Wierd typo: the author asserts that journeys that would have taken days w/out a car, take but minutes w/ the car. He meant "hours", and the editor should have caught it. Also, "slip over" was maybe intended to be "fall over" or just "slip", unless that is an Anglicism unknown to me.

1. Surrounded by Statistics:
This is really just an introduction to the introduction.

2. Simple Descriptions
"ordinal", "ratio", "absolute" scale. I do not know that I have ever cared.
Summary numbers: the mean, the median, and the mode. There is the range, which is a measure of dispersion. Combined with the mean and the median, it should tell you something. It would be nice to write a program that would generate graphs of the distribution of values given the number of values, the range, the mean, and the median. It would be fun to under specify, too, so that, given only the mean, you could show the variety. There is also the standard deviation, and you could play the same trick with that. We can subdivide the data using quantiles, like the median, only with smaller sections.

3. Collecting Good Data
"observational" vs. "experimental" studies. Sampling the data: "sampling frame", "stratified random sampling" and "clustered random sampling". If we just want to find the average of some value over a total population we can sample and, the author argues, it is the absolute size of the sample rather than the relative size that matters.

4. Probability
"subjective", "frequentist", and "classical". I guess I would call classical "formalist". The cumulative probability distribution graph is drawn as continuous. But I looked on Wikipedia and that is not a necessary property. It is, however, a cadlag function. That makes sense, and I like it better. The probability density function doesn't convey information by checking one particular point, but rather by finding the area under the curve between two points. Wikipedia has a nice graphic showing the relationship of the median, mean, and mode to the probability density function. There is also a probability mass function, useful for representing probabilities of continuous random variables. Comparison of distributions: Bernoulli, binomial extends Bernoulli over multiple trials. The Poisson distribution, according to this author, is useful if the number of trials is unlimited, but I thought that was true of the binomial as well. For continuous variables, we can have the "uniform", "exponential", or "normal" distributions.

I didn't read all of this, I was looking for statistical mechanicanics ( which is actually probability ) ( )

Completely inadequate explanations and details, didn't learn ANYTHING from reading this. ( )

I didn't read all of this, I was looking for statistical mechanicanics ( which is actually probability ) ( )