Is there a way for standard deviation to ignore negative numbers? In other words, how can you make it so that 2 or 3 deviations do not extend into negative territory?
When looking at a set of only positive numbers, within one or two standard deviations, I end up with negative numbers. It's essentially impossible for a data point to be negative. Am I doing something wrong?
Also, when using deviations from the mean, do you add or subtract the full deviation from the mean or divide by two and plus and minus that to get one deviation from mean
I'm sure this embarrassingly simple for this group but this is not my strong point!
Please let me know if I can clarify. My recent experience with statistics is relatively limited.
Thanks!
Best Answer
Standard deviation is kind of a unit of distance your numbers are from the mean. The problem with standard deviation is that when it's introduced it's usually done assuming a bell shaped curve. From that it's easy to see what 1 standard deviation means ( 68% of the data is within 1 standard deviation ). It sticks in your head easily. Problem is, when we observe counts, or other types of data values(like blood pressue, failure times, etc) their set of observed values dont follow a bell shaped curve. In those contexts the easy to understand idea you were given for the sd doesn't work as intuitively as you would hope. My best advice is to use other measures then a standard deviation. Learn about a concept called percentiles. For example the 25th percentile is the count for which 25% of your data falls beneath it. So if the 25th percentile was 8, then 25% of your data has 8 or less. In other words I'm telling you to find a different statistic to use to classify where your numbers fall, instead of using a standard deviation.
This was asked awhile ago and had some good answers as well: How to interpret two standard deviations below the mean of a count variable being less than zero?