Eliciting actual donations toward a public good has been proposed as a means of estimating a lower bound to individuals' compensating surplus, and can be accomplished using mail/phone surveys or field experiments. This study shows that when warm-glow is present, the elicitation instrument decreases the transaction costs of donating. This presents an obstacle to using the donation mechanism. As a remedy, we propose the use of a multi-donation mechanism where subjects can direct their donation to alternative public goods. Results from a field experiment confirm this instrument-induced bias can be large, suggesting field experiment practitioners should seriously consider how their experimental procedures may alter economic behavior.