Data mining is sifting through large amounts of data looking for anything that stands out, even if it happened by chance. Astrologers are often guilty of this, using one study to claim people born under certain signs are more accident prone. However, a single study doesn’t prove anything until it’s replicated. A followup study actually demonstrated that people born under different signs are more likely to have car accidents.

Coincidence is when two seemingly independent variables align in a way that seems to defy the odds. We may be tempted to assign meaning to this when there isn’t any. Mathematically speaking, when 23 people are in a room, there’s a one in two chance two of them will have the same birthday. Our intuition will tell us this is more improbable than it is, but our intuition about things is often wrong. Getting heads several times in a row when flipping a coin isn’t as unlikely as we think it is because randomness is more clumpy than we’d expect. Our memories also play a role. Since we forget ordinary occurrences, this makes interesting occurrences stand out more. Our memories also like to make the story better over time.

Methodological naturalism is the belief that the universe is real, nothing exists beyond the natural, and every effect has a cause. If an idea can’t be tested, it’s not scientific. This is that stance skeptics take. Postmodernism is the belief that science isn’t any more true than any other belief, however science isn’t a belief system. It’s method for eliminating bias and it’s the best method for eliminating bias there is.

According to Occam’s razor, when two hypotheses both explain the evidence, the one that introduces the fewest new assumptions should be preferred. People often think this means the simplest explanation is the correct one, however, the simplest explanation won’t always account for all the evidence and this needs to be taken into account.

Denialism (such as climate change denial) often masquerades as skepticism, however skepticism accepts science while denialism denies scientific evidence. Denialists are driven by ideology, exaggerate how much doubt there is, ask for more evidence than is possible, and don’t change their minds when proven wrong. Skeptics, on the other hand, are happy to change their mind when the evidence is compelling enough.

When it comes to scientific studies, statistical significance doesn’t mean as much as most people think it does. You also need to take other things into account like the prior probability. P hacking involves data mining to get to statistical significance. The best way to fix this is by only trusting studies that have been independently replicated.

While there are real life conspiracies such as the NSA spying scandal exposed by Ed Snowden or the Iran Contra affair, belief in grand conspiracies involving an improbable number of people and organizations that never get uncovered is absurd. The more people who get involved in a conspiracy and the longer it lasts, the higher the chance it will be uncovered by a whistle-blower or investigative journalist. A witch hunt is an unjust investigation that doesn’t use the regular rules of evidence. Examples include McCarthyism, the satanic panic, and facilitated communication.

There’s more than one placebo effect, so it’s more correct to refer to them as placebo effects. Subjective outcomes (such as a decrease in pain level) are susceptible to several psychological factors. Reporting bias may occur when subjects report they feel better even if they don’t. Also, researchers are biased in favor of a positive effect. People take better care of themselves when they decide to pay more attention to their health. Another placebo effect is regression to the mean which is the tendency for things to go back to normal on their own. Symptoms will usually get better over time even if you don’t do anything. Belief can have an effect on the subjective experience of pain as can distraction, exercise endorphins, and even cursing. Cancer doesn’t have a placebo effect because it’s not subjective, it’s more physiological. Placebo effects don’t require belief, so they also work on animals and babies. A placebo can also make you think you’re better when you really aren’t. Most alternative medicine is placebo.

An anecdote is a story offered as evidence. While personal experience is a form of evidence, it’s the weakest form of evidence and isn’t useful for testing hypotheses. A personal experience occurs outside of a controlled environment, so it’s subject to numerous biases and confounding factors. Anecdotes rely on memory, which is unreliable. Plus, it might be a fluke or the person might have been lying. Why would someone lie about something like seeing a ghost or an alien? Just having an interesting story to tell is motive enough.

The book also covers the Clever Hans effect, which is communicating through nonverbal cues. This effect was named after a horse who could supposedly do math, but the horse was really picking up on cues from the people asking him questions.

Observing someone may alter their behavior. Other factors such as expectation also contribute to the Hawthorne effect. The self help industry is based on hawthorn and placebo effects. Diets work this way as well. Simply paying more attention to what you eat will result in weight loss regardless of the specific diet you pick.

Psychics use cold reading (a collection of techniques such as making vague statements and employing high probability guesswork) in order to trick people (and sometimes themselves) into thinking they have access to unknown knowledge. Psychics and faith healers also use hot reading which involves getting specific information about people ahead of time using something like social media. Peter Popoff had people fill out prayer cards ahead of time and used these for hot reading.