How to Lie With Statistics
Catherine Davison
October 22, 2013
Summary #4
Chapter 8 – Post Hoc Rides Again The Post – hoc analysis which is the cause and effect problem. Methods of presenting cause and effect: 1) present a result without a significance value 2) use untestable assumptions 3) use precision and accuracy interchangeably 4) perform nonsensical test that sound good.
Keep in mind that a statistic is only worthwhile when it satisfies the assumptions on the test. Knowing whether the assumptions are met is dependent on the competence of the person running the test.
Just because two things seem to have a relationship, could it have been by pure chance? It cannot be determined by causation and effect. The two variables have no effect on each other at all.
Chapter 9 – How to Statisticulate Statisticulate is the process of misleading people using statistics. It is also misinforming with figures, or statistical manipulation might not be a mathematician purpose.
Lying with statistics – is this dishonesty or incompetence? Mostly dishonesty.
The author list various tricks – things like measuring profit on cost price, showing a graph with a finer Y-axis scale just to show the steep growth is, how income calculations mislead by involving children in the family as individuals for the average amongst a few.
Chapter 10 – How to Talk Back to a Statistic In how we talk back to a statistic one should ask themselves to find out if the statistic that you are reading being presented is it genuine or not. There are 5 simple steps, in Huff’s own words, “how to look a phoney statistic in the eye and face it down”. (page **)
Question 1 – Who says so? Find the bias before trusting. Does the one publishing this result have anything to gain from the result?
Question 2 – How does he know? How was the result measured and was the sampling biased?
Question 3 – What’s missing? Unstated averages not showing the original data, comparisons without stating the other side, you
October 22, 2013
Summary #4
Chapter 8 – Post Hoc Rides Again The Post – hoc analysis which is the cause and effect problem. Methods of presenting cause and effect: 1) present a result without a significance value 2) use untestable assumptions 3) use precision and accuracy interchangeably 4) perform nonsensical test that sound good.
Keep in mind that a statistic is only worthwhile when it satisfies the assumptions on the test. Knowing whether the assumptions are met is dependent on the competence of the person running the test.
Just because two things seem to have a relationship, could it have been by pure chance? It cannot be determined by causation and effect. The two variables have no effect on each other at all.
Chapter 9 – How to Statisticulate Statisticulate is the process of misleading people using statistics. It is also misinforming with figures, or statistical manipulation might not be a mathematician purpose.
Lying with statistics – is this dishonesty or incompetence? Mostly dishonesty.
The author list various tricks – things like measuring profit on cost price, showing a graph with a finer Y-axis scale just to show the steep growth is, how income calculations mislead by involving children in the family as individuals for the average amongst a few.
Chapter 10 – How to Talk Back to a Statistic In how we talk back to a statistic one should ask themselves to find out if the statistic that you are reading being presented is it genuine or not. There are 5 simple steps, in Huff’s own words, “how to look a phoney statistic in the eye and face it down”. (page **)
Question 1 – Who says so? Find the bias before trusting. Does the one publishing this result have anything to gain from the result?
Question 2 – How does he know? How was the result measured and was the sampling biased?
Question 3 – What’s missing? Unstated averages not showing the original data, comparisons without stating the other side, you