Population or sample standard deviation, with the actual spread drawn on a number line - not just a number that's easy to misread.
Use population standard deviation when your data IS the entire group you care about - every student's exact score on one specific test. Use sample standard deviation when your data is a smaller sample standing in for a bigger population you're trying to estimate - polling 200 people to estimate a whole country's opinion. The sample version divides by n-1 instead of n, which slightly increases the result to correct for the fact that a sample tends to underestimate the true spread of the full population.
Worked example: the data set 2, 4, 4, 4, 5, 5, 7, 9 (the default above) has a mean of exactly 5. Each value's distance from 5, squared and summed, is 32. Population standard deviation = √(32÷8) = 2 exactly. Sample standard deviation = √(32÷7) ≈ 2.14 - a real, meaningful difference for the same data, depending purely on which question you're actually asking.