I used to be very closely involved in running the Lib Dem Voice surveys of party members (until I stood down from the editorial team) and the FAQ I wrote two years ago gave then a pretty solid set of reasons why they were very likely to be accurately representing the party membership overall.
Since then there have been some snippets of further supporting evidence in extracts from the party’s own polling of its members I’ve unofficially seen.* But that gets completely knocked sideways by what Nick Barlow aptly describes as the LDV polls ‘Literary Digest moment’, a reference to the most famous example of a previously accurate poll getting an election hopelessly wrong, in this case involving another President, but one who could order the dispatch of troops rather than the dispatch of membership cards, Franklin D Roosevelt.
For Lib Dem Voice it was a case of getting the party President contest very wrong.
Does that mean the LDV polls are sunk irredeemably? I hope not, as I think there is great value in having reliable information about what party members thinks – and such information not to be wholly within the control of party HQ.**
So here’s my suggestion for trying to fix them: in the past weighting the results by age, region, gender, attendance at conference or length of membership didn’t seem to be necessary as it made little difference to the results.
Give it a go again and see what happens and even better try it out on the Party President poll (as some such questions were asked), as what seems to be the problem is failing to balance in the polling those activists who network heavily with other activists on the one hand with less connected or active party members on the other.
UPDATE: The online versus offline voting figures for the election show that the Lib Dem Voice survey was much better at predicting how online voters would cast their votes than postal voters.
* For example, during this Parliament, the party has twice telephone surveyed party members and asked for views about the party being in government. On both occasions over 80% of party members supported the coalition, which is in line with the Lib Dem Voice polls.
** But as an aside, double bonus points to Austin Rathe from HQ for his active participation on social media during and after the Party President count, helping adding accurate information to various discussions.