1

I am not statistically oriented, so I am hoping you might be able to help.

I have a small lab setup that mimics a larger natural process (natural disaster). I have designed and completed 46 experiments varying a few control parameters. I want to describe the wind speeds over which I tested in the little experimental setup, without listing all 46 of them. Wind speed is one of those parameters I had control of, so my choice of wind speed does not statistically indicate or correspond to the distribution of wind speeds found in the original natural phenomenon, but was chosen in a rather selfish way to test the limits of a model.

I do not know if my data set of 46 wind speeds is a population or a sample though. I must be over-thinking this!

I was leaning towards sample: While I do have all the data for all the experiments I conducted in this little setup, I do not have all the data for all the possible experiments this setup could produce, and I certainly do not have wind data for all the times the original process occurred in nature... which is what I was modeling to begin with. so sample?

But, I really just want to convey the range of wind speeds over which I tested (and I controlled the wind speed as a parameter). I am not saying these wind speeds are statistically similar to all the ones that occur during all these natural disasters, but just how to describe the data to people without giving them a list of 46 wind speeds.

super silly, I am sure. (sorry!) hoping someone can set me straight? thanks.

payton
  • 11
  • 2
  • 1
    Some of our threads, like [What is the difference between a population and a sample?](http://stats.stackexchange.com/q/269/930), provide some discussion around the definition of a sample and a population. – chl Jan 14 '15 at 17:38
  • Do you want to talk about the wind speeds *only in these 46 experiments* or do you want to make inferences about wind speeds in other circumstances? Your answer to this question determines whether you have a population or sample. – whuber Jan 14 '15 at 17:53
  • @payton, why don't you just dump the whole set of numbers (edit your question, do not do this in comments), and we'll try to figure out the best way to characterize these. – StasK Jan 14 '15 at 18:37

2 Answers2

1

The set of numbers that you have is the design, or, even more technically, the measure of the experiment design. It is neither the sample nor the population. The latter would be the distribution of wind speeds in the natural processes, and that itself is poorly defined, as we hear the news of record breaking hurricane winds from time to time. The former would be a set of observations from this natural process taken at specified or random points in space and time. Again, what you have matches neither of these.

In describing this set of numbers, you could give a histogram or a dot plot of your experimental wind conditions. I would definitely give a range, and if you set these speeds up as a grid (which would probably be a sensible thing to do if that's the only experiment condition) from say 0 m/s to 20 m/s with a step of 1, and then to 50 m/s with a step of 2, I would just give that description.

StasK
  • 29,235
  • 2
  • 80
  • 165
0

You are absolutely right. You have a sample. A population is "everything".

Yair Daon
  • 2,336
  • 16
  • 29