Monday, 29 September 2014

National 1938-2014 [updated]

David Farrar did two things this morning: first, he posted a carefully truncated graph that got people talking, showing what he claimed is evidence of Labour’s popular decline – with a trendline added in “to reinforce the obvious point” – and, second he didn’t post a similar graph showing the declining trend in support for National, his own party.

Lindsay Mitchell fixed the second omission. Below is her graph showing National’s long-term trend in support since 1938. She also added in a trendline “to reinforce the obvious point.” Like Labour’s, it too declines – in part due to MMP dividing up the vote between more parties than heretofore, and partly due to the slow rise of Helen Clark and the dramatic fall of Bill English in 2002:

So why did Farrar choose 1938 as a starting point rather than Labour’s actual starting point in 1919?

Perhaps because that 1938 election delivered an all-time high result for Labour, allowing the remaining results to show the dramatic decline he desired.

It’s a little bit of a statistician’s simple sleight-of-hand that most of the commentariat has bought almost without demur.

If however we were to do the job his selective starting point obscures, to start instead with the post-war 1919 election instead (the first election in which the modern Labour Party was a force) we’d see a somewhat different “trend.”


I’ll let you draw your own long-term trend line, or medium-term cycle lines, if you think any of them them appropriate.

And I”ll add a link to what must be one of David’s favourite books:

How to Lie with Statistics.

UPDATE: If there is a political party with a genuinely discernible trend, perhaps it is this one …


Answers for which on a postcard, please.


  1. 1938 was the first election after National was formed; that seems the obvious reason for starting there (ie: measuring the elections which National and Labour were both contesting).

    Like various government departments and people you don't like - these statistics are not wrong or made up either! haha! (*curses! foiled again!*)

  2. @Mr Lineberry: Well, no, because if you're going to measure Labour's support over its life then you'd show figures over the course of its entire life, i.e., back to 1919.

    That the United and Reform parties coalesced in 1931 and finally merged in 1938 to form the National Party has little or no relevance to that.

    The selective starting point has been chosen not to illustrate, but to dissemble.

  3. Your original point was whether David Farrar was fiddling the figures by choosing 1938 as a starting point - and no he isn't, and if there is one thing I am well and truly sick of lately it is 'conspiracies' (spying, dirty politics, whaleoil and Collins, massaging statistics)

    He may also have wished to only compare the elections which were Labour vs National.

    Elections prior to 1938 aren't entirely apples with apples for several reasons -

    1. Parties didn't contest every electorate (mainly due to cost)
    2. Labour's 1922 result everyone talks about was 24% from 41 candidates out of 80
    3. There was no such thing, prior to 1938 (when most people had radios) as election 'advertising' as we think of it today.
    4. For instance, the 1925 election (picked at random) was based on door knocking and public meetings for people to make up their minds - not from a news cycle, or 'advertising', as happens today.

  4. My original point is that Mr Farrar is using a well-known technique of how to lie with statistics. That remains my point.

  5. I immediately noticed the "from 1938" choice and thought the same thing you did, however even with the full time line it does support a conclusion that their vote share has been in long term decline, much like the production of things like fax machines rose, peaked, and then declined.

    Lindsay's point is more relevant however - the decline is also experienced by National, so the conclusion Farrar implys by omission about it just being Labour is flawed.


1. Commenters are welcome and invited.
2. All comments are moderated. Off-topic grandstanding, spam, and gibberish will be ignored. Tu quoque will be moderated.
3. Read the post before you comment. Challenge facts, but don't simply ignore them.
4. Use a name. If it's important enough to say, it's important enough to put a name to.
5. Above all: Act with honour. Say what you mean, and mean what you say.