Data Collection Dates start 2013-04-10 end 2013-05-05
Data Collection Mode Face-to-face [f2f]
Data Collection Notes
Pre-Test: A pre-test field exercise for the LFS April 2013 was conducted on Wednesday 26th January 2013 and Thursday 27th January 2013 in the Orange Walk District in the rural villages of Shipyard, August Pine Ridge and Trinidad and in the urban areas of San Jose Palmar and Orange Walk Town. Personnel from the Statistical Institute of Belize (SIB) were responsible for conducting the pre-test.
Training of Field Workers: Training of 49 Field Workers was conducted over a two-week period from 13th to 25th June. Training sessions were held in Belize City at the University of Belize, Faculty of Management and Social Sciences. Participants were housed in nearby hotels. Training included plenary sessions, intensive smaller group meetings, practice sessions, one day collecting real data in the field and a final summary of the entire process. Special attention was given to editors and field supervisors. Along with the training received by all participants, these persons received special instruction on handling questionnaires, editing questionnaires and field supervision.
Secondary Editing: Secondary editing programs were not in place during the data collection phase of the survey.
Field Checking: Field checking is designed to identify any problems in the fieldwork as quickly as possible to allow correction by SIB staff. The field tables were produced prior to data collection to yield counts of selected variables from which indicators were calculated to monitor the progress of the fieldwork.
Sensitization: A sensitization campaign was executed by the SIB. The efforts include newspaper advertisements and TV messages on local TV stations nationwide. The stations included:
- Wake Up Belize
- Open Your Eyes
- Love FM 4. Belize Watch
The Mennonite communities in Corozal, Orange Walk, Stann Creek and Toledo Districts were visited by SIB staff in the week of 4th July to sensitize the leaders of those communities to the benefits of LFS and to pave the way for interviewers to their communities.
Data Entry and Data Processing
Data entry and data processing was scheduled for 3rd April to 11th May, 2013. Preparation of tables and preliminary summaries were scheduled for 9th June to 11th July, 2013.
Questionnaires
The questionnaires for the LFS were structured questionnaires based on previous LFS 2012 questionnaires with very few modifications and additions.
A household questionnaire was administered in each household, which collected various information on household members including sex, age, relationship, ethnicity, country of birth and education.
The individual person’s questionnaire includes types of training, employment and underemployment, occupation and income.
Data Collectors
Enumeration:
In April 2013, there were 54 interviewers recruited country wide to complete data collection within 4 weeks from Sunday, April 7th to Saturday, May 4th for a total of 2,502 sampled households (1,601 rural and 901 urban). Eighteen (18) interviewers worked in urban areas while 36 worked in rural areas. Eight Assistant Statistician IIs and 4 HQ staff acted as field supervisors in each district. Belize City was split into Northside and Southside and Cayo district was split into 2 with each district having an ASII assigned to the area. Fluent Spanish-speaking field staff from Corozal and Orange Walk conducted interviews in 3 enumeration districts (EDs) in San Pedro mainly due to having several Spanish speaking households in the island and the availability of the field staff. These interviewers were led by a temporary Field Supervisor and the Belize District Assistant Statistician II (ASII) performed spot checks. In each district, both the Assistant Statistician II and one HQ staff each had a field check day for every week, so 2 field visits were conducted per week per district. Regular office checks were also done by ASIIs. There were some EDs, for example, Little Belize in Corozal district, Indian Creek/Indian Church/San Carlos/ Kakabish in Orange Walk district, Calla Creek in Cayo district, Burrel Boom Road/Hattieville/Western Paradise and Mahogany Heights/Mile 25 in Belize district, and Trio/San Pablo/Farm 21 and 22, San Miguel/Silver Creek, Jalacte and Santa Teresa/Mahbilha in Toledo district, that were completed in teams due to far distances needed to be travelled or wide coverage.
Table 7
Distribution of interviewers and households by district and Urban/Rural
| District | No. of interviewers | No. of Urban Households | No. of Rural Households |
|---|---|---|---|
| Corozal | 9 (2 urban, 7 rural) | 100 | 300 |
| Orange Walk | 10 (2 urban, 8 rural) | 125 | 325 |
| Belize (northside and southside | 7 (5 urban, 2 rural) | 300 | 100 |
| Cayo (Belmopan and Cayo) | 8 (5 urban, 3 rural) | 200 | 200 |
| Stann Creek | 10 (2 urban, 8 rural) | 100 | 300 |
| Toledo | 10 (2 urban, 8 rural) | 76 | 376 |
Data Processing and Editing and Coding:
In April 2013, data was collected using paper administered questionnaires. The information collected was scanned into a database with a software called Teleform. With the use of this software, precision and legibility were crucial in capturing accurate data, so interviewers were asked to use only 2B pencils to fill out questionnaires. The design of the questionnaires incorporated bubbles (to be properly shaded) and lines where necessary descriptions and specifications were needed. Interviewers were trained to write responses clearly and shade bubbles accordingly, otherwise the scanner would not pick up the responses.
After the questionnaire was checked thoroughly by the interviewer, it was checked by field supervisor, then by an editor/coder, then by an ASII, until it was scanned and verified by data entry staff. A check list was prepared for interviewers and editors/coders to follow in order to minimize errors. Eight editors/coders were stationed in each district and their role was to check the consistency of each paper questionnaire, code occupations and industries at the first digit level and advise the ASIIs of any inaccuracies. If the occupations/industries could not be coded, then questionnaires were sent back to the field for clarifications. Questionnaires that were signed by ASIIs or HQ staff were then sent to HQ to be scanned and processed through the Teleform reader. Editors/coders started the week after data collection and ended a week or two after. On average, editing/coding of questionnaires was completed in 5 weeks.
Temporary verifiers were also recruited to verify the data that had been processed by Teleform. About 2 verifiers were stationed in Belmopan and were supervised by a headquarter staff in the IT section. IT staff was responsible for trimming the edges of each questionnaire and prepare these to be processed through the scanner. Once these were processed, the software highlighted areas of the questionnaire which could not be read, so the verifiers’ role was to go through the scanned images and make necessary corrections. Verifiers worked during and after the data collection period, taking them a maximum of two weeks after data collection.
Further to the verification process, a data cleaning phase followed where excel error sheets were produced by the IT department whereby inconsistencies were captured and needed to be corrected and updated by ASIIs. This was a manual process with the use of images and cross-referenced with an Access LFS database.
There were many drawbacks to using this software since smudges or pencil marks or improper erasing would be read by the scanner. Interviewers were also strongly advised to keep questionnaires intact and avoid folding the edges. However, it was difficult for the interviewer to return a crispy and smudge free questionnaire due to the obvious working conditions: rain, mosquito swatting, walking through unavoidable mud, etc. Also, there were inconsistencies in data processing where description for a ‘main job’ spilled over the section for ‘other job’ when an individual only had one job. There were also few cases where individual questionnaires could not be tracked, even though each individual questionnaire had a barcode.
Estimates of Sampling Error None