The CEO and the CIS, publicly-owned opinion polling agencies, were both spectacularly incorrect with some of their predictions about the Catalan election results. Well, at least the Spanish government-owned CIS got the order of the parties right, but the Catalan government-owned CEO didn't even do that. Thus they overestimated ERC and underestimated Junts, among other political parties.
Where the CEO went seriously wrong
Directed by Jordi Muñoz, Catalonia's Centre for Opinion Studies (CEO) had placed the Catalan Republican Let (ERC) ahead of the Junts+ Puigdemont candidature. A major forecast error, given the results this Sunday: Carles Puigdemont's list did relatively well, obtaining 35 deputies and 21.62% of the votes, while those of Pere Aragonès sank like a stone, with only 20 seats and 13.68%. According to the CEO poll of April 26th, election night was supposed to end up the other way around: the Republicans were expected to get between 31 and 37 deputies (20-24%), while Junts support was predicted to be between 28 and 34 (18-22%). In the case of the election winners, the Catalan Socialists (PSC), the calculation was between 40 and 47 seats (28-33%). In the end, the Socialists obtained 42 deputies with 27.96% of the votes.
The matter does not end there, since the predicted results for the People's Party (PP) were also wide of the mark: 8-12 deputies (6-9%) predicted, when they actually got 15 (with 10.97%). In the case of Vox, the estimate also came up short: the far-right party won 11 seats (7.96%), even though the CEO estimated between 5 and 9 (5-7%). The same story with the left-wing CUP, which was predicted to win up to 8 deputies with 6% of votes (in the end they got 4, with 4.09%). And the Catalan polling organization predicted that far-right Aliança Catalana might not enter Parliament (it ended up entering with 2 seats and 3.78%). With the leftist Comuns, they came close: predicted to win between 3 and 6 deputies (3-5%), they ended up obtaining 6 with 5.82%.
The CIS also miscalculates
Similar is the case of the Centre for Sociological Research (CIS), directed by José Félix Tezanos and a public entity belonging to the Spanish government. Specifically, the CIS poll on May 6th also miscalculated: it predicted that Junts and ERC would tie and that the PSC would win by a margin of more than 14 points. Those on Puigdemont's list were expected to obtain between 15.4 and 18.1% of votes, far from the 21.61% they got on Sunday. In the case of the Republicans, the range was between 15.2 and 17.9%, but the reality is that they did not even reach 14% (13.68%). To make matters worse, they also inflated the Socialists: predicted to get 29.8 - 33.2%, when the real percentage was 27.96%.
Minimum (left) and maximum (right) estimates of vote percentages that Catalan parties would obtain, from CIS poll of May 6th
As with the CEO, the CIS also erred with the Spanish conservatives, the PP. The survey gave the PP between 15.4 and 18.1% of the vote, but in reality they only got 10.97%. With Vox, they were close: the forecast was between 5.8% and 7.5%, against the final 7.96%. With the Comuns, the same applied: the calculation was between 5 and 6.7%, compared to 5.8% this Sunday. More of the same with the CUP: the CIS gave the anti-capitalists between 3.2 and 4.6%, and they finally got 4.09%. Finally, they also did well with the percentage won by the Catalan Alliance (AC): they predicted between 3 and 4.4%, and the Islamophobic party obtained 4.09%.