Teach Yourself Statistics
A. Graham Understanding Statistics Teach Yourself, 2010.
This book is my attempt to present the basic ideas of statistics clearly and simply.
(I am interested to see the extent to which it helps the reader to avoid – or perhaps lures them into – some important pratfalls. I have only dipped in.)
11 Regression: describing relationships between things
What we can do now is to use the [regression] line to make predictions … . However, it is worth stressing that these really only are guesses which carry all sorts of assumptions that may not hold in practice.
… The dangers of inaccuracy increase when making predictions which involve having to extend the line beyond the range of values in the sample. Such predictions are known as extrapolations … .
[T]here is no guarantee that the observed pattern will continue to apply.
13 Chance and probability
Odds are the most common way of measuring uncertainty in situations where people are betting on an event whose outcome is unknown. In horse race, for instance … .
In statistical work, probabilities are usually measured as numbers … . Usually, in statistical work, outcomes are the results of a trial or experiment … .
Probability is about calculating or estimating the likelihood of various outcomes for particular trials.
- We are warned about extrapolations, such as assuming that statistics will be stable.
- Here, probability is only defined for gambles, trials and experiments.
There is clearly no warrant for deductions about, for example, economies, and particularly not for probabilistic judgements about the future.