Sometimes Using a Manipulation Check to Verify That Your Experimental

Sometimes using a manipulation check to verify that your experimental treatment had its intended effect only undermines the efficacy of your treatment. For example, when people are primed to feel self-uncertainty, having participants report how uncertain they feel completely undoes the effect of the uncertainty prime. In other words, given the same manipulation, we see effects in the hypothesized direction when a manipulation check for uncertainty is not used; however, if a manipulation check is used, these effects disappear. If this is the case, how do we know that our manipulation is actually manipulating uncertainty, rather than some other construct entirely? How might we be able to assess the degree of people's uncertainty caused by our manipulation without explicitly asking them how uncertain they feel?