The 2018 Tonga National Disabiltiy Survey was conducted jointly by the Tonga Department of Statistics (TDS) and the Ministry of Internal Affairs, Social Protection and Disability. It is the first population-based comprehensive disability survey in the country. Funding was provided through number of bodies including UNICEF, DFAT and Tonga Government. The Pacific Community provided technical supports through out different stages of the survey.
The main purpose of the survey is to desctibe demographic, social and economic characteristics of persons with disabilities and detemine the prevalence by type of disability in Tonga, and thus help the government and decision makers in formulating more suitable national plans and policies relevant to persons with disabilities.
The other objectives of the Disability survey were collect data that would determine but not limited to the following:
a. Disability prevalence rate at the national, urban and rural based on the Washington Group recommendations;
b. degree of activity limitations and participation restrictions and societal activities for persons with disability:
c. ascertain the specific vulnerabilities that children and adults with disability face in Tonga
d. establish the accessibility of health and social services for persons with disability in Tonga
e. generate data that guides the development of policies and strategies that ensure equity and opportunities for children and adults with disabilities.
An additional module was included to collect information on people's perception/experiences of service delivery of Goverment to the public.
Kind of Data
Sample survey data [ssd]
Version 01: Clean, labelled and de-identified version of the Master file.
National and island division coverage.
There are six statistical regions known as Divisions in Tonga namely Tongatapu urban area, Tongatapu rural area, Vava'u, Ha'apai, Eua and the Niuas.Tongatapu Urban refers to the capital Nuku'alofa is the urban area while the other five divisions are rural areas. Each Division is subdivided into political districts, each district into villages and each village into census enumeration areas known as Census Blocks.
Unit of Analysis
The survey covers all usual residents of selected households, all children 2-17 years and adults 18 years and above and undertake comparisons between persons with and without disability.
Producers and sponsors
Authoring entity/Primary investigators
Tonga Department of Statistics (TSD)
Ministry of Finance
Technical Assistance Sampling Methodolgy
Government of Tonga
United Nations Children's Fund
Department of Foreign Affairs and Trade
the total number of households to interview approximates 5,500 households based on the budget allocation available
the selection of the sample is based on different steps (see previous section)
Stratification: this sample design is a stratified multi stage random survey. Stratification happened based on the disability status of the households and their geographical residence.
STAGES OF SELECTION:
- the first stage of selection focussed on the selection of Enumeration Areas or Census Blocks as Primary Sampling Unit for households with disability. In total 334 PSUs have to be selected in order to cover the expected sample size.
- the stage 2 of the selection concerns only the households with no disability as all households with disability from the selected EA are selected for interview
Level of representation:
The survey will provide a comparison of the status between households with and without disability at the island group level.
All non-response have been replaced according to the disability status of the household. Disable households that had to be replaced were replaced by another household with disability from the closest block.
The sampling frame used was the 2016 population census. No additional listing were conducted.
The Sampling strategy is designed consistently with the purpose of the survey. The purpose of the 2018 Tonga Disability Survey is not to estimate the prevalence of disability in Tonga, which has been done on a very accurate way in the 2016 Population Census, but to compare the situation of the household with disability with the situation of households with not disability across the 6 geographical zones of Tonga.
The sampling strategy of the 2018 Tonga Disability Survey is based on 2 stages stratified random sample.
The stratification carried out in this survey is based on the disability status of the household:
- strata 1: households who declared at least 1 member in disability (according to Washington Group list of question)
- strata 2: households who did not report any disability member
The sampling frame used in this survey is the 2016 National Population Census that included the set of question on disability (from the Washington Group). In addition to the first set of stratification, the geographical breakdown of Tonga (by 6 island groups) has to be taken into consideration.
The overall idea is to equally split the total sample in both strata (1 & 2), which has been allocated to approximatively 5,500 households.
A replacement procedure is implemented in case of non -response.
The first step is to identify the households with disability from the population census. Households with disability are the households who reported at least 1 member as disable according to the 6functionning domains recommended by the Washington Group (see, hear, walk, remember, self-care, communicate).
In the strata 1, the sample distribution of approximatively 2,750 households was allocated using the square roots distribution of households across the 6 island groups.
The next step consists in determining the number of blocks (Enumeration Areas) to select as Primary Sampling Unit. Again, by getting from the census frame the average number of households with disability in each block by island group will generate the number of blocks to select as PSU.
Within each selected block, all households with disability will be selected for interview.
The strategy for strata 2 (non disable households) is to use the same blocks that have been selected for households in strata 1 and interview within those blocks the same number of households as strata 1.
Here is the final sample - after selection:
Tongatapu urban: 1336
Tongatapu rural: 1884
EA SELECTION (Primary Sampling Units labelled as blocks in the 2016 Tonga census):
The EA were selected using probability proportional to size (size means number of households with disability within the EA). Within all selected EAs, all households with disability are selected for interview, and the same number of household with no disability. Households with no disability to interview in the EA were randomly selected, using uniform probability of selection.
Deviations from the Sample Design
Deviation from the original sampling plan was observed due to challenges in the field:
The main fieldwork challenge was to trace the selected households (that were selected from the 2016 census frame) especially after cyclone Gita that hit Tonga before the field operation. Geography and composition of households have changed (and the household listing was not updated).
Under those circumstances, the total number of households interviewed has changed. Here is the percentage of modification between the original sampling plan and the survey achievements for each of the 2 stratas:
Finally, more households than expected were interviewed.
Looking at the original plan, based on the original selection, here are the response rates:
Tongatapu urban: 82.7%
Tongatapu rural: 81.1%
*in Niuas only 1 island was planned and finally both islands were included within the selection, this explains the reason why
On the overall almost 86% of the selected households were interviewed (4647 households out of 5668 interviewed). The difference represents changes that happened in the field due to:
- households that were classified as disable during the census who are no longer disable (576)
- households that were classified non disable during the census who became disable (440)
Including all the replacements due to the changes in the HH listing, the response rate are higher than 100%.
Sampling weights were calculated at person and household level.
The household weights were generated according to the probability of selection of each household (inverse of the factor of probability of selection of the blocks and probability of selection of the households within that block). The weighting computation process made a distinction between households in strata 1 and households in strata 2.
The household weights were then adjusted to match the total number of households at the island group level.
Due to a lower average household size within all island groups when comparing the census and the survey data, it is not possible to use the household weights at individual level. A specific set of person weights was computed.
The person weight results of an adjustment of the household weights by the age structure of the population in both strata. This set of person weights matches the total household population from the 2016 census.
Dates of Data Collection (YYYY/MM/DD)
Main data collection
Mode of data collection
Computer Assisted Personal Interview [capi]
4 permanent staff of the Tongan Statistics Department (TSD) were assigned as Headquarters for administration supervision of each Island Division. Eleven field supervisors were assigned to a team of roughly three or four enumerators to manage and supervise during the field work.
The Tonga Disability Survey (TDS) mobilised 13 teams to do the survey. Each team comprised of:
- 1 supervisor
This person is responsible for overseeing all activities of the team including management of workload of interviewers, organise field logistics, transport and travel, payments and other teams needs and requirement on the field. The supervisor assigned work to enumerators, maintain and monitor data quality control checks and provide progress report to office.
- 1 Team leader
Assist the supervisor by monitoring the interview progress and provide support to interviewers on the field. The team leader also assisted in the interview and also provide control check on the field.
- 3 Interviewers
The interviewers are the key persons in the field as their roles is to visit the houses and collect the information by carry out the interview with the respondents.
In addition to the fieldworkers, headquarters were also mobilised to undertake quality check of the data and monitoring the team's progress in the field.
- 4 Headquarters from TDS
- 3 Headquarters from SDD
Type of Research Instrument
Tonga Disability Survey 2018 used the CAPI system for the interview. However, the questionnaire was developed manually using excel and word software. The questionnaire was then converted to the CAPI using the Survey Solutions software. The questionnaire has two parts - the household and personal questions.
The Household questionnaire containing questions asking about characteristics of all household members of and about the household characteristics. It contains the following parts:
· Household schedule/roster - listing all members and recording other social and economic information
· Household characteristics - ask about household structure, characteristics, goods, assets and income.
The Personal questionnaire contains questions asking about child functioning among young children (aged 2-4 years) and older children (aged 5-17 years). Questions on adult functioning are also asked of adult aged 18 years and above. The personal questionnaire includes the following sections:
· Young Child functioning for children aged 2-4 years old
· Older child functioning for children aged 5-17 years old
· Adult functioning for persons aged 18 years and older
· Tools and service (2 years and above)
· Needs and availability (2 years and above)
· Transport (2 years and above)
· Health care and support (5 years and above)
· Education (5 years and above)
· Employment and income (15 years and above)
· Participation and accessibility (15 years and above)
· Other social issues (18 years and above).
The development of the questionnaire went through several consultations and review from key partners and stakeholders within and outside Tonga including Tonga National Statistics Office, Non disability and disability offices in Tonga, UNICEF, WG, PDF, UNESCAP and SPC. Though the questionnaire was originally developped in English, it was also translated to Tongan local language. The first draft of the questionnaire was tested during the Pilot training and fieldwork. The questionnaire is provided as an external resource.
The draft questionnaire was pre-tested during the supervisors training and during the enumerators training and it was finally tested during the pilot test. The pilot testing was undertaken on the 27th of May to the 1st of June 2018 in Tongatapu Urban and Rural areas. The questionnaire was revised rigorously in accordance to the feedback received from each test. At the same time, a field operations manual for supervisors and enumerators was prepared and modified accordingly for field operators to use as a reference during the field work.
Tonga Statistics Department
Ministry of Finance
The World Bank Survey Solutions software was used for Data Processing, STATA software was used for data cleaning and tabulation.
Data editing for the Tonga Disability Survey started from the CAPI Questionnaire Design stage where validation checks and enabling conditions were incorporated to ensure that data collected by the Interviewers were of good quality. Validation checks, filtering and enabling condition could be found in the attached questionnaire for reference.
The survey used the Survey Solutions Software for data collection and the monitoring processes involved in the use of its applications was also part and parcel of the Data Quality Assurance to ensure the capturing of good reliable data. Having two levels of data checking and editing through the Supervisor and Headquarter gives more reliability to the data being captured.
All completed questionnaires are synchronized by the Interviewers to their respective Supervisors who checks the questionnaires thoroughly and approves or rejects the questionnaires, depending on the errors found. This allows for validation of errors by the Interviewers while they are still in the field as they can revisit the households to verify the errors.
Headquarters has the final checks on the completed questionnaire which allows the questionnaire to be fully scrutinized before the final batch editing of the data when the survey is completed.
After the completion of the survey, data was then downloaded in Stata and final editing is done to the full data. Data editing file (.do file) is attached, showing what was done in the editing. The .do file is well documented to show what changes were done to ensure data is consistent and reliable. Corrections were done in the program and this is run against the raw data and saved in a different file, to ensure the raw data remain as the original version.
Editing program filename is 'TDS_2018.do'.
Upon the completion of the final editing process and before the generation of tables for the report, a RECODE process has to be done to the data, based on the recommendations of the Washington Group (WG) Short and Extended Sets of functional difficulties for the different age groups.
This is done in Stata as well and the .do file used for the recode is also attached for reference. Different recodes were done for the Age group 2-4, 5-17 and 18+.
New Recoded variable include:
1. Seeing_2to4 and Seeing_2to4_id
2. Hearing_2to4 and Hearing_2to4_id
3. Walking_2to4 and Walking_2to4_id
4. Communication_2to4 and Communication_2to4_id
5. FineMotor_2to4 and FineMotor_2to4_id
1. Learning_2to4 and Learning_2to4_id
2. Playing_2to4 and Playing_2to4_id
3. Behaviour_2to4 and Behaviour_2to4_id
4. SOME_2to4, ALOT_2to4, CANNOT_2to4
5. FunctionalDifficulty_2to4 and FunctionalDifficulty_2to4_id
AGE 18+ (Short Set and Extended Set)
1. vis_ss_adult, VIS_ADULT_id - VISION
2. hear_ss_adult, HEAR_ADULT_id - HEARING
3. mob_ss_adult, MOB_ADULT_id - MOBILITY
4. com_ss_adult, COM_ADULT_id - COMMUNICATION
5. cog_ss_adult, cog_1_adult, cog_1a_adult ,cog_2_adult, cog_3_adult - COGNITION
6. rem_indicator, cog_indicator, COG_ADULT_id - REMEBERING
7. ub_ss_adult, SELFCARE_ADULT_id, ub_1_adult ,ub_2_adult, ub_indicator, UB_ADULT_id - UPPER BODY
8. anx_1_adult, anx_3_adult, anx_3y_adult, anx_indicator, ANX_ADULT_id - ANXIETY
9. dep_1_adult, dep_3_adult, dep_3y_adult, dep_indicator, DEP_ADULT_id - DEPRESSION
10. pain_2_adult, pain_4_adult, pain_4y_adult, pain_indicator, PAIN_ADULT_id - PAIN
11. tired_1_adult, tired_2_adult, tired_3_adult, tired_3y_adult, tired_indicator, TIRED_ADULT_id - TIRED
12. wg_ss_adult wg_ss_adult_id - SHORT SET
13. wg_es1_adult wg_es1_adult_id - EXTENDED SET 1
14. wg_es2_adult wg_es2_adult_id - EXTENDED SET 2
15. wg_es3_adult wg_es3_adult_id - EXTENDED SET 3
16. FunctionalDifficulty_all, FunctionalDifficulty_all_id - INDICATOR FOR ALL AGE GROUP
Filename for the recode is 'TSD WG Disability Identifier.do', which created:
1. New variables for the recodes
2. Wealth Index variable
3. Merging the weight variables into the final data
After the completion of the RECODE, generation of tables were then done. These tabulations were done in Stata as well, based on the table specifications provided by the statistical officer in charge of disability surveys at the Pacific Community (SPC).
Estimates of Sampling Error
Sampling errors were computing in Stata using the survey set package.
Sampling errors were computed with the adjusted finite population corrector such as:
- Fpc1: total number of EA within the strata
- Fpc2: total population within the EA
Sampling errors were computed for the main tables: tables 3.1, 3.2, 3.7 and 3.9 of the document "Sampling Errors table - Tonga 2018 TDS" provided in the External Resources.
"Tonga Statistics Department, Tonga Disability Survey 2018 (TDS 2018), Version 01 of the licensed dataset (December 2018), provided by the Microdata Library. https://microdata.pacificdata.org/index.php/home"
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.
Version 01 (February 2019): This is the first attempt at documenting the Tonga Disability Survey of 2018. Done in Tonga by the Tonga Statistics Department.
Version 02 (July 2019): This is the review of the existing documentation of the Tonga Disability Survey of 2018. Done in Noumea, New Caledonia by the Statistics for Development Division (SPC).