Contents - Previous - Next

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

Demographic and socio-economic changes in families in four Guatemalan villages, 1967-1987

Patrice L. Engle, Suzan L. Carmichael, Kathleen Gorman, and Ernesto Pollitt



The Institute of Nutrition of Central America and Panama (INCAP) carried out a longitudinal study of the effects of nutritional improvements on growth and development in early childhood in four villages in eastern Guatemala, 1969-1977, with a preparatory survey in 1967 and a follow-up study of the participants in 19881989. This paper examines differences among the four villages in education, occupation, quality of housing, and demographic profiles over a 20-year period, focusing on comparisons between the two villages that received a high-energy, high-protein supplement and the two that received a low-energy supplement at two different times: before the initial longitudinal study and before the follow-up study. The results suggest gradual improvement in all the villages on a number of indicators. However, the two pairs of village were not comparable on all measures; of particular concern for the interpretation of effects on cognitive development are differences in education.


The INCAP longitudinal study, described in detail in this issue, was designed to test the hypothesis that supplementary feeding improves mental development and physical growth. The intervention took place in four villages in the Spanish-speaking, eastern section of Guatemala. Villages were randomly assigned to supplements. Two villages received atole, a high-protein, high-energy supplement and two received fresco, a low-energy drink. Because the treatments were assigned by villages rather than by individuals, it is important to test whether the villages differed in terms of characteristics that could affect the outcome variables. If important differences are found, these must be controlled for in the analyses before effects can be attributed to exposure to the supplement.

Many studies have shown that socio-economic status (SES) affects cognitive development [1]. Thus, differences in SES between the villages receiving different treatments have to be considered in interpreting differences in cognitive development associated with the treatments. Although initial equivalence between the children in the atole and the fresco villages in height, weight, and home diet was established [2], no analyses have determined equivalence on socio-demographic variables. Family background variables that are believed to have theoretical associations with developmental outcomes and that are available for study are parental education levels (literacy), house quality, father's occupation, and family size and composition.

Differences between the atole and the fresco villages at the time of the follow-up study also must be examined. Variations in the patterns of socioeconomic development since INCAP's departure from the villages in 1977, leading to differences in SES in 1987, would also have to be considered in the analyses and interpretation of results.


Description of the villages

The study villages are located in the department of El Progreso, a dry, mountainous area north-east of Guatemala City on the main road to the Atlantic coast. The large fresco village is closest to Guatemala City, at 36 km, and the small fresco village is the farthest, at 102 km. The latitude of all the villages is around 14°50'. The large fresco village is 1,250 m above sea level, both atole villages are at 860 m, and the small fresco village is at 275 m. The average temperature range for the small fresco village is 24°C-38°C, and for the other three villages is about 14°C-32°C. The rainy season occurs from June to October. The agricultural patterns, history, and infrastructure of each village are described elsewhere in this issue [3].



Census data from each of the four villages were collected before the initiation of the study (1967), in the midst of the supplementation study (1974), and before the follow-up study (1987). All families living in the four villages at the time of the census were sampled. For each census, every ever-united woman over the age of 15 years or mother in each village was interviewed, and information was obtained on each member of the nuclear family living in the household. A family was defined as a mother-child unit, although other family members living with the same unit would be counted as part of the same family. In almost all cases, the mother or primary caregiver was the informant. Table 1 shows the number of families per village included in each round of the census and the total number of families for whom there are data for at least one of the three assessments.

TABLE 1. Number of families with census data by year of census and village

Village 1967 1977 1987 Totala
Large fresco (SD) 174 225 369 471
Large atole (C) 178 236 368 464
Small fresco (ES) 132 187 254 325
Small atole (SJ) 110 136 237 281

SD = Santo Domingo, C = Conacaste, ES = Espíritu Santo, SJ = San Juan.
a. Number of families for whom there are data at any assessment date. New families formed in 1987 are included as separate families.


The 1974 census form was almost identical to the 1987 form. The 1967 form had similar information on most measures but small differences in the coding of occupation, literacy, and type of housing. The subject provided information about family structure, marital status, religion, number of pregnancies, number of live children, and the relation of the household head to the head of the extended family at the time of the census. The interviewer also observed and recorded the quality of the house (e.g., types of walls, floor, roof). For each family member, parity, relation to the rest of the family, birth rate, education, occupation (if over 10 years old), current status in the family, and the dates of changes of status (death or migration) were coded. In addition, in 1987, when many of the subjects had started their own families, the identification number of the previous family was recorded. For some subjects, at least two prior families had to be coded.


Construction of variables

The data from the censuses were used to construct various indicators of SES: house quality, mothers' and fathers' education, and fathers' occupation. Details of the construction of the variables are given elsewhere [3].

House quality

Income measures are usually used as proxies for a number of social and environmental variables that affect cognitive development. For example, income is associated with the availability in the home of materials that foster education such as books, toys, and pictures. Income also relates to adequate food supply and access to health care and education. In rural areas of developing countries, however, accurate information on family income is difficult to obtain. For example, in agricultural communities, quantifying household production and expenditures can be extremely difficult and potentially erroneous. As a result, house quality is often used as a proxy for direct measures of income [4].

Nine variables describing house quality were assessed at each of the three time periods: an overall rating of the type of house; ownership of the house; number of rooms; types of floor, walls, and roof; whether the kitchen is located in a separate area or is part of the house; type of toilet; and the number of major possessions such as a car, radio, or bicycle. Other variables for which data were available were the method of water disposal, the source of water, and the presence of electricity, but these were not used in the assessment of house quality because they tended to be village specific and contributed little to within-village variation.

To generate an index that would capture both within-village variation in house quality and socially meaningful between-village differences, the analytical strategy was a factor analysis to generate factor loadings for each year based on indicators standardized within villages. First, the nine variables were standardized within village to allow for comparability across villages. Second, for each year, the nine standardized variables were factor-analysed. The factor loadings are remarkably consistent across time, which improves their comparability across the three time periods [3]. Finally, to recapture between-village differences, house quality scores were constructed using the original individual raw scores multiplied by the factor loadings.

Parental education

Parental education has been positively related to the cognitive development of offspring in many studies throughout the world. It is positively related to frequency of educational opportunities, verbal stimulation provided to the offspring, and parental aspirations for children.

From the census data, the literacy of mothers and fathers was available at all three periods. In 1967 the code was simply a 0-1 distinction (0 "not literate"; 1 "literate, or some schooling"). In the 1974 and 1987 measures, literacy was evaluated as "not literate" (0), "can read or write a little or with difficulty" (1), and "can read and write" (2). All of these evaluations were reported by the informant; there was no testing to evaluate the accuracy of the informant's opinions.

In 1974 and 1987 the amount of schooling of both the mother and the father were also assessed. The informant indicated how many years of school each person had passed. This number may be very different from number of years attended, since grade repetition is common.

Parental occupation

Occupational status is a "carrier" variable that may be associated with income, community status, resource availability, and family socialization practices. Its indirect effects on the cognitive development of children are thought to occur through the resources available to the children, both stimulation and health, and what the characteristics of the person's occupation reveals about his or her skills. Occupation may also reflect parental availability for child care.

Both mothers' and fathers' occupations were assessed. In 1967 fewer than 5% of the women reported having an occupation, so these data are presented for 1974 and 1987 only. In 1967 only 5 occupations were coded, whereas in 1974 and 1987, 27 categories were recorded, including retired and incapacitated. For purposes of presentation and comparability across time, this scale from 0 to 27 was recoded to be similar to the 1967 scale. The original and the recoded scales are highly correlated (r = .88). The same scale was used for coding all family members' occupations.

Demographic variables

Although not used as SES indicators, demographic variables were used to compare population changes in the villages. They included the mother's report of her marital status, number of pregnancies, number of live births, number of children who had died, and fertility rate. The fertility rate was calculated as the total number of children born per potential childbearing year, estimated as ages 13-45. The total population in each village and the percentage change from 1967 to 1987 were also calculated.


Reliability of socio-economic measures

Table 2 shows the correlations of each SES measure across time periods. Some of these variables should be identical at the various time periods, such as the number of years of school a person had passed, since they reflect a characteristic of childhood. Other variables, such as occupation or house quality, reflect current status and would be more likely to change over time. As the table illustrates, most of these indicators show a high degree of consistency over time. As expected, the smallest correlations are found in the variables that reflect current status, and the largest are for those that should not change.

TABLE 2. Correlations of socio-economic and demographic measures over time

  1967-74 1974-87 1967-87
N Corr N Corr N Corr
House factor 466 .69 533 .50 360 .37
Mother's literacy 429 .72 491 .77 326 .63
Mother's education 488 .80
Father's literacy 380 .82 433 .77 282 .82
Father's education 397 .73
Father's occupation 371 .36 427 .37 282 .27
Mother's occupation 492 .43
No. of pregnancies 472 .69 537 .63 360 .41
Child deaths 439 .79 538 .77 336 .70
Fertility rate 411 .72 481 .71 321 .54

Correlations for literacy and occupation scores are Spearman rank-order; all others are Pearson product moment correlations. All correlations significant at p < .01.
a. Data for 1967 not available.
b. Number of children who had died per mother.

Contents - Previous - Next