Education as policy: the impact of education on marriage, contraception, and fertility in Colombia, Peru, and Bolivia.

Using data from the World Fertility and Demographic and Health Surveys of Colombia, Peru, and Bolivia, we model the effects of education on three demographic outcomes: the timing of first sexual union, contraceptive use, and fertility. These effects are examined over time and across geographic areas using a multivariate framework. We find substantial improvements in female educational attainment over the last fifty years and a strong relationship between education and the demographic outcomes. Each successive increment in education is associated with declines in the marriage rate, increased contraceptive use, and lower fertility. Education accounts for some of the changes over time in the demographic outcomes, but the pattern varies by outcome, time period, and geographic area. In support of the social diffusion hypothesis, our results indicate that educational differences in reproductive behavior are reduced as the level of development increases and societies pass through their demographic transition.

PIP: This study examined the effects of educational attainment on the timing of first union, contraceptive use, and fertility in Bolivia, Peru, and Colombia over the past 50 years. Data were obtained from World Fertility Surveys and Demographic and Health Surveys for Colombia (1976, 1986, and 1990); Peru (1977-78, 1986, and 1991-92); and Bolivia (1989 and 1993-94). Individual level data were used to examine the effects using various multivariate techniques: Cox proportional hazards models for age at first union; logistic models with controls for socioeconomic status for contraceptive use; and log linear techniques for fertility. Findings indicate a strong relationship between each demographic outcome: marriage age, contraceptive use, and fertility. In all countries, education influenced women's individual decisions about family formation. The strongest impact occurs between primary and secondary schooling, especially for marriage age. Increased educational attainment accounted for most of the decline in marriage rates over time. Increased education contributed to an increase in contraceptive use mostly in Peru. In Bolivia and Colombia, contraceptive availability was probably more important in the expansion of contraceptive use over time. Educational attainment over time only accounted for fertility decline in Peru and Colombia, and the effects were smaller than in other studies. Changes within educational categories appear to have contributed more to fertility decline than the expansion of educational opportunities. As countries progress through their transitions, there is a corresponding increase in contraceptive use and a later decline in actual fertility.

[Changes in fertility]. / Cambios en la fecundidad.

PIP: Mexican fertility levels did not change dramatically in the years 1988-97, but maintained a gradual downward trend. In the late 1960s and the entire decade of the 1970s, the fertility decline was spectacular. National fertility surveys beginning in 1976 have provided rich information on fertility trends and their determinants, which along with census data reveal the history of the past 3 decades of fertility change. The 1995 total fertility rate of 2.9 was less than one-half of the 1965 total fertility rate of 7.1. The age pattern of fertility has also changed, with the proportion of births to younger mothers progressively increasing. The average age at childbirth declined from 29.2 in 1965 to 27.4 in 1995. The 1995 age-specific fertility rate for women over 35 was one-fourth that of 1965. Sterilization is the most widely used contraceptive method among women over 30, and, at present, 70% of women over 35 are sterilized. The decline in adolescent fertility is due primarily to the rise in average age at first marriage. The marital fertility rate among women 15-19 years old remains at about 400/1000, a high level explained by the frequency of marriages to legitimize prenuptial conceptions. Comparison of fertility data for 5-year periods shows the impact of contraceptive usage and of the family planning program, especially during 1975-80, when the public family planning program was most active. The age pattern of fertility decline suggests that women used contraception more to limit than to space fertility. Fluctuations in fertility after 1980 appear to be linked to the relaxation of public sector family planning activities. The average number of children born by age 40 declined from 5.68 for women born in 1941-45 to 3.46 for those born in 1955-60.^ieng

[Fertility in Loja, Bolivar, Esmeraldas, Azuay and El Oro. Series: Provincial reports from the Demographic and Maternal-Child Health Survey, 1994 (ENDEMAIN-94)]. / La fecundidad en Loja, Bolivar, Esmeraldas, Azuay y El Oro. Serie: Informes provinciales de ENDEMAIN / 94.

PIP: During the periods 1965-69 and 1989-94, Ecuador's total fertility rate (TFR) declined from nearly 7 to 3.6, while contraceptive prevalence rose from 33.6% to 56.8%. The 15 provinces of Ecuador were divided into 3 equal groups on the basis of Bongaarts' index of contraception, and their fertility characteristics were compared. The groups discussed here showed moderate differences in the index of contraception from the national average of 0.435. 4 of the 5 provinces in group II had indices higher than the national average: Loja, 0.567; Bolivar, 0.566; Esmeraldas, 0.535; and Azuay, 0.526. El Oro's index of 0.398 was lower. El Oro's TFR of 3.46 was lower than the national average of 3.61, but it still was in the third phase of the fertility transition. Although Loja and Bolivar had nearly identical indices of contraception, the TFR in Loja (4.15) was significantly lower than that in Bolivar (5.12). The TFRs were 4.66 in Esmeraldas and 3.95 in Azuay. The differences in TFRs in provinces with similar indices of contraception are due in part to different age patterns of fertility, distributions of women according to educational and marital status, and other factors. Age-specific fertility rates for women aged 15-19 ranged from a low of 65/1000 in Loja to a high of 120 in Esmeraldas. Maternal education was the single greatest observed influence on the TFR. The TFR at the national level was 6.24 for illiterate women and 2.13 for those with higher education. The TFR in the 5 provinces for women with primary or lower education ranged from 6.1 in Esmeraldas and Bolivar to 4.02 in El Oro. The differential between the most and least educated women ranged from 1.36 in El Oro to 4.27 in Bolivar. The proportion of women single ranged from 29.3% in El Oro to 48.6% in Loja. Calculation of the indices of contraception, marriage, and postpartum infertility showed, as expected, that El Oro had the most favorable indices.^ieng

[Fertility in Imbabura, Los Rios, Carchi, Tungurahua and Manabi. Series: Provincial reports from the Demographic and Maternal-Child Health Survey, 1994. (ENDEMAIN / 94)]. / La fecundidad en Imbabura, Los Rios, Carchi, Tungurahua y Manabi. Serie: Informes Provinciales de ENDEMAIN / 94.

PIP: Ecuador's total fertility rate (TFR) declined from nearly 7 to 3.6 during 1965-69, while contraceptive prevalence rose from 33.6% to 56.8%. The provinces of Ecuador were classified into three equal groups based on Bongaarts' index of contraception, and their fertility characteristics were compared. The 5 provinces in group III with relatively homogeneous indices of contraception were examined in this article. The index of contraception was 0.503 in Imbabura, 0.469 in Los Rios, 0.457 in Carchi, 0.456 in Tungurahua, and 0.430 in Manabi. The 5 provinces were all in phase 3 of the demographic transition, with TFRs of 4.44 in Imbabura, 3.68 in Los Rios, 3.78 in Carchi, 3.68 in Tungurahua, and 3.43 in Manabi. The index of contraception and TFR, respectively, were 0.435 and 3.61 for Ecuador. The fertility differentials in provinces with similar levels of contraceptive usage were attributable in large part to differences in age patterns of fertility, maternal educational levels, marital status, and prevalence of contraceptive methods. Age-specific fertility rates for women aged 15-19 ranged from 76/1000 in Imbabura to 114/1000 in Los Rios and Carchi. The difference between the average number of children of illiterate women and that of women with higher education ranged from 1.66 children in Manabi and 1.69 in Carchi to 2.97 in Imbabura, compared to the difference of 2.44 for Ecuador. The proportion of women single ranged from 30.4% in Los Rios to 39.9% in Carchi. Contraceptive prevalence in the 5 provinces ranged from 50.9% in Imbabura to 56.9% in Tungurahua. The differences were reflected in Bongaarts' indices of nuptiality and postpartum infertility as well as contraception.^ieng

Using parity-progression ratios to estimate the effect of female sterilization on fertility.

In this article, a new methodology that employs parity-progression ratios to estimate the effect of female sterilization on fertility is described, and results using data from Ecuador are compared to those obtained using a previously existing approach that classifies women by marital duration. The methods differ in how they disaggregate marital fertility and in the assumption they make about what the subsequent fertility of sterilized women would have been if they had not been sterilized. The analysis of the Ecuadoran data shows that the estimate of births averted by sterilization has diminished over time, even as sterilization prevalence has been increasing. This situation is attributed to a decline in the fertility of nonsterilized women resulting from increased use of reversible methods of contraception.

PIP: Two methods are used to show the impact on fertility of sterilization: births averted in a single five-year period before the survey and the parity progression ratio approach for 1979-84 and 1984-89 used to estimate births averted by birth order and duration since first union. The births averted analysis revealed that fertility among nonsterilized women declined from 5.93 in 1979-84 to 5.01 in 1984-89. The births averted approach showed almost no change in total births averted and a shift in the distribution of births averted by marital duration. Births averted were greater during 1984-89 among marital durations under 15 years and during 1979-84 at marital durations of 15-24 years. The actual percentage of ever married women who were sterilized increased over the decade, but the hypothetical proportions being sterilized in a given duration remained almost unchanged. This approach underestimated births averted, because it did not account for future declines in fertility among nonsterilized women. The parity progression ratio (PPR) approach indicated a nonsterilized fertility decline of 1.28 births per woman and a marital fertility decline of 1.04 births. Between the two periods, sterilization contributed to a reduction of 0.24 births. The preferred PPR approach accounted for the effects of sterilization on individual PPRs and proportions of ever-married women. This method offered consideration of increased sterilization prevalence, but declining births averted by sterilization. A comparison of both methods shows that the parity progression approach yielded higher estimates of the average number of births averted by sterilization and higher total nonsterilized marital fertility rates (total marital fertility rates in the absence of sterilization). Marital fertility rates remained similar. Data are obtained from the 1989 Ecuador Demographic and Health Survey and the World Fertility Surveys of 1979 and 1989.

[The nature of the preceding birth technique]. / La naturaleza del metodo del hijo previo.

PIP: The author applies the preceding birth technique of infant mortality estimation to World Fertility Survey birth history data from a number of developing countries. (SUMMARY IN ENG)^ieng

[Unwanted fertility in Mexico]. / La fecundidad no deseada en Mexico.

PIP: Data from the 1987 National Health and Fertility Survey were used to explore trends and differentials in unwanted fertility in Mexico. Women were classified in three educational categories: illiterate or incomplete primary, complete primary or some secondary, and complete secondary or more. The four urbanization categories were: under 2500; 2500 to 19,999; 20,000 and more; and metropolitan areas. The place of primary socialization was the place of residence until age 12. Two categories of employment of household head were included, agricultural or nonagricultural. The categories for mother's employment considered employment until the union and employment between the time of union and the birth of the first child. The proportion of births that were unwanted was estimated by calculating the total fertility rate for the year before the survey and comparing it to desired family size. The "total desired family size" was 2.4 for the entire sample. In other words, nearly 37% of the total fertility rate was undesired. The greatest differentials in percentages of undesired fertility were found in educational categories. 40.7% of the total fertility rate in the least educated groups vs. 13.6% in the most educated groups was undesired. Differentials in the other categories ranged from 1.9 for a history of employment before the birth of the first child to 8.8 for agricultural or nonagricultural employment of the household head. When effects of education were controlled, the percentage of undesired fertility declined as educational level increased for all rural or urban residence categories, places of socialization, and employment groups except women who worked before the first child was born. Women with more education, urban residence, and with histories of employment had the lowest levels of undesired fertility.^ieng

[Decline of infant and child mortality in the Dominican Republic: an exploratory investigation on changes in the structure and composition of socio-demographic determinants during the mortality transition]. / Descenso de la mortalidad infantil en la Republica Dominicana: una investigacion exploratoria sobre cambios en la estructura y composicion de los determinantes socio-demograficos durante la transicion de la mortalidad.

PIP: Results of the Dominican Republic's four most important fertility surveys were used to analyze how relations between selected sociodemographic factors and infant mortality have changed in recent decades, and to evaluate the contribution of changes in the distribution of these factors to mortality decline. The empirical analysis was carried out in four parts, the first of which analyzed trends in infant mortality in the Dominican Republic between 1960 and 1990. The second part established the importance of each determinant in each period studied and used logistic regression to analyze changes in the structure of the relationship between the variables and infant mortality. The third section analyzed changes in the relative weight of the categories of determinants during the study period. The final section standardized the risk of death in 1964-69 using the composition of births in 1985-90 and vice versa to evaluate the influence of each determinant in the infant mortality decline. The number of study variables was limited to six by lack of comparability in the four fertility surveys. The six variables were region and zone of residence, education and age of the mother, and birth order and interval. The results of this research did not indicate any clear pattern of change in the structure of relations between determinants and infant mortality. It appears plausible that there are various routes toward the transition leading to low mortality. In the Dominican Republic, maternal education was most important from 1964 to 1969, when infant mortality was high, and became relatively less important as a determinant in subsequent years. The changing relationship between maternal education and infant mortality suggests that there are no generalized patterns but rather the weight of each variable differs from one context to another and depends on the influence of other socioeconomic, medical, and demographic determinants.^ieng

[Spacing and limiting components of the fertility transition in Latin America]. / Los componentes de espaciamiento y limitacion de la transicion de la fecundidad en America Latina.

PIP: Trends in marital fertility according to rural or urban residence, wife's educational level, and husband's occupation are analyzed for 6 Latin American countries which participated in both the World Fertility Survey and Demographic and Health Surveys. The countries were Colombia, Dominican Republic, Ecuador, Mexico, Peru, and Trinidad and Tobago. The principal methodological tool for this analysis is a statistical model of period marital fertility which expresses fertility as a function of spacing, which is assumed to operate equally in all durations of union, and of limiting, which increases in importance as the duration of union increases. The model permits a summary description of the levels and patterns of marital fertility and yields parameters that may be interpreted in terms of basic behavioral mechanisms, such as lactation and contraception. Total marital fertility in the 6 countries decreased in the recent past, with the magnitude of decline varying from .4 births/woman in Trinidad and Tobago to 2.2 in Mexico. The results indicate that the transition originated in an educated urban minority and has spread to almost all strata studied. Most of the observed fertility decline resulted from birth limitation, but spacing played a surprisingly large role. Despite the diversity of conditions in the 6 countries, the indices of spacing and limiting in the different social strata appeared to have followed a single pattern of increase over time. Although the trajectory followed by the indices of spacing and limiting is sufficiently broad to accomodate substantial differences between the countries, it is well defined, indicating that a common explanation exists. It is demonstrated that the pattern of increase is consistent with a simple mathematical model of social diffusion.^ieng

Contraceptive use and sterilization among Puerto Rican women.

A comparison of contraceptive use in the early to mid-1980s among married Puerto Rican women aged 15-49 in the New York City area reveals that island-born Puerto Rican women living in New York rely on female sterilization to nearly the same extent as do women living in Puerto Rico (45% and 41%, respectively) and that mainland-born Puerto Rican women use sterilization as much as do all women in the United States (19% for both groups). Puerto Rican women in New York use reversible methods to a greater extent than do women in Puerto Rico (22% v. 16%), but to a lesser extent than do all women in the United States (37%). Although mainland-born Puerto Rican women in New York use reversible methods more than do island-born women in New York (42% v. 23%), they tend not to adopt these methods to the same extent as do all U.S. women during the early reproductive years, when education and employment are critical to socioeconomic attainment.

PIP: Researchers compared data from 1985 on 1998 15-49 year old Puerto Rican women living in the greater New York City area with data from 1982 on 3174 15-49 year old women living in Puerto Rico and with data from 1982 on 7969 15-44 year old women living in the US to examine their contraceptive practices. Women who were born in Puerto Rico and later lived in New York City had almost the same sterilization rate as those women still living in Puerto Rico (44.7% vs. 40.5%). On the other hand, about the same proportion of Puerto Rican women born and raised in New York City used reversible methods (42.1%), especially oral contraceptives (18.5%) and the IUD (11%). Only 18.5% of the US-born Puerto Rican women had undergone female sterilization which basically equalled that for all US women (18.7%). A higher proportion of Puerto Ricans in New York City used reversible methods than did those in Puerto Rico (21.7% vs. 15.7%), but not as high a proportion as all US women (36.8%). US-born Puerto Ricans did not accept reversible methods as early in the reproductive years as did all US women (42.2% vs. 58.4% for 15-24 year olds). Puerto Rican women, regardless of residence or place of birth, reported fewer male sterilizations and less condom use by their partners than all US women (0% male sterilizations for all Puerto Rican women in New York City, 4.6% for those in Puerto Rico vs. 10.8% and 2.8% of partners using condoms for island-born Puerto Ricans, 5.1% for US-born Puerto Ricans, and 4.4% for those in Puerto Rico vs. 9.8%). These findings on partner responsibility for contraception may reflect the cultural definition of women as rearers of children. The researchers hoped that the results of this descriptive study would motivate others to conduct further research to determine socioeconomic correlates of contraceptive practice and cultural and religious variables.

A comparison of the fertility of Dominican, Puerto Rican and mainland Puerto Rican adolescents.

Data from three fertility surveys are used to examined the probabilities and determinants of adolescent births among Dominican and Puerto Rican women. Young women in the Dominican Republic are the most likely to have had a child by each year of age from 14 through 24, followed by young women on the Island of Puerto Rico; the probability of an early birth is lowest for Puerto Rican women on the U.S. mainland. Eighteen percent of Dominican women have had a child before their 18th birthday, compared with 13% of women living in Puerto Rico, and 10% of Puerto Rican women in metropolitan New York. The cumulative probabilities that Puerto Rican women will have borne a child before their 20th birthday are almost identical, whether the women live on the island or the U.S. mainland, but the difference between Puerto Rican and Dominican women widens. The order is reversed, however, in the analysis of premarital births: The probability of a premarital birth during adolescence is highest for Puerto Rican women in New York, and lowest for Dominican women. In a separate logistic regression analysis, education and age at first sexual intercourse are shown to be important determinants of adolescent fertility in all three populations.

PIP: Researchers analyzed 1980s data on female adolescents including Puerto Ricans in the New York metropolitan area, those in Puerto Rico, and Dominicans in the Dominican Republic to determine the probabilities and determinants of adolescent births among these women. Female adolescents from the Dominican Republic had the highest probability of having delivered an infant by each year from 14-24 years followed by those in Puerto Rico and then Puerto Rican immigrants in the US. For example, the cumulative hazard rate at age 24 was .65 for the Dominicans compared to 0.5 for Puerto Ricans in Puerto Rico and in the US. Moreover 18% of Dominicans delivered an infant before 18 years old whereas this figure for the 2 Puerto Rican groups was 13% and 10% respectively. Further the cumulative probabilities that Puerto Rican female adolescents will have had a child before 20 years old were essentially equal (37.4% and 37.3% respectively) compared to 56.1% for the Dominicans. On the other hand, the order was opposite for premarital births. Puerto Rican women living in metropolitan New York were at greatest risk of having a premarital birth (cumulative hazard rate at age 24=.2) then those living in Puerto Rico (.1) followed by Dominican women (.05). Age at 1st intercourse was the strongest determinant of adolescent birth in all 3 countries (p=.001). Indeed the younger the adolescent was at 1st intercourse, the greater the likelihood she would deliver a child while an adolescent. Another strong determinant of adolescent birth was dropping out of school (p=.001 in US and Puerto Rico; p=.01 in the Dominican Republic). The fact that most women did not use contraceptives before 1st birth was only significant for the women not living in the US (p=.001). Further research is needed to identify regional and cohort differences in contraceptive use and attitudes toward early sexual activity and childbearing.

[Fertility trends in Haiti]. / Tendances de la fecondite en Haiti.

PIP: 5 surveys and the 1971 and 1981 censuses have provided differing and apparently contradictory assessments of fertility levels in Haiti over the past 2 decades. This work surveys the published fertility estimates to ascertain whether any trends are discernable and analyzes to what extent variations in total fertility rates can be explained by variations in proximate fertility determinants. 3 methods of data collection were used in the surveys. The 1971-73 survey was conducted in multiple rounds. The 1977 Haiti Fertility Survey, the 1987 Survey of Mortality, Morbidity, and Service Utilization, and the 1989 Contraceptive Prevalence Survey involved partial or total reconstruction of the respondents' fertility histories. The 1983 Contraceptive Prevalence Survey only included the date of the most recent birth. The data corresponding to the 1970s suggest an overall fertility decline from 6 children per women in the late 1960s to slightly over 5 in the mid-1970s. Results of the 1983 and 1987 surveys, on the other hand, indicate a fertility increase, with total fertility rates of 6.2 for 1982-83, 6.3 for 1982-86, and 7.0 for the first half of 1987. The 1989 survey indicated a total fertility rate of 4.6 for 1984-89. Two somewhat different conclusions may be drawn from the data. The first is that, taking into account the variable quality and precision of the data, Haitian total fertility rates appear to have remained between 5.5 and 6.5 during the past 20 years. This conclusion is supported by the fact that almost all of the observed and adjusted total fertility rates fall within the range of 5.5-6.5. The second possible conclusion is that after a decline from around 6 in the late 1960s to 5.5 in the mid-1970s, the total fertility rate increased again to over 6 in the early 1980s. The linear adjustment of observed and adjusted results of the 5 surveys lends support to this conclusion. Analysis of results of the 1977, 1983, and 1987 surveys using Brass's method tends to corroborate this conclusion, as does analysis of the proximate fertility determinants using Bongaarts' method. It is likely that fertility increased from the mid-1970s to the mid-1980s, to decline slightly to a level equal to or under 6 children/woman by 1989. Control of reproduction in Haiti appears to be achieved through variations in the frequency and stability of unions and in the duration of breast feeding. Recent increases in contraceptive usage revealed by the 1989 survey may signal the beginning of a transition and new modes of fertility control.^ieng

Birth spacing and infant mortality in Brazil.

The effects of birth spacing on neonatal and post-neonatal mortality in Brazil were found to be very consistent with models based on data from other South American countries. The model for neonatal mortality simplified to three significant variables, whereas the model for post-neonatal mortality included four significant interactions.

PIP: Researchers used 1986 data from the Brazil Demographic and Health Survey to determine the effects of birth spacing on neonatal and post neonatal mortality then compared the results with other countries (World Fertility Survey). 1st births had a higher relative risk (RR) of neonatal death (1.49) than births of order 203 Brazil. A similar pattern existed for Peru, Ecuador, Costa Rica, Colombia, and Guyana. In Brazil, 4-6 birth order had the lowest risk (.62). High order births (7+) for all the countries had an RR almost the same as the baseline 2-3 birth order class. Birth order was significant (p.01). Post neonatal mortality for 1st births in Brazil was more favorable than other countries, but the RR for high order births was less favorable. The RR of neonatal mortality declined with mother's education for Brazil (p.05) as well as it did for Peru, Ecuador, Colombia, and Costa Rica. It did not fall in Guyana, however, where it was elevated (1.19) for mothers with 4-6 years of schooling. The positive effect of maternal education was even stronger for post neonatal mortality in Brazil (.62 for 4.6 years and .27 for 7+ years). The effect was basically the same for all the countries, but mothers with 4-6 years of education in Guyana (1.03) had slightly higher post neonatal mortality. The RR of neonatal mortality for the index child if there were 1 surviving sibling born 0-2 years earlier was only slightly increased (1.25) in Brazil, but if there were 1 dead sibling born 0-2 years earlier, RR stood at 2.4 and was significant. This RR rose further if 2 or more children were born in this interval (3.71). For infants in Brazil who had 2+ siblings born 2-4 years earlier, the RR of neonatal mortality was elevated (1.83). Indeed Brazil had higher neonatal mortality for these children than did the other countries. Birth spacing was highly significant (p.001).

Can women remember how many children they have borne? Data from the east Caribbean.

The number of births reported by 786 women of native ancestry on the 1954 census of St. Barthélémy, French West Indies, was compared to the number of births registered to these same women in the records of vital events. The magnitude of the difference between the two numbers was related to a woman's parity and marital status, but not to her age or literacy. In general, the agreement between the two numbers was very high, with the mean of reported parities within 5 per cent of the mean of registered births for all age groups except one, where the difference was 5.6 per cent. The women of St. Barthélémy were well able to remember the number of children they had borne. It appears likely that the discrepancies between reported parities and registered statistics in previous studies are the result of cultural misunderstandings in questionnaires, or perhaps sample decay, rather than of women's memory loss.

PIP: A study of women's ability to recall the number of children ever borne found a very high level of self-reporting accuracy. 786 native women of St. Barthelemy, West Indies, resident for the 1954 census, were specifically under consideration. These women were aged 15 years and older after the French began keeping vital records in 1878. Their recall of the number of children borne was compared to the extremely accurate number of births registered in records of vital events. A mean 2.00 children/woman were registered, with standard deviation of 2.81, and a range of 0-17 live births. In general, women were well able to remember how many children they had borne, with very high agreement observed between the 2 measures. 5.68% of respondents made reporting errors of any kind, and the average difference between reported and recorded live births was 0.031 birth/woman. SPSS-X regression was then employed to examine the effects of woman's age, marital status, literacy status, and number of live children registered borne on the magnitude of difference between the 2 figures. The difference was significantly related to the number of live children borne and marital status, but not age or literacy. Those of high parity and those unmarried were more likely to make omissions when controlling the other variables. Study results suggest that linguistic and cultural misunderstandings, and/or sample decay may be more important sources of error than memory lapses. These former sources may or may not be easier to correct.

Survey report: El Salvador.

PIP: Among the recent studies documenting trends in fertility and contraceptive behavior in El Salvador is the 1988 Family Health Survey (FESAL88). Conducted by the Asociacion Demografica Salvadorena with technical assistance from the US Centers for Disease Control, Division of Reproductive Health, FESAL88 (between May-July 1988) interviewed a total of 3579 women, ages 15-44. Total fertility declined from 6.3 to 4.5 lifetime births/woman between 1978-85, but no significant change was found between 1985-88. Similarly, contraceptive use increased by 26 percentage points between 1975-88, with a slight increase occurring after 1985. Approximately 47% of married women use modern methods of family planning, with female sterilization used by 1/3 of all married women. About 1/2 the women currently not using family planning say they intend to do so, but only 1/2 of these women know where to obtain a method. Other determinants of fertility also showed little change between 1985-88. In these years, the average age of marriage rose from 19.8 to 20.3 years, but almost all of the increase occurred in urban areas. 95% of women interviewed breast fed their last child; the average duration of breast feeding was 17 months. In focusing on Salvadorian women 15-24 years of age, FESAL88 found that their fertility rate was 139 birth/1000 women, the highest rate for any country in Latin America or the Caribbean. 9 in 10 women age 15-24 think there should be special family planning services for young adults. Finally, among the married women who had a birth in the 5 years prior to the survey, 2/3 received a prenatal examination, and 2/3 received an antitetanus toxoid injection. Even though only 40% of married women with a new baby report receiving postnatal care, 76% of infants received well-baby care.^ieng

[Contraceptive practice in Mexico: two five-year periods, two different patterns (1976-1977 to 1987)]. / La practica anticonceptiva en Mexico: dos quinquenios, dos patrones diferentes (1976-1977 a 1987).

PIP: This work uses data from the 1976-77 Mexican Fertility Survey, the 1982 National Demographic Survey, and the 1987 National Fertility and Health Survey to analyze fertility and contraceptive behavior among Mexican women in the 2 5-year periods from 1976-82 and 1982-87. The proportion of fertile-aged women in union using a method increased from 41.6% in 1976-77 to 58.8% in 1982 and 61.4% in 1989. The most significant gains in use of the more effective methods as in contraceptive prevalence occurred in the 1st 5-year period. The proportion of women sterilized increased from 3.73% in 1977 to 16.6% in 1982 and 21.7% in 1987, equivalent to increases over the 2 5-year periods of 345.7% and 30.7%. Female sterilization was the only method whose use increased constantly over the period. Use of the IUD was almost unchanged between 1977-82, while use of oral contraceptives and injectables both declined from 1982 to 1987 after increasing over the preceding 5 years. Contraceptive usage increased intensely in all subgroups of women studied between 1976-82, but in the 2nd period from 1982 to 1987 contraceptive usage declined significantly in some subgroups. Contraceptive prevalence among women in urban and metropolitan areas increased systematically after 1977. Rural women increased their use from 22.5% in 1977 to 48.6% in 1982. 5 years later the proportion of rural users had declined to 45.3% although it was still well above the level of 1976-77. The most significant increase in contraceptive usage among illiterate women and those with primary educations occurred from 1976 to 1982. In 1987 women with primary and higher levels of education demonstrated increases in contraceptive usage compared with 1982, but contraceptive usage among illiterate women declined from 40.8% in 1982 to 28.9% in 1987. Data from the 3 surveys demonstrate low but increasing rates of use among women 15-19 years old, who may be in the process of developing a new fertility pattern characterized by longer birth intervals. Women in the middle reproductive years had higher rates of use showing little change from 1982 to 1987 and unlikely to show great change in the future due to their already high level. Women aged 40-44 showed systematic increases in use from 1976 to 1987, but women aged 45-49 increased their use only from 32.6% in 1977 to 39.0% in 1987. The declines in age differentials in contraceptive prevalence from 1982 to 1987 demonstrate that age is losing weight as an explanatory factor. Between 1976-77 and 1982, women of all parities increased their contraceptive usage substantially. Women with 3 or fewer children continued to increase their usage through 1987, but usage among women with 4 or more children actually declined between 1982-87. The results suggest that a small proportion of rural, less educated, and high parity women may have returned to their former patterns of uncontrolled fertility. Nevertheless, fertility differentials were much smaller in 1987 than in 1976-77.^ieng

Mexico.

PIP: Results of Mexico's 1987 National Survey of Fertility and Health (ENFES) shows significant changes in total fertility rates (TFR) and contraceptive prevalence rates. These changes are due i large part to the institutionalization of a population policy enacted in 1972 that has continued to receive strong support from the government. The TFR declined from 6.3 to 3.8 with urban rates falling 50% and rural rates 3/4. Between 1976-86 use of modern contraception doubled, going from 23-45%. Use of the pill declined while female sterilization increased for 9-36%; IUD's remained the 2nd most popular method at 18%. Contraceptive prevalence rates mirror changes in desired family sizes; women between 15-19 now desire 2.6 children while women at the end of their reproductive cycle expect to have 4. Infant mortality rates dropped from 85 to 47/1000 between 1970 and 1987. 62% of illiterate women wish to stop childbearing as compared with 49% of women with secondary schooling. This difference is related to differences in the ages of the 2 groups; as education has spread, women without any schooling tend to be older and have higher parity; and in spite of wanting to stop childbearing, they are 10 times less likely to use contraception than their more educated counterparts. 67% of the women interviewed received prenatal care from a doctor, with higher rates among the urban population. Between 80-90% of women breastfed their children, with higher rates among the rural poor.^ieng

[Fertility Survey, Metropolitana region, Chile 1989]. / Encuesta de Fecundidad, Region Metropolitana, Chile 1989.

PIP: A fertility study was carried out by the Chilean Association for Protection of the Family (APROFA) in metropolitan Chile in 1989 to update data from the most recent fertility study in 1974. A random and self-weighted sample of 600 women aged 15-44 in 32 urban communes of the metropolitan region were interviewed in November 1989. 55% were married or in stable consensual unions, 39.1% were single, and 5.9% were separated, divorced, or widowed. 24.1% had primary educations, 15.5% had some form of higher education, and .2% were illiterate. 64.2% of the women had children, of whom 63.8% had 1 or 2. Only 2.1% had 6 or more children. 46.2% felt that the ideal family size was 1-2 children and 30.7% that it was 3 children. 40.4% of women with children wanted another child and 59.6% did not. 25.1% of the sample had never had sexual relations, 7.3% had not had sexual relations in the 12 months prior to the interview, and the remaining 67.6% were sexually active. At the time of the survey 6.1% were pregnant and 11.1% had had a birth in the preceding 12 months. 28.3% of these pregnancies were considered unwanted, usually for economic reasons or because the family was considered complete. 55.6% of the sample reported using contraceptives. 48.6% used IUDs, 26.0% used oral contraceptives, 6.9% periodic abstinence methods, and 1.8% barrier methods. Only 7.5% of women aged 15-19 used contraception. Rates of use stabilized after age 25 at about 72%. 80.9% of women married or in union used a method, as did 20.2% of single and 59.3% of separated or divorced women. The data for the survey are still under analysis and all statistics are provisional.^ieng

[Fertility and contraception in the rural areas. Important changes in contraception]. / Fecundidad y anticoncepcion en el campo. Cambios importantes en anticoncepcion.

PIP: National fertility surveys conducted in Mexico since 1969 indicate that fertility in rural areas (localities with under 2500 inhabitants) is declining. The total fertility rate of rural women in union aged 20-44 years declined from 8.2 in 1968 to 5.8 in 1985. The beginning of the steep decline in rural fertility apparently occurred between 1973-75, at a time when official family planing programs were just starting, especially those accessible to rural women. Knowledge of contraception spread rapidly between 1969-76. The number of women in union who had ever used contraception doubled between 1976-81 to 40.2%. Important rural health programs developed in the late 1970s may have influenced contraceptive usage. Fertility continued to decline between 1980-85. The proportion of women who had ever used contraception in 1985 increased by only 3.5% in comparison to the 1981 level, but by 1987 the use of traditional methods such as rhythm and withdrawal was almost nil, and the use of IUDs, injectables, and sterilization had expanded. The age specific fertility rates suggest that the fertility reduction between 1980 and 1985 resulted in large part from the behavior of women over 34.^ieng

Fertility change in Costa Rica 1960-84: analysis of retrospective lifetime reproductive histories.

Lifetime reproductive histories of a 1984-85 nationally representative sample of 870 women aged 25-59 years provided data to describe the evolution of fertility, contraception, breast-feeding, and natural fecundability in Costa Rica between 1960 and 1984. The contraceptive prevalence rate increased from 23% in 1965 to 58% in 1975 and 66% in 1984. Duration of breast-feeding was stable during the 1960s, decreased in the early 1970s, and increased after about 1976. Fecundability among women who did not practise contraception was lower than expected and declined between 1960 and 1975, probably because of selection effects. Despite a high consistency between estimations from the reproductive histories and other sources of data, some suggestion of omissions of short periods of contraceptive use in the distant past was detected. The survey may have reduced recall errors by using a calendar that summarizes major life events together. The analysis demonstrated the feasibility and usefulness of asking for lifetime reproductive histories in fertility surveys in developing countries.

PIP: Lifetime reproductive histories of a 1984 through 1985 nationally representative sample of 870 women aged 25 through 59 years provided data to describe the evolution of fertility, contraception, breast feeding, and natural fecundity in Costa Rica between 1969 and 1984. The contraceptive prevalence rate increased from 23% in 1965 to 58% in 1975 and 66% in 1984. Duration of breastfeeding was stable during the 1960s, decreased in the early 1970s, and increased after about 1976. Fecundability among women who did not practice contraception was lower than expected and declined between 1960 and 1975, probably because of the selection effects. Despite a high consistency between estimations from the reproductive histories and other sources of data, some suggestion of omissions of short periods of contraceptive use in the distant past was detected. The survey may have reduced recall errors by using a calendar that summarizes major life events together. The analysis demonstrated the feasibility and usefulness of asking for lifetime reproductive histories in fertility surveys in developing countries. Retrospective information gathered on 900 women permitted reconstruction of the results in a a period of dramatic changes. However, Costa Rica may be somewhat exceptional among developing countries because of its high levels of literacy. Some of the few, older, illiterate women may have found a life history calendar difficult to decipher, but others seemed to grasp the approach as easily as did the literate men.