¯\_(ツ)_/¯ Uncertain Times Need Clearer Words
Maybe? Probably? Unlikely? Give me a number.
Predictions are everywhere. Weather, meeting deadlines, elections, etc., and they are all described via common and casual words or descriptions of chance. For example:
It is probably going to rain this weekend.
He is unlikely to find a new job.
We are definitely going to complete the project on time.
If we think of a prediction as a value on a scale from 0% (the event absolutely never happen) to 100% (absolutely will happen), what do most people mean by “probably”? 75% chance? 51% chance? 90% chance?
To find out, I very unscientifically gathered data of a handful of these types of common words. This plot orders the words by the median response level of certainty.
Some observations:
Definitely and absolutely are the words to choose if you want to convey >99% certainty. These words also had low spread in responses (shorter ←→), which implies that most people are in agreement that these words imply high certainty.
Surely is a weird one. It implies relatively high certainty (90% median response), but it has a huge spread of responses (long ←—→). This implies that there is a lot of potential for variance in the interpretation of “surely”. Unless you’re Irving from Severance, you probably aren’t using the word “surely” too much anyways.
Maybe is the most decisively uncertain response with the vast majority of responders at 50% chance. If you have no idea what will happen, stick to “maybe”.
Uncertain is also a 50% chance word, but it skewed towards the negative side. In other words, while on average most people interpret “uncertain” as 50/50, plenty of people interpret it as between 30% and 50% chance. Depending on the context, using “uncertain” seems to imply some pessimism towards the chances of an event.
Unlikely and doubtful had similar survey responses but important differences. Both had identical medians (20% chance) and ranges (most responses fell from 10% to 30% chance). But, the most frequent response (mode) for “unlikely” was 30% while the most frequent response for “doubtful” was 10%.
I like the direction this is going in. Most normal people (i.e., those of us who are not economists) don't really understand or intuit confidence limits. So we need to figure out how to use language. But I've found it a bit hard to be precise because it depends on the audience. There's not substitute for not understanding who you are trying to influence.