Glossary

The ability to generalize the results of your study or clinical findings to other contexts (e.g., outside of therapy), behaviors, times, or clients (i.e., with the same or different problems).

Extraneous variability refers to fluctuations in a client’s performance that are due to influences external to the therapy situation (i.e., context) and not due to treatment. Internal validity threats (e.g., history, maturation, instrumentation, testing, mortality, regression) are likely causes of external variability.

The ability to extend the results of a specific intervention beyond the treatment setting to non-clinical daily living contexts, to apply to other behaviors or other individuals.

These data are collected outside the treatment context, during functional communication activities in applied and/or naturalistic settings. They address the issue of the application of skills acquired during therapy to real communication contexts and directly address to instrumental outcomes and ultimate outcomes.

Removing one or more data points from visual and quantitative analysis while maintaining the data in the client’s record. In Therapy-Science, this is accomplished by enclosing each data point with parentheses. A grayed-out version  of the datapoint will appear on the graph but will not be connected to the other data points in the series.

Related terms: ghost, ghosted data

The evaluation of an individual’s general level of functioning (i.e., psychological, physical, communicative, social, occupational domains).

A desired result or possible outcome that a person or a system envisions, plans and commits to achieve . In Therapy-Science, goals are restricted to individual phases and specified by criteria statements. 

A composite measure of a client’s progress by organizing a variety of performance evaluation and behavior tracking measures into a coherent description of a client’s improvement over time.

Related term(s): GAS

Defines the behavioral phenomena that you plan to target for intervention, as well as, the procedures, criteria, and other components that comprise the intervention process needed to evaluate clinically relevant behavioral change. In a well-built goal statement, the outcome is operationally defined and evaluated by an explicit criterion, thereby making it testable and falsifiable like a research hypothesis.

A dynamic, interactive, visual analysis tool that allows interaction, data exploration, and data analysis to provide both qualitative and quantitative analysis of the data. Similar to a data graph.

The scale or divisibility of a behavior into smaller and smaller parts.

A diagram or chart displaying the relationship between variables. A single-case graph shows the relationship between the evaluation of one or more treatments over time.

The timeline displayed across the bottom of a single-case graph. It is usually marked by sessions, days or weeks. Note that sessions may not represent equal intervals, as the intervals between meeting times may differ. This can impact the visual display of the client’s performance on the graph.

Related terms: y-axis

A prediction about the performance of the the phenomena under study. A scientific hypothesis makes a specific prediction and is testable, that is, it makes provisions for the ways the hypothesis can be rejected.

A dynamic, interactive, visual analysis tool that allows interaction, data exploration, and data analysis to provide both qualitative and quantitative analysis of the data. Similar to a data graph.

A variety of techniques used to gain information about a stakeholder’s behavior, values, attitudes, etc., summarize one’s own or a stakeholder’s perceptions about the client’s performance or to judgements about the client’s quality of experience. These include diaries, interviews, checklists, individualized rating scales and global assessments.

A set of categories designed to capture the distinct qualities of a client’s problem which can be used by the client and others to rate their success over time. They can be used to identify and assess individualized problems across a wide range of clients. Scale items describe an individual’s experiences and are rank ordered.

Related terms: IRS

Conclusions made about a behavior phenomena during the identification and coding process, based on evidence and reasoning. Low-level inferences involve identifying behavior based on their physical-perceptual attributes. Higher level inferences involve decisions based on the perceived intentions of the speaker/actor (He meant it as an insult.), value judgements, (that was a bad attempt, He didn’t try hard enough), assumed consequences of actions (he went home (the speaker based their assumption on standard practice).

Data characterized with a high degree of variability.

An observation time-sampling technique in which the observer only records the behavior that occurs at the endpoint of each time sample. It is good for recording behavioral states, that is, activities that are relatively long in duration such as attending/disattending, talking, reading, playing, hand flapping, etc.

Corresponds to the emergence of generalized performance changes that can be observed in other contexts and/or function independently outside the clinic context. An instrumental outcome may consist of the display of a target behavior in an applied context (e.g., client begins to correctly articulate the /s/ phoneme in initial position during conversations with his teacher) or represent a shift in cognitive, emotional or attitudinal stance (e.g., client may pause and reflect before responding to a difficult question). As such, an instrumental outcome may serve as the basis for the attainment of other functional outcomes and best monitored using generalization probes outside the clinic context.

The specific behavior or state that believed to be the prerequisite for the client to progress successfully through the intervention process. It represents the objectives that must be obtained in order to achieve the ultimate goal. An intermediate outcome is similar to a short term objective in that both may focus on a specifiable waypoint in a client’s progress. They differing that an intermediate outcome always represents a waypoint towards the achievement of an ultimate outcome.

Internal validity relates to the causal linkage between the independent (treatment) variable and the dependent (response or outcome) variable. An internally valid study is one which minimizes the threats to internal validity in order to confidently measure causation.

The interquartile range (IQR) is a descriptive estimate of variability. Dividing a dataset into 4 quartiles, the IQR represents the range between the 25th (1st) and 75th (3rd) quartile, or the inner 50% of the data. The median is the 2nd quartile.

Related term(s): IQR

A condition or phase in which the practitioner treats, trains and/or provides technology to resolve the target problem or improve performance.

A visual key to the data in the graph, linking a meaningful title to the corresponding data series.

The average performance within a phase or data path. Therapy-Science uses the median to calculate level because it is less influenced by extreme scores.

A treatment phase that closely follows the last intervention phase, in which the practitioner provides the client with the strategies for being independently successful, and begins to let the client manage their behavior by themselves. Progress involves both objective and subjective goals.

A measure of the average score, derived by summing up all the values then dividing by total number of values. Compared to the median, the mean influenced more by extreme scores, especially in small data sets. The median is used to calculate Level statistics in TS

Makes claims about the functional relationship between inputs (e.g., an intervention) and patient-relevant outcomes, based on assumptions made about the mechanisms involved (Mechanisms are arrangements of parts/features of an observed phenomena that describe a stable relationship between its ‘inputs’ and ‘outputs’). Such reasoning will involve an inferential chain linking the intervention with a clinical outcome.

A measure of the average which utilizes the middle value of a distribution of values to represent the average. It differs from the mean, which is derived by summing up all the values then dividing by total number of value. Compared to the mean, the median is less influenced by extreme scores, especially in small data sets. The median is used to calculate Level statistics in Therapy-Science.

Verbs that express the intellectual processes or mental states of the speaker, their addressee or anotjer individual who is being talked about (e.g., think, annoy, believe, realize, know, hate).

A type of systematic review that addresses a defined research question by using statistical methods to summarize the results of a set of studies that fit pre-specified eligibility criteria.

A statistical technique used to determine a trend in a time-ordered set of data (time series, single case design). In its simple form, a moving average is calculated by averaging initial subset of the data set (e.g., first 3 numbers), averaging successive subsets of the data series by excluding the first number of the previous subset and including the next value in the subset. The trend line that is produced by a moving average smooths out the variability of the data set, while being sensitive to changes in trend within a phase.

A single-case design that consists in making numerous systematic modifications in the goal criteria, task type and/or task difficulty over the course of an intervention. The changing criteria design is a good example of a multiple phase-change design.

A set of descriptive and inferential statistical techniques that do not rely on assumptions that the data are drawn from the normal or other probability distributions. Non-parametric methods are particularly suited for analyzing small and non-normally distributed datasets.

A source of measurement error in which accumulated observation experience begins to alter the criteria used by the observer to categorize or evaluate the correctness of observed behavior.

Defines the specific process in which behavior is observed and classified, and/or the procedures used in eliciting the behavior (e.g., therapy task) are objectively descrived.

The process of defining a behavior by its observable features and the specific techniques used to conduct the observation or measurement of the behavior.

The expected result of the intervention.

An datapoint that lies an distance from other values in a data path or data series judged to be abnormal or extreme by the analyst.

An adaption of single case design in which the client’s response to changes in magnitude levels throughout a single session are recorded.

A type of observation time-sampling in which observations are made throughout the entire time-sample. Partial time-sampling is sensitive to recording instances of short duration behaviors, as they will be noted whenever they occur.

A quantitative measure that evaluates the proportion of data points in the treatment phase that exceed the most extreme value in the baseline phase. The strength of this analysis is that it is simple and conservative. The two phase series must be largely non-overlapped in order to exceed at least the 70% non-overlap threshold to be considered noticeable.

The proportion of data points in a particular phase that do not overlap with the datapoints in the preceding phase. Typically this measure is used to assess differences between the treatment and baseline phases. Percent non-overlapping data can be calculated in a variety of ways and serves as the basis Percent-Non-Overlapping Data, Conservative Dual Criteria and Tau-U statistical tests.

The specific course of a treatment. A change in treatment conditions (e.g., shift from baseline to intervention) will result in a phase change.  The concept of phase is synonymous with the terms of  treatment and independent variable.

The number of data points comprising a data path within a phase.

A vertical line drawn in a single case graph that marks the boundary between treatment phases. It separates a data series into data paths. In TS a phase line is created by placing a pipe or “|” character between data points in the data window to create two phases.

Consecutive data points from a data series that fall within a specific phase.

Related term(s): data path

Data collected outside the treatment context that addresses the generalization of training to other situations and activities. Probe data may be collected in either structured or unstructured context and typically addresses a client’s instrumental outcomes.

The meaning and/or reasoning underlying client performance.

An interpretative investigation to identify the meaning underlying client performance.

The numeric values of a data set.

Measuring, describing and analyzing the numeric values of a data set.

In this family of research designs, the treatment and control groups are different at the beginning of the study. Although the study may be of high quality and the findings significant, the design alone does not control for other differences in the two groups which may explain the study’s results.

A study in which people are assigned at random (by chance alone) to receive one of several treatment conditions, including the experimental treatment and either a different type of treatment or no treatment.

Baseline data developed by interviewing stakeholders about the client’s past performance (e.g., stuttering essential unchanged last 6 months, speaks only 1 word utterances).

The overall consistency of a test measure or the dependability one’s observations. For example the consistency in which two observers give the same estimates of a given behavior or the accuracy in which a single observer was able to code similar behaviors over time using the same category would be a matter of reliability. Threats to reliability would include observational inconsistency due to lack of training, observer drift and/or vigilance problems.

Baseline formed using previous therapy data. Retrospective baselines can be valuable in determining changes over time while participating in therapy.

Related term(s): archives baseline

The simple phase-change design (A|B) design is comprised of two phases, typically a baseline (A) phase and a treatment (B) phase. The baseline phase serves as a control to compare against the client’s performance when receiving intervention services.

Related term(s): A|B design

Family of research and evaluation designs characterized by the systematic repeated measurement of client outcomes and regular, frequent, predesignated intervals under different conditions, baseline and intervention, and the evaluation of outcomes over time, and under different conditions in order to monitor client progress, identify intervention fax, and more generally, learn when, why, how, and the extent to which client change occurs. Also known as single–subject design, single–system designs, and N=1 designs, or sometimes time-series or interrupted time-series designs.

The single case graph utility that is included as part of the Therapy-Science subscription service. It is visually modeled after a traditional single case graph, but introduces the interactivity, visual analysis components, and data exploration and analysis capabilities associated with a graff.

The single case graph utility that is included as part of the Therapy-Science subscription service. It is visually modeled after a traditional single case graph, but introduces the interactivity, visual analysis components, and data exploration and analysis capabilities associated with a iGraph™.

Slope (or trend) refers to the direction and degree of change typifying a data path as it moves forward in time. Visually, it can be described as upward or increasing, versus downward or decreasing, or flat (i.e., no trend). Trend can also be gradual versus steep, and immediate versus delayed.

A style of thinking and reasoning that involves one’s rationalization about the world in ways that serve the group with which we are affiliated.

Those utterances or aspect of an utterance that has performative function in language and communication action, particularly with regard to its intention, purpose, or effect.

Often used in place of variability, emphasizing its functional application to the single case situation (i.e., stable data = low variability, unstable data = high variability).

Related term(s): spread, drift, dispersion, variability.

The variability of the most stable portion of the data series. It can be estimated using the interquartile range around the median or the variability cloud accessable through the level and trend line functions.

The process of determining whether the probability of the test statistic obtained is more extreme than the probability cutoff value of the null hypothesis (i.e., normal state of affairs). Statistical significance only pertains to the likelihood of occurrence not treatment effect size or real-world importance of the difference.

An external evidence procedure that addresses a defined research question by collecting and summarizing all empirical evidence that fits pre-specified eligibility criteria. A meta-analysis uses statistical methods to summarize the results of these studies. As with any other study, systematic analyses and meta-analyses may differ in quality and focus.

The target performance or target behavior consists of the behavioral elements relevant to your clinical intervention. It described well enough to be identified and discriminated from other non-target performances in the context in which the intervention is taking place. (link)

Statistical test that assesses the difference or non-overlap between two adjacent data paths. The advantages of the Tau-U over other tests are that it takes into account the trend of both the baseline and treatment phases, and limits the influence of the baseline trend across the treatment phase based on their respective phase lengths.

The ability to determine whether the treatment precedes the behavioral observation in time (i.e., the cause precedes the effect). This is a critical component to establishing internal validity.

A non-parametric method of trend for robustly fitting a line to a set of points that chooses the median of the slopes of all lines through pairs of two-dimensional sample points.

The duration in which therapy is conducted, event data collected, or the period of time referred to during an interview. Time frames bound the intervention, observation and coding of your client’s performance.

A family of methods in which an observation is organized into a series of equal duration time units. Results are reported as the proportion of time units containing at least one instance of the target performance.

Any measure that explicitly employs time as a property of the measure, such as duration, latency, or time sampling techniques.

A systematically formulated set of principles and procedures that are used to modify and improve client performance.

Data gathered while treatment is being conducted.

The specific course of treatment. A change in treatment conditions (e.g., shift from baseline to intervention) will result in a phase change.  

Related term(s): treatment, independent variable

A single-case design that consists of a single phase of treatment data, that is there is no baseline. Changes in intervention can be monitored, but it doesn’t provide information about the source of the behavior change.

The direction and degree of change typifying a data path as it moves forward in time. Visually, it can be described as upward or increasing, versus downward or decreasing, or flat (i.e., no trend). Trend can also be gradual versus steep, and immediate versus delayed.

Related term(s): slope

A nonparametric outlier detection method. It is calculated by creating a “fence” boundary a distance of 1.5 IQR beyond the 1st and 3rd quartiles. Any data beyond these fences are considered to be outliers.

Signals the accomplishment of the client’s treatment objectives, which lead to termination from the treatment program. It is similar to the more traditional long-term goal, but is oriented towards program completion and bears a closer hierarchical relationship to constituent objectives.

The extent to which a concept, conclusion or measurement is well-founded and corresponds accurately to the real world.

The fluctuation of performance in a set of data. In single-case design, the term stability is often used in place of variability, emphasizing its functional application to the single case situation (i.e., stable data = low variability, unstable data = high variability).

Related term(s): spread, drift, dispersion, stability.

Displays the numeric scale (e.g.,frequency, percent, rating) used in the measurement of the target behavior.

Related term(s): x-axis

An individuals ability to maintain sustained and focused attention.

Examining the effects of an intervention at different points over time in order to find discontinuities and other patterns that aid in the interpretation of behavioral change.

Representing numerical information visually, usually through graphs. Visual representation tools assist users in finding patterns in data, and making decisions based on those patterns.

The data being inspected or analyzed exists within a single phase of the design.