> The only semi-serious use of Twitter polls I've ever seen is stuff like "which game should I stream next". It's very obviously not a representative sampling

> The only semi-serious use of Twitter polls I've ever seen is stuff like "which game should I stream next". It's very obviously not a representative sampling of anything except your most engaged followers. Unless you manage to achieve global representation, most polls work by taking a sample from a specific population. In this case, it usually involves a segment of your followers or, occasionally, a broader Twitter audience. However, it clearly does not include people who refrain from participating in random Twitter polls. The polling conducted by your online newspaper represents a portion of their readers. When they perform on-the-ground reporting, the sample consists of people present at the actual physical location, and so on. This is the reason we aggregate multiple polls to obtain global numbers because individual polls inevitably have a limited audience, one way or another.