1. ## Help please- experiment design

Hi all,

I am having real trouble working out the best way to test what I want to test. Bear in mind that what I have to produce is a research proposal (not the actual research itself), so any discussion is entirely hypothetical.

I want to evaluate whether the implementation of a particular security system causes a reduction in theft in the shops where it is installed. I have 6 months to do the research. As far as I can tell, there are quite a few ways to go about this. Here are my ideas so far

Take two stores. Install the system in one, don't install the system in the other. Take several measurements of theft rate both before and after implementation, and use a T-test or similar to see if there is a significant reduction in the store where the security system is installed.

Alternatively, use some kind of time series analysis to compare the rate of theft in the two stores.

I could really use some help with this. Bear in mind that any proposal I make has to be realistic. For example, I couldn't perform an analysis of 100 different stores with daily theft measurements since that's just not possible.

Any help is greaty appreciated

2. ## Re: Help please- experiment design

The two stores should be in different areas. If the alarm is a deterrent for break ins for one store then those robbers will break into the other store if it is accessible. This falls under Stable Unit Treatment Variable Assumption.

3. ## Re: Help please- experiment design

Hey, thanks a lot for the reply.

I've actually been having further thoughts since my previous post, and have come up with a different (better?) experiment design.

The security system will be installed at 10 stores. This will be the experimental group. There will be a control group of 10 stores which don't have the system installed.

6 months after the system is installed, we will assess whether there has been a statistically significant reduction in crime at any of the stores.

I was thinking it would be best to use a Wilcoxon Signed-Rank test? Since there may be outliers? (stores with abnormally high crime rates to begin with)

What do you think?

4. ## Re: Help please- experiment design

How do you define crime?

Store theft

6. ## Re: Help please- experiment design

Help plz! [Design Experiment-different level parameters]

Hi all,

I need to design an experiment with 11 design factors , 4 variables with 2 levels and 7 variables with 3 levels. I use Statistica software. I wanted to use mixed 2 and 3 level designs method for design experiment but I figured out that this method just can design experiment for 4 variables in 2 levels and 6 variables in 3 levels.

Thanks

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts