KAP COVID Study Methods
This study was implemented in collaboration between John Hopkins Bloomberg School of Public Health’s Center for Communication Programs with the Massachusetts Institute of Technology (MIT), the World Health Organization and Facebook’s Data for Good. The methodologies described in this document are a cursory overview of the data coordination and sampling methodologies that were utilized. A more robust description of those procedures is forthcoming and will be made available in this location when available.
Scope of Survey. Sixty-nine countries were identified for recruitment of Facebook users at the outset of this study. Countries were selected where Facebook usage was sufficiently widespread and could plausibly result in samples representing a range of national characteristics. Twenty countries were identified as longitudinal (wave) countries and another 49 countries were identified as cross-sectional (snapshot) countries. In the end, two countries were removed due to inadequate response rates. See the full list of countries and designation below.
Recruitment. Individuals in the focus countries responded to an advertisement within their Facebook feed that was designed and implemented directly by Facebook’s Data for Good. When clicking upon the recruitment advertisement, individuals were pushed to a survey portal operated by the team of researchers at MIT. Individuals clicking on the link were presented first with consent information and age verification questions to restrict recruitment to adults over 18 years of age. Then, if they consented, they progressed through the survey with the option of withdrawing or not responding to questions at any time. No compensation of any kind was provided to participants. The Committee on the Use of Humans as Experimental Subjects review board at MIT provided ethical approval for this study (# E-2294).
The survey went live on July 6, 2020 and data will continue to be collected until at least October
2020 for the wave countries. The survey was translated to 51 languages and participants in many countries could choose the language in which to take the survey. As of August 8, 2020, over 300,000 completed surveys were received from 67 countries. The full survey instrument is provided below. Target sample sizes for snapshot and wave countries was 3,000 per data collection event (wave countries 8 x 3,000 = 24,000).
Weights & Data. All of the statistics generated in this dashboard were weighted using sampling weights determined by teams of statisticians at MIT and Facebook. A full account of the decisions surrounding the creation of the sampling weights is forthcoming.
External data were used to categorize the overall severity of the epidemic in the focal countries. Data were retrieved from the COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University to generate a case count and caseload trend variables. The date when case counts were retrieved coincided with the launch of the first wave or when the snapshot survey was launched. For the 47 cross-sectional countries, data reflects case counts on 7/20/2020 and for the 20 longitudinal countries it reflects 7/6/2020. The date when case counts were retrieved for the 47 cross-sectional countries, data reflects the caseload trend from 7/6/2020 to 7/20/2020, while for the 20 longitudinal countries it reflects from 6/22/2020 to 7/6/2020. Countries were classified according to which tercile it fell into (high, medium and low) according to their COVID-19 severity on both variables.
Survey. The survey content was divided into eight blocks. Every individual was presented with the first five blocks of questions which addressed: trust in information sources, knowledge, vaccine acceptability and healthcare, and demographics. In snapshot countries, all respondents are shown an information block and then three additional blocks that are randomly selected from the remaining blocks, which include: Information needs, basic knowledge about COVID-19, distancing familiarity, importance and norms, risk perceptions and locus of control, prevention behaviors in practice, behavioral measures taken, beliefs about efficacy, work, and intentions to visit locations if open and open with precautions. In multi-wave countries, respondents are shown four randomly selected blocks. By design, not every participant was asked to respond to every question in the survey.