What makes the wind blow?

How do birds fly?

Our world is full of curious phenomena.

To find answers or solve problems,

we can use a process, which was first acknowledged

by the scientist and philosopher

Ibn al-Haytham, in the 11th century.

Also known as Alhazen,

he is considered to be the father of optics

- and the scientific method.

There are six steps to it:

1.Observe and Ask Questions

2.Research

3.Formulate a Hypothesis

4.Test your hypothesis

5.Conclude

6.Share Results

The goal of the scientific method is find out the truth.

Let's try.

Step 1: Observe and Question

Observation helps us formulate challenging questions

that you will be able to test.

A good question converts the natural sense of wonder

into a focused line of investigation.

When is the best time to drive to school?

Which food is my dog's favorite?

For example

If you observe that women smile more often than men,

you might ask: why do women smile more often?

Step 2: Research

Find out if other people have asked the same

or similar questions.

If you research online, use search terms like

“study ...”, “research ...” or “meta-analysis ...”

- which is a summary of research for a specific topic.

Read as much about your particular subject

to see what you can find out about.

For example, Research happiness based on gender

or study the science of smiling

in different cultural contexts.

Step 3: Formulate a Hypothesis

A hypothesis is a theory that you can test

to see if your prediction is right or wrong.

From your observation,

you have noticed that woman smile more often

and that people who are smiling seem to be happy.

From your research

you know that there are different types of smiles,

shy, genuine and false.

In one paper you read that baby girls

smile more often than baby boys.

Here is a hypothesis:

Women smile more than men

because they are happier than men.

Step 4: Test Your Hypothesis

When you test your hypothesis,

you want to make sure to do this in a fair way

and that the conditions are constant.

For this hypothesis, we can design a test

where an interviewer talks with a set of men

and women for 5 minutes each,

counts how many times they smile,

and then asks each one to rate their level of happiness.

To get a good sample of the population,

we invite 300 women and 300 men.

Seems like a good test, right?

But wait, what if the interviewer is a woman,

and men tend to smile more at women?

Or vice versa?

Or what if the topic discussed

is one that interest women more than men?

And what if people aren't reliable reporters

of their actual level of happiness?

So clearly, we would need to be much more careful.

Step 5: Analyze and Conclude

Let's assume that you designed a very careful experiment,

controlling for as many variables as possible.

Now you can analyze the data

to see if your hypothesis is correct, or incorrect.

Depending on your findings,

you may want to change your hypothesis

or change the design of your testing.

Perhaps you have discovered an even more interesting question.

This stage of the scientific method

can be repeated as many times as necessary

until you find just the right hypothesis

and test method to find accurate results.

Step 6: Share the Results

When you are satisfied that you have

proven or disproven something important,

report your results.

In science, it is important to detail your methods

so that your peers can review your work

- which is a critical step to getting published.

If your results are solid,

your experiment can be repeated by other scientists.

Such reproducibility is a sign of good scientific work.

But failed results can also be interesting

- an incorrect prediction could prove to be

important and should always be reported.

To make sure you get it completely right,

here are 3 more things you can check before you publish:

A) Any scientific theory is falsifiable

Real scientists know that there is no such thing as a scientific proof.

In other words, you can never prove your theory to be 100% right.

All you can do is find A LOT of supporting evidence

that it could be correct.

Here is one example:

Say that someone says “hamsters CAN fly,”.

We cannot prove that this as false.

Yes, we have never seen a hamster fly,

but we can't test all possible conditions

or look in all possible places on the planet

to know that ALL hamsters NEVER fly.

Maybe a space hamster does?

So while we can often prove that a phenomenon exists,

it's much harder to prove the nonexistence of something.

If your theory can't possibly be proven wrong,

then it's not falsifiable and hence, not scientific.

B) Correlation is not Causation

When you analyze your results,

it is important to separate between two possible reasons:

correlation or causation.

Let's you hear that towns that have more churches

also have more bars.

Could it be that religion makes people want to drink?

Or that drinking helps people to find God?

If you add more facts, such as

“larger towns have both more bars and more churches,”

you can see that a larger population

is a more likely cause of higher numbers of bars

AND churches.

There is probably a correlation, but no causation.

If we compare men with women

and would conclude that woman smile more and are more happy,

then this still doesn't mean that its happiness that makes them smile.

Maybe they just eat more chocolate and cookies,

which makes them both: happy and smile a lot.

C) Avoid Selective Windowing

When you publish you got to show ALL relevant facts.

Colgate once ran a advertising campaign

claiming that “80% of dentists recommend Colgate”.

What they didn't tell us

is that when they asked dentists

to select their preferred toothpaste,

Colgate was just one of many other brands

they also also recommended.

Colgate was later sued and forced

to take down their misleading ads.

The purpose of science is always to find out

the truth and nothing but the truth.

To use science to mislead us is wrong

and terrible business practice.

Lets do a last example together.

I have two coins.

One is bigger.

Why?

The small coin says 1 Cent,

the bigger one says 5.

Aha!

Small coins are worth less money.

Bigger coins are worth more money.

I pull some more coins from my pocket.

2 more Pennies, 1 more Nickel,

and a Quarter Dollar, which is 25 Cents.

Great, my hypothesis seems true.

But wait, is the quarter worth more because it is bigger?

So is that a correlation or a causation?

Hmmmm…

My sample size is pretty small.

I don't think I am ready to report my results.

Can you help out?

Please apply the Scientific Method

to study you local currency.

Maybe you have a hypothesis

that we can test until we get solid,

repeatable results to report.

Please publish your findings in the comments below!