Friday, 29 April 2016

Self-serving Bias: Confusing Inputs & Outcomes Since 1967

This post was borne out of a small debate I recently got into online with a couple of people, which highlighted to me what I think is quite a common cognitive error.  I have no data to back up my view that this error is particularly prevalent, this is entirely conjecture based on personal experience.

The error is this:
Confusing outcomes with inputs.

Or perhaps it would be better thought of as:
Disregarding the process(es) used to take us from inputs to outcomes.

This came up in the context of discussing elections and election results.  Specifically, I had two people independently assert to me something along the lines of:

“At the last election, country X voted overwhelmingly for party Y.”

I challenged this statement as highly misleading*, on the basis that party Y, whilst winning an overwhelming majority of parliamentary seats, did not win an outright majority of votes, rather the party won only a plurality of votes.**

Party Y actually received slightly less than 50% of the votes cast in the election in question.

I suggested that it is far more accurate to categorise this level of support as a ‘pretty even split’ than as ‘overwhelming support’.

The response I got from both of my online adversaries was (paraphrasing):

“They won almost all of the seats.  I’d call that overwhelming.”


“They received more than twice as many votes as the 2nd placed party.  And they received more votes than the 2nd, 3rd and 4th placed parties put together.  Looks pretty overwhelming to me.”

Both of these responses have two things in common:

The factual statements made in them are entirely accurate.
The factual statements made in them are entirely irrelevant.

Both of these people have completely missed my point.  They have both made the mistake of confusing outcomes with inputs, or of disregarding the process(es) used to take us from inputs to outcomes.

I do not disagree that winning almost all of the available seats in an election is rightfully considered an overwhelming victory for any party.

However, this is a very different thing from the party in question having ‘overwhelming support’ or the electorate ‘voting overwhelmingly for’ them.

“The electorate of Country X voted overwhelmingly for Party Y.” is a statement about the INPUT to the election.

“Party Y won an overwhelming victory, taking almost all of the seats in Country X’s Parliament.” is a statement about the OUTCOME of the election.

These statement are talking about two very different things.  It is entirely possible for either one of these statements to be true and the other false simultaneously.

Statements about INPUTS do not directly translate into equivalent statements about OUTCOMES and vice-versa.

The link between INPUTS and OUTCOMES are PROCESSES.  In this case, the process is the particular voting system used: how people cast their ballots, how votes are counted, how the electorate is divided into constituencies, etc.

It’s not possible to turn INPUTS into OUTCOMES without PROCESSES.  Similarly, it’s not possible to turn a statement about INPUTS into a statement about OUTCOMES (or vice-versa) without considering the particular PROCESS(ES) that took us from those INPUTS to those OUTCOMES.

As an example, if a political party wins 100% of the seats, without receiving any votes (e.g. in an absolute dictatorship), it is correct to say that such a party has overwhelming power or authority, but an error to say that it has overwhelming support.

The reverse is also true.  If a political party doesn’t win any seats, despite receiving 99% of the votes, due to extreme gerrymandering or a rigged election say, most people wouldn’t argue against the notion that such a party has overwhelming support, but it would be an error to say that it had won an overwhelming electoral victory.

These examples are extreme to illustrate my point.  In modern democracies, things are rarely so one-sided.  This is probably for the best.

I think in these cases this error is a manifestation of self-serving bias.  The people I was discussing this with were clearly supporters of Party Y.  They like to think – as do we all – that other people mostly see the World the way they do (or would, if only the ignorant fools were better informed) and are broadly supportive of the same things, that their ideas are popular and that ‘their team’ is ‘winning’.

This was illustrated when I brought up another vote on which one of my debating partners was on the ‘losing team’ by a relatively narrow margin.  They were unwilling to characterise this as ‘overwhelming opposition’ to their position.  The lack of symmetry is striking.  They are happy to characterise a mite under 50% support for their position as ‘overwhelming support’, but refuse to refer to 55% opposition to their position as ‘overwhelming opposition’.  The cognitive dissonance is immediately obvious to anyone not emotionally invested in their ‘team’.

I knew I’d won the argument when I got a response to this point that began:


* Put aside for now the fact that a country, as a collective, has no agency or ‘will’ of it’s own and so cannot be said to vote for/against anything (unless the vote is unanimous, and even then there are caveats), as if the views and opinions and wishes of the entire populace were homogenous.  For our purposes here ‘Country X’ is convenient shorthand for ‘The electorate of Country X’.

** I’d go further than that and say that even if a party wins 50% + 1 votes, it’s still highly misleading to describe this as ‘overwhelming’ support.  However, reasonable people can and will disagree about exactly where to draw the line between what constitutes ‘overwhelming support’ and merely ‘support’.  Some people may draw that line at 66.7%, some at 75%, some maybe at 90%, or even higher.  However, something I do not think is at doubt is that the point at which something can be considered to have ‘overwhelming support’ is definitely somewhere north of 50% (i.e. at minimum has a majority behind it).

Monday, 25 April 2016

The Price of Steel in India

This is a follow up blog post to this one here, summarising some more thoughts I’ve had on Scott Alexander’s ‘The Price of Glee in China’ post and inspired by some of the discussion in the comments there.

I know this is a conversation from a month ago, so is now ancient history in blogging terms.  This is why I’m not a very good blogger.

There’s a comment in the middle of my post that I sort of regret making because it was poorly worded and it’s not really crucial to the point I was trying to make, but lots of people who are presumably sympathetic to protectionist arguments seem to have fixated on it.

The phrase in question is this one: there is nothing economically or morally relevant about political boundaries

I have to apologise for the extremely poor wording on my part here.  This phrase is not what I intended to say at all.  That was sloppy.  Sorry.

As pointed out by several other commenters on Scott’s post, politics does have a substantial effect on the economics of a country.  In the most obvious ways, politics and politicians influence tax (and subsidy) rates and hiring practices / regulations (e.g. occupational licensing, minimum wage laws, the EU working time directive, etc. etc.).  They also have a less direct influence through things such as health and education policy.

I don’t debate any of this.  If someone got the wrong end of the stick from my previous post, I can totally understand why and again can only apologise for my sloppy phrasing.

Anyway, this all detracts from the main point I was trying to make in response to Scott’s post.  So let me try to make that point again, hopefully in a clearer and more succinct way:

Scott summed up his argument as:

If we were to actively try to keep the Indians from industrializing, that would be pretty awful. But that’s not the argument at hand here. The argument at hand is ‘are we morally required to sacrifice our own economy in order to help the Indians industrialize?”

My response to that is:

But that is the argument at hand here.

Put aside for the moment whether protectionist policies are generally a net positive or a net negative for the population (as a whole) of the country that enacts them.  Put aside also whether protectionist policies are, in general, even effective at bringing about their stated aims.  These are debates for another day.

Let’s assume, for the sake of argument, that tariffs imposed by the US Government on imports of Indian steel helps American steelworkers, with no offsetting cost to any other Americans whatsoever.  In other words, the tariffs are a net positive if we consider only the welfare of Americans.

The cost of this policy is paid entirely by Indian steel companies, who now find it more difficult to compete in the American market.

What is this policy if not an active intervention by the US Government which retards industry in India and harms the Indian economy?