Implementation of the Haiti Phone Survey

1. Scope of procurement

1.1 Contracting authority

Flowminder Foundation (Stiftelsen Flowminder), a non-profit organisation based in Stockholm, Sweden, with offices in Sweden, the UK and in Switzerland.

1.2 Subject and description

Flowminder Foundation is calling for tenders for the implementation of a telephone survey among Haiti’s mobile phone users.

1.2.1 Background and Objectives

In partnership with Digicel Haiti, Flowminder is developing a privacy-secure web platform serving mobility and population indicators derived from anonymised and aggregated Call Detail Records (CDRs) to approved third parties to support work in the humanitarian and development sectors.

The objective of the phone survey is to validate Flowminder’s methods for producing mobility estimates based on CDR data, particularly the key concepts (home location, mobility, migration), and to learn about the socio-demographic structure of phone users in Haiti.

1.2.2 Detailed characteristics

Survey design will be drafted by Flowminder and finalised in collaboration with the implementing organisation. Survey guidelines, protocols, and reporting templates will be co-developed by Flowminder Foundation and the implementing organisation.

The implementing organisation will be responsible for:

• Translations of the questionnaire to French and Creole
• Questionnaire programming in CATI software
• Interviewer recruitment, training, coordination and management
• Pilot interviews and adaptations of the questionnaires
• Full data collection
• Initial data cleaning of raw data
• Internal data quality monitoring and assurance
• Transfer of data and meta-data to Flowminder compliant with EU-GDPR and UK-GDPR

Coverage

Geographic coverage includes all departments and communes of Haiti.

Sampling

The planned net sample size is n=7,300 respondents. Three different sampling approaches will be combined.

For subsample A, a pre-screened random sample of phone numbers from all Mobile Network Operators in Haiti will be provided. From experience with similar sampling designs, Flowminder expects the response rate to be at least 50% (of the gross sample). Stratification of the sample based on departments (or communes) will be required via the use of screening questions. The target sample size for subsample A is n=5,000.

For subsample B, to achieve a sufficiently large subsample of the mobile population, additional screening questions on recent trips and relocations will be asked. Quotas will be set to arrive at a subsample of at least n=300 respondents who have done a trip or have relocated in the past year. The share of this mobile population out of the general population in Haiti is known to be around 5% from a previous survey.

For subsample C, a random sample of Digicel users will be generated. The target sample size for this subsample is n=2,000, the expected response rate also >50%. This sampling approach is dependent on the prior approval from Digicel.

Questionnaire

The source questionnaire will be provided in English. Translations will be needed for French and Creole. Questionnaire programming will need to include complex filters, randomisation of modules and/or questions. Flowminder needs to approve the final version of the questionnaire as programmed. The questionnaire will at maximum include four open ended questions. Interview duration should be on average 20 minutes. An interviewer questionnaire of up to 10 questions will also be included.

Training of interviewers

Comprehensive training of all interviewers needs to be implemented (at least three full days). If the security and public health situation allow it, in-person training should be done for all interviewers. A person from Flowminder will join and assist in the training sessions (via online connection).

Pilot interviews

Testing of the questionnaire needs to be undertaken. At least 200 pilot interviews should be conducted, followed by revisions of the questionnaires and debriefing of interviewers.

Duration of data collection

The full survey is expected to take four weeks and is scheduled to start in August 2023. Data collection should be completed in August ideally, with a possible extension into September 2023.

Monitoring

During data collection, daily update reports must be sent to Flowminder. Agreed indicators have to be monitored and communicated to Flowminder. Preliminary datasets (all calls placed) should be shared every 3 days. Call-backs to 10% of respondents have to be implemented for quality assurance following an agreed protocol.

Data Cleaning

The implementing organisation is responsible for initial data cleaning of the raw data. Duplicate cases must be removed or flagged for removal. Text entries in languages other than English must be translated to English. Missing values must be consistently coded to missing. The encryption-protected dataset shall be sent in Stata format (.dta) and must contain complete and accurate variable labels and category labels.

Metadata

As part of the full data, date and time of the interview, interviewer ID and timestamps for each question have to be collected. In addition to the dataset of complete interviews, a (separate) dataset of incomplete or aborted interviews must be delivered.

Cooperation and Reporting

The implementing organisation will work closely with Flowminder Foundation’s Survey Statistician throughout all phases of the survey, and will report milestones and deliverables. Regular updates to Flowminder on progress of preparation and data collection, including agreed control indicators, are required. A codebook in English needs to be produced.

1.2.3 Deliverables and timetable

The telephone survey and all relevant reporting has to be completed by 15 October 2023.

Deliverable	Due date (2023)
CATI programming in English	17 July
Interviewer training	18-20 July
Pilot interviews	21-24 July
Pilot dataset sent to Flowminder	25 July
Revision of questionnaires	26-28 July
Final test of questionnaires	28 July
Debriefing of interviewers	31 July
Start of data collection	01 August
End of data collection	29 August
Survey dataset, initially cleaned	08 September
Field work report (difficulties, events, etc.)	15 September
Codebook for the survey dataset	22 September
Final narrative report & financial report	15 October

 

1.3 Place of performance

Bidders need to be based in Haiti (preferable), in the EU or the UK or a country with an adequacy decision on data protection. A Haiti-based implementing organisation with contextual knowledge will be preferred. After data collection is completed, the data will be transferred to the EU and UK. To this end, a Data Processing Agreement will be set up. Compliance with EU-GDPR/ UK-GDPR is a prerequisite.

1.4 Duration of the contract

12 June to 30 October 2023.

3. Evaluation of tenders and award

Tenders will be evaluated for completeness, technical expertise of the narrative proposal, acquired experience and reference work of the implementing organisation in the required areas of work, plausibility of budget positions and the overall volume of the submitted tender. The award will follow the method of best price-quality ratio.