Early Intensive Behavior Intervention - Criticisms

Criticisms

Gresham and MacMillar (1998) specifically cite a lack of a true experimental design in Lovaas' (1987) experiment on early intervention. They charge that he instead implemented a quasi-experimental design of matched pairs regarding the distribution of subjects within the experimental and control groups. Gresham and MacMillar (1998) also state a lack of a true representation of autism in that the subjects were neither randomly sampled from the cumulative autistic population nor were they randomly assigned to treatment groups. The internal validity of the study was also called into question due to the possibility of skewed data resulting from three influential threats. Instrumentation, changes or variations in measurement of procedures over time, was argued to have been altered in both the pre-test and post-test conditions which were confounded by a differentiation in ascertaining cognitive abilities and intelligence of the subjects. The pre-test utilized four measures of cognitive ability and mental development. Five of the subjects' intelligence was determined through a parent-reported measure of adaptive behavior. All of the subjects were post-tested three years later using five other measures of intelligence and cognitive ability. Long-term follow-up was assessed with three measurements of (1) intelligence, (2) nonverbal reasoning, and (3) receptive language. The original three measurements during the testing phase were determined by (1) IQ score, (2) class placement, and (3) promotion/retention. External validity was called into question concerning sample characteristics. Lovaas' (1987) criteria for acceptance into the program required a psychological mental age greater than 11 months and a chronological age less than 46 months in the case of echolalic children. Schopler et al. (1989) purport that if both the intellectual and echolalia criteria were rigidly adhered to at the North Carolina institute, approximately 57% of the referrals would have been excluded from the program.

Other criticisms include a failure to operationally define the use of the term 'reinforcement' for compliance, the use of a Pro-rated Mental Age, and the statistical regression of the child's IQ over time. Boyd (1998) addressed the potential impact of a disproportionate sex ratio of females to males on the control group's mean IQ score. Females with autism typically display slightly lower levels of functioning in comparison to their male counterparts.

In a rejoinder to Boyd's (1998) article that cited an unequal sex ratio as a source of error, Lovaas (1998) listed three reasons as to why the disporportionate ratio's influence on the data was negligible. The autistic population at the time had a ratio of 4:1. Lovaas (1998) argued that the ratios for the experimental group, control group 1, and control group 2 of 16:3, 11:8, and 16:5, respectively, were in fact near the expected ratio scale of the general population with the exception of control group 1. The second argument lay in the studies Boyd (1998) referenced in regards to low intellectual performance in females diagnosed with autism. One of the studies admitted to having a female subject with Rett disorder, a condition that showed little responsiveness to intensive early behavioral intervention. Lovaas (1998) concluded by proposing that males may more readily meet diagnostic criteria for autism because of certain salient characteristics inherent in the sex while the subtleties in their female counterparts may be overlooked.