Yes-No Grid (YNG) response formats have been found to yield higher self-repor­ted behavior than Multiple Response List (MRL - 'select all that apply') response formats (Thomas and Klein, 2006, Journal of Official Statistics). In this current stu­dy, we examined how response formats could affect behavioral intention measu­rement. 56316 respondents (U.S. 18+) participated in a web-based survey. To assess respondent satisficing, 2 manipulations varied respondent load. Respon­dents were randomly assigned to complete the experimental section either at the

beginning of the survey or at the end of the survey and we re randomly assigned to either 5 or 10 behaviors out of 20 behaviors (e.g. drinking coffee, attending a religious service, reading a scientific article). As a standard criterion, respondents were asked how frequently they engaged in the assigned behaviors in the past 30 day and then randomly assigned 1 out of 6 possible likelihood response formats (asking if likely to perform the action in the next 30 days):

MLR, YNG, Dichotomous unipolar likelihood (Not likely/Likely), Dichotomous bipolarlikelihood (Unlikely/Likely), 5 category unipolar likelihood, 5 category bipolar likelihood.

Replicating prior work, the YNG response format yielded significantly higher li­kelihood endorsement frequencies across behaviors than the MRL, regardless of survey position. Extending our findings, the dichotomous likelihood measures also had higher endorsement across behaviors than the MRL. Behaviors presen­ted later in the survey had somewhat lower rates of endorsement as being likely, especially for the MRL, though this effect was observed for all formats. Behaviors presented in a context of 10 behaviors showed somewhat lower likelihood endor­sement than the 5 behavior context.