High Frequency Cell Phone Survey on the Socio-Economic Impacts of Ebola 2014-2015
As of June 7, 2015, Sierra Leone had reported more than 12,900 cases of Ebola Virus Disease (EVD), and over 3,900 deaths since the outbreak began. The Government of Sierra Leone, with support from the World Bank Group, has been conducting mobile phone surveys with the aim of capturing the key socio-economic effects of the virus. Three rounds of data collection have been conducted, in November 2014, January-February 2015, and May 2015. The survey was given to household heads for whom cell phone numbers were recorded during the nationally representative Labor Force Survey conducted in July and August 2014. Overall, 66 percent of the 4,199 households sampled in that survey had cell phones, although this coverage was uneven across the country, with higher levels in urban areas (82 percent) than rural areas (43 percent). Of those with cell phones, 51 percent were surveyed in all three rounds, and 79 percent were reached in at least one round.
The main focus of the data collection was to capture impacts of EVD on labor market indicators, agricultural production, food security, migration, and utilization of non-Ebola essential health services.
Kind of data
Sample survey data [ssd]
v01 - edited, anonymous datasets for public distribution
The documented datasets - slhfcps_r1, slhfcps_r2, and slhfcps_r3 - correspond to each of the three survey rounds.
Due to differing characteristics between responding and non-responding households, the results should be considered “descriptive” rather than representative of the Sierra Leonean population. Overall the response rate was higher than expected given the nature of the survey and the difficult conditions under which it was conducted. In Sierra Leone, of the 4,199 households interviewed in the LFS, 65.8 percent (2,764 households) recorded a cell phone number for the household head, and, of those, 80.0 percent responded to at least one round of the cell phone survey. The unweighted sample was 59.1 percent urban (2,483 households) and 40.9 percent rural (1,716 households). Of urban households, 81.4 percent (2,021 households) listed a cell phone number for the household head, and, of those, 88.1 percent (1,780 households) responded in at least one of the three rounds of the cell phone survey. Of rural households, 43.1 percent (740 households) listed a cell phone number for the household head, and, of those, 58.1 percent (430 households) responded in at least one of the three rounds.
Unit of analysis
- Household Head
All households from the 2014 Sierra Leone Labor Force Survey which provided cell phone numbers.
Producers and sponsors
Statistics Sierra Leone
Innovations for Poverty Action
World Bank Group
Technical & Financial Assistance
The sampling frame for the cell phone survey was the Sierra Leone Labor Force Survey (LFS) 2014. The LFS is a nationally representative stratified cluster sample survey conducted in July and August 2014, and includes the oversampling of urban areas. As part of the LFS, a total of 4199 households in 280 enumeration areas were interviewed. Interviewers collected the phone number, if available, for the head of household, and 2,764 households interviewed in the LFS included phone numbers. All available numbers from the LFS were included in the cell phone survey. The phone numbers were reported for 43 percent of rural households and 82 percent of urban households. Those households reporting numbers are unevenly distributed across the sample though there is at least partial coverage in all districts, ranging from 93 percent in Freetown (Western urban) to 30 percent in Kailahun district.
Overall the response rate was higher than expected given the nature of the survey and the difficult conditions under which it was conducted. In Sierra Leone, of the 4,199 households interviewed in the LFS, 65.8 percent (2,764 households) recorded a cell phone number for the household head, and, of those, 80.0 percent responded to at least one round of the cell phone survey.
The unweighted sample was 59.1 percent urban (2,483 households) and 40.9 percent rural (1,716 households). Of urban households, 81.4 percent (2,021 households) listed a cell phone number for the household head, and, of those, 88.1 percent (1,780 households) responded in at least one of the three rounds of the cell phone survey. Of rural households, 43.1 percent (740 households) listed a cell phone number for the household head, and, of those, 58.1 percent (430 households) responded in at least one of the three rounds.
The base weights for the cell phone survey were the probability weights from the Labor Force Survey (LFS). Sampling weights for the LFS households were calculated by,
Household weight = 1/(PEA,strata * PHH,EA)
where PEA,strata is probability of EA being selected within strata, and, PHH,EA is probability of household being selected within the EA.
To account for higher likelihood of more populated EA’s being selected, PEA,strata is calculated as,
PEA,strata = (nEA,strata * NHH,EA)/NHH,strara
where nEA,strata is number of EA’s selected within the strata, NHH,EA is the total number of households within that EA, and, NHH,strara is total number of households across all EAs in that strata.
Household selection probability was calculated using,
PHH,EA = nHH,EA /NHH,EA
To compensate as much as possible for non-response and low coverage rates, an attrition adjustment was applied. A propensity score adjustment, which uses the available characteristics of the household head from the LFS (age, gender, location, and employment sector) to calculate an aggregate probability of response, was calculated. These calculations need to be done separately for each combination of data sets, meaning the attrition calculations between the LFS and round 1 would be different than those between the LFS and round 2, which would also be different than those between the LFS and households that answered in both rounds 1 and 2. As an example the results of this analysis between the LFS and round 1 of the cell phone survey are presented in Table A1 in the appendix of the Basic Information Document. The inverse of this probability is then applied to the probability weights, therefore increasing the weight for underrepresented groups. As a final step, a post-stratification correction was applied, adjusting the weights to match known population totals at the district and urban/rural levels.
Dates of collection
Mode of data collection
Computer Assisted Telephone Interview [cati]
As the survey was administered by telephone, the length of the questionnaire was targeted as 20 to 25 minutes. In Round 1, the questionnaire focused on employment and labor market conditions, non-agricultural business operations, agricultural activity, food security, health responses (covering only fever and pregnancy), remittances, travel, trust and knowledge about Ebola. In Round 2, questions were added on social assistance and education on the radio, and there were small changes to the existing questions based on the results from Round 1.
Questions on earnings were revised to match the Labor Force Survey questions more closely, in particular to account for earnings that were expressed in time unit other than months, and questions on the incidence and treatment of child diarrhea were adding using identical wording to the Demographic and Health Survey (DHS). The most substantial changes were to the migration section as the Round 1 analysis found inconsistencies in the migration reporting. Details of these changes can be found in the Round 2 report. In Round 3, the agriculture, social assistance, and education sections were expanded while the trust section was dropped due to limited variation between Rounds 1 and 2.
The only questions on Ebola Virus Disease (EVD) specifically were in Round 1 and focused on whether the respondent had heard of Ebola and what were their main sources of information were. This section was placed at the end of the questionnaire in order to elicit unbiased responses in other sections, since people may be distrustful of the government especially regarding Ebola, at a time of such emergency.
Questions related directly to incidence of EVD within the household were excluded for two reasons. First EVD is a relatively rare event and the sample was unlikely to yield sufficient observations for meaningful analysis, and secondly, the respondents will be called repeatedly as part of the high frequency survey therefore it was necessary to avoid sensitive questions that may increase attrition in later rounds. The included questions were worded in such a way as to facilitate differences-in-differences comparisons. The vast majority of questions were identical in their wording to those asked during the LFS or other nationally representative surveys for which detailed data were available including the DHS, the National Public Services Survey (NPS) and the Agricultural Households Tracking Survey (AHTS).
In a few cases, the time period over which the questions were asked was shortened to make it relevant to the last few months during which the outbreak has been growing. For example, the NPS asked about remittances in the last year whereas in November 2014, respondents were asked about remittances received in the last month.
Statistics Sierra Leone
The datasets were cleaned and compiled by teams from Innovations for Poverty Action and the World Bank's Poverty Global Practice and Social Protection and Labor Global Practice.
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including country, acronym and year of implementation)
- the survey reference number
- the source and date of download
Statistics Sierra Leone. Sierra Leone High Frequency Cell Phone Survey on the Socio-Economic Impacts of Ebola 2014-2015, Ref. SLE_2014-2015_HFCPS_v01_M. Dataset downloaded from [URL] on [date].
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.
Kristen Himelein Kastelic
Poverty Global Practive, World Bank
Nina Rosas Raffo
Social Protection and Labor Global Practice, World Bank