Geographic selection on any geographic variable down to the census block level (i.e., state, county, MSA, DMA, exchange, ZIP Code, ZIP+4, census tract, census block group, and census block).
 
Telephone exchange level estimates for over 48 demographic variables (e.g., age, income, home ownership, education, race, etc.).

Modified RDD samples that provide higher efficiencies while maintaining the integrity of the sample frame.

The standard GENESYS RDD methodology produces a strict single stage, epsem sample of residential telephone numbers. In other words, a GENESYS RDD sample ensures an equal and known probability of selection for every residential telephone number in the sample frame.

The database structure and the sampling methodology obviate the need to utilize a two stage or double sampling process to ensure representative telephone samples. Thus, GENESYS RDD samples will deliver the full statistical value of each interview, without the reduction in precision associated with other methodologies.

GENESYS RDD samples achieve their statistical efficiency through a highly structured database in combination with single stage sampling procedures.

A. Structure of the GENESYS Database

The GENESYS database is structured using twenty independent strata, ­the nine census divisions plus AK and HI split by metro and non-metro county definitions...

• Metro Strata
  • within each stratum, Core Based Statistical Areas (CBSA) are ordered largest to smallest.
  • within each, CBSA’s are ordered by Metropolitan Statistical Area (largest to smallest) followed by Micropolitan Statistical Area (largest to smallest).
  • within each, counties are ordered by those containing the principle city(s) (largest to smallest), followed by those serving each of the remaining component county(s) (largest to smallest).
  • and within each county, exchanges are ordered largest to smallest.
• Non-metro Strata
  • within each stratum, counties are ordered geographic serpentine pattern by state.
  • within each county, exchanges are ordered largest to smallest.

The rationale for ordering the GENESYS database in such a fashion is to ensure strict geographic representation and to increase the homogeneity within the implicit strata created by the GENESYS sampling procedures. The result is that expected sampling variations are actually less than those of an equivalent simple random sample (SRS) of the same size.

B. The GENESYS Single Stage RDD Sampling Procedure

The GENESYS RDD methodology utilizes an unrestricted random sampling process within implicit sampling intervals. Operationally, the number of intervals, or implicit strata, is equal to the total RDD sample size desired.

For example, let’s say a hypothetical sample frame contains 110 area code/exchange combinations, and of the 11,000 (110 x 100) possible two-digit (hundred series) banks, 60%, or 6600, are working. If we need a sample of 1000 random telephone numbers, the GENESYS process would proceed as follows:

1. Compute the implicit interval/size (C). C = M x 100/N Where: C = the interval size M = the number of "working banks" in the defined sample frame. N = the RDD sample size using the numbers from our example, C = 6600 x 100/1000 C = 660

2. Generate a random number between 1 and C (660 using our hypothetical numbers).

Note: A systematic, or nth selection process, may result in a bias if suitable precautions are not taken. For example, because the metro stratum is ordered from the largest to the smallest MSA, we would expect exchanges near the beginning of any interval to be associated with larger metro areas, on average, than those near the end of any interval. This means that the smaller the sample (and hence the larger the interval), the more likely it will be that such bias will occur.

3. Identify the actual telephone number associated with that random number.
This is accomplished by dividing the random number by 100 and adding one to the non-decimal portion. This identifies the sequential area code/exchange working bank combination, while the last two digits of the random number represent the actual random two-digit suffix.

4. Return to step 2 and generate a random number for the next interval (i.e., between 661 and 1320). Steps 2 through 4 are repeated until the sample file is exhausted.

Following completion of the generation process, the sample is replicated according to project specifications and a sorted sample file is available for output in whatever format and containing whatever information is desired.

As the above example demonstrates, GENESYS assigns to each and every number within every interval (and consequently to each and every possible area code, exchange, and four digit suffix) a known and equal probability of selection.

Note: There is, however, no guarantee that each and every telephone household has an equal probability of being selected due to the fact that roughly 16% of telephone households have two or more different telephone numbers

See also our Related Document - Bias in RDD Sampling.