An average woman worldwide now has fewer than half as many children as in 1950 (Bloom et al. 2025). This sharp decline in fertility rates has attracted much recent attention from researchers and policymakers as well as the general public. In many low and middle-income countries, where robust birth registry systems are not always available, these fertility estimates come from demographic survey data. These surveys typically interview “eligible women”, usually ever-married women aged 15 to 49 years in a randomly selected sample of households, and ask them to report all live births. These data are then statistically extrapolated to estimate birth rates for the broader population.

Fertility rates and errors of omission 

The accuracy of these estimates depends on whether respondents provide complete birth histories. However, several factors can lead to omitted births. Birth-history questions are often just one component of a long questionnaire that includes many other topics. Depending on where the question appears, respondents may become tired and skip reporting some births. This may be especially true in low- and middle-income countries where women have a relatively higher average number of children than women in higher-income countries. 

Respondents may also misinterpret the birth history question. For example, rather than reporting all live births, the respondent may think she only needs to report children who are still alive. Or she may feel that she needs to report only children who live with her currently, thus omitting daughters who have married and moved away.

Questions about deceased children can be emotionally sensitive. Some mothers may have suppressed those painful memories, and even if they remember, they may not want to relive them by talking about them.

Collecting birth histories, therefore, is more complex than meets the eye. It also takes a well-trained interviewer who understands the intent of the birth history questions, administers them carefully, and is attentive enough to probe the respondent when needed. 

Estimating omitted births: the role of sex and survival status

Given these issues, in a recent article we estimated the proportion of missing births and the factors associated with those omissions (Sharma et al., 2024). We used data from the India Human Development Survey (IHDS). The IHDS is a large-scale panel survey that follows the same households over time, allowing us to compare births reported by the same respondent across two waves. 

We tracked births reported by the same 25,474 women in the first and second survey waves, conducted in 2004–2005 and 2011–2012, respectively. Ideally, all 73,408 births reported in the first wave should also have been reported in the second wave. But this was not the case. We found that 2,913 births (4% of births reported in wave 1) were not reported in wave 2. Moreover, these were not random omissions; they followed a clear pattern, as shown in Figure 1. 

Figure 1: Survival status and sex of children reported in wave 1 (2004–2005) but not reported in wave 2 (2011–2012) of the India Human Development Survey

Percentages add up to 100%. Base: 2,913 births reported in wave 1 but not in wave 2.
Source: Sharma et al. (2024).

Approximately three-quarters of all omitted births were children reported as dead (solid areas in Figure 1) in wave 1, with male and female children contributing almost equally to the omissions. However, among the 27% of children who were reported alive in wave 1 (horizontal-line areas in Figure 1), girls were disproportionately more likely to be missing in wave 2.

Looking at these data from another angle, Table 1 shows the stark contrast in omission rates by children’s survival status recorded in wave 1. Nearly one-third of children reported as deceased in wave 1 were not reported in wave 2. In contrast, only a little more than 1% of children reported alive in wave 1 were omitted in wave 2.

Table 1: Births omitted in wave 2, by the child’s survival status recorded in wave 1 (India, 2004–2005 and 2011–2012)

Source: Sharma et al. (2024).

What are the probable factors behind births omissions? 

Figure 1 and Table 1 clearly show that mothers are much more likely to omit dead children. This is understandable, given the emotional sensitivity of these memories and the difficulty of asking and answering questions about deceased children.

To explore other factors associated with omitted births, we analyzed relationships between omissions and several geographic, maternal, child, and interviewer-level characteristics. Table 2 displays an excerpt of the results from this analysis, conducted separately for omissions among children who had died and children who were alive. What emerges is that birth omissions are a complex function of several survey variables.

Table 2: Relationship of different characteristics with birth omissions, analyzed separately for deceased and alive children 

Notes: ‒ : no evidence of an association of the variable with birth omission or no clear story.
°:  Child’s age for dead children is the purported child age at the time of interview.
Source: Sharma et al. (2024).

What can we do

Birth omissions are large enough to matter. Given the importance of fertility estimates to public debate and policy making, it is critical to ensure that demographic surveys collect accurate birth history data. 

Our analysis shows that birth omissions do not occur at random. This is, in fact, good news: if we can identify which births are most likely to be omitted, we can improve how fertility data are collected. For example, since we know that dead children have an elevated risk of omission (especially in urban areas), interviewers should be trained to administer these questions with care, to be aware of cues that suggest that the respondent is hesitant, and then probe such respondents sensitively. With the availability of Computer-Assisted Personal Interviewing (CAPI), survey instruments can be programmed to dynamically alert interviewers to those combinations of characteristics that are prone to omissions. 

Respondents need to be reminded that the survey is collecting data on all live births and not only on the children who are currently living with her. The data also show that interviewers with a higher workload tend to have lower omissions, both for alive and dead children. Whether this is because IHDS recruited better interviewers as the survey proceeded or because interviewers became better at collecting fertility data as they gained survey experience is a topic of future research.

References

Bloom, D.E., Kuhn, M., Prettner, K. (2025). The Debate over Falling Fertility. Finance & Development Magazine, June 2025, International Monetary Fund,  Washington, DC, USA. Accessed at: https://www.imf.org/en/publications/fandd/issues/2025/06/the-debate-over-falling-fertility-david-bloom 

Sharma, S., Desai, S., Barik, D., & Sharma, O. P. (2024). Gone and forgotten? Predictors of birth history omissions in India. Demographic Research, 50, 929.

The post Gone and forgotten? Omissions of birth histories in an Indian survey appeared first on N-IUSSP.