The methodological issues impacting on the design of developmental tests are central to the aim of this work. Some of them are particular to India and some to developmental tests per se.
Issues particular to developmental tests:
Developmental tests have distinctive characteristics that require consideration.
Change over time: -
Development is a continuous process and will be influenced by a host of extraneous factors that may either conceal or exaggerate the real changes due to maturation.(Harris 1990, Shelton 1987). Situ ational factors concerning the child or the examiner would influence performance and interpretation at any one time. Therefore tests need to provide data regarding the variability to be expected in test scores.
The framework employed in a test:
Shelton (1987) noted that psychological measurements often deal with intangible variables that are not amenable to direct measurement. Therefore, evidence provided by psychological or linguistic assess ments needs appropriate interpretation. This is dependent on the framework incorporated within the
assessment procedure and is therefore Implicit. For example, the medium used for evoking language in a child is integral to a test and will influence the type of result obtained. To better understand test interpre tations it is important to know the underlying rationale of test design
Developmental norms:
Many assessment procedures are based on an implicit comparison between the development of a child and that of non-disabled children of a similar age. The extent to which the child’s response differs indi cates the severity of any developmental problem. Rosenbloom (1994) stated while outlining difficulties in defining developmental disorders that there is "... a need to recognise that there must inevitably be a continuum between normal and abnormal development and that any cut off will be arbitrary and at best statistically or clinically rather than pathologically based.’ Normal children show considerable variation with respect to the speed of development, particularly language, in terms of the relationship between the structural, semantic and functional features (Brown 1973, Lieven 1978).
Establishing age norms has several problems. It is not easy to be precise about when any specific skill has been mastered by a child. Also, a child’s performance may vary from moment to moment.
Since children’s linguistic performance varies across situations it is important to replicate conditions that existed during data collection for age norms while testing (Shelton 1987). The knowledge and sensitivity of the observer are very important to the authenticity of results obtained (Harris 1990).
The competence and performance gap:
Developmental assessments assume that it is possible to sample a child’s ability and that, under certain conditions, this will provide a useful indication of the child’s potential. This may not necessarily be so since the testing situation demands use of communicative, social and other psychological skills that could cloud the child’s performance. Cronbach (1970) distinguished between measures of maximum performance and typical performance.
Others:
Some other issues are typical to language assessments. Mittler (1972) pointed out that language is a group of skills. These different aspects of language develop at varying rates. Language competence also depends on cognition, on some of its dimensions more than the others. Other psychological processes such as auditory perception, attention and aspects of memory also influence language development. All these factors may confound language assessment (Hardy 1987).
II.3.2C USES OF DEVELOPMENTAL SCREENING AND SCREENING
TESTS
Developmental screening is a system that aims to check that children are developing normally for their age, as measured against prescribed criteria. Failure to meet these criteria may indicate developmental delay or defect, or may be caused by environmental factors. Rosenbloom (1994) emphasized the rel evance of environmental factors to developmental disorders and suggested separate evaluation of them. They are of even greater importance to child development in Third World countries. When a possible problem is identified by a screen appropriate referrals, detailed assessments and subsequent management can be planned.
Screening tests can be used in several ways at different levels. Primary care personnel use them to identify children with problems requiring referral. Developmental or community paediatricians may use the same test as a first stage analysis of a referred child’s cognitive profile; in the case of the Bus Puzzle Test the profile will suggests global delay or specific language delay or disorder.
Developmental screening forms an integral part of child health surveillance in the Western World (Bellman and Cash 1987, Hall 1989, Committee on Children with Disabilities 1986). Prevention of possible prob lems, early identification of problems and management and care of children once a problem has been identified form the core aims of such programmes. Through early identification it is hoped to mitigate the effect of developmental disabilities on later education. Information from surveillance programmes can help to identify health trends, plan health and social service resources and even influence social and political policies.
In Third World countries screening tests have been used for establishing prevalence rates of specific developmental delays and disorders (Khan & Durkin - in press). They have also been used for comparing various subpopulations (Pankajam 1990) and to establish and evaluate intervention programmes (Murlidharan and Misra 1988 - cited in Murlidharan 1992) They can be used to identify areas requiring greater resources particularly in improving detailed diagnosis and management of such problems and research into aspects of development.
Lack of adequate personnel, screening tools and resources have made large scale child surveillance impossible in the Third World. However, staff in health clinics see a preponderance of young children, a proportion of whom will have developmental delays of significance for later progress (WHO 1992). An opportunity can be taken to assess the development of children who attend clinics for health reasons, particularly when parents themselves are uncertain whether their offspring are developing normally and ask for advice. Many will attend for physical complaints which turn out to be related to learning problems in children of low ability of whom too much is expected either at school or at home. Specific learning
impairments may present in the same way. In this context, several countries have added records of developmental milestones to children’s health record cards (WHO 1992). The availability of simple tools testing infant and preschool development will enable a substantial proportion of the child population to be covered.
WHO (1992) recommends that screening tests should be developed for the use by teachers to identify those who are likely to have severe learning problems. Such an exercise would help identify children with treatable medical conditions or those in need of remedial education. Expectations of their perfor mance would also become more realistic.
IL 3.3A PSY C H O M ETR IC STATUS OF TE ST IN G TO O LS IN THE
DEVELOPED AND DEVELOPING WORLDS:
Developmental tests and screening tools have been refined over the last two to three decades in the developed world. This is demonstrated by two reviews done a decade apart (Glascoe et al 1990, McCauley and Swisher 1984).
Glascoe et al (1990) reviewed 19 developmental screening tests for young children according to guide lines laid down by the American Psychological Association (1985). Four of the 19 tests had excellent reliable studies, 2 rated good, 5 rated fair, 2 were poor and 5 tests had no reliability studies done. Three of the 19 tests reviewed by Glascoe et al (1990) had excellent validity studies; two rated good; 4 rated fair; 8 were poor and 2 of the tests had no validity studies. Three of the tests reviewed rated excellent with regards to the sensitivity index; 1 rated good; 3 rated fair and the rest were either poor (3) or had no sensitivity index at all (12). Five of the tests showed an excellent specificity index, 1 was good , 1 was poor and for the rest no specificity index was derived. They concluded that several tests approached recommended standards for educational and psychological tests.
McCauley and Swisher (1984) reviewed 30 language and articulation tests for preschool children for their psychometric characteristics. Only 20 per cent had adequate sample size. Three out of the thirty tests analysed gave description of the normative sample. Only thirty per cent of language tests exam ined, had been subjected to any item analysis. Seven of the thirty tests gave the mean and standard deviation of raw scores . Test-retest reliability was reported in only one of the 30 tests. None reported interexaminer reliability. Sixteen per cent of the 30 tests reviewed had reported concurrent validity. No test had reported predictive validity. McCauley and Swisher concluded that most of the instruments available for testing the language of preschool children were not well developed and imposed important limitations on their use.
Third world countries are only recently developing an interest in psychometric tests on a large scale. This is in part due to the WHO initiative (1992) and in part due to the success of intervention programmes like the ICDS in India.
In her review of developmental test instruments in India, Murlidharan (1992) pointed out the lack of standardisation and established norms. Only one of the 19 tests had been standardised. Six of the tests had norms. Eight tests had test-retest reliability ranging from 0.56 to 0.99. Whilst only one reported inter examiner reliability. Four of the tests reported validity indices ranging from 0.73 to 0.81. Two tests reported sensitivity and specificity indices. She concluded that though ‘ a great deal of activity in testing’ existed in India, the psychometric properties of the instruments need to be improved particularly in regards to standardisation and norm-establishment. The information provided on socioeconomic target groups was inadequate and regional, ethnic, rural-urban-tribal variations need to be addressed in standardisation and norm- establishment. Item Analysis was not done in any of the tests reviewed.
Many western tests have been adapted for Indian children in the past - Bender Gestalt Test for Children (Dwivedi 1970), Standford Binet (Kulshrestha 1971) and Gesell's modification (Bevli 1978). Modifications have been minor without item analysis and are less than satisfactory. Pictures of a needle, a rose or a picture depicting a letter being posted in a letter box, which were employed in Bevli’s study, are in fact out of the realm of experience of rural or even slum children.