An Old but Still Relevant Note on Statisitics

A NOTE ON STATISTICS Crosbie Walsh 10.3.02 Pn22.

Quantitative material has shortcomings, as Teresia notes, but for the wrong reason. The major shortcomings of statistics published by ADB, Pacific Community, World Bank and the UN the poverty study are:

1. Few measure poverty directly. Most, including the composition HDI and HPI measures, are based on assumed indirect relationships (e.g., life expectancy , health, education)

2. Most Pacific Island nations do not collect/publish statistics on poverty.

3. Almost all statistics are national figures which show averages or simple percentages. Poverty is a condition of inequality. Considerable differences exists within countries. See examples below.

4. In-country statistics, where they exist, relate to crude political divisions such as provinces.

5. Measures of inequality are generally lacking.

6. Much data are of questionable accuracy, and most are dated.

7. Most comparisons between countries based on existing data lack validity due to differences in definition and vastly different geographies.

In My Opinion …

• The absence of appropriate statistics results in governments, NGOs and aid agencies basing their policies on assumptions, which cannot help but be coloured by personal ideologies and agendas.

• Appropriate statistics are essential to define, locate and monitor poverty.

• No serious study of poverty (and no serious programme of poverty alleviation) would proceed without the “best possible” database, and statistical information would be an important part of that database.

• Data should be relevant to the country it concerns. Not all data will be suitable for pan-Pacific comparisons. Data for pan-Pacific comparisons should be very carefully examined (see note 7, above).

• The data for appropriate statistics must be readily obtainable and verifiable. Some could be obtained from existing censuses and government and quasi-government annual reports. Other data may require small, recurring sample surveys of known populations, such as schools and localities.

• One or more “agency” should be commissioned to develop a poverty database and reference library on poverty (e,g, Pacific Community; USP through the Centre for Development Studies).

SOME MEASURES OF INEQUALITY CONCEALED BY NATIONAL AVERAGES

All of the examples are from my own research.

NUKU’ALOFA 1964 (My MA thesis). Urban households differentiated by six types of house showed major differences in: total living space (from 146 to 36 sq.ft.); the number of rooms (>6 to 1); household heads in farming or casual occupations (6 to 43%); and household annual income means for each housing type (from > T$800 to

NIUE 1976 (Niue 1976 Census of Population and Housing, Vol II.)

Niuean officials assumed little income variation between and within villages, and no “poverty” (taken in my analysis to be a 1976 annual household income of under NZ$200). The results from a sample of seven of the 13 villages showed just how wrong they were:

Village	Hakupu	Avatele	Mutalau	Liku	Tamakautoga	Toi	Hikutavake
Mean H’hold income	3366	2590	2197	1904	1876	1384	1341
% < NZ$200	24	33	49	44	53	50	44

FIJI 1976 (My PhD thesis). Differences in income between squatter and urban village settlements in Suva. The results show significant Fijian-Fijian Indian and City-Urban Area differences in the means and, perhaps more important, in the standard deviations from the mean. Contrary to “local knowledge”, the Fijian poor were significantly “better off” than the Indo-Fijian poor. Other data showed the Fijian “kin exchange” system to be more intact than the Indo-Fijian.

Fijian 1976 $	Fijian City Villagers	Fijian informal housing in Urban Area	Fijian City Renters in Squatter Areas	Indo-Fijian City Squatters	Indo-Fijian informal housing in Urban Area
Mean	63.3	50.9	48.4	39.4	38.1
Standard deviation	36.9	34.1	26.6	21.6	21.2
Standard Error	3.5	5.8	5.3	2.1	2.5
Rank:Best-worst	1	2	3	4	5

FIJI 1996 (My as yet unpublished monograph on housing for the Fiji Bureau of Statistics which was based on the 1996 census). Note: the ranges below concern 15 provinces, and 29 towns and urban areas.

• Safe Water

The ABD Fiji Discussion Paper (March 2001) stated 77% had “access to safe water”. No source was given for this figure. The 1996 Fiji Census show 60% had access to metered (piped) water. The percentage ranged from rural 31% (and provinces from one to 100 percent); urban 93% ranging from 61 to 100%).

• Cooking on Open Fire. Fiji 42.8%; provincial range 8 to 87%; urban range 2 to 62%.

• Wick lamps main source of lighting. Fiji 29%; provincial range 12 to 80%: urban range 2 to 62%.

• Flush toilets. Fiji 46%; provincial range 7 to 76%; urban range 20 to 95 %.

• “Inferior” housing (makeshift, old wood, old tin, etc.) Fiji 25; Provincial range 8 to 43%;

Urban range 4 to 41%.

I trust these figures demonstrate the importance of appropriate statistics in studies of poverty.

A comprehensive literature search would reveal far more information of this type by other researchers.

ACW