| Course | ACC 337 Introduction to Accounting Analytics |
|---|---|
| Topic | Topic 8 |
| Paper type | Reproducible analysis workpaper |
| Length | About 990 words, 2 pages |
| Format | APA 7 student paper |
| School | Grand Canyon University (GCU) |
| Program | BS in Accounting |
| Updated | September 2026 |
Free sample paper for ACC 337 Topic 8
From 31,400 Raw Claims to 29,788 That Count: A Reproducible Warranty Analysis Workpaper for Arbor Home Appliances
[Author Name]
Colangelo College of Business, Grand Canyon University
ACC-337: Introduction to Accounting Analytics
Topic 8 Assignment
[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.
From 31,400 Raw Claims to 29,788 That Count: A Reproducible Warranty Analysis Workpaper for Arbor Home Appliances
Purpose
This workpaper supports Arbor Home Appliances' analysis of warranty claim cost by product line for the calendar year, prepared for the controller's warranty reserve review. Arbor is a composite appliance maker. The workpaper is written so that a colleague with access to the same systems can reproduce every figure without asking the preparer anything. It starts where the analysis started, with the raw extract, and records every change as a numbered step with its count and dollar effect. All figures are illustrative.
Source
The extract was pulled from the service system's claims table on January 8, using the query in Appendix B with the parameters shown there. It contains 31,400 claims with a total claim cost of $4,210,000. The extract file is saved read-only in the analysis folder with the date in its name, and it is never edited; every step below works on a copy.
Preparation Steps
Table 1 records each step with its count and dollar effect, restating the running totals after each one.
Table 1
Preparation Steps and Running Totals
| Step | Action | Claims removed | Dollars removed | Claims remaining | Dollars remaining |
|---|---|---|---|---|---|
| 0 | Raw extract from the service system, pulled January 8 for the calendar year | 31,400 | $4,210,000 | ||
| 1 | Remove test claims entered by the training team, identified by the TRN prefix | 380 | 52,000 | 31,020 | 4,158,000 |
| 2 | Exclude open claims: a scope decision, since their cost is not yet final | 1,020 | 168,000 | 30,000 | 3,990,000 |
| 3 | Remove claims dated after December 31 that the extract picked up | 212 | 31,000 | 29,788 | $3,959,000 |
| 4 | Map 14 retired model codes to current codes using the crosswalk in Appendix A | 0 | 0 | 29,788 | $3,959,000 |
Notes on Each Step
Step 1. The training team enters test claims in the live system with claim numbers beginning TRN. They are not real claims. All 380 were confirmed as test entries with the training coordinator.
Step 2. Open claims, still awaiting parts or technician visits, are excluded because their cost is not yet final. This is a scope decision, not a correction: the warranty reserve review estimates open claims separately, and including them here would count them twice. The decision is recorded so a later reader does not mistake it for missing data.
Step 3. The query's date filter used the claim's last-updated date rather than its claim date, so 212 claims opened in the first week of January appeared in the extract. They were removed by filtering on claim date.
Step 4. Fourteen older model codes were retired during the year and replaced with new codes for the same products. The crosswalk in Appendix A maps each retired code to its replacement so that a model's claims are not split across two codes. The mapping changes no counts or dollars.
How the Data Is Organized
The prepared file keeps one row per claim and one column per field: claim number, claim date, status, model code, product line, part cost, labor cost and total claim cost. The crosswalk is a separate two-column table, retired code and current code, joined to the claims by model code rather than typed into the claims file. Keeping each kind of information in its own table, with one observation per row, makes the steps easier to repeat and the results easier to check (Wickham, 2014). It also means that if another model code is retired next year, one row is added to the crosswalk and the analysis runs unchanged.
Intermediate files are saved after each step with the step number in the file name. A reviewer can open the file after step 2, count its rows and confirm the 30,000 shown in Table 1 without rerunning anything, which makes the review faster and localizes any disagreement to a single step.
Query and Parameters
Appendix B contains the full query. Its parameters are stated explicitly: claim date from January 1 through December 31, claim status closed, exclude claim numbers beginning TRN. Every parameter that affects the result is written in the query rather than set in the tool's defaults, because defaults change between users and versions. Keeping the exact query, its parameters and the order of steps with the results is the core of making computational work reproducible (Sandve et al., 2013).
Output
Table 2 gives the output the analysis should reproduce. A rerun that does not produce these figures has differed somewhere in Table 1.
Table 2
Warranty Claims by Product Line, Closed Claims, Calendar Year
| Product line | Claims | Claim cost | Cost per claim |
|---|---|---|---|
| Refrigerators | 11,240 | $1,694,000 | $151 |
| Dishwashers | 8,915 | 1,030,000 | 116 |
| Ranges | 6,103 | 842,000 | 138 |
| Laundry | 3,530 | 393,000 | 111 |
| Total | 29,788 | $3,959,000 | $133 |
The Colleague's Rerun
A second analyst reran the workpaper from the raw extract. The first rerun came out 212 claims and $31,000 high: 30,000 claims and $3,990,000. The difference matched step 3 exactly. The colleague had used the original query, whose date filter on last-updated date had caused the problem, and had not seen that step 3 existed to correct it. Two changes followed. The query in Appendix B was corrected to filter on claim date, so step 3 now removes nothing when the corrected query is used, and the note to step 3 explains both versions. The second rerun matched Table 2 to the claim and the dollar.
The rerun is recorded here, including its first failure, because it is the evidence that the workpaper is reproducible. Audit data analytics guidance calls for documenting the data, the procedures and the results so that the work can be reviewed (American Institute of Certified Public Accountants, 2017). A workpaper that has never been rerun by someone else has only been asserted to be reproducible.
Agreement to the Ledger
Total claim cost for closed claims, $3,959,000, agrees with the warranty expense charged against the reserve for claims closed during the year in the general ledger. That agreement shows that the prepared data is complete for its stated scope, and it is the figure the controller will use in the reserve review.
References
American Institute of Certified Public Accountants. (2017). Guide to audit data analytics. AICPA.
Sandve, G. K., Nekrutenko, A., Taylor, J., & Hovig, E. (2013). Ten simple rules for reproducible computational research. PLoS Computational Biology, 9(10), Article e1003285. https://doi.org/10.1371/journal.pcbi.1003285
Wickham, H. (2014). Tidy data. Journal of Statistical Software, 59(10). https://doi.org/10.18637/jss.v059.i10
Reading the GCU ACC 337 Topic 8 instructions
The Topic 8 instructions in ACC-337 ask students to document an analysis so that someone else can reproduce it. Weak submissions show results without the path to them. The assignment rewards workpapers that start from the raw extract with its date and totals, record each change as a numbered step with counts and dollar effects, distinguish scope decisions from corrections, and include the query with every parameter stated. Students should organize the data so steps are easy to repeat, give the output a reviewer should reproduce, record an independent rerun including any failure, and agree the result to the ledger.
Inside the ACC 337 Topic 8 example
The finished workpaper begins where the analysis began, with the raw extract: 31,400 claims totaling $4,210,000, pulled from the service system on a stated date. Every change after that is a numbered step with its count and dollar effect. Test claims entered by the training team come out, 380 of them and $52,000. Open claims are excluded as a scope decision, 1,020 and $168,000, since their cost is not yet final. Claims dated after the period end are removed, 212 and $31,000, leaving 29,788 claims and $3,959,000. A crosswalk maps fourteen retired model codes to their replacements. The query is included with its parameter values, and the output table states the figures it should reproduce. A final section records the colleague's rerun, which first came out 212 claims high.
Where the marks sit in the ACC 337 Topic 8 rubric
Faculty grading the reproducible workpaper reward a path someone else can follow. Excellent workpapers keep the raw extract untouched, restate running totals after each step and label exclusions such as open claims as scope decisions. Credit goes to stating query parameters explicitly rather than relying on tool defaults. The rubric values an output table with figures to match and intermediate files a reviewer can check. The highest level records a colleague's rerun, including a first failure and how it was fixed. Workpapers that have never been rerun by anyone else usually land in a middle band.
Common ACC 337 Topic 8 mistakes, and how to avoid them
An analysis whose documentation stops at the final chart is the characteristic loss on this assignment, because nobody can rebuild the figures from it. Workpapers that describe preparation in general terms, such as removing bad records, give a colleague nothing precise enough to repeat. Mixing scope decisions with error corrections hides the judgment calls a reviewer most needs to see, and excluding open claims is exactly such a call. Omitting the extract date leaves a rerun on newer data looking like a failure when it is not. Documents that never state expected outputs cannot be tested, only read. The most telling gap is a filter applied by hand in a spreadsheet and never written down, which is why an actual rerun by someone else is worth more than any amount of careful description.
Get an ACC-337 Topic 8 example written to your instructions
Send the ACC-337 Topic 8 instructions, your classroom rubric and a description of the analysis your section is documenting. You receive a custom example written to those criteria, with the source identified, every step numbered with counts and dollars, scope choices separated from corrections, target outputs stated and a reproduction test recorded, in 24 to 48 hours. The first one is free of charge.
More ACC 337 papers and BS in Accounting samples
- ACC 337 Topic 1: Business Question Reframing Memo
- ACC 337 Topic 2: Data Preparation Log
- ACC 337 Topic 3: Dataset Question Triage
- ACC 337 Topic 4: Rule-based Anomaly Screen
- ACC 337 Topic 5: Pattern Versus Evidence Dq Post
- ACC 337 Topic 6: Chart Choice Rationale
- ACC 337 Topic 7: Management Findings Brief
- ACC 335 Topic 8: Compensating Control Plan
- ACC 260 Topic 8: Keep or Drop Recommendation
- ACC 250 Topic 8: Annual Report Evaluation
ACC 337 Topic 8 questions, answered
Who should perform the reproduction test?
Somebody other than the writer, working only from the workpaper and the raw file. The author knows every unrecorded step and will repeat them without noticing, which is precisely what the test is meant to catch. In a course setting a classmate or a study partner can do it, and where the assignment is individual, a rerun a few days later from the document alone is a reasonable substitute.
Does the query itself need to be included?
Yes, with every parameter value exactly as run, because a description of a query and the query itself can quietly differ. Note the tool and its version as well, since functions sometimes behave differently between releases. A reader who does not use the same tool still benefits, because the logic of the filters and joins can be followed and rebuilt elsewhere.
What if the rerun matches on the first attempt?
Record that it matched and what was compared, the counts and totals at each stage as well as the final output. A match on the final figures alone could conceal two errors that offset, so stage-by-stage agreement is the stronger evidence. The workpaper should also say who ran the test and when, since a claim of reproducibility with no test behind it is only a description.
Where can I find a free ACC 337 Topic 8 sample paper?
The full workpaper is above, with the purpose, source, preparation steps and notes, how the data is organized, the query and parameters, the output, the colleague's rerun and agreement to the ledger. Share your analysis for a free worked workpaper.
Why did the rerun come out 212 claims high in ACC 337 Topic 8?
The colleague used the original query, which filtered on last-updated date and picked up January claims, and had not applied the step that removed them.