Research and Evidence-Based Practice Flashcards

A music therapist reads a meta-analysis reporting a standardized mean difference (SMD) of 0.42 (95% CI: 0.11–0.73) favoring music therapy for procedural pain. A colleague argues the result is clinically insignificant because the p-value was not reported. Which response best reflects evidence-based interpretation of this finding?

Answer: The confidence interval not crossing zero provides evidence of a statistically significant effect, and an SMD of 0.42 represents a moderate effect size worth clinical consideration.

A 95% confidence interval that does not include zero indicates statistical significance without requiring a separate p-value. An SMD of 0.42 falls in the moderate range (Cohen's benchmarks: small=0.2, medium=0.5, large=0.8), and in procedural pain contexts this magnitude is clinically meaningful. Dismissing findings solely because p-values are absent reflects outdated statistical thinking; modern reporting guidelines (APA, CONSORT) actually prioritize effect sizes and CIs over p-values.

A researcher uses a single-subject alternating-treatments design (ATD) to compare live music, recorded music, and a no-music control for reducing agitation in dementia. After 12 sessions, live music shows the lowest agitation scores but the data paths visually overlap between live and recorded music conditions. What is the most appropriate conclusion?

Answer: The overlapping data paths between live and recorded music suggest insufficient differentiation to conclude one is more effective than the other, though both may outperform the control.

In single-subject research, visual analysis is primary. Overlapping data paths between two active conditions in an ATD indicate the conditions are not sufficiently differentiated to claim superiority of one over the other — a core interpretive standard in applied behavior analysis and music therapy single-subject research. Both conditions may still outperform the no-music control if those paths show clear separation. Relying solely on means while ignoring visual overlap violates the interpretive conventions of this design.

A music therapist is critically appraising a randomized controlled trial investigating music therapy for depression in older adults. The study reports an attrition rate of 34% in the music therapy group versus 8% in the control group, but uses intention-to-treat (ITT) analysis. Which statement most accurately characterizes the methodological implication?

Answer: Differential attrition this large suggests potential attrition bias; ITT analysis preserves randomization integrity but cannot fully correct for the systematic differences that may exist between completers and dropouts.

ITT analysis preserves the benefits of randomization by including all participants regardless of completion, reducing the risk that completer-only analyses inflate effects. However, it cannot fully address differential attrition bias — when dropout rates differ substantially between groups (34% vs. 8%), the missing data patterns may be systematically related to outcomes (e.g., participants who worsened dropped out of the active condition). This threatens internal validity even with ITT. Appraisers should examine whether reasons for dropout were reported and whether sensitivity analyses were conducted.

A board-certified music therapist is developing a new outcome measure to assess social engagement in children with autism spectrum disorder. Pilot data show excellent internal consistency (Cronbach's α = 0.91) but a test-retest reliability coefficient of 0.54 over a 2-week interval. How should this psychometric profile be interpreted?

Answer: The high internal consistency suggests the items are cohesive, but the low test-retest reliability raises serious concerns about temporal stability, which is essential for tracking change over time.

Internal consistency (α = 0.91) reflects how well items on a single administration correlate with each other — it tells you the scale is measuring one cohesive construct. Test-retest reliability (r = 0.54) reflects score stability over time; coefficients below 0.70–0.75 are generally considered insufficient for clinical measures. A measure with high α but low test-retest reliability may contain items that are highly interrelated but that produce inconsistent scores across occasions, undermining its usefulness for tracking outcomes over an intervention. Social engagement in ASD is relatively stable over 2 weeks, so the low coefficient is a genuine psychometric problem, not an expected trait fluctuation.

A music therapist conducting a phenomenological study on the lived experience of music therapy in hospice care is asked by the IRB to provide a fixed interview protocol to ensure standardization. Which response most accurately reflects the tension between research ethics and qualitative methodology?

Answer: The IRB requirement for a fixed protocol conflicts with the emergent, co-constructed nature of phenomenological inquiry; the therapist should negotiate with the IRB to use a flexible interview guide while ensuring participant protections are maintained.

IRBs primarily focus on participant protections (informed consent, confidentiality, minimizing harm), not methodological standardization per se. Requiring a fixed protocol for a phenomenological study misunderstands the methodology — phenomenology depends on open-ended, evolving dialogue to surface the essence of lived experience. The appropriate response is to engage the IRB in dialogue, explaining that a semi-structured interview guide will be used while ethical protections are fully upheld. Abandoning the methodology or blindly complying with a scientifically inappropriate demand would both compromise either the research or participant welfare.

In a systematic review of music therapy for pain management, a funnel plot shows asymmetry with a cluster of small studies reporting large effect sizes on one side and an absence of small studies reporting null or negative effects. What is the most likely explanation for this pattern and its implication for interpreting the pooled effect size?

Answer: The asymmetry suggests possible publication bias, where small studies with non-significant or negative results were not published, leading to an inflated pooled effect size in the meta-analysis.

Funnel plot asymmetry — specifically, the presence of small studies with large effects without a corresponding cluster of small studies showing small or null effects — is a classic indicator of publication bias. In the absence of bias, studies should scatter symmetrically around the pooled effect regardless of sample size. When only positive small studies are published (while null-result small studies remain in file drawers), the pooled effect in a meta-analysis is systematically overestimated. This does not disprove the intervention's efficacy but warrants caution in interpreting the magnitude of the pooled effect. Egger's regression test can formally evaluate funnel plot asymmetry.