I know there has been much interest in when the National Election Poll would release the raw data from the exit polls. I want to voice one note of caution to those who have been hoping for a quicker release. First, remember my point from a few days ago concerning the different tabulations that NEP did on Election Day:
Those who are concerned about the discrepancy between the exit polls and the actual vote count should focus only on #3, the tabulations prepared just before poll closing for each state. Unfortunately, these were not officially released [see the original post for details on #3].
Another potential problem is that, in the past, Voter News Services (VNS – the forerunner of NEP) did not include the weighting variables or documentation necessary to replicate the “just-before-poll-closing” tabulations in the archival data. I do not have access to the Roper Center Archives, which maintains the old data, so I emailed Ruy Teixeira, who does. His reply:
You are absolutely right that there are no "intermediate" weights provided by the exit poll releases typically. You get one big weight which reflects all the adjustments that have been made to the survey and no guidance at all about how to construct any of the intermediate adjustments. There is no reason to believe that things will be any different this year which is one reason why I’ve been puzzled about people’s fervent desire to see the "raw" data released. The raw data, if they are simply accompanied by the final weight that aligns the raw data with the election outcome, won’t solve the questions people are asking these days because it won’t allow you to recreate where the data were at different steps along the way. I think what people really want to know, at a minimum, is what the data looked like after standard sampling/turnout weighting was done and before it was finally weighted to the election outcome. If this year’s release of the raw data + final weight is like previous years’, it won’t tell you that. [Emphasis added].
What interests me most – and what should interest those concerned about the small discrepancy between the exit polls and the actual results – is the report that Mitofsky International and Edison Research ultimately deliver to their clients and the tabulations and analysis that support it. I also want scholars to have access to the data so they can replicate the analysis and build on it, but that process will take time. For now, I would just like to see the analysis.
A piece of advice for those who want to see more transparency on this issue: It will do little good to harangue the folks at Edison research or bombard them with emails or requests. They are ultimately just the vendors, and the more time they spend answering email, the less time they have to get their work done. Their clients – the ones who own the data – are the NEP partners: ABC, CBS, CNN, FOX, NBC and the Associated Press. Ultimately, they decide what gets released and when.
