Often we have to make decisions without enough hard evidence to determine the potential outcomes of the decision with full precision. While of course uncertainty must be taken into account, we deal with these situations enough that sometimes guesstimating is the best we can do. In this context, there is a simple procedure you can follow to improve the accuracy of your hunches.
Mortar fire is the best analogy for the process. When firing a mortar you simply can't sight in, fire once, and expect to be accurate. You can measure the wind and other situational variables, but there's so much involved that you have to go by guess and check. Once you fire a mortar short of the target, you can make adjustments and fire one long. This creates a bracket from which you can obtain greater precision for your next shot.
In the world of day to day decisions, we can't see where the mortar lands, but we can try to create similar brackets to narrow our focus and improve our accuracy. Let's say we are trying to estimate the percentage of the population who harbor racial hatred, a topic of substantial discussion recently that is essentially impossible to measure. In this case we have to use hunches on the brackets based on our own instinct.
I'd say 10% definitely feels too low, and 50% feels too high. The easiest thing to do is take the midpoint and say 30% as the guesstimate within a slimmer margin of error. That margin is still big, let's call it 20% - 40%, but at least there's a bit more precision in this process than just taking a stab at the number without trying to bracket it (or not trying to figure out the number at all).
Race is a touchy subject, particularly at this juncture, but the primary discussion points rarely bother to estimate the proportion of the population that falls under the 'racist' epithet. People either shout that racism is still alive or that things have gotten a lot better, but those are both obvious truisms. Very little progress can be made between those who say the number is greater than 0% and those who say the number has ticked down in the past 50 years. Of course everyone's positions on this are more nuanced, but those are the numerical interpretations of the statements. If we could make an effort to be more precise perhaps we would stand a better chance of making some progress, and the best way to proceed would be to try to come to a consensus on the brackets and narrow down to a better guess.
If we want to figure out how to improve health outcomes, we have to have a sense of how many people are sick. I believe to figure out how to improve race relations we need to have a better sense of how many people harbor indiscriminate racial hatred.