Using Facebook data to predict the 2016 U.S. presidential election

We use 19 billion likes on the posts of top 2000 U.S. fan pages on Facebook from 2015 to 2016 to measure the dynamic ideological positions for politicians, news outlets, and users at the national and state levels. We then use these measures to derive support rates for 2016 presidential candidates in all 50 states, to predict the election, and to compare them with state-level polls and actual vote shares. We find that: (1) Assuming that users vote for candidates closer to their own ideological positions, support rates calculated using Facebook predict that Trump will win the electoral college vote while Clinton will win the popular vote. (2) State-level Facebook support rates track state-level polling averages and pass the cointegration test, showing two time series share similar trends. (3) Compared with actual vote shares, polls generally have smaller margin of errors, but polls also often overestimate Clinton’s support in right-leaning states. Overall, we provide a method to forecast elections at low cost, in real time, and based on passively revealed preference and little researcher discretion.

Democratic Deficit or the Europeanisation of Secession? Explaining the Devolution Referendums in Scotland

This article deals with the variation in the demand for self-government in Scotland – as measured by the vote in the two referendums – between 1979, when devolution was rejected, and 1997, when devolution was endorsed. The existing literature mainly deals with each of the two referendums in isolation and does not offer an explicitly comparative analysis of them. However, implicit comparisons contained in analyses of the 1997 referendum tend to identify as the main cause of the variation the ‘democratic deficit’ created by Conservative rule between 1979 and 1997, which was consistently rejected in Scotland. I take issue with this explanation on theoretical and empirical grounds and advances an alternative account grounded in an explicit comparison of the two referendums. Based on a rationalist approach, the analysis presented here identifies three key elements in the voting dynamics at the two points in time – a gap between support for self-government and the actual vote in the referendum; an interaction effect between attitudes to devolution and to independence; and the role of the European context in shaping perceptions of independence. I argue that significant change in these three variables (rather than a ‘democratic deficit’) appear to have been the most important determinants of the different results of the two referendums.

L'élection québécoise de 1985: un bilan des sondages

A great number of polls have been made public by the media during the 1985 electoral campaign in Quebec. The polls correctly predicted the winning party, but were not very good for forecasting the actual vote. A difference of 3 percent has been observed between the prediction and the vote with respect to the Liberal party and third parties considered as a whole. It is shown that the discrepancy must be attributed more to those who do not answer polls at all than to those individuals not willing to reveal their vote intention. Finally, the campaign seems to have had a small impact on the vote, the Liberal party increasing its share by 2 per cent and third parties loosing many votes among those who were tempted to support them at the beginning of the campaign.

A PREDICTIVE STUDY OF VOTING BEHAVIOR USING LAMPERT'S POLLIMETER

A voting behavior model hypothesizing voting intention as an intervening variable which mediates the relationship between voter attitudes and voting behavior was examined in the framework of the general election in Israel Voter attitudes were assessed via a new attitude scaling device embodying a number of characteristics that make it a promising tool for attitude measurement. Voter attitudes and intentions were collected prior to the elections, while the actual vote was collected after the elections. The model tested primarily via discriminant analysis, has gained considerable support. Implications of the findings are discussed and avenues for future research are suggested.

Memory for Past Vote: Implications of a Study of Bias in Recall

For political scientists and pollsters the way the individual voted on previous occasions provides an important source of data. In the absence of longitudinal studies, recall of past vote tends to be taken as equivalent to actual vote cast. How accurate is such recall? How far does accuracy decrease with time, where recall concerns not one, but two, previous elections? How far do errors introduce a systematic bias in the conclusions drawn from such data?