METHODOLOGICAL NOTE
There are two primary methods of sampling Hispanic households: random digit dialing (RDD) and listed surname. Each has cost and statistical limitations. A dual-frama approach, combining both, can be an effective compromise.
Making the choice between a listed surname or a 100% RDD approach when sampling Hispanics can be difficult. Statistical issues argue strongly for the RDD approach, while budgetary issues can be a compelling reason for using a surname approach. The trade-off is basically between full-coverage at lower incidence levels and partial coverage with higher effiiciency.
The alternative methodologies discussed below differ in that they produce different levels of representation and sample efficiencies. That is, in order to increase the liklihood of reaching a Hispanic household one must compromise, to some extent, on representation and coverage of the target universe.
Sampling with RDD
The first alternative is to utilize an epsem RDD sample which will provide full coverage of all telephone households. GENESYS also contains exchange level estimates of Hispanic households. For any geographic area, a Demographic Distribution report can be produced, which will provide information for stratifying exchanges based on incidence of Hispanic households, and/or to identify very low incidence exchanges so that they can be removed from the sample frame.
Alternatively, the sample can be targeted towards exchanges based on their relative incidence of Hispanic households, thus increasing the sample's efficiency. This is achieved by using the alternative RDD methodologies within GENESYS which sample exchanges based on a defined Measure of Size (MOS).
The MOS for this application would be the number of Hispanic households in the exchange. While the use of these alternative RDD methodologies will not produce a true self-weighting RDD sample (since exchanges are sampled disproportionately), probabilities of selection for each phone number are provided for weighting purposes.
The net result of either RDD method is a sample that maintains total coverage of Hispanic households (except those excluded by eliminating low incidence exchanges).
Sampling with Listed Surnames
This approach utilizes a Hispanic surname database in conjunction with a listed household sample frame. Hispanic Surname information is used as the criterion for selecting records from the listed household database. This approach will produce very efficient samples, in terms of reaching Hispanic households. However, in achieving this efficiency, significant compromises in coverage result. In fact, more than 60% of Hispanic households will be excluded:
- Those with unlisted phone numbers
- Those with telephone numbers assigned after their phone book was published
- Those without Hispanic surnames by birth, marriage, or legal name change
There will, of course, be a minimal number of non-Hispanic households selected, in that some individuals may have Hispanic surnames byt not self-identify as, or be, Hispanic.
The Dual Frame Alternative
One excellent alternative to either an RDD or listed surname approach is a combination of the two methods. A Demographic Distribution report details the Hispanic incidence by exchange for the defined geographic sampling frame. The decision then becomes one of allocating resources between "high incidence" exchanges using an RDD methodology, and the lower incidence exchanges where a listed surname sample is used.
The exchange report provides estimates of Hispanic households from which the relative populations in the "low" and "high" incidence frames can be determined. With this information, an allocation of total interviews between the two frames can be accomplished. (Note: the information required for developing an optimal allocation of interviews is included in the exchange report itself, since both incidence and relative population size is included).
The fact that the sampling process always involves a series of trade-offs is nowhere better illustrated than with the alternatives of Hispanic sampling. Please feel free to direct any further questions or concerns regarding this or any other sampling issues to GENESYS Sampling Systems.