171. As the census process has evolved, so has the role of the field operation. Historically the role of the enumerator was to collect information on the doorstep about all those living at an address. Such an approach is still used in some countries though is inevitably both expensive and labour intensive.

172. All addresses would still be visited if the objective of the field operation is to enable households to respond. This could include field staff providing households with paper questionnaires.

173. A smaller field force is required where the role is to encourage non-responding households to respond by communicating the importance and requirement to complete. As well as encouraging, enumerators also provide guidance on how and who needs to return a questionnaire and where further support is available.

174. Whatever the role of the enumerator, field staff also need to have the means to record information about individual addresses they visit. This could include when visits took place, whether contact was made or whether an individual refused to participate.

175. Effective planning and preparation is essential given the complexity and expense of hiring large numbers of temporary field staff.

176. Field activity involves direct contact with the public so needs to be considered within the broader context of census communication and engagement plans. Media activity would for example normally begin before field visits took place, increasing awareness that the census is coming.

177. Some countries have developed a formal “wave of contact” model to effectively sequence where field activity takes place.¹⁰ This model is simply a timeline of when interventions are due to take place and provides an overview the different stages of how the census operation interacts with the public. It is essential that contact with householders is effectively sequenced to ensure efficient use of resources and to reduce the burden on the public. It would clearly be inefficient to begin field visits to remind householders without giving any time for mailed returns to have been completed and mailed back.

178. Small scale testing in the years prior to the census is used by many countries to refine field activities. By testing different approaches alongside each other a country can determine which are the most effective in a rehearsal or in the census itself. The value of testing can be seen with the evolving role of technology and particularly during the 2020 census round when changes had to be made in response to pandemic-related social distancing restrictions.

179. A large-scale rehearsal is used by many countries to evaluate readiness for the live operational period. An end-to-end rehearsal is most commonly carried out one or two years before the year of the census when technology and processes are in place. While the rehearsal generally is not used to test different approaches, it is necessary to incorporate lessons learnt in time for the census itself. Rehearsals are particularly important in countries who conduct a census every ten years where staff may have less experience of the census operation compared to countries with a more frequent cycle.

180. The emphasis of the rehearsal should be on evaluating overall readiness rather than on achieving the same levels of as in the census itself. It will also not be necessary to conduct a rehearsal in all areas. It is more useful to select a small number of areas which represent different collection challenges such as a densely populated conurbation in addition to a sparsely populated rural area.

181. Information from the previous census can be used to scope the size of the field force required and its distribution across the country. Taking a systematic approach to using this information is important. Assumptions need to be made about the level of response as part of the planning of the enumeration alongside information about the population density and socio-economic characteristic of each area. In the United Kingdom information from previous census and area characteristics are used in combination to create a “hard to count” index score for small geographic areas to aid planning field resource.¹¹

182. Modelling the size of the field operation based on such assumptions will contain inherent uncertainty. There will also be uncertainty about the ability to recruit and train field staff across all areas. “Mobile” field teams are recruited in some countries which can be moved across local areas or regions where there are unexpected patterns of lower response or where there are recruitment challenges. Rehearsals can be a useful opportunity to understand how such mobile resources are practically used.

183. Many countries reported that for the 2020 round the Covid pandemic had a significant impact on how they recruited and trained staff. Far greater use was made of online training which also brings significant savings.

184. As the live operational period is so short it is essential that timely management information is available to understand where response patterns are not as expected. Technology has improved the ability to understand how such a large and complex operation as the census data collection is performing.

185. The range and complexity of available information from a census operation is a challenge. Key metrics only should be identified and presented with deeper analysis and insight used to better understand response patterns when required. A rehearsal can be a useful way of refining such metrics. Online data capture has enabled a step change in the provision of up-to-date intelligence on response patterns with electronic field devices providing timely information on field activity.

186. Interpretation of field management information is difficult without an expectation of response patterns over the course of the collection period. Many countries have developed forecast models to provide a measure against which to understand performance. Assumptions underpinning these models are based on the previous census and can potentially be updated from rehearsal learning.

187. The ability to act on managing the information is as important as the information itself. Two elements are worth considering as part of this:

(a) What interventions are available (such as media, field or post); and

(b) How are decisions made.

188. Available interventions are dependent on the design of the census but are likely to consist of increasing:

(a) Publicity and associated media communication. Using existing or adapted messaging to promote the importance of completing based on need or what has been shown to be most effective during collection.

(b) Direct contact with households – Sending additional reminders using means such as postal mail, email, or SMS. Providing physical questionnaires for individuals who prefer non-digital methods.

(c) Optimizing field visits – Increasing the number of in-person visits in low-response areas, adjusting visit schedules and enhancing the training of fieldworkers to improve engagement with respondents.

(d) Targeted strategies for hard-to-count groups – Collaborating with local communities, religious and social organizations, and engaging community leaders to increase participation among specific population groups.

189. Decisions on using interventions need to be made in the context of the objectives of the census and balancing priorities across the country or region. Rehearsals are an opportunity develop an understanding of the decision-making process and census takers can also benefit from observing how other countries run their operational period.

190. During the 2020 round most countries experienced additional uncertainty because of the pandemic. Scenario testing approaches were used to good effect in understanding the intelligence, inventions and decision-making processes involved.

191. Increasingly complex and diverse populations mean the challenge of enumerating the entire population once, only once and in the right place has never been harder. Inevitably some elements of the population will be harder to count than others meaning adaptations to the standard collection design are likely to be necessary.

192. In the United States for example, a framework is used to consider hard-to-count population groups. The model recognizes segments of the hard-to-count population: Hard-to-Locate, Hard-to-Contact, Hard-to-Persuade and Hard-to-Interview (Figure 1).¹² An adapted version of this model is presented below.

Hard to Locate – Unknown addresses, gated communities where there maybe physical barriers to access

Hard to Engage – Difficult to make initial contact with including homeless and transient populations

Hard to Convince – Suspicious about need for a census, providing data to government or low levels of engagement with government generally

Hard to Respond – Includes language barriers or those with limited digital access.

193. An important aspect of this framework is that being “hard to count” is not simply about willingness. For example, a predominantly online census resulted in the England and Wales census identifying a “digitally hard to count” group including older age populations with less access to the internet. Historically this group has tended to be part of the population most willing to complete the census.

194. Adaptations of the standard design will depend on the groups identified but may include a mix of the following:

(a) Clear messaging on census website and through trained staff (in the field or at contact centres) stating that:

(b) No personal data will be shared with other organizations or government departments. Census data will only be used in the production of statistics.

(c) The importance of census data and the ways in which data are used for policies that affect all parts of society.

(d) Engagement with community groups and/or relevant sector organizations in the planning, research and operational delivery of the census. This could include the identification of preferences in how and when to engage communities.

(e) Recruitment of field staff from local communities. The United States Census Bureau for example has an overarching strategy for hiring people who will work in the communities they live in. They make explicit reference to the importance of being able to speak languages of local communities. Other countries go further, making explicit commitment about field visits only being made by members of the same community.

(f) Counts for people who are not sleeping in a house/flat (apartment) or other form of managed accommodation, potentially in partnership with local authorities and relevant charities. Integrated with broader design for collection of census data for homeless including at hostels and people living temporarily at an address where they have no permanent other address.

(g) Tailored media campaigns for specific sectors including radio, television and online channels. This can include the use of “trusted voices” to promote and encourage response.

(h) Availability of relevant resources such as paper questionnaires, translated questionnaires and support guidance to enable responses. Trained field officers able to point to these resources or to where further support is available.

195. This section deals with census activities that are or might be outsourced: that is, performed or provided by a supplier or organization other than the NSO or census office. While countries conducting a fully register-based census generally have little or no need to outsource any census services, those countries employing any form of direct enumeration for their census generally have a greater need to outsource some activities. A list of census activities which NSOs may decide to outsource is given in this section, along with recommendations for factors that should be considered when outsourcing any of these. In particular, important recommendations are given relating to data confidentiality and quality requirements when outsourcing.

196. The complexity of much of the software and infrastructure required for many new and emerging technologies used for censuses may exceed the current technical capabilities of a census agency. It may therefore be necessary to outsource significant components of such technical elements, as well as non-technical activities such as large-scale recruitment of field staff. The advantages of doing so are:

(a) External suppliers possess technical experience and expertise which would otherwise be unavailable within the NSO. Using this experience allows NSOs to focus on their main task of carrying out the census rather than on developing in-house procedures and skills that are not part of their core competencies.

(b) The decennial or quinquennial cycle of many census activities means that some activities only need to take place once every ten years, by which time the pace of technological change may imply that the nature of the activity has evolved. By outsourcing, the NSO does not have to invest in the infrastructure or knowhow required for something that happens very infrequently.

(c) Some census activities, such as enumeration with field staff or data processing, must take place within a short period, often requiring a concentrated period of peak staffing levels or the use of many electronic devices for data collection. Outsourcing can facilitate the management of these brief but intensive periods.

197. The appropriateness of outsourcing should be determined after subdividing the overall census operation into separate stages. Candidate activities for outsourcing might include (among others):

(a) printing of questionnaires and other documents/materials

(b) translations

(c) transportation of questionnaires and/or field materials before and/or after enumeration

(d) call centre/telephone help line

(e) publicity campaign, public relations, social media or other communication services

(f) design and/or provision of online response technology

(g) design and web functionality for online outputs

(h) recruitment and training of field staff

(i) primary data capture and coding

(j) post-processing questionnaire destruction

(k) design and provision of questionnaire tracking systems

(l) provision of mapping services

(m) technological solutions for data storage, including Cloud solutions

(n) data editing and/or imputation

(o) quality assurance

(p) evaluation

198. Many of these activities will be less relevant or not applicable at all to those countries that carry out a fully register-based census, in which the opportunities for effective outsourcing are clearly much reduced. The creation and maintenance by external agencies of the registers from which the census information is extracted, is not outsourcing in the generally-accepted meaning of the term.

199. The decision to outsource will depend on the requirements of the census agency, the availability of skills in-house and the ability of the agency to manage complex systems development projects. Total outsourcing might at first seem to be a simpler process to manage. However, it is unlikely that a census agency will choose to outsource the total solution to the census operation, but rather that components of the system will involve a combination of outsourced elements, different external service providers working as contractors on specific projects, and in-house developments. There can be advantages to having several linked activities provided by a single contractor.

200. A clear understanding of the NSO’s requirements is needed before any contracts can be tendered so that these can be specified unambiguously to the contractor. These include understanding the objectives of the project, the outputs to be achieved and the standards these outputs must meet (quality, timeliness, cost), and confidentiality and public sensitivity issues. Specifications must allow for the possibility of requirements and objectives changing over the lifetime of the project. How these changes will be agreed to and approved by the census agency and the provider needs to be determined from the outset.

201. Timetabling, including milestones for key deliverables linked to payment schedules, needs to be agreed with the contractor. Regular monitoring needs to be undertaken at the operational level. Processes and escalation procedures should be established to allow senior staff to monitor progress and to deal with any major issues that cannot be resolved at this operational level.

202. A mixed approach to systems development is one in which the overall solution consists of outsourced systems, systems developed by external contractors working alongside census agency staff and systems developed in-house. This approach can have many advantages, such as greater flexibility to adapt systems as more is learned through the systems development and testing programme and during the actual census processing operations. This can lead to improved data quality and savings in processing costs as systems are optimized. However, census agencies should be aware that, for such an approach, management becomes much more complex. The census agency must be skilled in the management of complex projects, have a clear understanding of business processes and be able to manage carefully the integration of both the technological and clerical processes. It is important, therefore, to set up a dedicated team within the NSO to draft and manage contracts, deal with contractors, monitor processes, oversee quality and liaise with statisticians. Such a team should also have sufficient familiarity with appropriate project management methods, both traditional and agile, since the latter are increasingly prevalent in the IT world and beyond. Team-based working, where external contractors work very closely with census agency staff, is essential to ensure the success of this method of systems development.

203. Throughout the overall process, activities should be conducted by a method that is tailored to meet the requirements of users (e.g. with respect to the accuracy and timeliness of the results), and that reassures the general public on matters relating to confidentiality and data protection. It is important that no part of the outsourced operations should be performed using a method that could result in loss of the trust of the general public. In judging the appropriateness of outsourcing, census agencies should therefore carefully consider the following factors:

(a) strict protection of data confidentiality

(b) a method of confidentiality assurance that satisfies the general public

(c) relevant government policies and procedures

(d) quality assurance measures

(e) ability to manage and monitor the outsourced census tasks/activities

(f) control over the core competencies of the NSO.

204. Assurance of confidentiality should be considered paramount among these, since it is both a Fundamental Principle in its own right (see Annex II) and necessary as a means of maintaining the public support required for efficient and high-quality data collection. NSOs are responsible for data confidentiality, both in terms of reality and of perception. Consequently, outsourcing tasks that pose an actual or perceived risk to security or data confidentiality must be avoided. For all data collection or processing operations carried out by external suppliers, safeguards should be put in place to ensure that there is strict protection of confidentiality. This may be achieved, for example, by requiring external suppliers to encrypt data collected by them; by insisting upon intrusion tests into IT systems external to the NSO; by requiring evidence of data deletion after the project or mandate is completed; and by requiring companies providing IT services to demonstrate that they have the relevant certifications, e.g. the ISO-27001 certification.¹³ The outsourcing of such services should be considered carefully so that public trust and confidence in the census is not eroded. Where, for example, temporary enumeration staff are engaged under contract, this should be done in such a way that they are subject to strict monitoring and control, with the final responsibility for oversight resting with the census agency. Such field staff should be engaged in a way that ensures that their activities are governed by the relevant statistical legislation to preserve the confidentiality of the data that they collect. The requirements and criteria applied to external suppliers should be made public.

205. Another key issue to be considered in outsourcing is quality assurance. The census agency must be satisfied that the goods or services paid for are provided to an agreed and acceptable standard. In this respect, cost should not, ideally, be the first priority in considering and judging the successful bidder. Poor quality work, or any perceived risk to confidentiality posed by the lowest-priced bidder, could cause a significant loss of trust among users and the general public.

206. As part of any contract allocation process, potential contractors should be required to provide samples of their work, to give details of sites at which previous work can be inspected and/or to list referees who can verify their claims and credentials.

207. More detailed discussion of the issues to be considered in outsourcing and in the evaluation of software and hardware can be found in the Principles and Recommendations for Population and Housing Censuses – Revision 4, and the Handbook on the Management of Population and Housing Censuses.¹⁴