While we discuss the value of premiums in association marketing, it's important to keep in mind that there are actually two kinds. Many of us who think of premiums think purely in terms of back-end premiums, items that are probably in a $3 to $5 per item production range: thank-you gifts for joining or registering for $160 memberships or $495 conferences.
Habitat for Humanity version at the right. Front-end premia almost always outpull back-end premia, and by a wide margin. This is because they help with the first part of the direct marketing equation (get opened) rather than the second part (read me/take action).
Since there is always a delay in fulfillment and we live in an instant gratification world, the back end item is also less effective because it's only peripherally related to the offer. It helps, but probably won't contribute that much to put prospects over the edge (i.e. 'I'm thinking about joining, but the bag closed the deal').
For these kinds of items, consider items such as a custom-struck token or cling window sticker—a test cell of these items affixed to the reply or inserted in your standard approach might boost open rates sufficiently to generate enough lift in response compared to the control package, and enough to offset the probable 5- to 7-cent incremental cost you incur at rollout quantities.
CRS I know the former lifted our donor acquisition response rates about 30% when we first tested and remained at that level at rollout, although we continued to test token designs and metal colors/finishes to be sure we had the right one. We also found the lift in response was offset by smaller gifts and somewhat lower renewal rates, but of course in your association, you can determine this through your own testing.