Phone surveys in developing countries need an abundance of caution

If you have looked at last year’s Nobel prize winners work in economics you will realize that it is heavily dependent on good quality primary data that is collected through in-person household interviews in developing countries. Over the last few weeks, as the COVID-19 crisis escalated, most researchers (including J-PAL) have suspended in-person field activities for now. This poses huge challenges for those who have data collection efforts planned this year, especially for junior faculty on a tenure clock. And phone surveys seem like a great alternative for these unique and difficult times. J-PAL has shared an excellent post here on best practices for phone surveys and has curated a list of strategies for improving survey response rates on the phone that include suggestions such as building a brand around the phone number, protocols for failed call attempts, making participation attractive through mobile money or airtime, random digit dialing, to name a few. These are great suggestions for researchers who plan to conduct phone surveys during ‘normal times’.

However, things are anything but normal right now. Just as we researchers have been thinking of our colleagues, students, families, children and giving everyone that extra slack (pass/fail option in grades for students, option to extend tenure clocks among junior faculty, and others), we also need to be extra thoughtful about the human subjects we work with. To that end, we urge everyone who is thinking of a phone survey right now in a developing country to exert an abundance of caution.

Here are some simple ways in which we can all do just that:

1. Benefits to the human subject: Ask yourself, “is your phone survey going to improve these respondents’ lives in the next few months during the COVID-19 crisis? Are you working with this country’s health ministry or some other NGO or government partner trying to give money or other help that directly benefits these poor people?” If yes, then proceed with your survey. For all other surveys, pause and be circumspect.
2. Informed consent: Your respondents, just like you, are going through a stressful time. Acute economic (and associated cognitive) deprivation can make it difficult for poor people to give consent. Even if you are able to get an agreement to a survey, the validity of consent obtained in unusual circumstances may be questioned. Short and simple consent scripts may not be enough. Rather, give your respondents the information they need and encourage them to ask questions.
3. Compensation:Many people in developing countries are losing jobs and have lost their stream of income during the last few weeks and days. Supply of essential commodities has been hit hard. Consider whether you can offer respondents compensation for their time. You may impute forgone earnings (if any) or opportunity costs, as well as costs of re-charging the phone, maintenance, and usage before you decide on the compensation value. Make sure your respondent has some way of receiving mobile money through M-Pesa, bKash, Paytm and similar apps that have facilitated this for many developing countries. But be aware that usage is still very limited and definitely the poorest have very little access to any or all of this. Do everything you can to pay your respondent. Pay for talk time and more.
4. Communication: A phone is a very important source of information and communication for many especially poor households. They are currently receiving notifications from the government on their phones regarding how to keep themselves safe. Very few poor people (including millions above the poverty line) have access to computers and emails so there are no other substitute forms of communication available to them. The time you keep them on the phone is also the time they may be missing out on important health and social advisories, so keep the survey short. Questions should be framed in an unambiguous manner so that it is easy for respondents to answer.
5. Shared phones: : Poor households will often share mobile phones. Even within a household, there will be multiple members using the same phone. Passing on the phone among households and even members increases health risks. Remember that not all members will have equal access to the phone. Younger males are likely to be the ones who will be able to operate phones and ‘control’ conversations. Acknowledge that this may lead to biases in your survey.
6. Electricity: Given that the poor often have no electricity at home, charging their phones will be difficult and cost money if a long survey leads to drained batteries. In addition to concerns over cost of electricity, often people go to their neighbor’s house to charge the phone, which increases the risk of contagion and is against the idea of social distancing.
7. Scarcity: In their book ‘Scarcity’, Mullainathan and Shafir explain how that scarcity impacts cognitive processes and choices therefore all survey responses will be susceptible to all sorts of biases right now that are beyond our understanding. So be wary of the policy conclusions of your findings.
8. Intimate partner violence: Recent news reports and publications are showing that domestic violence cases are increasing with economic and mental stress. Women and children are particularly vulnerable when they are forced to be in lockdown with abusers who may also be family members. Be extra careful about keeping women on the phone for long right now, there is a potential risk that they would be subject to increased domestic violence due to their prolonged interaction with you on the phone.

If your constraints still warrant a phone survey then make sure that you have informed consent, the respondent is able to charge their phone within their premises, remind them of the importance of handwashing and social distancing during the conversation, and make sure you compensate your respondent generously. Following these strategies can help mitigate the unintended costs that researchers might impose on human subjects.

Stay well!