So, finally, I have downloaded, analyzed, and posted the first data from my experiment for the world to see and use!
But, first, take a look at these notes:
- If you have not yet, make sure you read the original blog post of the series for the background information on the experiment.
- Please be aware that I am not an expert on radiation. Although I have tried to make the experiment as scientifically rigorous as possible, please handle my results accordingly.
- The computer measured the radiation hitting the collector every thirty seconds. Therefore, each datapoint corresponds to the amount of radiation detected per thirty-second interval (counts per thirty seconds).
- For some reason, the computer stopped collecting data at 146.633333 hours. The next datapoint afterwards is at t=282.979167, which is not representative of when I restarted data collection.
Secondly, I welcome reuse and reanalysis of my results by others. They are licensed under the CreativeCommons Attribution-NonCommercial-ShareAlike 3.0 Unported license (see LizardWiki:Copyrights for details). The original data is available in a few formats. You can download the original data here. Note that, although not necessary, I would appreciate it if you would let me know if you reuse/reanalyze my data (although you do need to get my explicit permission if you wish to reuse my data for commercial purposes).
So, with that out of the way, let’s get on with the analysis of the results:
Histogram of the results of the experiment
At the time I downloaded the data to my flash drive (before going on spring break) on Friday, April 8th, 2011, the experiment had been running for exactly three weeks (more or less). As indicated by a histogram of the data (see image at right), the data I collected is representative data, since the histogram displays – as expected – a bell curve. Interestingly, the mode of the dataset according to the histogram is 10 counts per thirty seconds, the same as the median of the dataset.
One hour graph excerpt from results. Note the variance in the graph.
At this point I feel that it is important to emphasize that, beyond the histogram, this data cannot really be analyzed by eye. Take a look at the graph to the right. Note the fluctuations in the graph. At first glance, you probably wouldn’t be able to see a trend in the graph. This is why it’s necessary to use computer software to analyze the data, because – especially when viewing larger portions of the graph – the data fluctuates so much that it’s nigh impossible to pick out trends in the data by eye.
To analyze the data, I asked the data collection program – Vernier LoggerPro 3 (of which I conveniently have a copy installed on my home computer) – to draw a linear fit through the data and perform a basic statistical analysis of the data. Note that I have made an important assertion here: That the trend is a linear trend. I have not taken into account the possibility that the trend is actually non-linear (parabolic or what-have-you); that’s analysis for another day. So, with this in mind….
Week 3 results of the radiation scan experiment
To the right is the complete graph of the results, with some annotations by the program. Yes, believe it or not, that massive red wall is data. Very visible in the graph is where the program mysteriously stopped collecting data (it wasn’t due to the Geiger counter failing; it was still running when I restarted data collection). Click on the image to view a larger versions, which will make the annotation text readable.
Summary of statistical analysis:
- Minimum: 0 counts per thirty seconds (CP30S)
- Maximum: 24 CP30S
- Mean: 9.809 CP30S
- Median: 10.00 CP30S
- Standard deviation: 1.1323
- Samples analyzed: 44412
Summary of linear fit analysis:
- Average rate of change of radiation levels: -0.0009512 CP30S/hour
- Correlation: -0.05076
- Root mean square deviation (RMSE): 3.128
My analysis of the results is that there has been no significant increase in the amount of background radiation since data collection began, and levels have been relatively stable. I hypothesized that if there were radiation from the Fukushima Nuclear Power Plant in Japan reaching the U.S. that this level would increase over time, however, the data has indicated a decrease, if anything, in background radiation levels. I am concluding, therefore, that there is no or an insignificant amount of radiation reaching the U.S. from Japan, and if there is radiation reaching the U.S. at all, it’s far less than what would get from, say, an airplane flight or an X-ray exam. I certainly won’t worry about taking potassium iodide tablets.
However, again, since I’m not an expert on radiation, I invite you to analyze the data and draw your own conclusions.
Note that the experiment is still running, and that I’ll be downloading a new set of data the week I go back to school after spring break.
