The Lecture Circuit
Monday, July 7, 2008
Accountability: Types of Data
![My plan for this entry was to give an overview of data collection and analysis methods for accountability in speech-language treatment. We want to show that our clients are responding to our treatment—rather than simply maturing or reacting to other environmental influences (like a wonderful preschool or a sudden change in home environment). My goal was to review basic clinical methods and, in keeping with the theme of “Tower to Trenches,” show that these methods were useable in real clinical practice.
To say I bit off more than I could chew is an understatement! I have found myself writing and writing, with the resulting entry being too unwieldy. I have therefore decided to break this topic into sections (no, I haven’t given up!)
We need to consider the following:
1) What types of data you can collect
2) When and how you can collect data
3) What you do with data once you have it
4) How you can make decisions for specific clients about how much/how little data you need
In this first entry, I will start with types of data.
Types of Data
The first broad distinction to make is between quantitative data and qualitative data. Quantitative data come from counting behaviors in a specified measurement. These are what most people think about when they think about data: number of correct productions of a phoneme, number of complex sentences in a sample, number of spoken words comprehended by a listener. Qualitative data come from observations and may include descriptions and impressions; for example, “child was eager to explore toys but vocalized little when playing.” Ethnographic methodology provides a way to understand and collect qualitative data. This type of data can be extremely helpful in understanding our clients, especially when wanting to understand their communication in varied contexts, but it is not the focus of this web entry. [Interested in qualitative methods? Check out an article (Stone-Goldman & Olswang, 2003) about ethnographic methodology in the 2003 ASHA Leader, “Learning to Look, Learning to See” ; other suggested readings are included in that article.]
Focusing in on quantitative data for the purpose of accountability, we can make things simple by considering two big subcategories:
• Behaviors you are treating (things you are working on directly in treatment)
• Behaviors you are not treating (everything else)
If you do nothing more than keep this distinction in mind, you will be on your way to collecting data in a useful way.
Behaviors you are treating. When you provide therapy, you are always treating something (the treatment targets) and always not treating (not directly working on) other things. Even for a client with a very limited problem—such as errors on the /r/ phoneme—there are multiple forms (all those different types of /r/!), so some forms or some word positions will be treated before others. When you treat a behavior, you practice with the client, so that the materials become familiar and the client’s responses are cued and rehearsed. Treated behaviors and the corresponding materials are sometimes thus called practiced.
Behaviors you are not treating. Of the behaviors you are not treating, some will be closer to what you are treating by virtue of sharing linguistic features or functions, and some will be more distant.
• Closer; Related: For example, phoneme cognates pairs (/p,b/, /s,z/) and allomorphs (plural [s,z,ez] are forms that are closely related. Subject pronouns (I, he, she) are closely related; different two-word semantic relations (agent-action; action-object) are closely related. When forms are related, you have reason to expect generalization within the group. Forms that are related are said to be in the same response class.
• More Distant; Less Related: Phonemes with very different distinctive features are typically regarded as less related, for example, /s/ vs. /r/. A pragmatic function (e.g., requesting) is considered less related to phonemes or syntactic forms than to other pragmatic functions. I use the term “less related” rather than the more typical “unrelated” because we are talking about a speech and language system that on some level is integrated. Articulation does interact with and relate to language; it doesn’t help us to pretend that these behaviors are isolated from one another. But at the same time we can recognize that when behaviors are less related, we have reason to expect that generalization will not occur across the more distant behaviors.
How do we know if targets are truly related or are less related? Linguistic theory gives us some information, and research studies that report generalization patterns give us confirmation. For example, Hegde (reported in Hegde, 2002) did a study to show that is auxiliary and is copula responded similarly to treatment (treating one brought changes in the other, and vice versa), thus confirming that they were related, i.e., in the same response class. (His study was with an older individual with developmental delays, and the study would not likely pass current human subjects guidelines, but happily, there are more acceptable methods available to us.)
What to Do With the Data (Peek Ahead)
I will have a separate entry to discuss integrating the data to determine accountability, but I’ll foreshadow that section with this brief comment now:
• We compare changes on treated behaviors to changes in untreated behaviors. This should seem logical—we should see more change on what we work on than what we don’t work on.
• Further, if we see changes in untreated behavior, those changes should be greater for the behaviors closer to the target (related) than farther away (less related). In other words, generalization patterns should make sense.
We have come to the end of the section on types of data. Next up: when and how you can collect data. As I conclude this section, I am reminded of something I know well but seem to forget until real life gives me another learning opportunity--any task is easier when it is broken down into manageable pieces. We’ll see if I maintain this re-learning in my up-coming entries.
References
Hegde, M.N. (2002). Treatment procedures in speech-language pathology (3rd ed). Austin, TX: Pro-Ed.
Stone-Goldman, J., & Olswang, L. (2003). Learning to look, learning to see: Using ethnography to develop cultural sensitivity. Asha Leader 8(5), pp. 6-7, 14-15.](38D52741-913C-48DA-9D51-DFCF7764C1FF_files/shapeimage_2.png)
