| Do storks bring babies? No...but if we are not careful, we could mistakenly conclude so. Do fireman cause damage at fires ? Again no. The point is that statistical methods have often been misused. If the problem is simple then most if not all statisticians and statistical tools will be quite adequate. However, as is often the case in the real world, the opportunities can be complex. Statisticians refer to two kinds of errors and modern time series analysis focuses on ways to help you avoid them. |
|---|
| 1. False Positive (Type I) is when the conclusion is that the variable or coefficient IS important, but the true state of nature is that it is NOT. 2. False Negative (Type II) is when the conclusion is that the variable or coefficient is NOT important, but the true state of nature is that it IS. |
|---|
| It is fair to say that "All models are wrong , but some are useful". It is also fair to say that there is no such thing as a bad statistic just bad statisticians or bad software. What we have learned is that there has to be aggressive testing of the underlying assumptions before proceeding to the conclusions. Simple use of Regression with time series is by far the greatest cause of bad conclusions. |
|---|
| If you are careful, you can greatly reduce the chance of making wrong conclusions. Some of the examples we present show how easy it is to make mistakes. You can rely on the aggressive assumption testing to minimize your risks of concluding incorrectly. Good software can perform as a "technology assistant" or a "productivity aid" and enable you to make better conclusions. |
|---|