sample="quota" bates="502455761" isource="rjr" decade="1980" class="ni" date="19820701" June 1, 1982 TO: Dick Scotson cc: T. Trumbull J. Cummins T. Markwood T. McDonald FROM: Paul Cousino SUBJECT: Selective Parking and Reduction Program The Selective Parking and Reduction Program which will be a part of the total effort to eliminate weighting is proposed to work as follows. Sample requested: 300 Salem FF85 300 FFM85 Females 18-34 (Excluding Salem FF85) Requested returned sample would thus look as follows: An outgoing sample would be designed using return rates available based on previous experience with switching and return behavior. Using Steve Tucholski's memo and figures which he and I calculated, a suggested sample would be: As you can see in this example, the returned sample is not exactly the same as the requested sample. Due to the choosing on Status Code 01 and weighting by expected return/switching rates, no cell is appreciably short. We note, however, that some cells are much larger than expected. For example, Salem 18- to 24-year-old males are 43% higher than expected. When using the returned sample in its current form, shortages (as in Newport 18- to 24-year-old females) are exacerbated by the overages occurring in other cells (such as Kool 18- to 24-year-old females). The program which I am requesting can be designed in one of two ways. All cells will be reduced or paired to the level of the requested sample or only those cells which are significantly different would be paired. The second method would results in less loss of completed interviews but greater accuracy. The first method may be easier to carry out. Some discussion on this topic may be indicated. The returned sample after being passed through the Selective Paring and Reduction Program would look like one of the following two samples: Sample 1 above would be derived by paring the returned sample to the extent that it exactly matches the desired quotas. In Sample 2, however, the program pares only cells that are significantly different. In this example, only one cell is unaffected (females 25-34). In cases where a cell comes up short of its desired quota, the following procedures will be implemented. 1. If the ration of desired completes to actual completes is either less than 1.5 1.2 or greater than 2.5, the cell will stand short as it is and will not be weighted. 2. If the ration of desired completes to actual completes is between 1.5 1.2 and 2.5 inclusive, and the actual cell is greater than or equal to the sample size divided by the total number of cells in the sample, standard weighting procedures will then be implemented for that cell. 3. In the case of Number 2, but where the actual completes cell is 3% less than the sample divided by total cells, the cell will stand as is. Dick, if you have any questions about this particular program or need to discuss it further (I hope more discussion will be forthcoming), please give me a call. jks 1/14.45 National Family Opinion, Inc. f