I fucking hate giving estimates.

Just ask: "Does that estimate include testing?" Then double it regardless of the answer

Does it include the time to make the estimates? If so, it might diverge and explode.

After doing estimates for the last 7 years, I can honestly say I've never once felt like they have done anything for me other than providing me with a sense of dread. They are misused, misinterpreted and a huge source of anxiety for my team.

It's terrible indeed! My current strategy is to estimate a) The best-case estimate, b) The most-likely estimate, c) The worst-likely estimate, sum those three and get the average. All including testing time.

I’ve found overestimation to be more beneficial to general well-being then honest estimation. You never know what you’ll run into, and I’d prefer not to stress to hard over it

Remember to give them the marigin of error and accuracy of your prediction :p

-Can you give us the estimated time needed to complete this project?
-Yes it's going to take 6 months +/- months, and this will take 1 year with proper tests, coordination, seciurity, maintanable code with +/- 10 months. Accuracy of my prediction is about 20%

How many estimated do you estimate per week?

Start giving them intervals instead.

"This will take between two weeks and nine months."

This'll tell them to get their shit together and be more specific with their descriptions. Note: only works for salaried in-house devs.

Write a script to estimate the work for you based on your previous performance.

For that reason I advocate to not estimated in hours and instead opt for estimating in days. Who the hell needs estimation in hours or minutes? Fine grained estimation is just bogus.
Over estimated? No problem just pull another story and do as much as possible and do the remaining in the next sprint. Underestimated? Nah unlikely if you consider the correct facts when estimating but that would go far here and now...

I always try to give my best estimate and multiply it with Pi.