Academic journal article Behavior and Social Issues

Paradoxical Patterns in the Measurement of Hyperactivity

Academic journal article Behavior and Social Issues

Paradoxical Patterns in the Measurement of Hyperactivity

Article excerpt

ABSTRACT:

The use of rating scales rather than direct observation procedures to identify various behavior problems is a popular trend in the public school system. We compared a popular rating scale with a direct observation system to measure "hyperactive behaviors" in a classroom setting. The subjects were 30 seventh grade students chosen at random from heterogeneously grouped students at a public middle school. A Pearson product-moment correlation was used to determine the magnitude and direction of the correlation between the data obtained by way of direct observations and the rating scale. Reliability coefficients were obtained, and a regression equation and coefficient of determination were calculated. A statistically significant but relatively low correlation was obtained between behaviors recorded by way of the direct observation instrument and the behavior rating scale. Interestingly, a strong negative correlation was found between the level of hyperactivity obtained on the rating scale and the rating scale reliability coefficients.

Behavior analysts who work in the public school system are under increasing pressure to employ rating (Likert) scales in their assessment strategies. In fact, behavior rating scales have become one of the most popular types of assessment instruments in determining the extent and frequency of a wide range of maladaptive behaviors exhibited by children (Achenbach & Edelbrock, 1983; Gresham, & Elliot 1990; Quay & Peterson, 1983; Shaywitz & Shaywitz, 1988; Sparrow, Balla & Cicchetti, 1984). Most of these instruments were developed with the intention of employing them at various times during the school semester to summarize informal observations of problem behaviors (e.g., Katz, Kellerman, & Siegel, 1980). McConaughy and Ritter (1995) have praised the use of rating scales as a "best practice" in the assessment of emotional and behavioral disorders. In fact, many academic reviewers and professionals have pointed to the advantages of the time-efficient, low-effort simplicity of these instruments, (Barkley, 1987; Knoff, 1995; Ross & Ross, 1982; Whalen & Henker, 1976). Rating scales have been used to describe and quantify a wide range of problem behaviors including the characteristics representing Attention Deficit Hyperactivity Disorder (McCarney 1995), and it has been reported that the consensus of agreement among raters gets stronger as more profiles on student hyperactivity are performed (see Hinshaw, 1987, for a discussion). Irrespective of the types of behaviors under consideration, rating scales remain one of the most frequently employed tools in the psychologist's, diagnostician's, counselor's, and teacher's assessment armamentarium (Lahey et al., 1988).

Nevertheless, Sattler (1992) points out that rating scales may not correspond with data obtained by way of direct observations. He suggests that the internal consistency and interrater reliability are important features of behavior rating scales, and he has concerns about the adequacy of these measures. Sattler is not alone in his concern; Quay and Peterson (1987) looked at the Revised Behavior Problem Checklist and found that several studies demonstrated a general lack of interrater agreement between the ratings of teachers with parents and teachers with other teachers (cf. Simpson, 1989-1990; Simpson, 1991). Likewise, Elliot, Busse, and Gresham (1993) forwarded the notion that while interrater reliability is a desirable psychometric quality for rating scales, it may be difficult to establish.

The Behavior Assessment System for Children (BASC) is one example of a popular rating scale. The BASC manual (1998) reports three types of reliability data for its Teacher Rating Scales: test-retest, interrater and internal consistency. Interrater reliabilities for the Preschool Level Hyperactivity Scale are reported to be at 0.81; however, these reliability measures are obtained by correction estimates based on statistical strategies for interpolating reliability coefficients. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.