19 Labour Force Survey, Methodology LFS, Methodology © NISR, 2024 where,
20
© NISR
LFS, Methodology
The questionnaire of the Rwanda Labour Force Survey in its present form contains questions organized into
sections dealing with following topics:
A:
Household roster
B:
Education
C:
Identification of employed, time-related underemployed, unemployed and potential labour force
and outside labour force
D:
Characteristics of main job/activity
F:
Past employment
G:
Own-use production of goods and services
H:
Subsistence foodstuff production
S:
Special section (Questions to be added based on requests)
Not all questions are addressed to every household member. For children under 14 years of age, a minimum
number of questions is asked. For individuals aged 14 years and above, the number and type of questions
depend on their activities and he labour force status during the reference period. The basic reference period is
the seven days preceding the interview date, though certain questions employ alternative reference periods,
which are explicitly stated in the questionnaire.
The questionnaire was prepared in both Kinyarwanda and English. An initial version of the Kinyarwanda
questionnaire was tested during the February 2016 LFS Pilot survey. The field test conducted in selected
urban and rural areas, assessed the integrity of the instrument, including question clarity, interview duration
, coding and data processing. Insights from this pilot were instrumental in refining the final questionnaire.
Further modifications were made based on lessons learned during the pilot survey. The revised questionnaire
was subsequently tested through mock interviews conducted during the training of supervisors and
interviewers ahead of the August 2016 and February 2017 LFS rounds. Additional minor adjustments were
made based on fieldwork experiences in previous LFS rounds.
Beginning in 2024, the questionnaire was further updated to incorporate questions aligned with the
International Classification of Status in Employment (ICSE-18). This revised standard is being piloted in
preparation for its full adoption into the LFS starting in the 2025 survey rounds.
It is important to note that the LFS questionnaire was programmed in CSPRO by the NISR data processing
team to ease field operations. The 2024/25 version of the LFS questionnaire is provided in the annex for
reference.
Questionnaire design
2
21 © NISR LFS, Methodology 3.1 Preparations The main pre-survey activities conducted in preparation for field operations include the recruitment and training of interviewers, as well as the preparation of tablets and other logistical preparations. The recruitment begins with the preparation of formal request memos to relevant units, contacting previously trained enumerators and confirming their availability to participate in the upcoming round of data collection. Simultaneously, logistical arrangements are initiated to secure data collection devices (tablets or smartphones), vehicles and other materials. Another essential part of the preparation involves configuring the tablets for data collection. This includes installing of the LFS Computer-Assisted Personal Interviewing (CAPI) application, along with necessary lookup files to be uploaded to the devices, sampled household lists, deployment plans, and Enumeration Area (EAs) maps. These materials help enumerators in navigating and accurately identifying assigned households or units during fieldwork. Before the start of each quarterly LFS field data collection, enumerators participate in a refresher training session to reinforce their understanding of survey instruments and procedures. The training provides detailed feedback from the previous round, highlighting common errors and inconsistencies identified by Data Quality Monitors (DQMs) and Data Analysts. Enumerators are guided on how to avoid these mistakes and improve data quality, with a focus on proper use of the CAPI tool, question interpretation, and strict adherence to survey protocols. 3.2 Fieldwork Data collection Following preparatory activities, LFS staff take their respective responsibilities. As of 2024, the LFS team consists of 2 national coordinators, who oversee field operations and coordinate the work of supervisors; 9 monitors; 4 auditors; 32 team leaders; and 96 interviewers, bringing the total number of field personnel involved in interviewing to 140. Supervisors worked closely with monitors and auditors, conducting field visits across the country to ensure adherence to protocols. Monitors control data quality dashboards, provide feedback to enumerators and team leaders, and collaborate with auditors to review interview recordings when data quality concerns arise. Team leaders, typically the most experienced interviewers, provide real-time guidance to enumerators and hold daily debriefings to review collected data and address issues while still in the field. Fieldwork operations 3
22 Labour Force Survey, Methodology LFS, Methodology © NISR, 2024 Figure 1: LFS Fieldwork organization The quarterly field data collection is conducted in the mid-month of each quarter, specifically February, May, August and November, from the beginning to the end of the month. Listing of households in new rotation groups is also carried out alongside data collection. Additionally, procedures were developed to facilitate the selection of sample households from the updated lists as part of field operations.8 3.3 Data quality control As with any survey, the real time monitoring of data quality is essential to minimize response errors and other errors such as coding and editing mistakes, as well as errors in data entry and processing. to allowit is also essential to allow enumerators to address issues while still within the assigned cluster. To support this, online different data quality monitoring platforms (discussed below) were developed to track data quality and provide timely feedback. Listing application Before selecting households for interviews, a fresh listing is conducted to update the sampling frame. This operation is carried out using the CSPro CAPI application, integrated with high-quality maps generated by the GIS section at NISR. These maps produced in GeoJSON format, are georeferenced and include clearly defined cluster boundaries and are specifically designed to be compatible with the CSPro application used by enumerators in the field, ensuring streamline listing operations. During the household listing phase of the quarterly Labour Force Survey (LFS), a systematic approach was employed to enhance efficiency, accuracy, and balanced workload distribution of the enumerators. Each cluster was assigned to a team of three enumerators. To support this setup, the GIS section subdivided each 8 Mehran, F., GIZ Consultant, “Rwanda Labour Force Survey February 2016. Selection of households without data entry as part of the field operations.” 30 December 2015.
23 Labour Force Survey, Methodology LFS, Methodology © NISR, 2024 cluster into three manageable segments, guided by spatial data and operational requirements. An online listing data quality control (DQC) platform was developed to monitor the completeness and accuracy of the listing activities. This platform is accessible to team leaders, monitors, supervisors, and coordinators. It enables the identification of potential errors, such as structures listed outside predefined cluster boundaries, and ensures full coverage so that no eligible households are missed. This image below illustrates key features of the online listing platform. Figure 2: Online data quality control platform-Listing Source: National Institute of Statistics of Rwanda Main survey application Following the listing and selection of households for interview—currently 12 households per sampling unit—interviewers visit the selected households to conduct interviews using the Labour Force Survey (LFS) questionnaire, programmed in the CSPro CAPI application. To ensure data quality during fieldwork, an online data quality control platform was developed and is accessible to all LFS staff involved in monitoring. This platform is automatically updated with data from the field when enumerators synchronize their devices with the NISR server. The platform provides several key features, including: – Overall interview progress – Progress tracking by team, enumerator, province, district, and date – Number of completed interviews – Refusals categorized by reason Distance between the interview location and the household listing coordinates Duration of interviews per household These features enable survey supervisors and coordinators to closely monitor the progress and performance
24 Labour Force Survey, Methodology LFS, Methodology © NISR, 2024 of individual enumerators and teams, helping to identify and address issues promptly and minimize potential non-sampling errors.The images below illustrate some of the platform’s key functionalities. Figure 3: Online data quality control platform-Main Survey Source: Source: National Institute of Statistics of Rwanda Audit platform Audit Platform is another crucial online Data Quality Control (DQC) application developed to enhance the monitoring of interviews conducted by enumerators. It is primarily used by DQC monitors and auditors based at NISR to ensure the quality and integrity of the data collected during fieldwork. All household interviews are recorded using the LFS CAPI application, and both the data and audio recordings are transmitted to NISR servers. These recordings are then indexed by question, allowing auditors to easily navigate to specific questions and listen to how the enumerators conducted the interviews. This enables a direct comparison between the recorded responses and those entered by the enumerators, helping to identify any inconsistencies. When discrepancies are detected, auditors can either correct the errors directly or communicate with the respective enumerator for clarification or correction. Below is an illustration showing some of the key functionalities of the Audit Platform.
25 Labour Force Survey, Methodology LFS, Methodology © NISR, 2024 Figure 4: Online data quality control platform-Audit Source: Source: National Institute of Statistics of Rwanda Power Bi dashboard This is another essential online platform that receives real-time data directly from the CSWeb environment. It automatically displays both the main Labour Force Survey (LFS) indicators and quality control indicators. This enables daily monitoring of data trends and facilitates timely discussions whenever unusual patterns or issues arise. Here below are some
26
Labour Force Survey, Methodology
LFS, Methodology © NISR, 2024
Figure 5: Online data quality control platform-Power BI dashboard
Source: Source: National Institute of Statistics of Rwanda
Other data quality control mechanisms
In addition to the online DQC platforms discussed above, other data quality checks performed by NISR staff
were summarized as follows:
High-Frequency Checks : Inconsistencies and errors in the LFS data are identified daily and feedback is
provided. This process follows a step-by-step approach, where STATA syntax developed by LFS analysts is
executed and errors are compiled. These are then given to monitors, who forward any issues to the team
leaders in the field for investigation and correction. Once the team leaders addressed the issues, they sent the
files back with comments on each correction. Supervisors then reviewed these comments and, if necessary,
reached out for further clarification.
Spot-Checks: During the quarterly LFS data collection, coordinators and supervisors conduct unannounced
visits to various locations countrywide. They visit different teams to assess team organization and follow up
on any issues, ensuring the quality of the data collected.
Overall, the quality assessment of labor force survey data is conducted thoroughly to minimize sampling
errors, response errors and non-response errors to the greatest extent possible.
27
© NISR
LFS, Methodology
Data security is a critical component in protecting confidential information, ensuring the privacy of respondents,
and complying with relevant protocols and regulations. In the Labour Force Survey (LFS), various technologies
are employed across different stages—data collection, transmission, and reception—to uphold data security
and privacy.
Initially, LFS processes relied on File Transfer Protocol (FTP) to transmit data from the field. However, this
approach had several limitations, including delays in data retrieval and extended processing time due to
manual editing requirements. To address these challenges, FTP was later enhanced with additional security
features, and a more robust system—CSWeb—was introduced.
CSWeb offers a significant upgrade by integrating Transport Layer Security (TLS) to encrypt data during
transmission. This ensures secure communication between enumerators' tablets and the central data
reception system. The adoption of CSWeb brought several benefits, including real-time data transmission from
the field, minimized delays in accessing data, and reduced the need for time-consuming editing procedures.
Since tablets are the primary data collection devices in LFS, stable internet connectivity—provided by major
service providers in Rwanda—is crucial for uninterrupted data transmission. The data reception and hosting
environment is built around a centralized CSWeb server, which enables seamless, real-time ingestion and
monitoring of incoming field data.
Data security and processing
4
28 © NISRLFS, Methodology The remote analytics environment for LFS was established on a Windows-based platform, supporting simultaneous access by multiple users. To ensure secure data access and remote connectivity, a Virtual Private Network (VPN) was configured. This setup provides a secure and encrypted communication channel, allowing only authorized users to access the system. LFS data are accessed via Remote Desktop Protocol (RDP), with analysts, coordinators, and consultants granted individual login credentials to connect securely. This method significantly enhances data security by preventing local storage of sensitive datasets on personal devices, thereby minimizing the risk of data loss, theft, or damage that could compromise respondent confidentiality. Once the data collection is complete and all data are successfully received, analysts perform additional data cleaning and validation using STATA software. This process results in a finalized, high-quality dataset. Subsequently, sampling weights are applied, and a series of derived variables are generated based on the collected responses. The enriched dataset, now containing both sampling weights and derived variables, is used to compute key labor force estimates as specified in the survey’s tabulation program, along with other analytical tables included in the main report. For transparency and reproducibility, a schematic overview of some STATA syntax used to construct derived variables is presented in the figure below. In addition, the questionnaire, highlighting the questions used to generate derived variables, is provided in Annex 4. Figure 6: some derived variables from LFS questions Derived variable: status1 (Employment status) Derived variable: employed16(Employment to pop ratio) and UR1 (unemployment rate) Derived variable: PLF (Potential labour force) Data analysis and reporting writing5 29 Labour Force Survey, Methodology LFS, Methodology © NISR, 2024 Derived variable: LUU (Labour underutilization) Derived variable: TRU (Time-related underemployment) Tabulation: STATA syntax for summary labour force indicators Source: National Institute of Statistics of Rwanda 30 © NISR LFS, Methodology Annex 1: Rwanda LFS theoretical Sample Rotation Scheme 2-2-2 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 1 1D 2 2C 2D 3 3C 3D 4 4C 4D 5 5B 5C 5D 6 6A 6B 6C 6D 7 7A 7B 7C 7D 8 8A 8B 8C 8D 9 9A 9B 9C 9D 10 10A 10B 10C 10D 11 11A 11B 11C 11D 12 12A 12B 12C 12D 13 13A 13B 13C 13D 14 14A 14B 14C 14D 15 15A 15B 15C 15D 16 16A 16B 16C 16D 17 17A 17B 17C 17D 18 18A 18B 18C 18D 19 19A 19B 19C 19D 20 20A 20B 20C 20D 21 21A 21B 21C 21D 22 22A 22B 22C 22D 23 23A 23B 23C 23D 24 24A 24B 24C 24D 25 25A 25B 25C 25D 26 26A 26B 26C 26D 27 27A 27B 27C 27D 28 28A 28B 28C 28D 29 29A 29B 29C 29D 30 30A 30B 30C 30D 31 31A 31B 31C 31D 32 32A 32B 32C 32D 33 33A 33B 33C 33D 34 34A 34B 34D 34D 35 35A 35B 35C 35D 36 36A 36B 36C 36D 37 37A 37B 37C 37D 38 38A 38B 38C 38D 39 39A 39B 39C 39D 40 40A 40B 40C 40D 41 41A 41B 41C 41D 42 42A 42B 42C 42D 43 43A 43B 43C 43D 44 44A 44B 44C 44D 45 45A 45B 45C 46 46A 46B 47 47A 47B 48 48A 48B 49 49A 2028 2029 Group/Qs 2024 2025 2026 2027 2030 2031 2032 2033 2034 Note: letters, A, B, C, D means first, second, third and fourth appearances, respectively of a PSU in the quarterly sample Annexes 6