Attitude-focused interventions guide public opinion. Assessing their efficacy relies on pretesting. Previous research shows that inquiring on participants’ attitudes prior to an intervention has mixed effects on intervention efficacy. Given the implications of attitude-focused interventions, it is important to understand the mechanism behind pretest effects and offer solutions. A recent network approach to attitudes proposes that when attention is low attitudes vary with information, i.e., become more negative with counter attitudinal information and more positive with pro-attitudinal information. However, increasing attention ‘freezes’ attitudes, so that they can no longer vary with new information. Assuming pretests increase attention and interventions provide information, the Ising attitude model can explain the mixed results of pretesting on intervention efficacy. Specifically, prior to pretesting attitudes can be normally distributed on a negative to positive continuum. Depending on the type of pretest attention may (or may not) increase, thereby ‘freezing’ (or not) attitudes to their original position, resulting in a mix of intervention outcomes. Despite its explanatory potential, the model lacks empirical support. To this end, we tested how individual differences in initial information and pretest attention impact intervention-driven attitude change, as well as how much different pretest types increase attention.