After the problems of 2015, and the need to deal with the issues raised by the subsequent Sturgis enquiry, the pollsters were inevitably going to be under considerable scrutiny in the 2017 election.

The polls proved very good at predicting the Conservative vote, with an average of the final polls being only 0.2% higher than the actual result. (The actual result shares are based on all seats except Kensington).

They did, however, prove much less good at predicting the Labour vote, with the pollsters’ average being 5.2% below Labour’s actual share. This is only the second election since 1987 when the pollsters have underestimated the Labour share.

The average poll figure for the Liberal Democrats was very close, being only 0.3% different from the result.

While the final polls were not ideal, the BPC does not feel there is a need for another formal inquiry; the detailed findings of the Sturgis review are available here and provide a reference point for understanding the issues and challenges. Instead all BPC members who produced final polls will produce a “lessons learned” report, and these will be presented at a conference before the end of this year.

	CON	LAB	LD	UKIP	Green	Other	Method	Sample Size	Fieldwork
Opinium	43	36	8	5	2	6	Online	3002	June 4
Survation	41	40	8	2	2	6	Telephone	2798	June 6-7
Ipsos MORI	44	36	7	4	2	6	Telephone	1291	June 6-7
ICM	46	34	7	5	2	6	Online	1532	June 6-7
ComRes	44	34	9	5	2	5	Online	2051	June 5-7
YouGov	42	35	10	5	2	6	Online	2130	June 5-7
Panelbase	44	36	7	5	2	6	Online	3018	June 2-7
Kantar Public	43	38	7	4	2	6	Online	2159	June 1-7
BMG	46	33	8	5	3	6	Telephone / online	1199	June 6-7
Average	43.7	35.8	7.9	4.4	2.1	5.9
Result	43.5	41	7.6	1.9	1.6	4.4
Difference	0.2	-5.2	0.3	2.5	0.5	1.5