The strategy that we described above assumes that there are at least some common items across studies and that a reasonable subset of the common items are invariant (i.e., do not show DIF by study or other covariates). But in some desired IDA applications, there may be no common items (i.e., no items can be harmonized) or too few common items (or common items without DIF) to confidently link measures together. One option to facilitate IDA in such circumstances is to conduct what we refer to as a bridging study. The basic idea is to embark on a new primary data collection for the express purpose of linking together the measures used in the original set of studies. A bridging study would involve recruiting new participants, ideally from a similar population as that sampled in the contributing studies intended for the IDA, and administering items from all of the original studies to these participants. Not all of the original items would have to be administered; a subset of items is sufficient (reducing participant fatigue and eliminating redundancies between items). Pooling
