**5. Studies about tics**

As expected in behavior-analytic research, direct observation has been the preferred method for quantifying tic severity. However, researchers in psychiatry, neurology, and even the

Epidemiology of Tics 171

The indirect measures of tic severity could seem inadequate and, as such, research that has relied exclusively on instruments designed for collection of indirect measures should be

One of the hallmarks of behavior analysis is the use of direct observation to quantify behavior. This preference is based on the premise that direct observation is more objective than indirect methods such as self-report or clinician ratings of tic severity. Several studies, published in behavior-analytic outlets, demonstrate the value of using direct observation to quantify changes in tic frequency when evaluating behavioral treatments for tics59. Still, many researchers outside behavior analysis have largely preferred indirect measures over direct observation60. Among the foremost concerns raised by these researchers is that observations conducted within a clinical or research context may not generalize to other settings such as school or home. Generalization between settings is an important issue in both research and clinical practice. Indeed, it is not uncommon for parents to report that a child's tics are more or less severe while at the clinic compared to when the child is at home61 Such reactivity to setting has been attributed to several factors including natural fluctuation, reinforcement contingencies, children's ability to volitionally suppress or temporarily withhold tics, reactivity to observation and internal

Regardless of the reason for contextual variation in tics, such fluctuations have important implications for the measurement of tics. If the scientific and clinical community is to have confidence in the results of behavior-analytic work utilizing direct observation methodology, observations conducted within a research setting must be generalizable to other settings. Clinic- and homebased observations are highly related, suggesting that, in general, clinic observations correspond well with home observations63. However, examination of individual data shows that generalization should not necessarily be assumed; many children exhibited differential tic frequencies across the two settings, suggesting that, whenever possible, observations should be conducted in multiple settings. Lack of consensus regarding the most reliable, valid and feasible methods for collecting and coding direct observation data has also been cited as a reason for the preference of indirect measures over direct observation64. Practitioners and researchers in disciplines outside behavior analysis may be more likely to use direct observation methods if the effort associated with their use can be reduced, without any sacrifice of their validity and capacity to generate representative samples of target

Direct observation to longer samples and event-frequency coding to a less arduous timesampling method (*i.e.* partial-interval coding) have been used to evaluate outcomes in tic research65, although partial-interval coding is more user-friendly because it does not require the observer to record each occurrence of the tic; thus, it might be preferred over the event frequency method. However, partial-interval coding cannot be recommended as an alternative if it does not yield a reliable measure of the behavior. Because simulation studies have suggested that partial-interval coding may underestimate the frequency of high-rate short duration responses, especially if they occur in rapid succession or as bouts, as is the

interpreted with caution.

**5.2 Direct observation** 

states such as anxiety62.

behaviors.

case with many tics66.

broader field of behavior therapy have preferred indirect measures, such as clinical impression, self-report inventories and clinician-rated scales. The most commonly cited reasons for not using direct observation include concerns about generalization of observations made in clinic or research settings to other relevant settings, such as home or school53 and disagreement about the best methods for collecting and scoring direct observation data54. Although the empirical basis for these concerns is not firmly established, acquisition of data supporting the use of direct observation methods may encourage those outside behavior analysis to use direct observation as a primary assessment method rather than relying on potentially biased verbal self-reports.

Studies in tics may be divided into three groups: 1) Studies made in clinical grounds, 2) Large-scale screenings and 3) Studies involving selectively school population.

Studies of in-hospital population comprise patients with most severe symptoms, in different age groups and different methods of final diagnosis confirmation are used. Procedures used in large-scale screening studies make possible the elimination of potential selection bias. Large populations are studied using transparent and repetitive confirmation of diagnoses. Their validity is additionally checked in parallel validity studies. The highest prevalence of tics is obtained in studies involving schoolchildren. Data are gathered from multiple sources: from parents, teachers, and children, as well as videos, from classroom observation and diagnoses made by experienced clinicians. Epidemiological surveys of school-age children have shown tic rates ranging from 4% to 50%55. This instability in reported rates is perplexing and is probably more artifact than truth. For example, prevalence of tics increases if transient tics are taken into account56, if studies were made just in public awards and when children attending special education schools were studied57. Inversely, prevalence of tics lowers after direct observation extends for a wide time58.
