There is an election in a week, you want to poll the citizenry before the election, and your financial resources are limited. What should you do? Should you (A.) Give up because it is simply not possible to get a full-fledged poll out in the field. (B.) Beg your donor to give you a last minute cash infusion to bring on more staff and a polling company. (C.) Join the 21st Century and leverage technology to generate a fully randomized national telephone poll using a platform like Voto Mobile. Voto Mobile's goal is to make interacting with an audience via mobile phones - either one-way via broadcast or two-way in an interactive fashion -- easy and inexpensive.
Over the last few weeks I have had the opportunity to sit down twice with developers and staff from the socially conscious start-up Voto Mobile. Based out of Kumasi, Ghana, Voto Mobile has the straight-forward goal of “Mobile Engagement, Simplified.” The company is leveraging the ubiquity of mobile phones around the world to enable both research and social engagement that offers CSOs, NGOs, Political Parties and other organizations new capabilities.
Voto Mobile is a platform that is accessible from anywhere in the world. While the set-up and management of an account requires Internet access, the individuals being surveyed or having information broadcast to them need little more than a basic telephone connection, either mobile or land-line. The platform is robust and offers the ability for administrators to engage citizens via SMS or voice. (SMS communication requires mobile phones). Administrators can connect with citizens via defined user lists or they can conduct nationwide calling campaigns. The platform provides real-time response data, and for voice calls information about the time individuals were on each call. Voto Mobile also offers multi-lingual support and conditional questioning. The system is capable of callbacks if an individual is disconnected or unable to answer the phone. The nation-wide survey capability is based on using all possible phone numbers within a country. If a number doesn’t answer or there is no live number there is no cost for that call.
Voto Mobile charges $.02/min above per minute rate of the local telco provider. Thus if a phone call typically costs $.05/minute, the cost of using votomobile would be $.07/minute. If your survey is 10 minutes long and completes successfully your charge would be approximately $.70/survey with some margin for individuals who answer the call but then drop off part way through. Since the duration of a call is tracked it is possible to determine where calls are being lost. To create a robust statistical sample population of approximately 1500 individuals it is likely that you will have to make a significantly higher number of calls to achieve the desired number of survey completions. However, if all calls were answered or hung up based on call completion or call drop, a full 1500 person 10-minute survey with a local telco cost of $.05/minute would cost approximately $1,050. If you factor in partially completed calls the total cost of the program would be more but still in aggregate would be significantly less than fielding an-on-the ground survey or call bank.
Having demoed the platform three times now I find it takes about an hour to set up a basic survey in English. You can either have the system call your phone to record the questions and configure the options or you can pre-record the questions and answers and upload them. They suggest using Audacity, an open source audio software, to record MP3 files for upload. If you wish to have multiple languages it takes longer to set-up.
The timeline for completion of a survey depends on a variety of factors, however on average it is a much faster way to field a survey and receive results in return. Voto Mobile estimates that a full sample would complete within 24-72 hours with a size similar to the one above.
I have just started talking about this platform with some of our program staff and there are a series of recurring questions that have come up. I chatted with Dr. Max Boots, Voto Mobile’s general manager about some of these questions:
In conducting typical surveys we are able to over sample particular populations to ensure minority representation. Is there any way to do this using Voto?
The short answer is 'No." However, because you can increase the size of your sample far more than you could doing a standard survey, you can sample within the sample based on questions focused on aage, ethnicity, language, sex, etc. Thus, while the platform itself does not know if a particular phone number is associated with a man or a woman because it is randomly calling all numbers in a given country, you can create a fully randomized sample using targeted questions and a big-enough sample size.
Aren’t conventional surveys simply more accurate?
No. There is significant evidence that during in-person or over the phone surveys with an actual person there is a biasing factor imparted by the survey giver. This can manipulate how truthfully individuals respond to questions. This becomes even more important in societies where gender roles can push preconceived notions to the forefront of a respondent's mind. Because every call recipient receives the exact same message or can have a message tailored specifically to alleviate bias, by say having a gender question determine whether the remainder of a survey is conducted using a male or a female voice, we can actually reduce bias.
Adding to this is a factor of anonymity often not present during in-person surveys. It is possible to get respondents to provide detailed information that they might not otherwise provide.
How is this really different from traditional SMS based messaging?
It is different in several ways. First, whether we like it or not many people are not literate and simply cannot read well enough to respond to messages accurately. Voto Mobile’s use of voice alleviates this problem through voice communication. Second, because it largely only requires pressing the number keys on a phone it is highly accessible and reduces user error with a maximum of 10 options for response, or, alternatively with a voice response. Third, Voto Mobile enables highly accurate statistics for how long individuals remained on a given phone call. This feature is particularly important when determining how to improve a survey during a beta testing phase. Fourth, you can establish credibility and trust by using respected voices from radio, television or other fields that can improve response rates and make individuals feel engaged in a valuable process.
What are the drawbacks?
Obviously like SMS-based platforms this product only works where people have access to telephones, mobile or land-line. While access to phones is growing in ubiquity, it is not completely universal and this must be accounted for.
What about security? Could polling on controversial issues be disrupted?
Because the calls can originate either in country or externally via voice-over-IP or VOIP the only way to prevent a successful poll is literally to shut down the phone service to a given country.
On the whole I have seen many platforms over the last nine months of working at NDI andam particularly impressed with Vot Mobile. It has been widely used in Africa to date and Voto Mobile has been partnering with such organizations as the World Bank, UNESCO, and local and national governments. There are a wide variety of capabilities that I simply do not have the space or time to write about, but there are a large number of potential use cases for this platform in our work. If you would like to try it out simply go to their website and sign up for a free demo and play around with it or be in touch with me!