Umfrage-Decoder: So lesen und interpretieren Sie öffentliche Umfragen (Abacus)

New to me. I think it’s a good guide that seeks to temper the „horse race“ dialogue around polls.

Summary:

– … polls are also widely misunderstood. Many people read them like sports scores. Who is ahead? Who is behind? Who gained a point since the last poll? … A well-designed survey is not simply a snapshot of who is winning — it is a window into how Canadians think about issues, how they evaluate leaders, and how events shape the public mood.

– Contents:

– 01 Start with the Purpose;

– 02 Understand the Sample;

– 03 Consider the Survey Mode;

– 04 Pay Attention to Field Dates;

– 05 Understand Margin of Error;

– 06 Examine Question Design;

– 07 Look Beyond the Topline;

– 08 Why Polls Can Differ;

– 09 Snapshots, Not Predictions;

I’ll add the following to this guide. Composite or aggregate or consensus polls (e.g. Canada 338) that average the results of several poll companies cut out the 2-3% random variation (i.e. noise) you find in a timeseries of polls. This variation is more like 5-10% for the regional data. When looking for trends, it’s a good idea to wait for 2 or 3 polls to come out before talking about „movements“ in public opinions.

Their disadvantage is that they are slow to pick up on any sudden movements in public opinion, so they are of limited use to political strategists compared to individual polls.