DBA 831 Topic 8 pre-specified analysis plan example

Analytic Foundations for Business Leaders Grand Canyon University Free custom sample in 24 to 48h

This DBA 831 Topic 8 sample paper is an analysis plan, dated before any data arrive, for testing a safety incentive program across a composite building products maker's twenty plants. It was prepared for GCU DBA 831, Analytic Foundations for Business Leaders, catalog code DBA-831, the course in Grand Canyon University's DBA that closes on research commitments. The primary outcome is recordable injuries per 200,000 hours over twelve months; because Kerr warned about rewarding one thing while hoping for another, near misses and clinic visits guard against suppressed reporting. Ten plants start early by witnessed draw, a Poisson model is fixed, one subgroup is allowed and each result pattern is interpreted in advance.

CourseDBA 831 Analytic Foundations for Business Leaders
TopicTopic 8
Paper typePre-specified analysis plan
LengthAbout 1,000 words, 2 pages
FormatAPA 7 student paper
SchoolGrand Canyon University (GCU)
ProgramDBA
UpdatedOctober 2026

Free sample paper for DBA 831 Topic 8

1

Fixed Before the First Injury Report: A Pre-Specified Analysis Plan for Testing a Safety Incentive Program Across Twenty Plants

[Author Name]

College of Doctoral Studies, Grand Canyon University

DBA-831: Analytic Foundations for Business Leaders

Topic 8 Pre-Specified Analysis Plan

[Instructor Name]

[Date]

The organization, setting and figures below are a composite written as a model document. No real employer, client, colleague or patient is described.

What this page is doingThe title states when the plan's choices were fixed and what it tests. The manufacturer and every figure are a composite, labeled illustrative; the sources are real.
2

Fixed Before the First Injury Report: A Pre-Specified Analysis Plan for Testing a Safety Incentive Program Across Twenty Plants

Research Question, Hypothesis and Date

This plan was written and dated before the study began, and it was filed with the company's internal audit group so that any later change can be compared with the original. A composite manufacturer of building products operates twenty plants with about 4,600 production employees. It intends to introduce a safety incentive program that pays each crew a quarterly bonus for every quarter without a recordable injury. The research question is whether the program reduces injuries. The hypothesis is that plants operating the program will record fewer recordable injuries per hours worked over twelve months than plants not yet operating it. Simmons et al. (2011) showed that undisclosed flexibility in choosing outcomes, exclusions and comparisons can make almost any hypothesis appear supported; this plan removes that flexibility by fixing each choice now. Nosek et al. (2018) described the same commitment as the defining feature of preregistration.

Outcomes

The primary outcome is the recordable injury rate: recordable injuries per 200,000 hours worked, the standard denominator for about 100 full-time employees over a year, measured over the twelve months after each plant's start date. Hours come from payroll records, injuries from the plant's injury log as reviewed by the corporate safety office.

Paying crews for injury-free quarters gives them a reason not to report injuries, a version of the mismatch between what is paid for and what is wanted that Kerr (1975) analyzed. A fall in recorded injuries could therefore mean fewer injuries or fewer reports. To tell these apart, the plan names two secondary outcomes that the incentive does not reward and that tend to move with true injury frequency: near-miss reports per 200,000 hours, and first-aid visits to the plant clinic per 200,000 hours. Clinic visits are logged by contracted nurses, not by crews.

Design and Unit of Analysis

The twenty plants are randomized, by a draw witnessed by the audit group, into ten that start the program at the beginning of the year and ten that start twelve months later. The later group serves as the comparison during the first year and receives the program in the second, so no plant is denied it. The plant is the unit of analysis because the program is introduced plant by plant and plants differ in processes and hazards; individual injuries are not treated as independent observations.

Statistical Model and Data Handling

Injury counts will be analyzed with a Poisson regression of each plant's monthly count on program status, with hours worked as the exposure, the plant's injury rate in the prior twelve months as a covariate and standard errors clustered by plant. If the data are more dispersed than the Poisson model allows, judged by a dispersion statistic above 1.5, a negative binomial model will be used instead; that rule is fixed now. The test is two-sided at the 5 percent level.

Months with no hours worked are excluded. A plant that closes during the study contributes the months it operated, and its closure is reported. Missing monthly injury counts will not be imputed; if more than two months are missing for any plant, the analysis will be repeated without that plant and both results reported. No observations will be removed as extreme values; a single serious incident is part of the outcome.

Subgroup Analysis

One subgroup analysis is permitted: plants operating three shifts compared with plants operating one or two, because night shifts carry different hazards and supervision. Any other comparison, by product line, plant age, region or crew size, will be labeled exploratory and reported as a basis for future study, not as a finding.

Interpreting the Results in Advance

Each combination of primary and secondary results has an interpretation fixed now, so that no result can be read selectively once it is known.

Table 1

Pre-Specified Interpretation of Each Pattern of Results

Recordable injuriesNear misses and first aidInterpretation
FallFall or steadyProgram likely reduced injuries
FallRise or no change while injuries fall sharplyPossible suppressed reporting; not a success
FallFall more steeply than injuriesReporting suppressed overall; investigate
No changeAnyNo evidence of effect at this sample size
RiseAnyProgram may have harmed safety; review immediately

With ten plants per group, the study can reliably detect only a large effect, about a 30 percent reduction in the injury rate. A smaller true effect may produce a null result, which the report will describe as inconclusive rather than as evidence that the program fails.

Who Sees the Data, and When

Results will not be examined while the first year is under way. The corporate safety office will collect monthly counts, but the comparison between early and later plants will be computed only once, after the twelfth month, by an analyst who did not design the program. Plant managers will see their own plants' figures as usual, since safety cannot wait for a study, but will not see the comparison. This rule matters because a program that looks promising after four months invites pressure to extend it early, which would destroy the comparison group, and a program that looks poor invites pressure to change the outcome measure. The one exception is safety: if any plant's recordable rate doubles over three consecutive months, the safety office will review it at once, whatever the study's status.

Allowed Deviations and Disclosure

Three deviations are allowed. If a plant cannot start on its assigned date for operational reasons, it may start up to two months later, and the analysis will use the assigned date. If a serious incident requires a plant to suspend the program, the plant stays in its assigned group. If corporate policy changes the definition of a recordable injury during the study, results will be reported under both definitions. Any other change to this plan will be dated, explained and reported alongside the results, with the original analysis also shown, so that a reader can see what changed and why.

References

Kerr, S. (1975). On the folly of rewarding A, while hoping for B. Academy of Management Journal, 18(4), 769-783. https://doi.org/10.2307/255378

Nosek, B. A., Ebersole, C. R., DeHaven, A. C., & Mellor, D. T. (2018). The preregistration revolution. Proceedings of the National Academy of Sciences, 115(11), 2600-2606. https://doi.org/10.1073/pnas.1708274114

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

The DBA-831 Topic 8 assignment instructions

Topic 8 in DBA-831 asks doctoral students to write an analysis plan fixed before data collection begins. Strong plans state the question, hypothesis and date, define the primary outcome with its denominator and period, and name secondary outcomes that guard against the most likely distortion. Students should describe the design and unit of analysis, fix the model and the handling of missing data, closures and extreme values, and limit subgroup analyses in advance. The plan should state how each combination of results will be read, who sees the data and when, and which deviations are allowed and how they will be disclosed.

Inside the DBA 831 Topic 8 example

The finished plan reads as a document written before the study starts and dated to show it. It states the hypothesis, that the incentive program reduces recordable injuries, and the primary outcome, recordable injuries per hours worked over twelve months. Because paying workers for injury-free periods can discourage reporting, the plan names near-miss reports and first-aid visits as secondary outcomes, and it states in advance that falling injuries alongside falling near-miss reports would be read as suppressed reporting, not success. Plants are assigned to start the program in a randomized order. Exclusion rules, the handling of plants that close and the one permitted subgroup analysis are all fixed. Simmons, Nelson and Simonsohn's demonstration of how flexible analytic choices produce false positives is cited as the reason for the plan's rigidity.

Where the marks sit in the DBA 831 Topic 8 rubric

Faculty reward choices fixed while results are unknown. An Excellent plan cites Simmons, Nelson and Simonsohn on analytic flexibility and Nosek and colleagues on preregistration. Credit goes to secondary outcomes chosen because Kerr showed incentives reward the wrong behavior. The rubric values plants randomized as the unit, a fixed Poisson rule with a dispersion switch and one permitted subgroup. Strong plans tabulate how each result pattern will be read, including suppressed reporting. Top marks go to blind interim handling and dated, disclosed deviations.

DBA 831 Topic 8 help before you submit

Plans lose the most ground when they leave the primary outcome open. Naming injuries as the outcome, without a denominator, period or definition, leaves room to choose among several versions once the data arrives, which is the flexibility pre-specification exists to remove. Papers that measure injuries alone miss the reporting-suppression problem the case plants, and so cannot tell a safer plant from a quieter one. Subgroup analyses listed without limit invite the search that finds something by chance. Exclusion rules left for later are decisions that will be made while looking at results. Some plans specify everything except how results will be interpreted, which reopens the door at the last step. Gelman and Loken's point about the many paths an analysis can take is misread when cited as an accusation of fraud rather than a description of ordinary practice.

Get a DBA-831 Topic 8 example written to your instructions

Send the DBA-831 Topic 8 instructions and the rubric listed in your classroom, with the study scenario or proposal your section assigned. We write a custom example to them, with the primary outcome defined precisely, secondary outcomes chosen for known distortions, exclusions and subgroups fixed and interpretation set before data, in 24 to 48 hours. The first one is free.

More DBA 831 papers and DBA samples

DBA 831 Topic 8 questions, answered

Why commit to an analysis plan before collecting data?

Because once results are visible, decisions about outcomes, exclusions and subgroups tend to drift toward whatever looks significant, often without anyone intending it. Simmons, Nelson and Simonsohn showed how combining a few such decisions can make false findings likely. Fixing them in advance means a positive result was not selected from many possible analyses, which is what gives it evidential weight.

Can a pre-specified plan ever be changed?

Yes, with disclosure. Plants close, data systems fail and definitions turn out to be unworkable. A good plan states in advance which kinds of deviation are acceptable and requires any change to be reported with its reason and timing. What it forbids is quietly revising the analysis after seeing results and presenting the revised version as if it had been planned all along.

Where can I find a free DBA 831 Topic 8 sample paper?

The analysis plan is printed here in full: question and date, outcomes, design, model and data handling, the subgroup, pre-set interpretations, who sees the data and allowed deviations. Send your study's question and setting and a plan follows at no cost.

Why track near misses and clinic visits in DBA 831 Topic 8?

Paying crews for injury-free quarters gives them a reason not to report injuries. Near misses and nurse-logged clinic visits are not rewarded, so a fall in injuries without a fall in them would point to suppressed reporting.

Why will results stay unexamined until month twelve in DBA 831 Topic 8?

Early figures invite pressure to extend a promising program, which would remove the comparison group, or to change the outcome of a weak one. Only a safety alarm, a plant's rate doubling for three months, triggers an earlier review.