From UBC Wiki

Abductive arguments, or inference to the best explanation, begins with a set of observations that can be explained by competing hypotheses, and claims that one of these hypotheses can best explain the observations.

Really, it’s inference to the better explanation, or the best explanation we have come up with so far. The truth of the premises does not guarantee the truth of the conclusion.

Example of Inference to the best explanation argument:

Suppose you observe (O) the following: Joe did not show up to work You form the following hypotheses to explain O:
H1: joe is sick
H2: joe was abducted
H3: Joe fled the country

Based on one or more factors outlined below (parsimony, predictive power, coherence), you determine that the best explanation is H1. You may then form the following argument:
1. Joe did not show up to work today
2. The best explanation of (1) is that Joe is sick
3. Therefore, Joe is probably sick

How do we determine what makes the best explanation?

That is, how do we adjudicate between competing candidate hypotheses for the 'best' explanation? Some factors that allow us to favour one hypothesis over another include predictive power, coherence and parismony.

Predictive power: How well a given explanation predicts future events, or how many future events it is able to predict

example of two hypotheses, where one should be favoured as a better explanation than the other because it provides greater predictive power:

Consider a situation in which we are trying to adjudicate between two hypotheses about when birds lay their eggs. One hypothesis is that the birds will always lay their eggs during spring. Another holds that the birds lay their eggs whenever they feel particularly happy. The first hypothesis should be chosen based on the principle of predictive power. It is difficult to determine when birds are happy, and therefore hard to predict when the second hypothesis would suggest birds should be laying their eggs. It is easy to tell when it is spring, which allows us to predict when birds will be laying their eggs.

Parsimony: the simpler explanation

For Example: Joe is not in class. His classmate Kyle has come up with two hypotheses for why he is not present. The first is that he overslept. The second is that he is a North Korean spy who has been discovered by the CIA and is currently fleeing the country. Kyle favours the first hypothesis because it is simpler.

Coherence: the explanation or hypothesis coheres with other things we justifiably believe

example of two hypotheses, where one should be favoured as a better explanation than the other because it is a more coherent explanation:

Observation: There is a penguin on an isolated chunk of ice. Hypothesis one: The penguin swam over to this chunk of ice to get there. Hypothesis two: The penguin flew over to this chunk of ice to get there. We would favour hypothesis one over hypothesis two because otherwise we would have to reject the belief that penguins can't fly, which we have good independent reason to believe. Another way of putting this is to say that the first hypothesis coheres better with other beliefs we have.