After reading the draft Bill no. 38, the Electoral System Referendum Act, recently tabled in the House and receiving first reading there are two obvious and essential changes required to make the Bill fairer. It has yet to be debated, so hopefully both these changes will be suggested and garner the full support of all MLAs….or at least they will if government MLAs are sincere in stating that they want a simple and fair referendum on our electoral system.
The draft Bill states at the outset: “The purpose of this Act is to make the process for the referendum transparent and fair…” then goes on to outline various measures by which the government believes that purpose will be achieved. If the government is serious about achieving a truly “transparent and fair” referendum, it will make at least two significant changes to its draft legislation.
1. Fix the implicit response order bias in the form of the Question.
Are people more likely to select the first or second option presented in a “yes – no” question? Short answer: the first option, referred to as the “primacy effect”. There is a wealth of credible scientific research that has demonstrated a clear and significant “bias” in favour of the selection of the first option presented visually.
I’ve reviewed a number of these reputable scientific studies and they have all arrived at the same conclusion. I won’t get into this research here since it is unequivocal and conclusive. I’ll simply offer a link to one article, along with a list of those studies taken from that article:
Similarly, the research has found that when options are presented orally, the last option is more often selected (i.e., referred to as the “recency effect”).
Despite the normal practice of listing “yes” first and “no” second with “yes – no” questions, the ballot design for the electoral referendum has listed “no” first and “yes” second (see the first example in the image at the top of this article), giving a clear bias in favour of the “no” option. This primacy effect bias is even stronger when options are listed “vertically” rather than horizontally, for various impressionistic and psychological reasons – e.g., “winners” are seen to be on “top”; what’s on top is “dominant” (greater); what’s below is subservient (lesser), etc.
This double-bias favouring “no” over “yes” can be easily corrected in the following manner (see bottom example in the above image):
(a) Place the Yes – No options horizontal rather than vertical; and
(b) Print half the ballots with “No” first, and half the ballots with “Yes” first
2. To be fair, only ballots “cast” for the referendum should be counted in the calculation of 50% + 1
Section 4 (1) of the draft Bill reads:
“4. When referendum is binding (1) The result of the referendum is binding only if more than 50% of the electors recorded in the poll book as having voted in the general election vote the same way on the question.”
I can’t say that I fully understand this section of the Bill because I’m not exactly sure what is intended with the wording: “recorded in the poll book;” however, it appears there will be just one poll book for both the general election and the referendum.
Given that the referendum question is structured in such a way as to require a vote for MMP, anyone recorded in the poll book but not casting a vote in the referendum will apparently be counted as a vote for the contrary option (FPTP), despite the fact that they don’t actually vote in the referendum. That is patently unfair.
On the other hand, if some people show up to vote but – for whatever reason – decide not to vote in the general election, but only in the referendum for the MMP option, it is not clear whether their name be recorded in the poll book and count as a vote for MMP. Here the wording of 4(1) is both curious and confusing, because it specifically says that the result of the referendum is to be calculated on the basis of “electors recorded in the poll book AS HAVING VOTED IN THE GENERAL ELECTION VOTE…..” This clearly requires clarification. A lawyer friend of mine commented that this wording is “nonsensical”.
If the referendum is to be fair, only the actual ballots in the referendum should count. If the threshold for the successful option is 50% + 1, then that number should be 50% + 1 of the physical referendum ballots – since a “yes – no” set of options is contained on the ballot – and have nothing whatsoever to do with the number of people who vote in the general election, but may or may not vote in the referendum.
What would be preferable is if there were two options on the ballot positioned horizontally (FPTP & MMP) and you simply check the box beside the one you want. And half could be printed with FPTP first, and half printed with MMP first, in order to give each option the same degree of “primacy effect” response order bias.