PRB

On May 1, 2010, Indonesia started its sixth population census since gaining independence in 1945. Obstacles encountered in previous enumerations have led to modifications in field operations and modernization of the technologies used in data processing. Of course, many of the problems were beyond the control of statisticians. Geography is always a challenge for enumeration, with nearly 1,000 inhabited islands and Indonesia’s many isolated settlements on mountaintops or deep in swampy forests. While most Indonesians now live close to centers of government, a significant minority are far from schools, roads, and facilities, putting them outside the easy reach of statistical counts. Politics and economic crises can also disrupt census activities. The 1961 Census analysis collapsed when political upheaval swept through Jakarta in the years leading up to a massive coup. Regional outbreaks of violence meant that portions of the population were uncounted in 1971 and 1990, and the Asian financial crisis of 1998 led to devastating budget cuts in 2000 with subsequent crippling of field operations. These challenging experiences were at the forefront of the thinking of senior staff of Statistics Indonesia as they planned the 2010 Census.

Avoiding Previous Errors in Data Collection

The 2010 Population Census of Indonesia differs from previous censuses in the revolutionary way modern communications technologies were used in fieldwork, data processing, and quality control. However, while there were plans to use text messaging in the field operations, the real breakthroughs came almost by accident.

When the 700,000 interviewers were recruited to do the house-to-house listings, the officials from Statistics Indonesia were most concerned about two problems that had arisen in the 2000 Population Census—the low education and poor handwriting of many interviewers. This had caused huge errors when machine scanning misread poorly formed numbers, with 8s being read as 5s, or 7s appearing as 1s. Many of these problems were undetected by the computerized consistency checks, causing a great deal of frustration for census analysts.

The officials planning the 2010 enumeration wanted to ensure that the codes were written clearly and scanned accurately so that such errors could be avoided. They required that interviewers have high educational requirements and neat handwriting. Veterans of the 2000 Census were pleased by the great improvement in the quality of field enumerators, and when they went to the field they were confident of much improved coverage and quality of data collection. What these older officials had not anticipated was another characteristic of the enumerators: Almost all carried a cell phone and were highly practiced in text messaging—Short Messaging Services (SMS)—as part of their daily social lives. Over the course of May and into June the daily operations of the census was enhanced by a broad, tight network of communications technologies. The “SMS Census” emerged with several distinct advantages over previous surveys.

Advantages of SMS-Based Surveys

The fieldwork would be done in two stages—a listing of the number of residents in each building on a form called L1, followed a few days later by a detailed questionnaire (called the C1 form) to collect information on households and individuals. Field supervisors were instructed to send via SMS message the daily L1 tally of males and females in each census block according to a standard format. These would be automatically uploaded into a spreadsheet that would sum the totals by subdistrict, district, and province, and would show the sex ratios of each government unit as a way of verifying the numbers. Each day the central office team was able to check the interviewers’ progress down to the level of the census blocks. Summary tables showed the number of census blocks covered, the numbers of households counted, and total numbers of male and female residents expected to be present for the C1 detailed enumeration. This iterative, reflexive system of monitoring raised hopes that the coverage of the census would be higher than ever before achieved in Indonesia.

The field coordinators had a range of phone numbers they could contact to resolve questions arising from field operations. For example, in Bali, a respondent to the C1 question on “district of birth” gave a district name that was unfamiliar to the interviewer. A call to the field coordinator was passed to the district office where a Google search revealed that the name was an old district name. The search provided the full details of the current name and the new province where the district is located. This information was referred back to the interviewer in real time to be discussed and verified with the respondent. Such corrective actions can only occur if the interviewers and the field supervisors are comfortable asking and answering questions, and this is actually more likely with texting than in face-to-face communications that in this hierarchical social system can be formal and limited. The growth of cell phone cultures are changing the ways Indonesians communicate and the census may be the first big application of free-form communication up and down the administrative pyramid.

Texting has provided instant reports of problems. Throughout the data collection process, there have been floods, motor vehicle accidents, demonstrations, fires, dog attacks, and occasional threats from community members. The provincial and central office ability to resolve these problems has been enhanced by the ease of two-way communication, and the local staff reported their growing confidence to deal with such difficulties knowing that their superiors were fully supporting them.

This technology allows mass messaging to all 700,000 field workers during the data collection. One example occurred on May 7 when the central office of Statistics Indonesia was asked if they could identify the oldest Indonesian in the census. Normally this would have been very difficult because the code for age had a maximum of two digits, and there are many thousands of people in the category 99+ years. The space for year of birth could have been used to identify the person with the earliest birth year, but many people are unable to give a year of birth. However, by sending out an SMS to all the field teams it was possible to immediately collect information directly from interviewers who had visited houses, and who could remember the relatively few cases of extraordinarily old people they had encountered. They returned to those houses to verify ages and record detailed information from respondents, friends, and family members. Sometimes the results were startling, like one woman claiming to be 157 years old or a woman said to be 120, whose oldest child was 70, implying she was a mother at age 50. The exercise was not an accurate determination of age, but it did generate enormous interest in the press and on television, and raised the profile of the census for tens of millions of Indonesians.

Finally, the use of SMS in the census operations fostered communication across the community. Thousands of messages flowed into the Statistics Indonesia supervisors with complaints and suggestions as citizens took the advertised motto to “Pastikan Anda Dihitung” (Be Sure You’re Counted) seriously.

There is a downside to all this communication. When the head of BPS holds a meeting, he has to remind his staff to switch their phones to silent and only read the messages that are urgent. The census operation must distinguish between the different priorities constantly coming in from the field. Focus on the highest priorities will define the degree to which the “SMS Census” becomes the “Full Coverage Census.”

Terence H. Hull is professor of demography in the Australian Demographic and Social Research Institute.