They never tell you enough on the statistics on the news to actually get much solid info. Like sample size, error bars, did they fit a particular distribution, what was tossed as outliers and more.
I always think about the chance of rain report and the different explanations of what it might mean (I don’t recall which is correct, but it illustrates how confusing it can be) :
Chance of rain today based on a predictive model
Is different from
It WILL rain today in this region and this percent of land area will get rain
Is different from
Over all the historical data for this region, we got rain on this percent of days when the conditions were “the same” in the past.
But all can be reasonably said to be 59% chance of rain today.
It does vary from place to place but in North America at least, most often it’s that it will rain in __% of the land area.
Additionally, I would add that knowing about statistics at least allows you to understand that without that additional info, any stat is essentially meaningless, or at least easily misleading.
Agreed. But part of the reason why they never report the numbers is that most of the readers don’t understand them anyway.
And a lot of the time it is not even just complex studies people don’t understand. There are also those who can’t even judge the magnitude of numbers, e.g. when it comes to spending or the magnitude of percentages (e.g. that 25000 occurrences of a minor crime in a year is not a lot if the country has dozens of millions of inhabitants)
They never tell you enough on the statistics on the news to actually get much solid info. Like sample size, error bars, did they fit a particular distribution, what was tossed as outliers and more.
I always think about the chance of rain report and the different explanations of what it might mean (I don’t recall which is correct, but it illustrates how confusing it can be) :
Chance of rain today based on a predictive model
Is different from
It WILL rain today in this region and this percent of land area will get rain
Is different from
Over all the historical data for this region, we got rain on this percent of days when the conditions were “the same” in the past.
But all can be reasonably said to be 59% chance of rain today.
It does vary from place to place but in North America at least, most often it’s that it will rain in __% of the land area.
Additionally, I would add that knowing about statistics at least allows you to understand that without that additional info, any stat is essentially meaningless, or at least easily misleading.
That is true, I just think it might be hard for many people to understand.
Agreed. But part of the reason why they never report the numbers is that most of the readers don’t understand them anyway.
And a lot of the time it is not even just complex studies people don’t understand. There are also those who can’t even judge the magnitude of numbers, e.g. when it comes to spending or the magnitude of percentages (e.g. that 25000 occurrences of a minor crime in a year is not a lot if the country has dozens of millions of inhabitants)