GB891 Unit 7 Collection Log Example

A Dated Row for Everything: A Unit 7 Collection Log for a Twelve-Week Bakery Safety Study

[Student Name]

Purdue University Global

GB891: Doctoral Project II: Project Design and Implementation

Unit 7 Collection Log

[Instructor Name]

[Date]

Corliss Baking Company, its plants and every log entry are composite and illustrative; the methods sources cited are real.

Why a Log, Not a Folder of Files

A study that collects data from twenty-four crews at four plants every week produces hundreds of separate pieces of evidence. Without a log, the only record of what was collected is the data itself, which cannot show what was attempted and failed, what was collected late or what was collected under unusual conditions. A log records each event at the time it happens, so that later analysis can account for conditions the data do not show. Shadish et al. (2002) stress that threats to validity, such as events occurring during a study, can be assessed only if they are recorded when they occur.

The Columns

Each row records one data collection event, with twelve columns.

Date and time. When the event occurred. Plant and crew code. Where, using crew codes rather than names. Event type. Pulse draw, feedback sheet sent, observation session, survey administration or interview. Collector. The research assistant, observer one or observer two. Planned. Whether the event was scheduled in the protocol. Completed. Yes, partly or no. Count. Answers collected, intervals observed or surveys returned. Language. For pulse draws and surveys, the languages used. Crew size that day. From the shift roster. New members this week. Associates joining the crew since the previous week. Conditions. Anything unusual, such as a line down, a merged crew or a supervisor absent. Deviation link. A reference number if the event departed from the protocol, linking to the deviation report.

Representative Entries: Week Two

In week two, all 24 crews had pulse draws completed on Wednesday or Thursday. Twenty-two crews provided four answers each; two crews at Calder provided three each because the night shift was short. Feedback sheets for all twelve intervention supervisors were sent by 9:20 a.m. on Monday. Observers completed forty observation sessions across the four plants, two per station per plant, and four joint sessions for agreement checks. One condition was recorded: the Ashford packaging line was down for two hours on Tuesday, so its jam-clearing observation was moved to Wednesday.

Representative Entries: Week Six

Week six recorded the first serious departure. On Monday, the night supervisor of one Brierly crew resigned without notice. The crew was covered by a line lead for the rest of the week. The feedback sheet for that crew, already sent, was received by no one in the supervisory role, and the log records it as sent but not received. Pulse draws for the crew continued, since the question concerned the crew's experience of supervision, and the line lead's name was noted in the conditions column without being used in any analysis. A deviation reference links to the full account in the Unit 9 report. Elsewhere that week, collection was complete, and the crew size column showed that six of the twelve intervention crews had at least one new member.

Representative Entries: Week Nine

Week nine recorded a comparison plant event. The new Dunmore manager, who had taken over in week eight, held a one-hour safety stand-down for all Dunmore crews on Thursday, a plant-wide meeting on hand safety prompted by a near miss at a slicer. The weekly check for new safety initiatives, added after the Unit 6 seminar, caught it. The log records the date, the plants affected and the content, and links to the deviation report, because a safety event at a comparison plant could narrow the difference between intervention and comparison crews.

Without the weekly check for new initiatives, the Dunmore stand-down would have looked in the data like nothing at all.

Totals the Log Supports

At the end of twelve weeks, the log supports a set of totals that the implementation chapter reports. Pulse draws were planned at 24 crews for 12 weeks, 288 crew-weeks, with a target of four answers each, 1,152 answers. Feedback sheets were planned for twelve supervisors for twelve weeks, 144 sheets. Observation sessions were planned at two per station per plant per week. Each total can be compared with what was completed, and each shortfall traced to its rows.

Observer Agreement in the Log

Every fourth observation session was conducted jointly, and the log records both observers' interval codes for those sessions. Agreement is calculated weekly by station using the kappa statistic Cohen (1960) introduced. The log flags any station-week with kappa below 0.60, the lower bound of moderate agreement, so that definitions or training can be revisited before disagreement accumulates.

Later Uses of the Log

The log serves three later purposes. It supports the fidelity measures in Unit 8, which follow the elements Carroll et al. (2007) describe, since sheets sent and received, crews reached and sessions completed are all recorded. It supplies the evidence for each departure in the Unit 9 report. And it allows the analysis to account for conditions, such as weeks with a supervisor absent or a stand-down at a comparison plant, that the outcome data alone would not reveal.

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

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1), 37-46. https://doi.org/10.1177/001316446002000104

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