Journal of evidence-based social work (2019)最新文献

Purpose: Community-based organizations (CBOs) are increasingly providing violence prevention programming and services. However, researchers have found that many of these organizations struggle to implement evidence-based practices or face administrative burdens of managing the grants that fund the organization. The current study sought to further explore facilitators and barriers of CBOs receiving state funding.

Materials and methods: Participants included 29 individuals who led grant writing or implementation efforts at a CBO receiving at least one violence prevention grant from a large state agency. Online focus groups were held, and reflexive thematic analysis was conducted.

Results: We divided the major themes into two categories: facilitators and barriers. Facilitators included: 1) these grants were a financial lifeline for organizations providing violence prevention services, helping expand organizations' capacity; and 2) grantees expressed satisfaction with the funder when there was clear and regular communication. Barriers included: 1) rigid application, reporting and budgeting requirements; 2) delays in reimbursement that hindered grants administration; and 3) an under-resourced state agency impacting the overall grant process. Grantees discussed the importance of and recommended fostering community and collaboration among violence prevention organizations.

Discussion: These results have implications for funders to consider when administering funds, supporting organizations, building community, and reporting requirements for violence prevention efforts. Recommendations for funders are provided, including streamlining budget requirements, providing training and technical assistance, and creating more opportunities for collaboration.

Conclusion: As more community-based organizations are funded to implement violence prevention services, funders need to review procedures to ensure equity throughout grant applications and implementation.

Purpose: This is a scoping review of low-intensity, home-based interventions promoting learning for 11-17-year-olds in foster care. Research shows that children in out-of-home care fall behind in school compared to their peers. Interventions to improve learning outcomes for these children typically i) target young children or children aging out of care, and ii) require many resources to implement. Low-intensity, home-based interventions for foster children have shown promise.

Materials and methods: We used scoping review methodology to comb through the research literature in four languages across 15 databases.

Results: From the resulting 1355 studies, we identified nine studies comprising seven interventions meeting our inclusion criteria, which we categorized into three groups: interventions with tutors as delivery agents (n = 3), interventions with foster carers as delivery agents (n = 3), and interventions based on distributed materials (without delivery agents) (n = 1). We describe the interventions in each category, their key similarities and their outcomes.

Discussion: Interventions vary in delivery agents and session structure, with tutor-led programs often aiming to improve math and reading skills. While such interventions may have limited efficacy compared to high-intensity alternatives, they offer advantages in feasibility, reach and sustainability, potentially yielding broader population-level impact. Geographic differences suggest contextual factors influence delivery models. Matching interventions to children with fewer support needs and ensuring fidelity is key to meaningful outcomes.

Conclusion: There is a need for more well-evaluated low-intensity, home-based interventions that address the unique contextual factors of out-of-home care.

Purpose: BIPOC and LGBTQ+ youth experience disproportionately worse mental health outcomes, partially due to a lack of culturally responsive practices and multicultural counseling self-efficacy among behavioral health professionals. Separately, professional quality of life (ProQOL) has also been linked to better client outcomes. However, little is known about the relationship between behavioral health professionals' multicultural counseling self-efficacy, culturally competent behaviors, and ProQOL. This study aimed to examine the relationship between cultural competence and ProQOL, as well as a hypothesized indirect role of multi-cultural counseling self-efficacy among these associations.

Materials and methods: Data were collected from a workforce training program among graduate level social work, psychology, and nursing students on culturally responsive practices whose practicum settings included working with BIPOC/LGBTQ+ youth. At baseline (T1), graduation (T2), and 9-months post-graduation (T3), participants (N = 113) completed measures on anti-racism, self-reflection and insight, intercultural communication, multicultural counseling self-efficacy, and ProQOL (compassion satisfaction, burnout, secondary trauma).

Results: Intercultural communication at T2 had a significant indirect effect on the relationship between multicultural counseling self-efficacy at T2 and burnout at T3. Self-reflection and insight at T2 also had a significant indirect effect on the relationship between multicultural counseling self-efficacy at T2 and burnout at T3. Mediation models examining secondary traumatic stress and compassion satisfaction were not statistically significant.

Discussion: Results suggest the importance of cultural competencies and multicultural counseling self-efficacy in facilitating culturally responsive practices and promoting clinicians' ProQOL.

Conclusion: Future trainings should emphasize these areas to minimize mental health disparities among LGBTQ+ and BIPOC youth.

Purpose: This study develops a systematic benchmarking framework for testing whether language models can accurately identify constructs of interest in child welfare records. The objective is to assess how different model sizes and architectures perform on four validated benchmarks for classifying critical risk factors among child welfare-involved families: domestic violence, firearms, substance-related problems generally, and opioids specifically.

Materials and methods: We constructed four benchmarks for identifying risk factors in child welfare investigation summaries: domestic violence, substance-related problems, firearms, and opioids (n = 500 each). We evaluated seven model sizes (0.6B-32B parameters) in standard and extended reasoning modes, plus a mixture-of-experts variant. Cohen's kappa measured agreement with gold standard classifications established by human experts.

Results: The benchmarking revealed a critical finding: bigger models are not better. A small 4B parameter model with extended reasoning proved the most effective, outperforming models up to eight times larger. It consistently achieved "substantial" to "almost perfect" agreement across all four distinct benchmark categories. This model achieved "almost perfect" agreement (κ = 0.93-0.96) on three benchmarks (substance-related problems, firearms, and opioids) and "substantial" agreement (κ = 0.74) on the most complex task (domestic violence). Small models with extended reasoning rivaled the largest models while being considerably more resource-efficient.

Conclusion: Small reasoning-enabled models achieve accuracy levels historically requiring substantially larger architectures, enabling significant time and computational efficiencies. The benchmarking framework provides a method for evidence-based model selection to balance high accuracy with practical resource constraints before operational deployment in social work research.

Purpose: The concept of evidence-based practice (EBP) has been under academic discussion for its suitability to social work, but the perspectives of social workers have received relatively little attention. The purpose of this study is to provide updated knowledge of social workers' perspectives on EBP, facilitators and barriers to EBP, and possible differences between social work contexts.

Materials and methods: We conducted a scoping review using the JBI methodology and identified 54 relevant peer-reviewed studies published after 2013.

Results: Perceptions and attitudes toward evidence-based practice were positive. Still, the conception appeared unclear and was confused with other related concepts The main barrier to implementing evidence-based practice was lack of time, and the main facilitators were staff skills and leadership. Only a few studies reported differences between social work contexts in terms of perceptions, attitudes, experiences, facilitators, or barriers related to EBP.

Discussion: Compared to previous research, there has been no change in social workers' understanding of EBP. At times, the discussion related to EBP within the discipline has been critical. However, according to our results, social workers did not share this opinion. The use of research evidence as part of evidence-based practice remains limited in social work, and social workers feel that research does not address their needs.

Conclusion: Because social workers perceive that research evidence does not meet their needs, it is essential to consider how research dissemination can be improved to increase the perceived usefulness of research. Adopting a more pragmatic research approach could help legitimize academic social work and strengthen the professional status of social work.

Purpose: This theoretical analysis aims to develop a comprehensive rights-based framework for navigating artificial intelligence integration in social work practice while addressing the ethical implications of AI deployment across micro, meso, and macro practice levels.

Materials and methods: The study synthesized interdisciplinary research drawing on social psychology, group dynamics theory, and institutional analysis. The conceptual framework integrated the I-C-E (Ingroup Identification, Cohesion, Entitativity) model with socioecological systems theory. Analysis was conducted on existing literature and documented case examples to examine how AI systems mediate interpersonal relationships and construct meaning in social work contexts.

Results: The analysis demonstrated that AI systems profoundly impact vulnerable populations by mediating interpersonal relationships and constructing meaning in AI-mediated environments. The developed framework successfully bridged social work theory with interdisciplinary insights to provide evidence-based guidance for AI implementation in social services.

Discussion: The proposed framework offers concrete strategies for social work education and provides research methodologies that center community voices. The analysis reveals how AI integration can be guided by evidence-based practice while maintaining focus on vulnerable population needs and democratic governance principles in social services.

Conclusion: This work provides evidence-based guidance for practitioners to harness AI's potential while safeguarding social work's core values of human dignity, self-determination, and social justice. The framework includes policy recommendations for democratic governance of AI in social services and establishes a foundation for ethical AI deployment across all levels of social work practice.

Purpose: This study investigates the inter-rater reliability between human experts (a forensic psychologist and a social worker) and a large language model (LLM) in the assessment of child sexual abuse statements. The research aims to explore the potential, limitations, and consistency of this class of AI as an evaluation tool within the framework of Criteria-Based Content Analysis (CBCA), a widely used method for assessing statement credibility.

Materials and methods: Sixty-five anonymized transcripts of forensic interviews with child sexual abuse victims (N = 65) were independently evaluated by three raters: a forensic psychologist, a social worker, and a large language model (ChatGPT, GPT-4o Plus). Each statement was coded using the 19-item CBCA framework. Inter-rater reliability was analyzed using Intraclass Correlation Coefficient (ICC), Cohen's Kappa (κ), and other agreement statistics to compare the judgments between the human-human pairing and the human-AI pairings.

Results: A high degree of inter-rater reliability was found between the two human experts, with the majority of criteria showing "good" to "excellent" agreement (15 of 19 criteria with ICC > .75). In stark contrast, a dramatic and significant decrease in reliability was observed when the AI model's evaluations were compared with those of the human experts. The AI demonstrated systematic disagreement on criteria requiring nuanced, contextual judgment, with reliability coefficients frequently falling into "poor" or negative ranges (e.g. ICC = -.106 for "Logical structure"), indicating its evaluation logic fundamentally differs from expert reasoning.

Discussion: The findings reveal a profound gap between the nuanced, contextual reasoning of human experts and the pattern-recognition capabilities of the LLM tested. The study concludes that this type of AI, in its current, prompt-engineered form, cannot reliably replicate expert judgment in the complex task of credibility assessment. While not a viable autonomous evaluator, it may hold potential as a "cognitive assistant" to support expert workflows. The assessment of child testimony credibility remains a task that deeply requires professional judgment and appears far beyond the current capabilities of such generative AI models.

Purpose: This research explores clinical social workers' perceptions of the usefulness of generative artificial intelligence (AI) in clinical practice, with a particular focus on large language models (LLMs).

Materials and methods: This qualitative reflexive thematic analysis explored the interviews of 21 clinical social workers and how they experience their work in the context of growing LLM use. Participants shared their perceptions and experiences with LLMs following a collaborative case consultation exercise using ChatGPT and a video demonstration of a client using ChatGPT.

Results: Social work practitioners described both benefits and concerns with LLM use in their practice. Two overarching themes emerged: (1) factors that enhanced social workers' perceived usefulness of LLMs in clinical practice, including support for administrative tasks and client engagement, and (2) factors that diminished perceived usefulness, such as concerns about confidentiality, loss of nuance, and limitations in conveying empathy and contextual understanding.

Discussion: Practitioners shared that they are using LLMs as idea generators in clinical work, while simultaneously expressing concern about the quality of information and the need for a human‑centered approach. They also noted that their decision to adopt LLMs is shaped by professional ethics and relational values, reflecting a preference for augmentation rather than full automation to preserve therapeutic depth and client wellbeing.

Conclusion: Future AI implementation should focus on practitioner training and clear ethical guidelines to support responsible integration of LLMs. Ongoing evaluation will be essential to ensure these tools enhance clinical practice without compromising the therapeutic relationship or core social work values.