Empiricism
A way of thinking about the world rooted in the precise observation of what you can verify with your own senses, and investigate through experience and observation.
Abductive Reasoning
Sometimes known as ‘inference to the best explanation’, this seeks to establish the best possible explanation for something believed to be true.
8 Steps To Apply Abductive Reasoning
- Begin with as precise an account as possible of something that needs explaining. 2. Suggest why it would be significant or interesting to explain this. 3. Present a possible explanation in the form of a theory or hypothesis. 4. Suggest either an experimental method or a non-experimental approach, drawing on diverse sources of evidence, suitable for testing your theory/hypothesis. 5. Investigate whether your explanation does manage to account for (or has successfully predicted) the evidence you have gathered. 6. Acknowledge whether any other explanation or explanations might more convincingly account for your evidence or results. 7. Acknowledge the limitations of your research. 8. Outline possible future investigations to further test and refine your theory – or to seek something different if it has proven unsuccessful.
Explanation
Any attempt, formally or informally, to explain something.
Theory
A general explanation of the underlying nature of a phenomenon.
Hypothesis
A precise, testable prediction designed to allow the rigorous investigation of a theory.
Scientific Method
The systematic empirical investigation of the world through observation, experiment and measurement, together with the development, testing and reformulation of theories.
3 Ideas of Scientific Method
- Replication: can the results we’re basing our theory on be reproduced? 2. Prediction: what predictions can we make on the basis of this theory? 3. Falsification: what evidence is capable of falsifying this theory? This commonly means making use of a null hypothesis in order.
Null Hypothesis
The exact opposite of the hypothesis you’re testing – seeing whether you can falsify a null hypothesis is a common way of ensuring rigour in research.
Occam’s Razor
The principle that, when choosing between explanations, the simplest one is usually best – while more assumptions make something less likely to be true.
Standard of Proof
The threshold beyond which you have decided to accept something as proven, meaning you will not accept something as true if this standard is not met.
Statistical Significance
The probability that a particular result was achieved entirely by chance, as opposed to having a noteworthy cause; setting a threshold for significance is the usual way of establishing a particular standard for proof in an experiment.
P-Value
The probability that an experiment’s results came about through pure chance, expressed in the form of a decimal between one (certainty) and zero (impossibility).
Correlation
Two trends that follow each other closely; the exact degree of correlation between two sets of information can be calculated through a variety of statistical methods.
Causation
The assertion that one thing is the direct cause of another.
When correlation is not causation
- A third factor: this is one of the most common sources of confusion around correlation, and it occurs when a third factor is the underlying cause of two other things that look closely related. For example, the price of your car and the size of your house may be closely correlated, but this doesn’t mean that one has caused the other. Both are themselves probably caused by a third, underlying factor: your wealth. 2. Advantage but not cause: one thing does have a substantial influence on another, but it’s not causal. For instance, it is not just a coincidence that lots of very tall people play professional basketball. Being very tall makes you more likely to succeed as a professional basketball player, while it’s tough to succeed if you’re short. Yet height is neither absolutely essential nor suffi- cient: most tall people aren’t good at basketball, and becoming good requires plenty of other factors to be in place. 3.Entwined cause and effect: the relationship between two things may be real, but both elements may be continually affecting one another. For example, there is a close relationship between inflation and unemployment, but they are both causes and effects at the same time, that is, each continually affects the other. 4. Pure chance: plenty of things are correlated simply by chance, and no meaningful relation- ship of any kind exists between them. For instance, the number of different beers brewed in America has increased over the last decade, and so has national debt. But neither one is likely to have a meaningful relationship with the other. 5. Statistical manipulation: an apparently impressive correlation can be the result of the selective use of statistics, where only the data that shows a desired result is discussed. For example, a small sample of dieters who lost huge amounts of weight might be widely discussed in a diet company’s promotional literature, while several other studies in which dieters didn’t lose weight are never published. 6. Confusing cause and effect: it’s possible for two things to be correlated, but for you to confuse which one is cause and which one is effect. For example, feeling depressed as a result of losing your job could be misidentified as its cause, rather than as a symptom (‘you probably lost your job because you were sitting around full of negative thoughts’).
Conduct Meaningful Research
- Recognize the conditions under which it is possible to meaningfully suggest the presence or the absence of causation. 2. Recognize the conditions under which it is not detailed and suggestive accounts of what is actually going on are the most valuable form of research.
Good Research Tends to…
- Be interested in establishing new knowledge or in rigorously testing existing knowledge. 2. Be conducted openly and transparently within a community of practice. 3. Invite replication by others, and the checking of all its raw results and analysis. 4. Thoroughly investigate an area in sufficient depth and detail. 5. Aim at a fair and balanced account.
Flawed Research Tends to…
- Be interested in seeking confirmation of a particular, favoured explanation. 2. Be conducted secretively or in isolation. 3. Be difficult for others to replicate, or to check and analyse in its entirety. 4. Rely on a superficial, selective or insufficiently detailed investigation. 5. Be overtly influenced by personal biases or other distorting pressures.
Example of Deduction Reasoning (Newton’s Apple)
All objects that are denser than air fall directly downwards, towards the Earth. All apples are denser than air. So the apples in this tree will fall directly downwards, towards the Earth.
Example of Inductive Reasoning (Newton’s Apple)
All objects that are denser than air fall directly downwards, towards the Earth. All apples are denser than air. So the apples in this tree will fall directly downwards, towards the Earth.
Example of Abductive Reasoning (Newton’s Apple)
The apples in this tree, like all other falling objects I’ve seen, are falling directly downwards to Earth. Why is this? Perhaps because the matter that makes up all objects – including apples and the Earth – itself generates a force that creates this attraction.
Example of Deduction Reasoning
The conclusion is a direct, logical consequence of the premises. If the argument is valid and the premises are true, then the argument is sound: the conclusion must also be true.
Example of Inductive Reasoning
The conclusion is supported by the premises, but cannot be proved to be true. If the argument is well structured and the premises are true, it is inductively forceful: it’s reasonable to accept it as true.
