Plenty of possibilities exist to explain that, but there are four you need to be very careful to consider.
- Sample variation can definitely be a problem, even for high-end Nikkor lenses. It doesn't take much to make a lens underperform in tests. Miscentered elements, slightly offset elements, an internal cam that's not quite positioned right, the list of things that could be wrong are nearly endless. Fortunately, sample variation problems are almost always mechanical issues, which means that the maker can usually fix them by disassembly and careful reassembly.
- Most lens testing is done at relatively close distances. Lens designs can be better when focused close rather than far away, or worse. It's a rare lens that is at its best from its closest focus distance all the way to infinity. This problem is especially true for wide angle lenses, where the size chart you'd need to fully test a lens at a "normal" distance would be huge. You can do various things to deal with this problem as a tester, but not every tester does the same thing.
- Many lenses aren't designed to focus on a flat field (e.g. test chart). Many lenses have a curved field of focus that varies in some complex ways. Thus, if you test one of those lenses on a flat chart with only central focus used, you get very bad results for the edges, where the focus has curved away from the plane of the chart.
- Many testers, including myself, are using digital cameras with automated interpretation software (Imatest). Unfortunately, over time, the cameras change, so you can't always compare today's results directly with yesterday's. So you have to be careful to understand whether a lens test report was done with a D2x or a D5, a D500 or a D810. Each camera would report different numbers due to their different pixel densities and antialiasing filters.
Personally, I run a wide variety of tests, often at different focus distances, on both flat and non-flat targets. I don't report numbers on lens tests for two reasons: first, to fully describe a lens there would be a massive amount of data to report, and second, you'd have to know how to interpret that data. I prefer to just present concise written summaries of what I found in my testing. I think that's just a better way for most people to understand the strengths and weaknesses of a lens.
