3076 — Qualitative and Quantitative Techniques for Mixed-Methods Studies that Promote More Effective Intervention Development

Pope CA (Ralph H. Johnson VA Medical Center, Charleston, SC), Davis BH (University of North Carolina at Charlotte), Jenkins CH (Medical University of South Carolina College of Nursing), Magwood GS (Medical University of South Carolina College of Nursing), Mason PR (Linguistic Insights, Inc.)

Objectives:
Although diabetes self-management (DSM) within the VA system improves patient outcomes, African Americans continue to experience poorer glycemic control and less DSM. Increasingly, qualitative studies investigate barriers and facilitators of DSM for African Americans, but often lack a credible means to link components of human behavior to interventions. Objectives: 1. Contrast how differing qualitative coding approaches affect information and observer biases. 2. Demonstrate how computer-assisted quantitative coding of changes in speaker stances in interviews complements qualitative observations and targets areas for intervention. 3. Propose a multidisciplinary mixed methods approach that confirms sources of affect, agency, evaluation, intensity, and certainty vital to culturally tailored interventions.

Methods:
A qualitative analysis of semi-structured interviews with 20 African-American patients with diabetes explored reasons for avoidable emergency room visits. Researchers presented differences in findings from qualitative coding using grounded theory and situational analysis. As a form of triangulation, the interdisciplinary team applied Stance Shift Analysis, a quantitative computer-supported coding method. A demonstration will show how this method identifies and measures 24 language-category variables per 100-word standardized sections in transcripts, using factor and cluster analysis, scaling each segment into four established scales. Segments which are one or more SD above the mean mark significant areas where the speaker signals confidence or doubt, indicates strong affect, offers opinion or rationales, elaborates with crucial details, personalizes choices, or exhibits agentive behavior.

Results:
Stance Shift Analysis identified dominant themes found in the qualitative approaches and additional themes which clarified otherwise puzzling findings. Keyword analysis of semantic domains highlighted cultural understandings of patients’ concepts of disease and self-management. Examples will show how Stance Shift Analysis combines techniques from content analysis and corpus analysis to measure how word use patterns signal speakers’ opinions, rationales, details and feelings, and personalization or projection of agency which speaks to possible actions.

Implications:
Interdisciplinary collaboration with innovative analytic techniques can enhance understanding of cross-cultural communication and strengthen interventions in VA health disparities research.

Impacts:
Greater confidence in qualitative methodology by using complementary quantitative techniques for coding data can increase researcher applications and support mixed methods studies that produce more effective interventions.