What is an explanation backed by results from repeated tests or experiments?

Repeated tests, using the same sample, same protocol, etc., guarantees you to minimize your errors.

Think that you have a soil sample and you want to detect its (lets say) organic carbon using modified Walkley Black test. 1 gram of fine (size) sample you used for the test and you report this to your lab manager. How can you be sure that it is definitely correct and precise?

Now think that you prepared three samples (each one is 1 gram) and you measure their organic carbon independently. You have three results and they are very close to each other (let's say within plus or minus 3% difference). Now you are confident that your measurement is dependable and precise.

A theory explains why something happens. A theory to be generally accepted needs to be supported by repeated tests and experiments.

A theory that makes predictions that do not come true is a bad theory and is generally rejected.

A scientists made a theory that the granite boulders and giant water falls and erosion was caused by an huge flood starting in Missoula Montana and spreading across Idaho, Washington and Oregon.

This theory went against currently accepted theories of these geological features being caused by slow uniform process. The theory was rejected for lack of experimental evidence supporting it.

Then when electro microscopes were invented, the crystals of the granite boulders in Oregon and Washington could be examined and compared to the granite in Missoula Montana that the theory said had been sheared off by the ice dam holding back the water that caused the flood. The granite samples matched perfectly. These tests and experiments confirmed the flood theory.

Today the Missoula flood theory is accepted because of the tests and experiments that support the explanation that seemed unreasonable when it is was first proposed.