Why “Evidence-Based” Isn’t Just a Buzzword

Sep 23

5 min read

Communication. Collaboration. Creativity. Flexibility. Critical thinking. Despite the diversity of these skills, they all often get lumped into the category of “social skills.” What unites them is a shared understanding that these competencies are critical for students’ futures in a world of constant change. Turn-taking, reading a room, repairing a misunderstanding, and knowing when and how to ask for help are skills that let students fully participate in the classroom. Ultimately, these competencies lead to greater happiness and success in life.

While pragmatic language and social communication skills determine whether a student can access everything a school offers, direct instruction in these areas is rarely provided systematically in a manner that is built on what research shows works best for students.

Dr. Maggie Mosher, Assistant Research Professor at the University of Kansas and Co-Principal Investigator on the U.S. Department of Education-funded VOISS and iKNOW grants, has spent her career at the intersection of special education, artificial intelligence, extended reality (XR), and research-based solutions. We asked her what “evidence-based” social skills instruction means; why it matters for students, especially for those with disabilities; and how AI and XR can help deliver research-backed instruction at scale.

“Evidence-based” is a term that gets used loosely. What does it actually mean for social communication instruction?

It means the practice has been tested and found to work in a similar situation, population, or need. These practices are not just piloted or well-liked but tested and shown to produce the outcomes they claim.

In our field, that usually means randomized controlled trials, peer-reviewed publications, and replication across multiple studies and sites. A lot of what’s marketed as ‘social skills training’ is often a program, practice, activity, or idea someone thought made sense. That’s not evidence. Families and educators deserve to know the difference, especially when the stakes involve a child’s ability to make friends, navigate a job interview, and self-advocate well into their futures.

Why is this distinction especially important for students with disabilities and pragmatic language challenges?

Students who have difficulty learning using traditional methods of instruction and observational learning are exactly the students for whom social communication doesn’t develop incidentally. Many students in the general education classroom can pick up turn-taking and read facial expressions simply by being immersed in typical peer interactions and seeing which behaviors receive which kinds of feedback.

For a student with a diagnosis of autism, a behavior disability, a language-based disability, or an intellectual disability, social skills frequently have to be explicitly taught, broken into steps, and practiced repeatedly with feedback, the same way we’d teach decoding or a math algorithm. If the instruction hasn’t been validated to actually build that skill, we’re spending a limited, precious resource (instructional time) on something that may not work. For students who already have the least time to lose, that’s not a risk worth taking.

You’ve researched extended reality and AI for years. Why bring technology into something as human as conversation?

It sounds counterintuitive at first, but the goal was never to replace human interaction. The goal was to create a safe, low-stakes place to directly learn and rehearse it. Practicing a difficult conversation with a real peer or adult carries social risk: embarrassment, judgment, or the fear of getting it wrong in front of people whose opinions matter. Direct instruction in social skills by an adult is the very form of instruction students with autism and learning differences prefer least. However, technology offers a safe space that is motivating and delivers systematic instruction.

Extended reality lets a student rehearse that same scenario (advocating for themselves, resolving a disagreement, joining a group) as many times as they need, with immediate feedback and natural consequences, and we provide re-teaching with individualized feedback, so the instruction can be redesigned in real time for that specific learner. AI provides responses dynamically, instead of following a rigid script written years ago by adults. The scripts go through a RAG (Retrieval-Augmented Generation) of research-based social narratives to ensure the responses retain the effective direct instruction while also resembling the user’s vocabulary and the unpredictability of a real conversation. Used well, the technology is a bridge to more human interaction, not a substitute for it.

What has the research actually shown when this instruction is delivered through VOISS?

We’ve now tested this across two randomized controlled trials with more than 3,000 students in roughly 300 classrooms, and the pattern holds: Students show measurable gains in social skills knowledge and in applying those skills in practice, not just recognizing the ‘right answer’ on paper but displaying it in various real-life situations.

What I find most meaningful isn’t just the large effect size; it’s what teachers and related-service providers tell us. They see a student who used to sit out of group work now participating and being asked questions by fellow students. They see a student who couldn’t handle a disagreement without shutting down, responding to peers, saying, ‘I see it differently, and here’s why.’ They find a student who hated school excited to come in the morning. That’s the evidence translating into a lifelong skill.

For a teacher or IEP team building a defensible, effective program, what should they actually look for?

Ask for the research, not the testimonial. Ask whether a tool or curriculum has been evaluated with a comparison group, published in a peer-reviewed journal, and studied with a sample similar to your students with successful outcomes. Then look at implementation: Does it require a trained adult to be present for every single practice opportunity, or can a student build reps independently while a teacher’s time goes toward the highest-need moments?

Evidence-based instruction doesn’t help anyone if it’s unsustainable. Educators serve a lot of students and parents have extremely busy days. Interventions need to be effective but also feasible to implement into a student’s busy day. The instruction has to be both rigorous and realistic for the classrooms it’s meant to serve, and it must use educational techniques that we know work, such as social narratives, repeated exposure, scaffolding, direct instruction and observational learning, generalization strategies, modeling and multiple practice opportunities.

What would you tell an educator who feels they don’t have time to prioritize this right now?

I’d tell them this is not a separate subject, it’s the subject that allows for every other subject to make sense. It’s the access skill underneath every other skill. A student may be extremely bright, but if they can’t communicate that knowledge or effectively accept feedback, they will not get or keep a job. Every group project, every classroom discussion, every conflict at recess is an opportunity a student either can or can’t fully use, depending on their social communication abilities. Why would we waste these opportunities or allow them to engage in them incorrectly? That would cause students to develop habits, which then need more instruction and time to break, which is actually more difficult and time-consuming than teaching it right the first time. Investing in these skills isn’t a detour from academic goals. For many of our students, it’s the shortest path to reaching their goals.

Dr. Maggie Mosher

Assistant Research Professor at the University of Kansas, course creator for Johns Hopkins University, and Chief AI Consultant, where her work focuses on artificial intelligence, extended reality, Universal Design for Learning, and evidence-based practices in special education. She served as Co-Principal Investigator on the U.S. Department of Education grant that developed the extended reality social skills scenarios behind VOISS.