Evidence: What Evidence?

You probably are aware of ‘Likes’  on places such as Facebook, Twitter and Instagram. These ‘Likes’ are put into an algorithm and then advertisements come up due to what you apparently ‘Like’.  I love to confuse and add in random ‘Likes’, bell-ringing, star-gazing, flower arranging, shoe repairing etc. My favourite response was from a computer scientist on an Open University programme about computers and statistics. He said that when he bought a book online a recommendation would come up telling him: ‘People like you like the following books…..’ To which he said to camera, ‘I don’t like people like me’!

My mother used to tell me that one day they will tell you that Salmonella is good for you. They will produce statistics to prove it. They will say that this is scientific proof. But statistics are not scientific proof. Statistics are mathematical proofs.

Statistics measure the correlation between two events. How likely they are to occur together. Statistics tries to remove confounding variables; things that also happen at the same time, but are not really correlated. But they do not prove cause and effect.

Science is interested in cause and effect. It is not empirical, another mistake made by non-scientists. Empiricism does not measure cause and effect.

Science measures the link between cause and effect. It does this by experiment. That is why most other disciplines are not scientific. They call themselves scientific (Social Sciences, rather than Sociology, for example) but they cannot do the experiment to test the theory. They can only do the stats.

Example A: If you want to prove that ‘watching violent television makes children violent’, you would need to

  • take two groups of children 24 hours a day, 7 days a week and
  • put those two groups of children in exactly the same environments (food, clothes, rooms, etc) and
  • let one group watch violent TV programmes and
  • not let other group watch violent TV programmes
  • for a number of days/weeks and
  • see if there is a difference in behaviour afterwards.

This experiment would never be allowed, it is unethical; so you cannot do the experiment to prove the link. You can do the stats on children that watch violent TV, but you cannot be sure other things have not had the effect.

Example B: If you want to say that the ‘killing of Archduke Ferdinand in Sarajevo started the First World War‘ you would have to

  1. travel back in time to Sarajevo 1914 and
  2. ‘un-shoot’ Archduke Ferdinand and
  3. see if the First World War still happened.

Good luck getting the grant for that research!

A number of academic disciplines that say they are using scientific methodology are not doing so and a number that say they are using statistics are still not doing scientific methodology.

The example I give my students are two excellent papers by very good teams.

  • one paper says that feeding babies on breast milk makes them more intelligent and
  • one paper says that feeding babies on breast milk doesn’t make them more intelligent.

How do they  find opposite results? By statistical analysis. There is no experimental proof to either claim. There are a lot of confounding variables which get in the way of the correlation between milk and intelligence.

All the scientific evidence points to Salmonella not being good for you, so whatever the stats say, please don’t try the Salmonella.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s