Chapter 3 Costs and Errors of Covering the Population

 

Definitions of Populations Relevant to the Survey

-          Coverage error

o   Arises from failure to give some units in target population any chance of being included in the survey

o   Function of both proportion of target population not covered by frame, and difference on survey statistic between those covered and those not covered.

o   Eg, lower coverage of young black males

-          Population of inference

-          Target population

-          Frame population

-          Survey population

 

An Approach to Coverage Error In Analytic Statistics

-          Inferential populations in analytic statistics

-          Treatment of inferential populations in survey analysis

 

Components of Coverage Error

-          A perfect frame is signified by one to one correspondence between frame population cases and target population cases.

-          Under coverage / non coverage

-          Existence in frame population of elements that are not members of target population / foreign elements

-          Duplication / over coverage

o   Can be eliminated at frame construction phase

o   "is this the only telephone number assigned to this household"

-          One frame element corresponds to several target population elements

-          In one sense, the frame population is the researcher's window into the target population and all efforts necessary to measure the purity of that window must be executed prior or during the survey itself.

 

 

Coverage Problems with the target population of households.

-          Household survey

-          Coverage of housing units

-          Definition of housing units – the almost infinite number of situations that people have for their day to day living arrangements challenges any definition of a target population based on residences.

-          Coverage errors for persons within housing units

o   Extra probing questions

-          Telephone number frames for household population

o   Coverage bias due to telephone frame when describing full household population will be sensitive to nature of survey statistic

 

Measurement of and Adjustments for Non Coverage Error

-          Multiplicity sampling and measurement

-          Multiple frame methods

o   Combine 2 or more inadequate frames in same survey to offer coverage rates that exceed that of any single frame

o   Eg, telephone based survey, combined with area frame

-          Use of post survey adjustments for coverage error

 

Survey Cost issues Involving Coverage Error