The 5 statistical errors in the media
The 5 Worst Statistical Errors in the Media: With Surprising Examples...
5. Selection Bias or Sampling Bias.
A good poll must be based on a representative sample. In 1936, the American presidential election pitted Republican Landon against Democrat Roosevelt. A major magazine conducted a massive poll. Ten million people were contacted via the telephone directory, and two million responses were collected. The result was clear: a landslide victory for Landon with 57% of the vote, and only 43% for Roosevelt. However, the actual election result defied common sense: only 38% for Landon and 62% for Roosevelt. A 20% error. How can this discrepancy be explained?
4. The Law of Runs.
The media loves to highlight run-of-the-mill results. But this is a statistical illusion. Gerald Bronner calls it the "rake effect." Human beings have a flawed understanding of chance. He imagines that chance doesn't follow patterns and that it's homogeneous, evenly distributed, and spread out.
3. The Average Rather Than the Median
In general, the media assumes the public understands the concept of the average, but not the median. Yet it's not that difficult to grasp. The average salary is the sum of all salaries divided by the number of individuals. And the median salary is the salary of the person in the middle of the scale. The salary of the average citizen. Unfortunately, the media almost always use the average rather than the median. Every year, the average wealth of Swiss people is published, which is around 560,000 francs. That's a lot. But this number doesn't mean much because the distribution is very skewed. It would therefore be much more accurate to use the median wealth, not the average wealth, as a point of comparison.
2. Relative and Absolute Indicators
The media mix up two types of poverty indicators without explanation. Absolute indicators, like the $2-a-day threshold. And relative indicators, like the poverty line. When are we below the poverty line? Generally, when we earn less than 60% of the median wage in the country. This relative poverty has nothing to do with poverty in the absolute sense.
1. Causality or Correlation
This is the source of most media errors. A correlation does not necessarily imply causation. Eating chocolate doesn't guarantee Nobel Prizes. The most likely explanation is that there is an external factor, a confounding factor.
Those who live near a high-voltage power line are more likely to be ill than others. Causality or simple correlation?
We will never know for sure, as the media wildly infers unproven causal relationships, and I hope I have made you aware of this.