Join Curt Frye for an in-depth discussion in this video Introducing statistical analysis, part of Excel 2007: Business Statistics.
- View Offline
Numbers are everywhere, especially in business where your company's health and future depend on your ability to make good decisions. You can use statistical methods to analyze your data to support that work. Statistics is the practice of collecting, analyzing, and interpreting data. There are two major branches of statistics, descriptive statistics and inferential statistics. You use descriptive statistics when you know all of the values in a dataset. For example, if you examine the olives on a tree, you can count the number of olives, in this case 4,000.
Then you can find the number of green, olives 3,000, the number of black olives, which are 1,000, and then find the proportion. In this case, the green olives are 75% of the total, and the black olives are 25% of the total. For inferential statistics, you take information from part of your population and extrapolate it to make guesses about your entire population. So, for example, let's say that you have a grove of 100 olive trees. You take a sample from five of those trees. You have 21,000 olives, of which 14,070 are green and 6,930 are black.
You see that the percentages are green is 67%, and black is 33%. Now that's true for those five trees, but it might not be in effect--almost certainly will not be--exactly true for the entire grove of 100 trees. So what you need to ask is how close am I to the true value, and how confident am I in that estimate? When you ask how close, you're asking how close your measured value is to the true value. So, for example, black olives make up 33% of your sample.
Well, the total might be 33%, but most likely it varies a little bit. So, for example, it might be 33% plus or minus 2%. So the total percentage of black olives might be between 35% and 31%. Indicating how close gives you a range. Next, you ask, how confident am I in that range? So you're asking what percentage of random samples taken from this population would fall within the range of, say 31% to 35%. The statement of how confident you are is what's called a confidence interval.
I'm going to switch from olives, and use a more familiar example perhaps, when it deals with political polling. So if we say that Governor A has an approval rating of 60% with a margin of error on the survey of 3%, how confident are we of that? Most of the time, the confidence level that you will find in statistics is 95%. So in other words, 95% of the samples that we take from the entire population should produce an approval rating of between 57% and 63%.
The other 5% would generate values outside of that range, for example, 55%, 54%--or on the high side, 70% or 75%. But we know from inferential statistics that 95% of the time, the approval rating for a survey will be within 3% plus or minus of the actual true approval rating. So the best way to phrase that statement is, we are 95% certain that Governor A has an approval rating of 60%, plus or minus 3%.
When you read an article online or pick up a newspaper, try to find the story that doesn't refer to numbers in some way. You probably won't be able to do it. Numbers are everywhere. But once you master the techniques in this course, you'll have an easier time making sense of your own data and interpreting data that others give you.
- Understanding statistical terms
- Creating a basic Excel table
- Auditing formulas
- Creating frequency distributions for qualitative data
- Calculating a running total
- Creating a histogram
- Using PivotTables
- Calculating mean, median, mode, and other numerical data
- Using probability distributions
- Population sampling
- Testing hypotheses
- Developing liner and multiple regression models