PAC 799B Module 4 Organizing and Analyzing Data Example

From Logs and Transcripts to Findings: Organizing and Analyzing the Pine Hollow Data

Student Name

Psychology and Addiction Studies Program, Aspen University

PAC 799B: Addiction Studies Capstone

Instructor Name

Month Day, Year

From Logs and Transcripts to Findings: Organizing and Analyzing the Pine Hollow Data

By the end of the six-month opt-out period, Avery had eighteen monthly release extracts from Pine Hollow, a county jail that exists only in this capstone, and transcripts of fourteen interviews: ten with released people and four with the two nurses, each interviewed twice. This paper describes how those data were organized and analyzed. The results themselves are reported in Chapter 4.

Cleaning the Release Records

The extracts listed 622 entries. Checks against the booking screen and the nurses' notes found three duplicate entries, each a release logged twice when a person was rebooked the same day, and eleven entries with the kit outcome missing. Duplicates were removed, leaving 619 eligible releases: 412 across the twelve opt-in months and 207 across the six opt-out months. For the eleven missing outcomes, the nurses' notes resolved nine; the remaining two, both night releases in the baseline year, were treated as declined, since the nurses confirmed that a missing entry at night usually meant no offer was made. The originals were kept unchanged, and each correction was recorded with its source.

A First Look: The Run Chart

Before any modeling, Avery plotted monthly uptake. Perla et al. (2011) recommend reading such a chart with rules rather than by eye, watching for signals such as a shift, where six or more consecutive months all fall above the center line or all fall below it, or a trend, where at least five months in a row each move the same way. The twelve baseline months varied between about 32 and 44 percent around a median near 38. All six opt-out months sat around 80 percent, so the six-in-a-row condition was met with room to spare. Within the baseline, no shift or trend appeared, which suggested the pre-change level was stable rather than climbing on its own.

Regression Decisions

The segmented regression planned in Chapter 3 was then fitted. Because the run chart showed no seasonal pattern and the baseline was flat, no seasonal terms were added. Residuals showed no meaningful autocorrelation, so a standard model was used. Following the implementation report, the model was run twice: once with all eligible releases and once restricted to releases whose kit was packed.

Coding the Interviews

Avery worked through the phases set out by Braun and Clarke (2006), first reading all transcripts twice and then coding every passage relevant to the research questions, producing forty-one initial codes. Codes were grouped into five candidate themes. On review against the full data set, two candidate themes, one about fear of officers and one about wanting privacy, proved to describe the same experience and were merged. A candidate theme about cost was discarded: it rested on one person's remark and did not recur.

Sensitivity Checks

Two decisions could have changed the quantitative result, so each was tested. First, the analysis was rerun with the two unresolvable missing outcomes excluded rather than coded as declined; baseline uptake changed by less than a tenth of a percentage point. Second, the model was run on packed-kit releases only, as the implementation report recommended; the estimated jump in uptake was slightly larger, as expected, since unpacked kits in months one and two could not be taken. Neither check changed the direction or rough size of the findings.

Checking the Themes

To test whether the themes fit participants' experience, Avery asked a peer mentor from the recovery organization, who had not seen the transcripts, to read the theme definitions and a sample of anonymized quotations and say whether they rang true. The mentor agreed with three themes and suggested that the fourth, about the kit being for other people, be renamed to reflect that participants often meant family members rather than friends. Avery revised the name and definition and recorded the change in the audit trail, following the trustworthiness steps planned in Chapter 3.

Setting Up Integration

Fetters et al. (2013) describe the joint display as a way to bring quantitative and qualitative results together so that they can be compared and interpreted as one. Avery's joint display has one row for each quantitative pattern, such as the gap in baseline uptake between night and day releases or the rise after the switch, a column for related interview material, a column judging whether the two strands confirm, expand or conflict and a column for the interpretation that follows.

Preparing the Baseline Comparisons

For RQ1, the baseline records were grouped by release time, release type and prior bookings, as defined in Chapter 1. Each group's uptake was calculated with a 95 percent confidence interval. Groups were compared with simple tests of proportions, without adjustment, since the purpose was description rather than prediction. Avery decided before looking at the numbers that differences smaller than five percentage points would not be discussed as meaningful, given the sample sizes.

Organizing the Qualitative Material

Transcripts were stored in a single folder with numbered files, and codes were applied in a spreadsheet with one row per coded passage, listing the participant number, whether the person was released under opt-in or opt-out, whether they took a kit, the code and the passage. This layout made it possible to see, for each theme, how many participants contributed to it and whether it appeared among people who took kits, people who declined or both, which matters for the joint display.

Analysis Decisions

Conclusion

The data are cleaned, checked and documented, the series has been read with run chart rules before modeling, the regression has been fitted under recorded decisions, the interviews have been coded into four themes and the joint display is ready. Chapter 4 will report the results.

References

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101. https://doi.org/10.1191/1478088706qp063oa

Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods designs: Principles and practices. Health Services Research, 48(6, Pt. 2), 2134-2156. https://doi.org/10.1111/1475-6773.12117

Perla, R. J., Provost, L. P., & Murray, S. K. (2011). The run chart: A simple analytical tool for learning from variation in healthcare processes. BMJ Quality & Safety, 20(1), 46-51. https://doi.org/10.1136/bmjqs.2009.037895