Generalization is a crucial part of Applied Behavior Analysis (ABA) intervention. Individuals with autism often struggle to apply skills across different situations, which means we must carefully plan how to transfer skills from training environments to real life. When working on generalization, we need to address two key questions: how can we encourage generalization during teaching, and how will we measure whether the skill has truly carried over to relevant settings?
When preparing to teach a skill, first consider how you’ll assess its generalization. What’s your ultimate goal? For instance, when teaching direction-following, I want learners to follow instructions from all caregivers and teachers, so I assess generalization across different people. With object identification, I want learners to recognize all examples of an item, so I assess generalization using various materials. For community behavior, I evaluate whether learners can demonstrate skills in different locations and with different people—ensuring they can apply what they’ve learned in any setting.
After determining how to assess generalization, you’ll need to plan how to promote it. Stokes and Baer in 1977 outlined nine strategies for programming generalization: train loosely, train and hope, sufficient exemplars, train “to generalize,” program common stimuli, sequential modification, introduce to natural maintaining contingencies, use indiscriminable contingencies, and mediate generalization. In home programming, I most frequently use programming common stimuli.
Programming common stimuli means using a consistent trigger that prompts the desired response. For example, when teaching appropriate community behavior, you might use a behavior contract that outlines store rules. This contract guides the learner’s behavior regardless of location or accompanying person—promoting generalization across both settings and people.
To measure generalization effectively, collect data before and after teaching using the same stimuli. This provides a clear comparison point. Focus on assessing one aspect of generalization at a time—whether that’s new materials, locations, or people not used in teaching. When testing responses to questions, keep all teaching elements constant except for the person asking. This helps identify specific areas where the learner might be struggling. If you need to collect data for different conditions, do it separately and use distinct symbols for each data set.
Here is an example to tie it all together.
Before teaching appropriate community behaviors with a behavior contract, gather generalization data in locations you won’t use for teaching. After initial data collection, begin teaching the skill. Once the learner meets the criteria, collect new data in those same generalization locations. An increase in performance indicates successful generalization. If improvement is evident but below target levels, you can extend teaching to those locations—but remember to gather baseline data at any new generalization sites first.
Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10(2), 349–367. doi:10.1901/jaba.1977.10-349
