MobileActive.org

On May 21 the National Democratic Institute (NDI) provided technical assistance to a Montegrin NGO called the Center for Democratic Transition (CDT) that successfully monitored their country's independence referendum using SMS as the primary observer reporting tool. We believe this is the first time an election monitoring group has employed text messaging to meet all election day reporting requirements. Details about the program follow.

General Information:

• Election monitors from the Center for Democratic Transition (CDT) in Montenegro conducted the first ever election monitoring project where SMS was used as the primary tool for reporting election information with NDI technical assistance.
• 200 observers reported approximately 11 times each throughout the day including voter turnout and results data, transferring over 2000 reports to the reporting center in Podgorica. All reports were automatically entered into the reporting database where they were immediately included in analysis reports.

Turnout

• The referendum required 50% turnout to be legitimate. CDT monitors reported turnout via SMS hourly, allowing CDT to report that the turnout threshold was met by 1pm.
• Monitor reports received via SMS were automatically entered directly into an Access database, allowing CDT and NDI officials in Podgorica to see "real-time" turnout reports with no data entry or phone operators involved.
• CDT held hourly press conferences to report turnout using "up to the moment" turnout reports.
• For each turnout report observers received automated confirmation text messages double-checking their turnout figure to ensure accuracy and giving confidence that their report was received. Any errors could be corrected by sending a second SMS message.

Observer Communication

• CDT and NDI officials in Podgorica used SMS to communicate regularly and directly with observers throughout election day. Instant, direct and simultaneous communication with the 200 observers on election day is unprecedented. This enabled CDT and NDI to keep observers abreast of planning, logistical and political events as they occurred.
• Observers sent SMS messages upon arrival to polling stations to confirm their participation and report polling station eligible voter figures for comparison with official voter roll figures.
• CDT sent hourly SMS updates to observers with national turnout results and any necessary logistical or operational updates. CDT could also call or text individual observers as needed. This was useful in keeping observers engaged and motivated throughout the long election day

Results Reporting (Parallel Vote Tabulation)

• The referendum was an "up-down" vote on independence with 55% required for the independence referendum to pass. Votes were counted in polling stations and CDT monitors reported results via SMS immediately as they became available.
• Observers received an automated response via SMS confirming their results figures. Errors could be corrected by sending additional text messages.
• Results data was available to CDT and NDI officials in Podgorica in real-time as it was reported. There was no lag time between when observers reported and when their results were included in the analysis reports by officials in Podgorica .
• Dynamic reports were available to officials for results analysis with data that was constantly updated via incoming SMS messages from observers.
• The referendum passed with 55.5% of the vote, reflecting a very slim margin of only 2046 votes. CDT's projected result was well within the margin of error of 2%, within .6 % of the official result and thus was "too close too call". This assessment was made public on election night. The speed of CDT's reporting system allowed them to be the first to know that the results would be within the margin of error of any national observation exercise.

Conditions

A series of conditions made SMS reporting for the Montenegro referendum possible. Those included

• CDT was an experienced NGO, having observed several elections in the past
• Montenegro has wide cell phone coverage
• CDT observers and Montenegrins in general are experienced SMS users
• The reporting requirements were simple: cumulative turnout figures and basic yes/no election results data are well geared toward short SMS text message report format.

Technical Overview

• The technical solution involved a single cell phone attached to a Windows laptop (USB).
• We used Clickatell aggregator and SMS Reception Center PC software with custom scripting for message and Access database management. The aggregator solution worked great but we're still investigating SMTP gateways and shortcodes for future projects.
• The software and scripts worked flawlessly, very flexible and could be adjusted "on the fly" during the day
• The toughest part was receiving nearly 200 messages concurrently, as observers were trained to report at the same time at hourly intervals throughout the day. One of the cellular providers had limits on Inbound messages to a single phone number that required that we find various workarounds, but we were able to clear blockages and all messages from observers were delivered to the PC.
• We learned quite a bit about working with different cellular providers, some of which were fairly tough lessons but we managed to receive every text message sent by observers in real time or very near it.

Lessons

• Pilot test at least twice, we ironed out significant bumps during the two pilots. Conduct the pilots several days in advance in case you need to work with cell providers to address problems.
• Thoroughly test throughput from / and interoperability between all providers beforehand
• Consider using multiple phones/SIMs for receiving, one from each provider. This would have allowed us to "load balance" and also have monitors send to the phone in their cell provider network and avoided interchange issues if any were anticipated.
• Accurately anticipate your outbound (sending) needs. If it's more than a few messages per minute plan on getting an aggregator or gateway (for our program the costs were not too prohibitive, about a nickel per sent message with Clickatell).
• Worth considering GSM modems to remove phone hardware and software from the equation (although we had no problems with the phone, an unlocked Motorola v188).
• Collect not only observer cell numbers in advance, but also their cell provider if not identifiable by the prefix (area code) or other means.
• Focus group your message format strategy to better anticipate user format errors so that scripts can handle data input in formats you'd not necessarily anticipate. Especially important when working in languages and cultures other than your own.

For more technical information on the approach, feel free to contact me.