Before beginning the study, we will test the procedure to make sure that it works efficiently and to make sure that we are not encountering floor or ceiling levels of noticing in both conditions. For this pilot testing, police trainees will individually perform a simulated traffic stop. We will alternate subjects between the hostile and cooperative driver. Funnel debriefing will be used to determine whether and when people noticed the gun. Depending on the rates of noticing, we may move the gun to make it more or less conspicuous. Once we have settled on a method, we will register the formal design for the primary study. No data from this preliminary work will count toward the official study. I will add notes about the results of pilot testing here before finalizing the full study plan. **Preliminary testing - August 29, 2013** We tested 7 people, with the driver acting compliant for half 5 and resistant for 2. Only 1 subject in the resistant condition noticed the gun and none of those in the compliant condition did. The gun was positioned on the passenger floor and was visible from the officer's perspective, but it was a black gun against a dark floor mat and did not stand out on an overcast day. We then swapped in a set of light-brown floor mats that made the gun more visible. In that condition, we tested 4 additional participants with half in each condition. Three missed the gun, with the one noticer in the compliant condition. Finally, we tested two participants with the gun positioned on the passenger side dashboard, just above the glove box where the driver goes to retrieve the insurance card. From that position, the gun is highly visible throughout the interaction. The actor leans forward initially to partly obstruct it from view, but it is visible when the driver gets his license and registration and for most of the interaction. We tested two participants in this condition, with one missing the gun and one noticing. In all cases, the experimenter read questions and marked answers. The procedure worked well, but some of the questions were redundant. For example, if they don't report any weapons, there's no reason to ask about a gun. It's also apparent from the subject's reaction whether they noticed the gun — the actions they are required to take during a traffic stop differ substantially if they spot a gun in plain site (e.g., they draw their own gun or retreat to their car or require the driver to exit the car). If they don't notice, they return to their car casually to write a ticket. We also will add a question about whether the officer has an actual patrol experience (some of the trainees do). **Changes planned for day two of pilot testing** 1. The gun will be placed on the dashboard to see if we can estimate the noticing rates 2. The questioning procedure will be reduced to the minimal set of questions, and they will be printed on a single page on which the experimenter will record answers. 3. We will test enough subjects to tell whether a sizeable percentage will miss the gun on the dashboard. Assuming they do, we will determine the sample size for the full study. 4. We will draw subjects from the group doing other types of traffic stops for this pilot. For the final study, we will draw them from a classroom setting so that their other experiences at the same time don't affect their performance of this task. 5. We have updated the procedure to reflect the changed process. **Preliminary testing - September 3, 2013** We tested 8 people with the driver acting entirely compliant for half and what we will call "aggressive" for the other half. (Technically, as long as the driver does what the officer asks of him, he is being compliant, even if he is insulting and obnoxious). In this day's testing, it was a sunny day and visibility was good in the car. All testing was done with the gun in plain sight on top of the dash in front of the passenger seat. 3 of the 8 participants missed the gun, with 2 of those coming from participants in the "aggressive" condition. We refined the questioning method, with the experimenter obtaining consent and then asking each of the questions verbally and recording answers and comments. This streamlined process worked well. Based on the questioning, we eliminated the specific question about the gun and the hypothetical questions about what would be more likely to be noticed. The reason is that we can tell immediately from the officer's behavior whether or not they noticed the gun. If they do, they are required to take immediate action. If they don't, they calmly return to their car to issue a ticket. The response sheet will ask them whether they noticed anything of potential danger. If they notice the gun, they say that immediately, so we skip the question about weapons. If they don't note the gun, we continue by asking about the weapon. Based on this pilot testing, we have finalized the response sheet and uploaded it to the Files section and Materials Component of this project. The protocol component reflects the finalized design and plans. Note: In all pilot testing, the experimenter was in sight of the stopped car and was also observable by some of the other trainees, meaning that they might know there was something other than a traffic stop involved. For the final version, trainees will be recruited from a classroom setting and will not see the experimenter until after the stop has been completed. Also, the pilot testing was conducted on the same day that the trainees were involved in practicing other high-risk stops. That might make them more likely to expect a gun. For the actual study, they will have completed routine traffic stops and some stops with illicit objects, but will not yet have conducted high-risk stops (e.g., apprehending a murder suspect, pulling over a driver who has a warrant out for their arrest, etc).