Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

9. Methods for the behavioural assessment of the consequences of malnutrition

Ernesto Pollitt

This chapter is divided into two parts. The first includes a brief description of behavioural assessment methods that have been used in published field studies on the functional consequences of energy protein malnutrition and iron deficiency. The second part includes a more detailed description of an information processing paradigm, and the methods derived thereof, that has been used in studies in Cambridge, Massachusetts, and Guatemala.

Description of methods in published studies
An information-processing approach
Annex: Behavioural test battery

Description of methods in published studies

The methods for behavioural assessments can be divided according to the age of the children sampled and the categories of behaviors evaluated. The breakdown by age may be as follows:

- infancy (0-24 months),
- preschool (2-6 years),
- school age (7-18 years)

The behaviours that have been assessed can be classified as:

- activity and play behaviour,
- motor development,
- social and emotional responses and development,
- cognition:

  1. general composite measures of intelligence,
  2. specific congitive measures.

Assessments of home stimulation will also be briefly described because they are relevant to the purposes at hand, although they do not deal directly with the behaviour of the child.



Activity and Play Behaviour

A distinction must first be made between attempts to measure activity as energy output, and measurements used for the assessment of activity as an indicator of curiosity, or motor involvement in play behaviour. In some instances there may be an overlap between the two types of measurements. However, given our primary concern with behaviour, the emphasis will be on the latter type of assessment.

Perhaps the most detailed description of activity measurements in infants is found in the study by Chavez and Martínez (1) on the effects of food supplementation on the behavioural development of infants and children in Mexico. They attempted to establish frequency and duration of motor movements of infants in bed asleep; and during time involvement in interactions with the mothers, and in play behaviour. They also measured activity of children in a room (3 x 3 metres) equally divided into squares where there were toys with which the children could play. This last procedure has also been used in a study on iron deficiency (2) which will be described in more detail in the section on the preschool child.

Motor Development

Assessments of motor development have been primarily restricted to the use of the Bayley Scale of Motor Development (3; 4; 5).

This scale samples fine motor manipulations of hands and fingers, and also the more gross motor capabilities such as walking, climbing, standing, and sitting.

The Gesell Schedules (see the section on cognition, below), which are primarily used for assessment of mental development, also include a motor component. This instrument has likewise been used by other investigators (1; 6; 7).

Social and Emotional Responses and Development

Two types of procedures have been used for assessments of social and emotional responses and development. One includes indirect assessments of the time at which the child reaches certain landmarks of social development; the other is based on direct observations of the child in specific social situations. Among the former, the only scale that has been used is the Vineland Social Maturity Scale (8; 9).

The purpose of this scale is to measure social maturity and responsibility from birth to maturity. It requires the participation of an interviewee such as the parents or perhaps a grandmother, who is intimately familiar with the infant. The Vineland scale consists of eight categories of social maturity: self-help general, self-help eating, self-help dressing, self-direction, occupation, communication, locomotion, and socialization.

This scale shows a high correlation (approximately 0.08 per cent) with developmental and intelligence scales. One problem with the use of this scale in the field, however, is that it requires an interviewee who is verbally fluent and at the same time sufficiently perceptive to give detailed information about the infant's social development (8).

Measurements of social response in a social context can be subdivided, in turn, into: (a) those which assess the infant's response to an experimental stimulus provided by the investigator; and (b) those which require observations of the child in a natural situation or in a social situation created by the investigator, but where there is no placement of an experimental stimulus. Among the former, a measurement that deserves particular attention because of its novelty and the fact that it is used with neonates is the Brazelton Scale (10, 11). Strictly speaking, this scale does not evaluate directly the social responsivity of the infant, yet this is one of its objectives. It uses 26 behavioural items to assess the infant's responsiveness to external stimuli (i.e., social interest in the examiner), motor organization (i.e., pull-to-sit), and ability to modulate state of consciousness (i.e., hand-to-mouth activities). These behavioural items, according to the authors, reflect the integrity of the infant's central nervous system, and the infant's capacity to elicit caretaking from an adult. It has been used in Guatemala with nutritionally deprived infants (11).

Most assessments of social-emotional responses in social situations have focused on the relationships between the infant and the mother, either in a playroom situation or during feeding (1; 12; 13). The methodology of some of these studies includes the placement of the mother and the infant in a playroom while their observer is behind a one-way vision screen. The coding of the interaction is done either immediately after the observation or it is facilitated by recording a TV tape. Although the latter strategy is more costly, it permits a much higher degree of reliability in the coding system.

In one study (14), for example, the behaviour of the mothers and infants were recorded on a Sony AV-3400 portable record playback video recorder and a Sony AVC-3400 video camera. The camera was located in one corner of a 12 x 15 foot experimental room, about 10 feet from the chair in which the mothers sat while feeding their infants. In this study the code used for the assessment of the feeding interaction was an adaptation of Brown and Bakeman's coding methodology (15).


Composite measures of development such as the Bayley Scale (4), the Gesell Schedules 17), or the Griffith Scale (16) have been the most popular tools for the assessment of mental development in studies of malnutrition and behaviour. The use of these scales in field studies in developing countries, however, is often made difficult by problems of normative and construct validity because they were conceived and standardized in highly industrialized countries. To a large extent these validity problems are similar to those found in crosscultural investigations of cognition in children. Because of this similarity, investigators interested in the use of these scales can take advantage of the enormous body of literature on cross-cultural methodology, and of the excellent discussions that exist on the transfer of test from developed to developing countries (17; 18).

In addition to the issue of validity, two other theoretical issues need to be considered by an investigator planning to use developmental scales in the field. These are:

- stability and change of intellectual functions,
- specific versus aggregate measures of cognition.

The issue of stability and change deals with the problem that the developmental scales are not good predictors of measures of intellectual performance in the school age period. This lack of test stability is related to the developmental transformations in the nature of intelligence from infancy to the preschool age years. Here again, a series of extremely useful and informative papers published in recent years that can be of great aid to the investigator (19, 20, 21).

The issue of specificity versus aggregate measures of intellectual competence has to do with the fact that the developmental scales generally give aggregate measures (developmental quotients) of different psychological processes involved in cognition. Accordingly, it is difficult to infer from these data what specific psychological processes, if any, are more likely to have been affected by the independent variable in question. (This issue has been discussed at some length in connection with iron deficiency; see 22.)

The Gesell Schedule was the first assessment tool developed for the purpose of the evaluation of mental abilities in infants. The initial scale, first published in 1925, consisted of 144 items divided into four general fields. Motor behaviour included postural control, locomotion, comprehension. drawing, and hand control. Language behaviour was assessed by means of vocabulary, word comprehension, conversation, and word production. Adaptive behaviour was comprised of eye-hand coordination, imitation, object recovery, comprehension, discriminative performance, perception and completion, and number conception. Personal and social behavior included reactions to persons, personal habits, initiative and independence, play responses, and acquired information.

The composition of the scales varied; while motor items comprised 45 per cent of the four-month schedule, language items comprised only three per cent of that schedule. Yet, by the twenty-fourth month, motor items were reduced to only 11 per cent of the schedule, while language items comprised 21 per cent of the schedule. On the other hand, the percentage of adaptive behaviour items remained relatively unchanged throughout the six schedules, ranging from 25 to 47 per cent of the total number of items (23). Essentially, there have been no basic changes in the scales since 1925. The classification of items into motor, adaptive. language, and personal-social categories is still used (24).

The Bayley Scales are divided into a "mental" and a "motor" area. Performance in the mental scale is expressed in the form of a mental development index. It contains 163 items and is normed at two-week intervals from two to five months of age, and at one-month intervals from six to 30 months of age. The infant scales also include an Infant Behavior Record containing a set of ratings which provide quantitative and qualitative information regarding the infant's social, emotional, and stylistic behaviour during the administration of the mental and motor scales (23).

One advantage of developmental scales such as the Gesell Developmental Schedule, the Bayley Mental and Motor Scales, or the Griffith Scale of Mental Development is that they yield scores that can be treated statistically as interval scales. This permits the use of parametric techniques for comparative analyses and allows taking advantage of the statistical packages now available for computer use.

Another recently-developed developmental scale that has been used in malnutrition-behaviour studies is the Albert Einstein Scale of Sensory Motor Intelligence (24). In contrast to the previously described psychometric techniques that are more empirically based, this new developmental tool is guided by Piaget's theoretical frame-work. The Einstein Scale of Sensory Motor Intelligence (24) is designed for infants between one month and two years of age. It consists of three scales, containing a total of 54 items. The three scales include a comprehension scale, which covers the development of adaptive reflexes (primary circular responses) and early systematic behaviour (secondary circular responses) in the lowest age range. Objects in the environment exist permanently and independently of behaviour. Finally, the Space Scale involves the infant's ability to function effectively in three-dimensional space.

The assessment of specific cognitive processes in infancy has been difficult because of both conceptual limitations and methodological constraints. Our understanding of the modus operandi of the infant's intellectparticularly in the preverbal stage-has not yet advanced to a point where we can define the operations of all the most critical processes involved and their inter-relationships. This is even more apparent if the objective is to identify levels of operation having clinical significance. Moreover, because of the obvious restrictions in verbal communication with infants, the tools for behavioural testing must be significantly reduced. Probably the most sophisticated attempts to measure infant cognitive function are found in those studies of habituation and pre-verbal concept formation.

The habituation paradigm has been used with infants as early as 15 days old (25). A detailed description and successful use of a habituation paradigm is that of Lester (26). In this investigation attentional processes were assessed by measuring heart rate deceleration to novel auditory stimuli. Heart rate deceleration has been found to correlate with the orienting reflex to a novel stimulus. When the individual becomes accustomed to the stimulus, habituation occurs and cardiac deceleration decreases. These physiological responses measure responsivity to the environment and to attention. In this particular investigation the responses of malnourished infants to 750 and 500Hz at 90b showed no cardiac orienting to the onset or change of the stimuli tones.

Brockman and Ricciuti (27) used a pre-verbal test of concept formation with severely undernourished children that also deserves attention. The test calls for the classificatory behaviour of children in response to eight objects (two groups of four identical stimuli) that vary according to size, color, form, and sound. The objects were presented in a standardized form and the child was simply allowed to play with them as he pleased. During testing, two types of behaviours are assessed. One, selective ordering, tapes the manipulation of similar objects in sequential order; and the other, grouping, assesses the classificatory system used by the child according to the complexity of the groups. The clinical use of this test followed Ricciuti's years of work (using the same approach) with normal children.

One concern of investigators in the area of malnutrition and mental development has been how to assess the stimulation that the home environment provides to the infant. The objectives are to determine the extent to which this interaction enters into the causal chain of malnutrition, and whether it explains part of the variance in the developmental measurements of the malnourished infant.

Perhaps the greatest effort to assess household stimulation in a systematic way, has been made by Cravioto and collaborators in Mexico adapting the Betty Caldwell Inventory of Home Stimulation (28, 29, 30). There are two versions of this inventory: one is for use with families of children from birth to age three, and the second one is for use with children from three to six years old. It contains items that essentially assess the amount of mental stimulation in the home through toys, games, and reading materials; the nature of language stimulation in the physical environment; and systems of punishment used in the household.

One methodological problem with the Home Inventory is that it requries an observer in the household itself. This is a costly procedure, especially in large-scale studies with large sample sizes. Moreover, in field studies in rural areas where the house is often quite small, it is an extraordinarily complicated mechanical problem to have a visitor in the household for a number of hours to collect substantive and representative information of the interaction between the mother and the child.


Preschool Years

Activity Level and Play Behaviour

As previously noted in the section on activity and play behaviour in infancy, activity level has been measured in various studies on malnutrition and behaviour by placing a child in a room with toys and observing directly his movements throughout the room and his degree of involvement with these toys. In one particular study (2) play activity was observed by dividing a 15 x 19 foot room into 16 equal parts by drawing lines on the floor. Each of three walls contained a bench with toys (the same toys on all benches). In the remaining wall there was a one-way observation mirror, behind which two observers recorded a detailed account of the child's behaviour. Activity variables were recorded in 1 5-second intervals. The following were some of the variables analysed: (a) manipulation of toys-the number of times a child picked up a toy; (b) the time the child spent playing with (acting on) one or more toys; (c) the number of times the child crossed a quadrant in the floor; (d) the amount of time a child gave clear behaviour evidence of gross motor movements within one or more quadrants; (e) the time the child verbalized without giving evidence of addressing an adult in the room.

This procedure generates useful and interesting information. However, it is costly both in terms of setting up the adequate room conditions for direct observations, and in terms of amount of time for data coding and analysis. Another problem with this technique is that there are no clear guidelines as to what may be the biobehavioural significance of the movements observed. Its use therefore calls for a specific hypothesis to be developed prior to data collection.

Motor Development

One procedure used for the assessment of motor development among preschool children is the LincolnOszeretsky Scale (31). This is a revision of the Oszeretsky Test of Motor Proficiency that was originally published in Russia. This scale is appropriate not only for preschoolers but also for school-age children, as the ceiling age is 15 years. It is constructed to measure all different forms of body movement from facial and finger musculature to postural reactions. It is not a popular scale and has been utilized in only one study related to malnutrition (9). Its use in the field is made difficult by a lack of appropriate standards.

Social and Emotional Responses and Development

Except for the assessment of mother-child interaction that have been conducted by the group in Colombia (13), there are no other known procedures that have been used in the field for the assessment of social and emotional responsivity in preschool children. This area needs further research, since it is very necessary to determine whether or not malnutrition leaves any functional deficits in the area of social competence. Some areas that, for example, could be investigated are the child's ability to obtain and maintain the attention of adults in socially acceptable ways or his ability to use adults as resources-that is, to obtain information, assistance, food, etc., by verbally requesting it, demanding it, or physically demonstrating a need for it.

Moreover, it would also be important to determine the extent to which the malnourished child is capable of expressing affection or hostility to adults through verbal or physical means.


The same theoretical issues raised in connection with the use of infant developmental scales are also relevant in the area of composite measures of intelligence for preschool children. One issue that may be different, however, is that of constancy and change in intellectual function. Whereas the scores from developmental scales are poor predictors of later intellectual function, this problem is significantly reduced with the preschool intelligence scales. After a child reaches about four to five years of age, the IQ measurements that are obtained from tests such as the Wechsler Intelligence Scale for Children or the Stanford Binet Intelligence Scale are good predictors of school age IQ. These scales, moreover, are the most popular at this age period.

There are two types of Wechsler Intelligence Scales that may be applied to preschoolers. One is the Wechsler Intelligence Scale for Children (WISC) revised version (33) and the other, the Wechsler Preschool and Primary Scale of Intelligence (34). However, the latter form is made to test much younger children than the WISC and therefore, it is more suitable for the assessment of preschoolers. This is an individuallyadministered scale that requires a very well-trained examiner and features a special kit of materials. Administration takes approximately 50-75 minutes, and testing may have to be spread over two different testing sessions. The test is structured into two types of tasks. The first involves verbal tests that include the following five subtests: information, vocabulary, arithmetic, similarities, and comprehension. The other set includes five performance subtests that include: an animal house, picture completion, mazes, geometric designs, and block designs. All five of the regular verbal subtests are quite similar in design to those that appear in the WISC. Moreover, three of the performance tests have counterparts in the WISC, while the subtests of animal house and geometric designs are different.

One of the problems in its use in the field is the time it takes to administrate, and another is that some test items may not be culturally suitable to populations from developing countries. There have been some attempts to adapt the WISC to particular cultural milieus, such as for example the new South African Individual Scale (35). The WISC has been utilized in a number of studies establishing relationships between malnutrition and mental performance (36; 37; 38; 39).

Another popular intelligence measure is the Stanford Binet Intelligence Scale (40; 41). This is also an individually-administered scale and it requires a highly-trained examiner. Test administration lasts for approximately one hour. Moreover, the test session requires an acute observer because the examinee must be observed and rated in such items as attention, reactions during test performance, problem-solving behaviour, and independence of examiner's support.

One of the useful features of the Stanford Binet Intelligence Scale is that it has been standardized for children from two to eighteen years of age. However, in contrast to the Wechsler Scales that provides a verbal IQ, a performance IQ, and full-scale IQ, the

Stanford Binet provides a single score. Validity information on the Binet is substantial; quantities of data have been collected since its origin early in the century. Moreover, it has proven consistently to be an excellent predictor of school performance.

Utilization in the field, however, also requires an adaptation of many of the tests and stimuli to the particular cultural conditions of the population where it is going to be used. For example, Cabak and Najdanvic used an adaptation of this scale for malnourished children in Yugoslavia (42). It has also been utilized in Lebanon (43) and in Chile (44).

One other simple test that merits attention because of its shortness in administration and the fact that it only requires a paper and pencil is the Goodenough-Harris Drawing Test (Draw a Man) (45). It has been used in at least one study trying to establish the effects of kwashiokor on intellectual development (35) and on the effects of malnutrition secondary to neonatal disorders on later behaviour (47). One of the particular features of this simple test is that it can be used in groups as well as with individuals and does not require a highly-trained examiner: The child is simply instructed to make a picture of a man. Two scoring procedures are available for this test. A rapid evaluation of the children's drawings of men and women can be made by comparing them against the quality scale that has 12 sketches of men and 12 of women rated from least mature to most mature. The second scoring method, which is more precise but also more time-consuming, uses a point scale. The child's drawing of a man is examined for the inclusion of 73 separate items, such as body parts, items of clothing, and proportions. It has been proven to have a relatively high correlation (~.80) with either the Stanford Binet Intelligence Scale or with the Wechsler Scales.

A problem in the administration of this type of test in preindustrial societies is the lack of familiarity that children, particularly preschoolers, may have with pencils and paper. However, in some instances the use of this test may be perfectly justified, especially with school age children and if they are enrolled in the formal educational system.

In addition to the aggregate measurements of intelligence, a number of specific measurements have been used for tapping particular cognitive processes in preschoolers. Among the tests for attention span, one that deserves special consideration is the Continuous Performance Task, which has a long history in the assessment of the attention of hyperactive children in the United States. This test consists of the presentation of a series of colored pictures (for example, a car) projected on a small screen (for example, 3 x 3 inches) and viewed one time at a high speed (i.e., 50 or 60 per minute) in random order. The subject has to identify (by pressing a hand-held button) one of the eight figures presented every time it appears. For each run, the score for subjects represents the total number of correct responses (47).

One difficulty in using this particular procedure in field studies may be that it requires electricity. However, it has been used quite adequately in a rural village in Guatemala for the assessment of attention in iron-deficient children. The investigator has to be careful to have test stimuli that are familiar to the examinee.

Tests for the assessment of specific cognitive processes in preschool children have been used by a number of investigators. The reader is here referred to some of the appropriate references, as a summary description of each of these tests goes beyond the scope of the present review (see 48; 49). It is important to note, nevertheless, that in most instances these tests have been used as part of a battery that was constructed without taking into account the interrelationships that may exist between these processes. Test selection was generally based on the investigators "hunch" of which tests might have a significant pay-off; or in a comprehensive approach, inclusions of as many tests as possible.


School-age Years

Activity and Play Behaviour, Motor Development

No data are available, as far as this author knows, for assessment of these behavioural variables among schoolage children.

Social and Emotional Pesponses and Development

Barrett, Radke-Yarrow, and Klein (50) have recently completed a paper on the assessment of the effects of early calorie supplementation on social-emotional functioning among school children in Guatemala. They observed children, in small, six-person group sessions, in nine different social situations that included free play in a novel environment, group problem-solving tasks, construction activities, competitive games, and impulse control situations. Assessments were made using a time-sampling procedure, determining frequencies for specific categories of behaviours such as aggression, seeking attention, and prosocial behaviour. Ratings were also made of responses to controlled manipulations, such as introduction to a novel environment, frustration and response to prohibitions. Moreover, after each session observers made global ratings on behavioural dimensions such as attention, energy level, disposition, and work orientation.

This approach is novel and is a welcome addition to existing testing tools because, while much information has been collected on intelligence testing performance of malnourished children, we know little, if anything, about their social behaviour.


The issues raised above in the section on cognition for preschool children are also relevant to this section. Moreover, the tests therein described are also applicable to the school-age child. Therefore, in this section we will only describe one additional intelligence test that deserves attention because it is allegedly a culturally-unbiaised test and may be suitable, in some instances, to populations in preindustrialized societies. The Raven Progressive Matrices test (51), developed by J.C. Raven in Great Britain, requires choosing the correct missing part from a matrix, from about six to eight possible answers. There are two forms of the test: one that is useful for children between the ages of 8 and 14, and the other for subjects between 18 and 65 years of age. The test is regarded as a measure of "9" (general intelligence factor). In some cases, like the Bacon Chow Study in Taiwan (52), it was used for the parents of the malnourished children studied.

Educational outcome is obviously an important variable in connection with the early nutritional history of the child. What, if any, are the effects of this early experience on the behaviour, performance, and general adaptation of the child in the school setting? A comprehensive coverage of this topic, in terms of both results and methodology, is found in a paper by Mushkin (53). She identifies three main variables for analysis: achievement, aptitudes, and attitudes and attributes. Among achievement indicators it is possible to use either results from standardized tests, such as the Wide Range Achievement test or results from specific school tests (e.g., reading, writing, spelling). Richardson, Birch, Grabie and Yoder (54), for example, obtained permission from the Ministry of Education in Jamaica to obtain results from the latter type of tests for children with a history of malnutrition. Recently, Galler et al. (55; 56), working in Barbados, also looked at the results of the eleven-plus exam that is used, following the British system, for determining the type of school in which the children should enroll after the sixth grade.

In the aptitude category, Mushkin (53) makes reference to the child's intellectual capabilities that could be tapped by IQ scales. Finally, in connection with attitudes and attributes, she suggests looking at behaviours that may be relevant to the teacher's educational task. Richardson, et al. (54), for example, developed a teacher questionnaire seeking information on the child's behaviour related to classwork, on social relationships, conduct problems, and peer choices. Moreover, they also developed a sociogram to establish the degree of social contacts among these children.

In summary, we find that a large number of instruments have been used with infants, preschool, and schoolage children for the assessment of different behavioural categories among children with a history of malnutrition. Some of these instruments were prepared specifically for the purposes of the studies; some were taken, with or without adaptation, from what was already available. A researcher planning to work in this area may well profit from the experience we have gained over the years by looking at all the bibliography that now exists. What we have done is nothing but a cursory coverage of a body of literature that has become enormous in the past 15 to 20 years.

Contents - Previous - Next