Bias is good - Necessary even

There is a popular conception that one should evaluate the evidence, free of personal biases, and determine what it indicates. This is a highly intuitive idea. It’s also flat-out wrong.

Evaluating data without recourse to bias is both practically impossible and undesirable, even if it would be practical. To be biased is not to consider the data in front of you on its own merits, but rather holistically as part of a construct that includes your preconceived notions and preferences.

There are, in fact, a class of people who evaluate the evidence in front of them largely free from bias – namely, very young children. Children, free of bias, often reach erroneous conclusions, necessitating the need for biased parental guidance. Presented with a missing tooth, and its replacement under their pillow by some money – and taking no other information into account – concluding that the tooth fairy is real is not particularly unreasonable. There is another word that means much the same as bias – experience – and it is only with experience that children come to realize that the tooth fairy is an extremely unlikely hypothesis for what they are experiencing. This bias allows them to reject the apparent explanation and look for an alternate one that fits their learned biases about how the world works.

Adults, by and large, if not universally, operate from a position of bias when making any decision. Data is generally limited and incomplete. If you take it ‘on its own terms’, you’re basically guessing. Bias is what allows data to be meaningful. If you are solving any remotely advanced science or math question, you start by laying out what is ‘given’ – i.e. declaring your bias. If you start from nowhere every time, you never get very far. Everyone fills in the gaps with their biases – some biases are simply more accurate than others.

Weeding out prejudice and other incorrect bias

This, then, leads to the obvious question. Obviously not all bias is true. It can’t be, as your bias and my bias may be mutually exclusive. So how do we weed out the accurate bias from the inaccurate? Fortunately, there is a tried and tested method – the scientific method.

There is another misconception that the scientific method is used to evaluate data. It’s not, not really. As far as evaluating data, the scientific method is the diametric opposite to the idea that one should, “Examine the evidence, and see where it leads.” This is exemplified by the very core of the scientific method – namely that it requires a hypothesis in advance. To execute the scientific method, you have to think you know what will happen before you even run the experiment. The method’s core purpose, in fact, can be seen to be to avoid evaluating the evidence solely after examining it. Doing that has proven time and again to simply lead to the examiner deciding the evidence supports their preconceptions, no matter how badly they have to mangle the evidence to make it fit. Laying your cards on the table, as it were, in advance, helps mitigate that.

But it goes further. Anyone who has ever designed an experiment knows that the data therein is anything but pure and unfiltered, at least if you’ve designed the experiment properly. The data is selected, pruned, and massaged in order to eliminate confounding factors before it is ever considered as part of the experiment. As a way of evaluating the evidence in an unbiased way, it seems a rather backward method.

The scientific method exists for the sole purpose of examining bias

The reason for all this is that the scientific method is NOT in fact a way of evaluating the evidence in front of you. It is a way of evaluating biases. You start a scientific experiment with a hypothesis – declared bias – not with data. You decide what kind of conditions you are rather sure will definitely lead to the expected result. You then carefully construct the situation where those exact conditions are met. Only that data is considered. It is precisely because you have declared your expected outcome, and then done everything in your power to guarantee that outcome, that you are forced to conclude that your bias is incorrect if the expected outcome fails to materialize. That is your null hypothesis.

This is what scientists mean when they say that experiments can never prove anything. They simply disprove the alternatives. Because scientific experiments exist for one reason, and one reason only. To evaluate whether a bias can be shown to be false. Even if you get the result you expect, it’s only because every properly constructed experiment stacks the cards in the favour of the expected outcome. It’s entirely possible that a more messy, or just a different, collection of data would yield a different result. It is precisely because the cards have been stacked to the greatest extent possible that a result different than that expected is convincing. And this is just about the best one can do. No other method has proven nearly as powerful as the scientific method for evaluating truth.

So, to recap, bias is a critical element of logical thinking. Data taken alone is, generally speaking, useless. It needs to be fit into a broader framework to make sense. That larger framework is simply bias by another name. But not all bias is created equal. To be a logical thinker, one must constantly examine their biases through application of the scientific method – a method that exists for the singular purpose of weeding out false bias.