DDBA 8307B Week 11 Statistical Decision Brief Example

Memorandum

To: Plant Manager, Bluestem Dairy

From: Operations analyst

Date: October 13, 2026

Re: Which changeover procedure to standardize on the filling lines

Recommendation

Standardize on Procedure B for product changeovers on all three filling lines. In a 90-changeover trial over six weeks, Procedure B was the fastest on average, the most predictable from one changeover to the next and wasted the least product. It saves about six and a half minutes per changeover compared with the current Procedure A, which across roughly 1,400 changeovers a year adds up to about 150 hours of filling time.

What Was Compared

Bluestem Dairy, a composite cooperative-owned plant in Wisconsin, fills milk, cream and flavored milk into gallons, half gallons and pints on three lines. Each switch from one product to another requires a changeover: the line is stopped, flushed, sanitized where required and restarted. The plant has used Procedure A for years. Two alternatives were proposed: Procedure B, which prepares the next product's settings and flush water while the current run finishes, and Procedure C, which adds a longer flush to reduce the risk of carryover between products. Each procedure was used for 30 changeovers, assigned in rotation across lines and shifts so that no procedure got only the easy changeovers.

What the Trial Found

Procedures B and C were both clearly faster than A. B and C were close to each other on time; the trial cannot say which is faster with confidence. But B was more predictable, which matters for scheduling, and it flushed 13 fewer gallons per changeover than C, about 18,000 gallons a year. C's only advantage was a perfect sanitation record, compared with one failure for B; that difference is too small to rely on from 30 changeovers each.

Table 1

Changeover Results by Procedure

Trial of 90 changeovers, August 18 to September 26, 2026.

What This Does Not Show

The trial measured six weeks in late summer, when the product mix leans toward flavored milk. Changeovers in winter, with more cream runs, may behave differently. Crews also knew they were being timed. Shingo (1985), whose work on quick changeovers rested on separating setup work that can be done while a line runs from work that cannot, would predict B's advantage, which is reassuring, but the plant should keep timing changeovers for the first two months after standardizing to confirm the gain holds.

Next Steps

Train all three line crews on Procedure B over the next two weeks, update the standard work sheets, and review changeover times and sanitation results monthly through December.

Appendix: Statistics

A one-way analysis of variance found that changeover time differed by procedure, F(2, 87) = 8.00, p < .001, η² = .16. Tukey's post hoc comparisons showed A slower than B (difference 6.4 minutes, p < .001) and slower than C (5.1 minutes, p = .009); B and C did not differ clearly (1.3 minutes, p = .72). Levene's test did not indicate unequal variances. Field (2018) describes these procedures. The 95 percent confidence interval for the B minus A difference, about −9.8 to −3.0 minutes, is reported so the size of the saving can be judged and not just its significance, which is the practice Cumming (2014) recommends.

References

Cumming, G. (2014). The new statistics: Why and how. Psychological Science, 25(1), 7-29. https://doi.org/10.1177/0956797613504966

Field, A. (2018). Discovering statistics using IBM SPSS Statistics (5th ed.). SAGE Publications.

Shingo, S. (1985). A revolution in manufacturing: The SMED system. Productivity Press.