 When 100 people were polled with the following question, are you happy with your social life? We get 73 yes answers, 22 no answers, and 5 undecided or no response. We asked another 100 people the following question instead. Are you unhappy with your social life? This time we get 65 yes answers, 27 no answers, and 8 undecided or no response. So some of these people are clearly both happy and unhappy with their social life. How could that be? This is a graphic illustration of something called confirmation bias. When we go looking for a particular response we are much more likely to arrive at it. What probably happens in the simplest terms is that the people ask the question, compare it against an internal test of consistency. Since most people can recall a time in the recent past when they were unhappy and a time when they were happy with their social life, the response is weighted towards the affirmative for both questions. This is called a positive test strategy. So we're not getting a real satisfaction level, we're just exploiting a psychological quirk of human decision making. Some groups use this phenomenon to their advantage. Our out of 5 doctors agree, 85% express satisfaction with our services and so forth. Confirmation bias is one of the most pernicious and dangerous pitfalls of scientific analysis of data. It's used unconsciously in some cases to arrive at a predetermined conclusion. My previous video referred to the psychological phenomenon of selective acceptance, the biased rejection and acceptance of evidence to support one's preconceived position. Now I want to explore the mechanism used to justify these types of beliefs. After the video, I hope the viewer will be able to spot these fallacies and debates with people with all sorts of different bias perspectives. I'd like to focus in on three areas, but I think that once you start looking for it you'll find confirmation bias in all sorts of places. The areas I want to focus on are alternative health practices, intelligent design creationism and HIV AIDS denialism. Let's start with the broad field of alternative health practices. This includes all the ways that people seek an alternative to scientific evidence-based medicine, so-called Western medicine. We can include a lot of things here that claim a scientific basis but tend not to hold up well under the scrutiny of third-party testing. This includes homeopathy, aromatherapy, diets based entirely on raw vegetables, and here's one that's sure to elicit some angry comments, chiropractic treatment. Insofar as these types of things are used in addition to the care of a licensed physician and don't introduce additional risks, I have very little objection to them. They make people feel good, they foster a positive mental attitude which has its own powerful physiological effects, but they aren't useful as alternatives to demonstrably effective medicine. Let's take as an example colloidal silver. There's a popular idea that a solution of silver produced by a homemade electrolysis rig from a pure silver coin can be effective against a wide variety of illnesses. The proponents point to historical use of silver as a germicidal agent and its present use in topical acne treatments. Much of what will be used as evidence for colloidal silver are anecdotes, stories told about instances where the treatment worked for a personal relation, a friend or family member. Anecdotal evidence is the worst kind of evidence, specifically because of confirmation bias. If you ask someone who has gone to the effort of making silver solution in their kitchen, if they feel better after ingesting a heavy metal with known toxic effects, they will almost always say yes, because at some point since taking the silver solution, they have had a good day and that day is selectively recalled as confirming the desired outcome. When we apply a more rigorous testing to these same people, and especially when we add a blinded control group that take only distilled water, the effects disappear. As a simple example, suppose the person was having trouble with indigestion in the last three weeks. They began the silver, and the indigestion went away two weeks later. Did the silver cause the change, or was this caused by some other change of diet, exercise, sleep pattern or physiology? If you ask the person, they will insist that it was the silver. More confirmation bias comes into play when the alternative therapy is tested by proponents. The data produced is combed through for any positive effects, ignoring the negative or neutral outcomes. The same is done for literature searches and authority figures. The critics of an alternative therapy are attacked as unreliable or biased, but the source of information that supports the preset conclusion is held up as unrefutable evidence. This is most graphically illustrated by the recent anti-vaccine groups. Many still cite a few studies that have largely been discredited, such as the Wakefield paper, where the investigator was found to be in the pay of trial lawyers currently engaged in anti-vaccine lawsuits, as fair and unbiased. And the dozens of other studies on very large populations done by international agencies such as the CDC, NIH and WHO are discounted as biased. The same level of scrutiny is never applied to the sources on the internet, as are applied to major studies in peer-reviewed journals. In my own experience, I find these people to be anything but skeptics. They are merely selectively reinforcing a previously held belief by combing the internet for websites of like-minded individuals. It's unfortunately easy to find such sites, and they are filled with pseudo-scientific claims of evidentiary support, or long lists of anecdotal stories of the dire consequences of vaccines or the miraculous cures of alternative medicine. What is needed and what is used in real scientific studies is a null hypothesis. The null hypothesis is an alternative to the tested hypothesis, where both can't logically be true. In our colloidal silver example, the null hypothesis might be that the health effects of the silver solution are no different than the distilled water control group who were told they were getting silver. If we go looking for either hypothesis to be true, and they both are, we can assume that our results are biased and inconclusive. If only the hypothesis but not the null hypothesis is clearly supported by the evidence, we have determined something useful about the therapy. Most people don't think in these terms. We tend to think about the scientific process as searching for support for a hypothesis. Real scientists have learned to think in terms of falsification. Can a hypothesis be falsified from available evidence? When I see the claims of alternative medicine, I notice that the approach is always the reverse. They go to great links to prove their technique, but never put it to the falsification challenge. The most powerful tool in experimental design of new drug evaluations is the randomized double-blinded large population study. In this case, double-blind means that neither the subject nor the experimenter administering the therapy have any idea whether the subject is receiving the drug or a placebo. It is only after the results have been reported that we allow the researchers to reveal which group the result is from. In this arrangement, there is no way for confirmation bias to be applied. The large population and randomized selection are also important elements to avoid experimental bias. There are cases where it is not feasible, or sometimes not even ethical, to design human experiments in this way, but it remains the highest level of evidentiary support.