Science Buddies: "Ask an Expert"

I am doing a project titled the Effect of Visual Evidence on Eyewitness Account. In this project, I am staging a small crime, and then, 45 minutes later sending out a survey to the eyewitnesses about the crime. Half of the eyewitnesses will get a mugshot book with the criminal in it, and the other half of the eyewitnesses will have to fill out the survey completely from memory. About how many eyewitnesses should I have per trial?

WOW...this sounds very interesting. I think you should try to have at least 10 people that way you can get better results.

Are you talking about 10 in each group (one with a mug shot, and one without), or alltogether have 10 people?

Hi!

oh... sorry about that . I think you sohuld try to have 10 people in each group, but if you can't get that many people to participate try to have at least 5-7 people in each group (mug and survey)

Hi there,

In general, whenever one is looking for a statistically significant difference, it's a good idea to have as many subjects as possible. Particularly when the data can be somewhat subjective (like taking a survey after an event as you've described here), you'll want as many subjects as possible. That way, if you see a difference, you'll know that it's statistically significant.

By the way, you'll probably use a t-test to figure out whether the differences are significant. Along these lines, when you perform a t-test, the more subjects you have, the higher your confidence interval. If you need help performing a statistical analysis of your data, please post back here.

I guess how many subjects you use will depend on how many times you'll be able to stage the crime, and how many subjects you can get. I think 10 for each condition is the bare minimum. If you could get 10 subjects (for each condition) a couple of times, that would be better. Failing that, I think increasing the number to 20 or more would be best. But I"m not sure if you'll be able to get that many individuals.

Your topic sounds very interesting. Best of luck!

Linh

Thanks for all of the tips! I am going to do my experiment with 20 people total during each trial. I will be doing 3 different trials. Therefore, in total I will have 60 people, 30 of whom will have used a mug shot book, and 30 of whom will have not.
After talking with my teacher, I have yet come up with another question. I will be doing this in three different classrooms at the school that I attend. We do not want the students, (eyewitnesses) to know that the purse snatching is part of a science experiment. We feel that this would create a biased result in the answer, and the students would not care how the answered the question, and ultimately, not be serious in the answers. I need your help in figuring out a way to introduce this, and make it a big deal, but remember I do not want the students to know what is actually taking place. We have to come up with something because normally if the teachers purse was stolen, she would be upset and freaking out, and that is not going to go on in this case. So we want something so that after the snatching the students and teachers can return to instruction, and then 45 minutes later, have them fill out the survey. Thanks a lot for helping, and I am really sorry if I confused you!

Hi Steph,

I'm a little confused about the purse snatching part. Are you going to steal the teacher's purse? And then she's not supposed to freak out afterwards? If so, I think you're correct in that it doesn't seem all that realistic.

Did you read the information about informed consent? Have you looked over it with your teacher? You'll have to get informed consent from all of your subjects, and the problem is that will inform the students of the nature of your project. Typically, you would have to deceive the subjects and then get informed consent afterwards. For example, you could gather all the students and teachers for a "special event", like a movie showing, a special lecture, or a reading. Then you would perform the experiment, and then after everyone fills out the questionaire, you would get informed consent to let them know that they were part of an experiment. BUT, there's a big problem here. I'm fairly certain that according to SRC guidelines, you're supposed to get informed consent before the experiment, and the paperwork would include a brief description of their role in the experiment. But you're right, this coudl bias the students. I think it's a good idea to speak with your teacher and find out whether you can get informed consent after the experiment.

If you can get informed consent after the experiment, I had a couple of suggestions about getting the students together. You could tell them it's a special lecture, a movie showing, a small performance, or maybe even a scientific demonstration. Then while the performance is going on, you could stage your small crime.

There's another issue in that you don't want to alter the subject's perceptions. However, if it is really a small crime (such as a purse snatching), would it be reasonable for you to have a mug sheet ready? Maybe you can get a student to perform the crime, and instead of a mug sheet, it can be a few selections of yearbook photos or something. The problem here is that you need a student that doesn't go to your school because you *don't* want anyone to recognize the perpetrator.

With regards to the victim's reactions, if you want something that a victim would not react strongly to, what if the "crime' is something even more minor? Like stealing the hall pass? Or maybe moving a TV from the room, and then finding out later that it was stolen. For example, in my high school we would have office aides move TVs from one room to another for special purposes. But what if you found out 45 minutes later that the TV never arrived like it was supposed to? And that now it was missing? Then you wouldn't expect a big reaction during the time period of the crime.

These are just a couple of thoughts. Please let me know what you think. In particular, let me know if you've thought about how to get informed consent.

Best of luck,
Linh

I had already thought about the student from a different school, and I have got that accomplished. I was also talking to the Board of Science, or something in my district. She told me that since it was in school, during school hours, and I had teacher supervision I may not need the consent forms. I do like the tv idea though. I was also thinking that I could possibly get the officer on duty at my school, to maybe be running through the school after the criminal and through the classroom. That way the officer could say something like, "We have this situation under control, so please just stay calm". That way there would be a reaction, but it would not be major, and the teacher could continue on with instruction. Do you think that this sounds like a good idea? Thanks!

It looks like you have your planning well under control!

You WILL need consent forms, however, if you are planning to take your project to a local or state science fair. Many fairs also have forms that need to be filled out when the project involves other people. Not doing this could disqualify you.

I think that, because of the nature of your project, you would need to give the consent forms to the students after the crime has taken place. You would need to take out the data sheets of anyone who did not consent to let you use the information.

Sonia

Any other suggestions?

Hi Steph,

It sounds like you are paying attention to the details of the project.

What data will you be collecting from your subjects? Are you familiar with the student t-test? This is a way to determine if two sets of data are statistically different. You will want to make sure you will collect usable data.

Also, how are you going to collect data from so many subjects after the crime in a short period of time? Will you have someone helping you?

Donna Hardy

I am not familiar with the student t-test. I will be collecting data in the form of a survey about the events of the crime. I will then use that data along with the mug shot book from one-half to know the accuracy of each subject. The questions I have in my survey are these so far:
1) Was the criminal a male or female.
2) Describe the criminals physical appearance.
3) Describe the criminals clothing.
4) Describe the event of the crime.

Any other suggestions? I also need help understanding how to do the student t-test being that I am a freshman and have not take Stat yet.

Hi Steph,

The student t-test will allow you to determine if two sets of data are significantly different from each other. Here is a site that will give you the information for doing the calculations. You don't have to wait until you take statistics to do this.

http://www.physics.csbsju.edu/stats/t-test.html

In order to use the t-test, you need to have some numerical data. You should probably consider revising your questionnaire so you can score the results. Perhaps you could ask more specific questions or plan to score results on a scale of 1 to 10 (or any range), depending on the number of details correctly reported by each observer. This will give you quantitative results.

Let us know if you have any questions.

Donna Hardy

Are there any more suggestions that can help me?

Hi Steph,

If you haven't already done your experiment, you should do it as soon as possible. You want to allow plenty of time to write up your results.


Donna Hardy

I am doing it at school with my science teacher and a few other teachers, we are going to do the experiment in school right after the holidays, and I will be able to collect the data and write everything up as soon as that is completed. The due date has also been pushed back, so I am still in good shape. But any other suggestions of what should be on the survey? I already have:
1) Is the criminal a male of female
2) Describe the criminal's physical appearance including hair, weight, height, race, and any other details that you may remember.
3) Describe the clothing of the criminal
4) explain, in detail, the events of the crime.

Hi Steff,

It sounds like you are well-prepared for your experiment. Good luck.

Donna Hardy