Simon Hughes and Miaoqi Zhu The Scientific Method

slide1. Simon Hughes and Miaoqi Zhu The Scientific Method<br>
slide2. What is Science? Origin: Comes from the Latin word “scienta” – to know

Mirriam Webster:
a: Knowledge or a system of knowledge covering general truths or the operation of general laws especially as obtained and tested through the scientific method
b : such knowledge or such a system of knowledge concerned with the physical world and its phenomena : Natural Science<br>
slide3. The Scientific Method The process by which scientists acquire a systematic and reliable body of knowledge about the world. Often described in 4 steps:

Characterizations\Observations. Definition of the problem from what is already known. This could take the form of observations that cannot be explained using existing scientific knowledge, or the identification of a gap in our current knowledge that needs filling.
Formulate a Hypothesis. Develop a testable hypothesis that could explain the problem and provide a solution.
Make Predictions. What logically follows if 3 is true? What would disprove 3?
Experiment. Design and conduct some experiments to test the consequences of your hypothesis stated in 3. Do the data support or reject the hypothesis?<br>
slide4. Real-Life Example We use the scientific method in our daily lives, we just don’t know it .
For example: your car breaks down:

Characterizations\Observations: Your car won’t start
Formulate a Hypothesis: Maybe my battery is dead (one of many possible explanations) – Often you want to start with the most likely hypothesis
Make Predictions: If my battery is dead, then my lights and car radio also won’t work
Experiment: Try the headlights. They don’t work. Try the radio. It doesn’t work. Thus it is likely my battery is dead

Solution: Replace the battery.
Often the results of the experiment lead to a solution to the problem, prompt
further experiments. Once you answer one question, you are often left with
many more unanswered questions.<br>
slide5. Scientific Example – Discovery of DNA’s Structure Note that Watson and Crick didn’t discover DNA, rather they determined it’s double-helical structure

Characterizations\Observations : Significance of the gene was known, but the structure of DNA was not
Formulate a Hypothesis : Crick and Watson hypothesized that the gene had a helical structure
Make Predictions: Crick formulates the transform of a helix showing an X- shape
Experiment : (1953) Primed by Crick’s transformation, Watson sees an X shape in Photo-51 (an X-ray diffraction image of the DNA molecule)

Class Exercise: An other example<br>
slide6. The Scientific Method – Broken Down Research an area of interest to you
Define the question
Gather information and resources (observe)
Form a hypothesis
Perform experiments and collect data
Analyze and Interpret data, and draw conclusions
Define future work that follows on from your experiments
Publish results
Retest (frequently done by other scientists)

There is an iterative process inherent in this approach that goes from steps 3 to 7<br>
slide7. 1. Research an area of interest to you Find an area of interest to you
Review the literature on that area
Focus on high quality sources of information:
What is the source?
Books
Journals
Conference Proceedings
Workshop papers
Websites of researchers
Is it recent?
How many citations does it have?
Who funds the research?<br>
slide8. 1. Research an area of interest to you Follow the trail of references
Start with a well regarded and recent paper, preferably a review paper
Obtain and read the references of that paper
Read the references of those papers, and so (bearing in mind criteria above)
If you start with a reputable source, you can somewhat rely on them to have validated the quality of their own refences<br>
slide9. 2. Define the Question What do you want to know?
What do you want to find out about your subject?
Clearly define the problem statement as a question that can be investigated.
Examples:
How does the amount of water effect plant growth?
What limits the efficiency of solar panels?
How fast does Mars orbit the Sun?
Are there gaps in the existing knowledge?
Are there areas that a certain technique has not been applied?
Has all of the major work been completed in this area?
Are there flaws \ limitations with the current approaches?<br>
slide10. 3. Gather information and resources (observe) What do you need to know so you can solve your problem?
Where \ how can you find this information?
What phenomena or processes exist that have not been explained, or possibly even researched?
No-one even thought to ask the question ‘what is gravity’ before Newton
Have observations been made that do not align with the current hypothesis?<br>
slide11. 3. Gather information and resources (observe) E.g. Scientists observed that the perihelion (edges of the elipse) of the planet mercury exhibits a precession (rotation about its axis) that does not agree with Newton’s laws of motion
This is one of the observations that led Einstein to his theory of relativity<br>
slide12. 4. Form a hypothesis Now you have done your research, you develop a hypothesis
A hypothesis is a prediction of the answer to you question
– abductive reasoning (covered later)
A hypothesis is an educated guess at the answer to a problem
This hypothesis also needs to make testable predictions so that it is falsifiable and can be tested through experimentation
Examples:
If a plant does not get enough water it will die
A neural network will out perform a decision tree on the task of image classification<br>
slide13. 5. Perform experiments and collect data How will you test your hypothesis?
What tests will answer your original question?
How much data do I need to collect?
How do I collect the data so that I can perform the analysis I want to do on it
What are the appropriate metrics to use?<br>
slide14. 6. Analyze the data What does the data tell you?
What conclusions can you draw from the data?
What patterns are present in the data?
What techniques can you use to help understand and visualize the data
Table
Graphs
Diagrams
Are there any gaps in your data?
Are there limitations as to what it tells you (indicates potential future work)<br>
slide15. 7. Define future work Did your results support or reject your hypothesis?
If they supported your hypothesis, do you:
Need to gather more data in support of your hypothesis?
Need to address shortcomings or limitations in your experimental design?
Answer questions that could not be answered in your original work
Have questions that your results raise that warrant further investigation ?
If your results rejected your hypothesis:
Are there alternate hypotheses that could explain the problem?
Do your results suggest alternative approaches or solutions?
Negative results are just as valid as positive results
If your results, or some of your results, were inconclusive, then do you:
Need to gather more data?
Need to apply different techniques for adequate verification?<br>
slide16. 8. Publish results Are your results significant enough to publish in their own right, or do you want to include them as part of some more significant piece of work?
What is the most appropriate medium to present your results?
Journal?
Book?
Conference?
Workshop?
etc 9. Retest Other scientists will likely want to verify your experimental results by replicating them, and continuing your work<br>
slide17. Standing on the Shoulders of Giants- Sir Isaac Newton Science is a cyclical process, and builds on what has gone before
Einstein’s theory of relativity built on Newtonian mechanics , it enhanced it, it did not refute it
One experiment normally leads to many more, as some questions are answered, more new questions emerge<br>
slide18. Questions Beget More Questions E.g. When atoms were discovered, it was thought that they were indivisible, and the most granular form of matter (atomos: Greek for indivisible)
We later discovered that atoms are formed of particles – protons, neutrons and electrons
We now know that protons and neutrons in turn are made up of sub-atomic particles, quarks and gluons
String theory posits that all sub-atomic particles are made up of strings of energy vibrating to form some cosmic symphony
“It’s turtles all the way down!” – Stephen Hawking<br>
slide19. The Atom and Beyond - Illustrated<br>
slide20. Logic Logical reasoning is important to the scientific process.
To formulate hypotheses, make predictions from your hypotheses, and interpret and analyze data, you need to use logical reasoning skills
There are two main forms of logical reasoning:
Deductive logic
Inductive logic.<br>
slide21. Deductive Reasoning Starting with one or more premises, or axioms (self evident truths), deductive logic proceeds as a sequence of deductions
If A then B, if B then C, and so on.
“Elementary My Dear Watson”
For deductive reasoning to be sound, the conclusion must always follow from the premise.
If the premise is true, then the conclusion is true
The opposite case cannot be assured (a common fallacy )
e.g. Every human that is alive has a beating heart. If something has a beating heart, it does not mean that it is human.
‘Affirming the Consequent’<br>
slide22. Deductive Reasoning Deductive logic is important for conducting scientific research
We use deduction primarily to
Formulate hypotheses
Form predictions based on those hypotheses
Test theories (our own, or other’s)
For scientific results and conclusions to be valid, the reasoning process must be sound (i.e. logical)
Facts based on unsound or illogical reasoning cannot be relied upon, and should not be incorporated into the body of scientific knowledge<br>
slide23. Deductive Reasoning - Illustrated<br>
slide24. Inductive Reasoning Inductive reasoning begins with an observation, not a pre-existing premise or assumption
In inductive reasoning, people draw conclusions about broad classes of objects or phenomena based on some more specific observations.
Scientists examine a sample and then draw broader conclusions about the population from which the sample was drawn.
In the scientific method, this approach often comes into play when making initial observations, and when analyzing data looking for patterns
The process involves formulating generalizations based on some observed patterns or principles
If A has property B, and A is a C, then other C’s also have the property B<br>
slide25. Inductive Reasoning Illustrated General Principle Observation Observation Observation Observation Observation<br>
slide26. Inductive Reasoning For example:
Observations:
You drop a ball, it falls to the ground
You let go of a fork, it also falls to the ground
You observe a bike leaning against a railing fall to the ground
Conclusion
You may then conclude that all objects with mass fall to the ground if unsupported, a force we call gravity<br>
slide27. Inductive Reasoning Suppose you bite into a number of small, green apples and they all taste sour
From that observation you would surmise that all small, green apples taste sour
(from http://www.scientificmethod.com )
Or, as a child, you may notice that every dog you meet barks, so you surmise that all dogs bark<br>
slide28. Deduction Versus Induction Principles derived from inductive reasoning don’t come with the same guarantees as those from deductive logic
In a deductive propositions, the consequent always follows the antecedent
So “If A then B” implies A entails B, i.e. if A is true then B is always true
Inductive statements are generalizations, they state general principles that are commonly true, but not always
E.g. ”Things that go up come down”
This holds for most objects, but not for those travelling under their own locomotion, such as a plane or rocket.
Or “fruit generally tastes sweet”
This holds for strawberries, raspberries and so on, but not for lemons and grapefruits.<br>
slide29. Abductive Reasoning Abductive reasoning involves proposing potential causes for an observation, often selecting the most probable
Similar to deductive reasoning in reverse, i.e. given B, A is the most likely cause
Particularly prevalent in science, this is the reasoning process used to derive hypotheses
Normally there are many potential causes for a given event, abductive reasoning is providing an educated guess as to what they might be, and often selecting the most probable cause so you can test it<br>
slide30. Deductive Reasoning Abductive Reasoning A B Cause Effect Explanation Observation<br>
slide31. Can Science Prove Anything? We have this extensive, wide-reaching system of knowledge about the world acquired through the scientific method<br>
slide32. Can Science Prove Anything? We have this extensive, wide-reaching system of knowledge about the world acquired through the scientific method
Is any of it proven?<br>
slide33. Can Science Prove Anything? We have this extensive, wide-reaching system of knowledge about the world acquired through the scientific method
Is any of it proven?
Is there a proven scientific theory?<br>
slide34. Can Science Prove Anything? We have this extensive, wide-reaching system of knowledge about the world acquired through the scientific method
Is any of it proven?
Is there a proven scientific theory?
How do we know it’s true?<br>
slide35. Can Science Prove Anything? We have this extensive, wide-reaching system of knowledge about the world acquired through the scientific method
Is any of it proven?
Is there a proven scientific theory?
How do we know it’s true?
In science, we can never know anything for certain, we can only develop theories, and then test them in the real-world<br>
slide36. The Nature of Proof Science and mathematics differ in one fundamental way - you can definitively prove something from first principles using mathematics or formal logic
You can never prove anything definitively in science, you can only falsify, or disprove.
There are no “proven theorems” or facts in science.
This is why it is important to develop theorems that are falsifiable
As a consequence, a good scientific theorem is testable i.e. it makes predictions about the real world that can be tested (and disproven if wrong)<br>
slide37. The Nature of Proof We believe theorems such as evolution and relativity which have withstood extensive attack and scrutiny from the scientific community, and for which a substantial body of evidence has been accumulated
It can take years to amass evidence to fully support a complex theory
E.g. the global warming debate – ‘Has the activities of mankind led to global warming’
In contrast, it only takes one piece of reliable evidence to disprove or refute a theory<br>
slide38. The Nature of Proof One of the principle purposes of the peer review process is to validate a theory by ensuring it can withstand scrutiny by the scientific community
Research that disproves existing theorems is just as valid and useful as research that develops new theorems.
Once a theorem has been disproved, science can proceed in other more fruitful directions.
Through this process of critical evaluation, we acquire the system of ‘reliable knowledge’ that we talked about during lectures 1 and 2<br>
slide39. Null Hypothesis As we can only refute theories, we cannot prove them, scientists often create a null hypothesis and try to disprove it
The null hypothesis is normally the opposite hypothesis to the one you believe is correct
By disproving the null hypothesis, you lend support to you original hypothesis
This uses a technique logicians call “proof by refutation”, or the ‘law of the excluded middle’
A can only be true or false.
By proving Not A = false, then A = true
Failing to disprove the null hypothesis may indicate problems in your original hypothesis, or you just may need to take a different approach<br>
slide40. Null Hypothesis E.g. You may want to attempt to demonstrate that people learn more effectively when studying in the morning as opposed to in the evening
You would then state your null hypothesis as “there is no difference in how effectively people learn between studying in the morning or in the evening”
And then perform some experiments and gather some data to test this theory
Often very little data is needed to disprove a theory, but to ‘prove’ something, you need to gather a lot of data<br>