Secondary sampling comes after primary sampling, while tertiary comes third. For example, if you are sampling an initial group of 1,000 voters on a political issue, the first round of results for your test or questions would be a primary sample. Analyzing that sample, you find an interesting trend, approach that section of the initial group and run another test. This is secondary sampling. Finally, after compiling the results from the primary and secondary sampling, you narrow your focus again and complete a final test, resulting in a third or tertiary sample.
The goal of secondary sampling is to compile and analyze the information set out in the primary sample. This level of sampling tends to be argumentative or agenda driven in nature.
The best way to picture tertiary sampling is to imagine a filtering system that has three stages. For example, as you pour pure maple syrup through each progressive level, the syrup becomes more and more refined. In this same fashion, information is filtered, with the tertiary level displaying the most refined and holistic look at that information.
When you think of sampling, you may think of scientific testing and methods to establish results or analyze a topic of focus. However, there is an additional type of sampling called sourcing---specifically, primary, secondary and tertiary sources. These sources can be books, reports, scholarly journals and reference books to name a few. The same rule applies to sourcing, that the tertiary level tends to be the most factual (reference books) where the primary and secondary sources tend to contain raw or lightly analyzed information (newspapers, novels, journals, et cetera).