4.1. Summary of Findings
The enumeration methods described in this report are used to conduct national censuses around the world, including the Nigeria 2006 Population and Housing Census. Previous studies used mapping infrastructure (such as Digital Globe [9
]) to develop public health maps for advocacy, training, and emergency response [26
]. Others combine geolocated cluster survey data with geospatial layers to predict target population proportions and coverage of public health interventions [28
]. The novel aspect of this activity was the combination of satellite imagery, geocoding methods, and navigational tools to develop a precise micro plan and guide the enumeration teams in the field.
The first objective of this micro-census was to obtain an accurate count of the total population living in Magarya ward, in Sokoto state. We counted 8601 persons in the ward. Of these, 240 (2.8%) are children younger than 12 months and 1599 (18.6%) are younger than five years. The age heaping we noticed around “round” numbers might be related to the uncertainty of precise age by many in the population [30
]. The sex distribution was roughly equal among men and women at all ages. The only exception was observed for men in the 20–40 year age group, who seemed to be under-represented. This absence might be consistent with labor migration outside the ward. As this is a de facto enumeration, persons not present on the day of the micro-census were not counted.
In each household, the average number of members was 7.2. The average number of children per household head was 3.0. Among the 1704 households, 1378 (80.9%) household heads describe themselves as monogamous and 237 (13.9%) as polygamous. However, approximately 25% of household heads reported having two or more wives, suggesting that there may have been some errors in the classification of household type. About 15% of household heads were female. This categorization is likely because of the absence of the male household head at the time of the enumeration, rather than the recognition of a woman as head of the household.
Finally, the distribution of households across buildings and dwellings/flats suggests the following: (a) almost half of the buildings in the ward were uninhabited; (b) inhabited buildings contained multiple dwellings/flats, some of which were not occupied; and (c) many of the dwellings/flats might have been occupied by more than one household.
The second objective was to compare the results of the micro-census with those from other sources of denominator data. In all comparisons, we found our number and proportion of children younger than 12 months and younger than five years to be smaller compared with other estimations. Multiple interpretations explain this discrepancy. The micro-census might have counted fewer persons in each age group compared with other enumeration exercises because we strived to avoid duplicate counts. Also, the uncertainty around the reported age—especially among younger children—might have contributed to the creation of imprecise age groups that make comparison between data sources difficult. It is possible that some children younger than 12 months were reported as being older (Pullum, T. The Demographic and Health Surveys Program, ICF, Rockville, MD, USA. Personal communication, 2018). However, as more people were added to each age group, the discrepancies seemed to smoothen. Another possibility is that variations in age trends within and across wards might be greater than those estimated by the GeoPoDe exercise. The use of assumed fixed proportions to estimate age group size does not reflect the reality in the field, and inflates the count of infants in some settlements.
When interpreting our findings, we must consider the shape of Magarya ward. Its boundaries are irregular, and the southern portion is enmeshed with Marafa ward, where the LGA capital Wurno is located (see Figure 2
). There are no physical barriers between the two wards (e.g., river, main road), so enumeration teams may easily overstep the established boundaries if the micro plan is not exact. This may lead to inflated results for other enumeration methods. This interpretation gains strength when we compare our results to those revised from GeoPoDe. These estimates were calculated using the exact same boundaries as the micro-census, using the geo-coordinates of each building visited. As shown in Table 3
, our total population count is only 107 persons fewer (1% of the micro-census population). This comparison underlines the need for standardized boundary shape files, so each enumeration exercise observes the exact same area and results are analogous. Our precise micro plan also minimized the chance of duplicate or erroneous counts. Because no other enumeration methodology achieved this level of precision, we would expect our count to be lower.
4.3. Strengths and Limitations
Planning: This micro-census required five months of planning (February through June 2018) and seven days for implementation (11–17 July, 2018). This long planning period allowed us to include key best practices that strengthen our implementation operations in the field.
Involvement of state and local authorities: The early engagement of officials from the Sokoto SERICC, the Sokoto EOC, and the LGA- and ward-level authorities ensured their approval and support when conducting trainings, field preparations, and the enumeration itself.
Involvement of community leaders: The collaboration of community leaders ensured that residents could be informed of the proceedings of the exercise and participate freely.
Satellite images: We used recent satellite images of Magarya ward, where each building was clearly visible. These images are already available to the Nigerian National EOC.
While the execution of the activity was successful, we faced multiple limitations during planning and execution. Some limitations could not be partially or fully addressed.
Single ward enumeration: This exercise included a single ward in Sokoto state because organizers wanted to assess the validity of this methodology before investing time and resources in larger areas. This dataset does not allow for any extrapolation to a larger area (e.g., LGA, state), as the activity was not designed to perform this task. The results of this micro-census are pertinent to Magarya ward alone, but they do suggest issues that may be relevant to other enumeration exercises.
Imprecise information on age: On the one hand, we collected information on persons of all ages, so no one was excluded based on erroneous age information. On the other, the estimated age groups may not be precise. This problem remains a concern when trying to estimate the number of children eligible for RI services.
Age displacement: Younger children may systematically be reported as older than they actually are. This phenomenon has been observed before in other enumeration exercises in developing countries [21
]. Therefore, some of the children reported in older age groups may actually belong to the youngest group. Without precise information on date of birth, it is impossible to reconstruct the correct groupings.
Data collection in the household: A local field guide accompanied each enumeration team to facilitate interactions with household members. All enumerators were women, allowing easy access to dwellings and families. Nonetheless, the enumeration teams may have missed persons temporarily absent from the house or the ward. Also, obtaining information from women when the household head was not present was challenging at times, especially when the information concerned the husband. This is a restriction in Sokoto communities, and might have affected the enumeration of men aged 20–40 if they were outside the house at the time of the micro-census, but residing in the ward.