Referendum opinion polling after a tragic event

This post attempts to identify the impact of the tragic death of Jo Cox MP on opinion polling prior to the outcome of the 2016 UK EU referendum (“Brexit”). 
If any reader considers this to be distasteful, please accept my apologies and go no further. 

Jo Cox MP died on 16 June 2016 and her death was announced by West Yorkshire Police at 17:00 that day. Referendum campaigning was suspended and Parliament was recalled on 20 June. The campaigns resumed on 21 June. The polling took place on 23 June with the results as shown below:

Among those who voted (the turnout was 72%), Leave had a lead of 3.8%; among the total electorate this would be equivalent to a 2.7% lead.

On Monday 20 June in a Guardian interview Nicola Sturgeon, First Minister of Scotland, is quoted as saying (my emphasis on her first two, apparently contradictory, sentences):

I think it inevitably will [affect voting decisions]. It’s too early to say whether it will have a direct impact on the result. I think there was a bit of disgust setting in on Thursday morning about the Farage poster. I started to detect a sense of ‘if you’re voting leave, are you associating yourself with that?’. 

Obviously nobody knows whether the debate around the referendum had anything to do with what happened to Jo, but the sense that the debate had become a little bit poisonous and a little bit intolerant and focused on fear of foreigners as opposed to legitimate debate about immigration, I suspect what happened will have intensified those feelings.

The chart below shows the data from opinion polls conducted in June 2016 in terms of the lead for Leave. Polls which did not report “Don’t know/Won’t vote” numbers are excluded. The horizontal lines through the data points indicated the duration of the polling in each case. A typical polling error of ±2.5% is indicated as well. The extent to which polls adjust for voting intention is not always clear so both the leads identified above are shown.

Looking across the month, the polling prior to 17 June can be regarded as being spread around the actual outcome as would be consistent with their inherent margin of error. Polls after this date tend to indicate more support for Remain as the preferred outcome.

One explanation for this might be that, while Cox’s death had little effect on most voters’ real intentions, there was a tendency for them to conceal their intention to support Leave in responses to opinion polling between her death and the vote. So perhaps there was not such a contradiction, after all, in canny Sturgeon’s two sentences.


UPDATE 22 JULY 2016

YouGov's Anthony Wells has now posted on UK Polling Report an interesting review of what lessons can be drawn from opinion polling for the UK EU referendum.  The possible impact of the tragic death of Jo Cox is not discussed.