Understanding Replicates

What is a Replicate?

During the preliminary data period, the term “replicate” will be used as a way of describing the incremental increase of the LPM sample size, as the launch date approaches. The following are some pointers in understanding what replicates are and how they affect the sample.

A sample replicate is a smaller version of the full sample. The full sample is selected from the DMA universe, based on a specific sample design. Smaller samples are then selected from the full sample, using that same sample design. These smaller samples are called replicates.

Having multiple replicates allows the Field Staff to install the sample in smaller segments, providing the opportunity to monitor progress as the sample grows to the final size.  Each replicate as it is completed adds stability and reduced standard error to the estimates.  However, not until the final replicate is in place and reviewed by our Statistical Research Department for representation of the market, should any estimates be considered up to the high quality standards of Nielsen Media and our industry.

Number and Size of Replicates is Determined by Market Sample

 In Los Angeles, New York, and Chicago, there will be three replicates; the sample size will start at 540 households (Replicate 1), then increase to 640 households (Replicate 2), then increase to 800 households (Replicate 3).

Replicate sizes are measured in terms of specs issued to our Field staff. This will vary slightly from market to market, based on the census vacancy rate in the market.  The following spec numbers are required in order to get Los Angeles, New York and Chicago up to their final sample size goals of 800 households installed: In Los Angeles, Replicate 1 has 675 specs, Replicate 2 has 125 specs, and Replicate 3 has 200 specs. In New York, the spec sizes are 680, 125, and 200, respectively. In Chicago, the spec sizes are 675, 125, and 195.

Incorporation of Additional Replicates into the Sample May Affect the Sample Characteristics or the Viewing Estimates:

Each replicate is selected using the overall sample design, therefore changes in sample characteristics or viewing estimates that occur after a replicate has been added are due to a reduction in sampling error. Specifically, as each replicate is added, the overall sample size (as well as the sample size of each characteristic) increases, and the sampling error for the viewing estimates goes down.

When is the Last Replicate Added to the Sample?

The final replicate is added into the sample just prior to the “launch” date for each market.  The actual date may vary from market to market and is still being determined for each of the LPM samples.

Replicate and Spec Details for Three Current LPM Markets in Preliminary Period:

LA – 1000 Specs –3 Replicates

675 to get to 540 Installed

125 to get to 640 Installed

200 to get to 800 Installed

NY – 1005 Specs – 3 Replicates

681 to get to 540 Installed

124 to get to 640 Installed

200 to get to 800 Installed

CH – 995 Specs – 3 Replicates

675 to get to 540 Installed

125 to get to 640 Installed

195 to get to 800 Installed