MAKING IT USEABLE    Weighting       Impact of Using Unweighted Data

If unweighted data are used, THE WHOLE COUNTRY WOULD SEEM TO LOOK LIKE NEW YORK AND MASSACHUSETTES BECAUSE:

• There are approximately 21,120 households from across the nation in the sample
  
  
• There are approximately 20,900 additional households in the add-on samples
  
  
• Therefore, the add-on samples will overpower the national data if unweighted data are used

The following example shows that because of add-on samples for Massachusetts (in the New England Census Division), New York (in the Mid-Atlantic Census Division), and Tulsa and Oklahoma City (West South Central) the share of the population based on the unweighted sample would not reflect actual population patterns.

Table 1

Population	Number of Persons Sampled	Actual Number of Persons (000)	Percent of Actual Persons
New England	19,056	11,827	4.9 %
Middle Atlantic	29,128	35,695	14.8%
East North Central	8,439	39,294	16.3 %
West North Central	3,264	17,306	7.2%
South Atlantic	9,083	45,943	19.0 %
East South Central	2,672	12,783	5.3 %
West South Central	13,107	26,062	10.8 %
Mountain	2,648	14,744	6.1 %
Pacific	7,963	38,022	15.7 %
All	95,360	241,675	100.0 %

---