By Katie Zaback, JVA research associate
Have you ever been in a room where two evaluators are having a discussion about the results of an evaluation? It’s not uncommon for those of us who love statistics to start using what sounds like our own language. But, have no fear—although the work that evaluators do seems complex, when you begin to learn the terminology and speak “stats-speak” you will start to notice that many of these concepts are equivalent to things we all consider on a daily basis.
So in the spirit of sharing, I would like to decipher one of the key technical terms that many find confusing: standard deviation.
Many researchers will give the standard deviation when they report an average or a mean. Most of us simply look at the average and ignore the standard deviation, or wonder why the researcher insisted on providing a measure that only they really understand.
However, standard deviation, although it is somewhat complex to calculate, is actually a pretty simple concept and it can be really useful to know. Standard deviation is essentially a measure of variance. When you think about variation, what do you think about? Differences, right? Standard deviation tells us if there are differences within our data.
Let’s break it down a bit further. When you calculate an average you always end up with one number, no matter how many numbers you start with. Calculating an average breaks data down data into one, digestible number that is easy to understand. However, when you use one number to represent many numbers, you lose information about those other numbers and how they relate to the average. Standard deviation gives you this information.
For example, I recently did a survey where respondents were asked to rate their overall satisfaction with a program. About half of respondents rated a program a 5 (Very satisfied) while the other half rated the program a 1 (Very dissatisfied), no one rated the program a 3 (Neither satisfied or dissatisfied). Despite the fact that no one rated the program a 3, the average of these ratings was 3.
Taken by itself, an average of 3 would suggest that participants were neither satisfied or dissatisfied and that they were neutral about the program. However, I know this was not the case, in fact, people were actually either very satisfied or very dissatisfied, no one was neutral.
By looking at the standard deviation, this would be clear. The standard deviation tells me the average distance of each response from the mean. Here, my standard deviation would be around 2. This high standard deviation tells me that there are differences within my data and suggests that I should look more closely at individual responses. So next time you look at an average, you can also consider what the standard deviation might be telling you.
Want a change to learn more “evaluation-speak?” Join me and the rest of JVA’s research and evaluation team at Evaluation Intensive on March 30 and 31. We will decode evaluation terms, teach you more about how and when to evaluate, and knowledge and give you the tools you need to implement program evaluation in your organization. Today is the last day to register, so click here to sign up.