I've been chasing inaccurate readings for a while now, and have developed a method that works well for me. After sparging, I give it a couple of good, gentle stirs, then use a wine thief to take samples from the bottom and near-middle of the kettle. I put each in a separate wide-bottom container and leave them on the counter to cool while I prepare the boil outside. By the time I've got the boil going, the samples on the counter have cooled. The last few times I've brewed, I've taken both a hydrometer AND a refractometer reading from each sample, compared all the readings and tried to addressed any inconsistencies--I tossed one hydrometer and recalibrated my refractometer to the point of being in sync with the GOOD hydrometer.
For OG gravity I pretty much do the same thing: stir well, take sample, set aside and allow to cool. For a long time I didn't realize that those temperature correction charts for my $5 hydrometers didn't really work for ish. Gotta cool it to within ~20F of the hydrometer calibration temp for a good reading, in my experience.