis could still be run, reporting its result is the strongest evidence that a change was not chosen for its outcome. Place it in an appendix or a table note beside the adopted estimate. Where the original cannot be run at all, as with an outcome that has no data behind it, say so plainly instead.
Who approves a departure from the analysis plan?
Usually the chair or committee, and sometimes the review board if the change touches data collection or consent. A departure log records what was requested and when, and leaves the approval cell blank until someone has actually approved it. Your program's procedures decide who must sign which kinds of change.
Is a sensitivity analysis a departure?
When the plan did not call for it, yes, and it should be logged and labeled exploratory. Planned sensitivity analyses are part of the approved design and need no row. The distinction matters because unplanned analyses multiply the chances of finding something, and readers weigh them accordingly when judging the primary estimate.
Where can I find a free GB892 Unit 9 sample paper?
This page holds a full GB892 Unit 9 departure log for a dock appointment analysis, recording seven departures from the proposal and the estimate each one moved.
Why report both the planned and adopted estimates?
Readers can then see how much a change in method moved the result. If only the adopted estimate appears, no one can tell whether the change made the finding look stronger.