New Technologies in Census Data Collection Part 1: Planning for Mobile Data Capture Select Topics in International Censuses1
Released October 2016
INTRODUCTION MOBILE DATA CAPTURE VS. SCANNING OR KEYING As mobile technology, like tablet PCs or smartphones, become more widely available, many statistical agencies The decision on whether to use mobile data capture should are considering using mobile data capture for the 2020 begin in the initial census planning stages. round of population and housing census data collection. Computer-assisted personal interviewing (CAPI) is when Adopting mobile data capture means changing from paper enumerators use electronic questionnaires on laptops, tab- to electronic questionnaires and affects the entire census let PCs, smartphones, or other handheld electronic devices life cycle. This is part one of a set of technical notes, New to conduct face-to-face interviews. Technologies in Census Data Collection, that presents an overview of what to consider when transitioning to mobile Mobile data capture allows users to take advantage of data capture. This document discusses planning consider- added features that can be programmed into mobile ations for mobile data capture. Part two deals with develop- devices. These include: ing an electronic questionnaire. • Integrated maps and Global Positioning System (GPS). • Computerized case management. • Automated skip patterns. • Simultaneous process for data collection, data capture, editing, and consistency and nonresponse checks. • Automated coding. • Ability to preload questionnaire responses and custom- ize questions.
There are both advantages and disadvantages to using mobile data capture, which should be weighed carefully. Be sure to consider the practicalities of using mobile devices, such as electricity supply, Internet availability and speed, and how much time is available until census enumeration. Preparing for a census using mobile data capture requires more preparation time than for a paper-based census.
1 This technical note is one in a series of “Select Topics in International Censuses” exploring matters of interest to the international statistical community. The U.S. Census Bureau helps countries improve their national statistical systems by engaging in capacity building to enhance statistical competencies in sustainable ways.
Advantages of Using Mobile Data Capture º Requires more technical support in the field to replace or repair devices and troubleshooting of •• Improves data quality through: hardware or software application problems. ºº Logical validation rules in the questionnaire. Errors and inconsistencies may be resolved at interview • Segmentation effect may offset improvements in data time, closer to the source. quality. ºº More accurate progression through the question- º Enumerators only see a portion of the questionnaire naire due to automated skip patterns and manda- at a time, so they may miss the bigger picture of tory items. the questionnaire. This may be alleviated through ºº Reduced manual data entry errors. design features that allow enumerators to see the ºº Reduced number of unlinked forms. Reduces the overall questionnaire structure. risk of one section of the form getting separated • Technological problems could interfere with from the other. enumeration. •• Improves field management and real time monitoring of º The battery could run out during enumeration. enumeration activities. º The equipment could break leading to data loss or lost enumeration days. •• Reduces time and costs by eliminating a separate pro- cess for data capture, automated coding, and reducing • There are more infrastructure constraints to consider. the volume of post-enumeration data checking. º Electricity needs to be available to charge the devices; challenging in areas with limited electricity. •• Decreases time between data collection and release of the results. º Limited Internet availability may cause difficulties in data transmission and other functions of the mobile •• Reduces costs of data capture, questionnaire printing, data capture software application that rely on the storage, and transportation. However, the increased Internet. A realistic assessment of Internet connec- equipment costs may outweigh the savings (see section tivity in all areas should be conducted during the on “Disadvantages of using mobile data capture”). planning stage. •• Can utilize GPS features and digital mapping capabilities. º Needs a system to transfer the data from the handheld devices in a timely and secure manner. Disadvantages of Using Mobile Data Capture This could be the Internet, local area network, or •• High equipment costs with limited long-term use. through manual data transfer. ºº A census generally employs a large number of º Data security must be assured during data collec- enumerators. Purchasing a mobile device for each tion, transfer, and storage. enumerator is costly. º Needs a system for backing up the data to prevent ºº The equipment may become obsolete quickly. data loss. ºº Devices may get lost, stolen, or break during the Internet-Based Data Capture census enumeration. •• More time needed during the preparation stage. Many national statistical agencies also are exploring using Internet-based data capture. The main advantage of ºº Need time for programming the application, setting Internet-based data capture is that it can be cost effective up systems for data transfer and storage. since enumerators are not used. It can also be used for resi- ºº Requires extensive testing. dents who are hard to reach in person. However, in order •• Needs more skilled programmers who are able to do for it to be cost effective, there needs to be a large num- sophisticated programming specific to the mobile device ber of people with reliable access to the Internet. Another (e.g., Android) and all the necessary components of an disadvantage is that a highly sophisticated control system electronic questionnaire application. is required to avoid duplications and undercounting and •• Requires technologically skilled enumerators with more to ensure the security of the data. Since the focus of this training and field support. brief is on CAPI, we will not go into detail on Internet-based ºº Requires that enumerators can use a mobile device. data capture. However, some questionnaire design consid- ºº Requires intensive training of enumerators on use erations for CAPI are also applicable to Internet-based data of device. Training may be more complicated than capture. with paper questionnaires.
2 U.S. Census Bureau PLANNING CONSIDERATIONS FOR MOBILE developed for charging and backing up the devices. It is DATA CAPTURE critical to have contingency plans for when electricity and/ or Internet access are not available. In order to ensure the success of the census, it is critical to identify all requirements for carrying out the census using Differences in Questionnaire Development Process mobile data capture and develop plans early in the census life cycle. As with a paper questionnaire, developing an electronic questionnaire is an iterative process. It must be developed, Census Timetable tested, revised, and then tested again, repeating the cycle until the questionnaire works as intended (see Figure 1). The census timetable should be adjusted to fit the needs With an electronic questionnaire, the technical aspects of of implementing mobile data capture. Although the pro- the electronic questionnaire application must be tested and gramming of the questionnaire cannot happen until the revised, in addition to content. questionnaire specifications have been developed, it is a good idea to plan for the programming in the early stages Further, after the subject matter specialists finalize the of questionnaire development. With mobile data capture, questionnaire content, specifications are needed to serve previously separate processes may be integrated, or may as a blueprint for programmers to design the application. need to be carried out earlier in the census life cycle. For Adequate documentation of the questionnaire instrument example, data collection, capture, and editing can be done is also required. Not only does the lack of paper documen- simultaneously in mobile data capture. However, generally, tation make it difficult to discuss the questionnaire with more time is needed to develop and test the application, set various stakeholders, it also makes testing more time- up the data transfer and processing systems, and procure, consuming and error prone since skip patterns are less program, and test the mobile devices. If edit checks are to obvious. In addition, an electronic questionnaire may con- be added to the application, the edit programming must tain additional features not included in a paper question- be completed before the enumeration begins, rather than naire, such as data validation and error messages. These programming those separately in the post-enumeration features also must be specified so that programmers can stage. Further, more time should be allotted for training design the application as intended. the enumerators since the training must include the use of the mobile devices. Therefore, it is critical to determine all When designing an electronic questionnaire, the subject the steps needed to take place in order to set up a mobile matter specialists may lose control over the wording, lay- data capture system and allow sufficient time prior to the out, and design of the instrument. Therefore, it is critical enumeration in the census timetable. that the subject matter specialists work closely with the programmers to make sure that there is clear communica- Budget Considerations tion regarding the questionnaire content, layout and design, data validation, and other specifications. It is also important Using an electronic questionnaire may save costs in the to have a good understanding about the timeline, ongoing printing of the questionnaire and data capture, but the changes to the questionnaire content, data security, and costs of the electronic equipment may be higher than the quality assurance. savings. Carefully consider all costs of conducting mobile data capture and account for them in the budget. These issues are discussed in more detail in New Technologies in Census Data Collection Part 2: Developing Infrastructure Considerations an Electronic Questionnaire. Infrastructure issues such as availability of electricity and Internet access can affect the success of mobile data cap- CHOICE OF MOBILE DATA CAPTURE APPLICATION ture. Early in the planning stage, identify places that lack electricity and/or Internet access. If relying on a cellular There are many existing software applications for mobile or Wi-Fi data connection to transmit data, it is important data capture. One such program, CSPro, is free software to research the speed of the data transmission and plan developed by the U.S. Census Bureau. It has been used in accordingly. Mapping the areas without electricity and/ over 160 countries for censuses and surveys. or Internet access is helpful in planning. Plans should be
U.S. Census Bureau 3 Figure 1. Steps in Electronic Questionnaire Development
Develop Program Specifications Questionnaire
Initial Write/Revise Planning Questions Test
Evaluate
Some questions to ask when selecting software are: •• Local area network (LAN), a computer network that con- •• How easily can someone modify the software to meet the nects computers and mobile devices within a limited area questionnaire’s specific needs? (such as an office building).
•• How easily can the enumerators use the software? •• Peer-to-peer (P2P) network (for example, tablet PC to tablet PC connection through Bluetooth or Wi-Fi). •• How much does the software cost? •• Cable connection (for example, connecting an enumera- •• What platform/operating system does the software tor’s tablet PC to her supervisor’s laptop). require? What are the cost implications? •• Wide area network (WAN), or a private network that •• What are the hardware requirements for the software? extends beyond the geographic reach of LAN, but is What are the cost implications? not connected to the Internet (for example, a dedicated •• In what format does the software output the data? mobile network).
•• How well does it work with the existing data processing It is important to consider what is feasible in all areas of the system? country when setting up a system. Reliable Internet may be available in cities, but not in rural areas. It may be neces- •• Is there technical assistance or training available? What is sary to establish more than one system according to infra- the cost? structure constraints. For example, data may be uploaded DATA SECURITY, TRANSMISSION, AND to the server through secure Internet connection in cities, STORAGE CONSIDERATIONS but in rural areas, data may be transferred using cable con- An important consideration is how to transmit data securely nection. Be sure to allocate time and resources to test the from mobile devices to a central server. Secure data storage data transfer and storage system. Failure to do so may have is also critical at every level where data are stored (for costly consequences. example on mobile devices during the enumeration, on IN-HOUSE PROGRAMMING VS. OUTSIDE supervisors’ devices, and on regional and central data serv- CONTRACTOR ers.) This topic should be discussed with the programmers One issue to consider in adopting mobile data capture and IT specialists. is whether to do the programming in-house or to use an Most mobile devices can store data either locally or on a outside contractor. One reason to hire an outside contractor network. Storing data locally risks data loss if the device is is if the agency does not have the technical skills or exper- lost or stops working. Storing data on a network requires tise to program the software application. Another reason a way to transfer the data securely from each device to the would be if an outside contractor already has developed a main server. Some ways to transfer data securely include: software application that can be adapted for the census. On •• Secure Internet connection. the other hand, an advantage of using in-house programing is that skills and devices used during the census would be
4 U.S. Census Bureau available in the post-censal period. Required programming REFERENCES skills depend on the selected software application. Bethlehem, Jelke, and Silvia Biffignandi, Handbook of Web CHANGES IN STAFF SKILL REQUIREMENTS Surveys, John Wiley & Sons, Hoboken, NJ, 2012. Adopting new technologies often requires a new set of Brancato, G., S. Macchia, M. Murgia, M. Signore, G. Simeoni, skills, which the existing staff may not have. Extensive K. Blanke, and J. Hoffmeyer-Zlotnik, “Handbook of training for existing staff may be needed or new staff may Recommended Practices for Questionnaire have to be hired. For enumerators, additional training will Development and Testing in the European Statistical be needed to familiarize them with the technical aspects of System,” European Statistical System, 2006,
COMPUTERIZED FIELD CASE MANAGEMENT Tarnai, John, and Danna L. Moore. “Methods for Testing and Evaluating Computer-Assisted Questionnaires,” in Stanley One feature of using an electronic questionnaire is that it Presser, et al. (eds.), Methods for Testing and Evaluating can be linked to a computerized field case management Survey Questionnaires, John Wiley & Sons, Hoboken, NJ, system. As data from each device are sent to a central 2004, pp. 319–335. database, the data can be used to monitor the progress of the enumeration and identify which households the United Nations Department of Economic and Social Affairs interviewers need to visit. Quality metrics and audits also Statistics Division, Census Data Capture Methodology: can be incorporated in the field case management system. Technical Report, United Nations, New York, 2009, Field operations and programmers should work together to
U.S. Census Bureau 5