5 Things to Consider When Doing Mobile Data Collection
Jul 6, 2017
This is a guest post by Nafessa Kassim, Director of Business Development, and Al Ismaili, CEO and Co-Founder of Bamba, a mobile data collection company in Nairobi, Kenya.
Development organizations now have a plethora of mobile data collection tools at their disposal, and choosing between the various options available can often seem like a daunting task. But even after finding the right tool, it is imperative to think through and address the most common challenges that arise in the mobile data collection process. In the last five years of helping development actors integrate mobile phones into their data collection processes, we’ve identified five challenges they run into most often.
Photo credit: David Staten
1. Sampling Technique
Development organizations must ensure their sampling techniques meet high quality standards. When we refer to sampling, we mean ensuring that the target beneficiaries who are surveyed accurately represent the population as a whole. Market research teams often use these techniques, but those techniques also apply in the development aid world, especially in organizations conducting programming and research. Keep in mind there is no such thing as a perfect sample. Statistical sampling will always have a margin of error that can be minimized by increasing the sample size.
Although a number of sampling techniques are available, the most commonly used is random sampling. Other sampling techniques include convenience, cluster, quota, and systematic. You can find additional pros and cons of these methods here.
However, if you’re simply looking to reach specific beneficiaries (for example, beneficiaries who have received direct aid), then sampling methodologies do not need to be taken into account.
2. Wide Geographic Span
Many sampling techniques can often result in a wide geographic span of beneficiary locations. The advantage of using mobile data collection tools is that they can often reduce the cost of reaching geographically dispersed sample populations. However, it is important to ensure your approach remains technologically agnostic. For example, some beneficiaries may be in regions of a country that have no internet but have mobile reception. Your data collection tool should then be able to use interactive voice response (IVR) (which is also highly effective for illiterate beneficiaries), text messages, or unstructured supplementary service data (USSD) to gather the appropriate data. In regions where no mobile coverage exists, beneficiaries can only be reached by deploying mobile-enabled field workers.
3. Many Languages / Dialects
Language is always a concern when collecting data remotely, since many languages and dialects must be accounted for. Artificial intelligence is bringing a wave of affordable translation technology, such as Google Translate, to our fingertips. However, they have a long way to go before they can claim 100 percent accuracy; moreover, they don’t cover many dialects. It is important to select a data collection platform that can handle language translation. Mobile data companies often accomplish translations via a hybrid of automated translation software and manual translation.
4. Fraudulent Data
In our work with multiple development clients, the subject of data fraud often arises. Fraud, like margins of error, can only be minimized, and not entirely eliminated. Choose a mobile data technology that uses fraud detection techniques such as pattern recognition, speeding, and red herring questions. Another technique to alleviate fraud is to ask beneficiaries a previously asked question a second time some days later where the software knows the expected response. This and other techniques including national ID checks (where government databases are available) are all ways of decreasing fraud.
5. Beneficiary Drop-off
A final challenge that development organizations face is when conducting tracking studies (i.e. when the same beneficiary is surveyed more than once) is that the beneficiary is non-responsive the second time. This is often handled by incentivizing the beneficiary through a number of mechanisms such as raffles, instant airtime, and instant cash, among other techniques. Various techniques from the market research world including gamification (using game techniques such as points and leaderboards) are being used increasingly more to incentivize participants during data collection.
Ultimately with a strong data collection technology, effective sampling, and accounting for these challenges, a development organization will satisfy the growing demand for data-based metrics, while also keeping up with today’s technology revolution.
Reader, do you work in mobile data collection and have great insights to share? Give us a shout at digital(at)dai(dot)com, we’d love to feature your guest post on our blog!