When the function for testing archives is invoked via the shell context menu (PowerArchiver > Test) then all the files in the archive get extracted to the current folder.
The test dialog reports as many errors as there are files in the archive but it fails to give any hint as to which files are supposed to be erroneous or what the nature of the problem might be. Comparing the extracted files to the originals shows no differences at all.
The .7z in question was produced with maximised compression settings in 7zip (taking forever but resulting in smaller archives than .7z produced by PowerArchiver with maximised settings). Therefore I wanted to see whether PowerArchiver can at least test .7z that it produced itself. Hence I had PowerArchiver convert a .pa with the same contents to .7z. There weren’t any errors reported but the resulting .7z contained fewer than half of the files contained in the .pa (137 of 366), so I scratched that test.
Performance is abysmal when testing via the context menu (e.g. almost 2 minutes for testing a .7z that 7zip tests in 4 seconds), but that is most likely due to the fact that the extracted files are written to disk. Testing the same .7z in the PowerArchiver GUI takes only 8 seconds but causes the mysterious appearance of a UAC dialog, as reported elsewhere.
The testing function is vital because PowerArchiver has a history of producing archives that it cannot unpack without errors or that do not conform to the respective file format standards (e.g. ZIP) so that other programs report them as erroneous.
The point of creating archives is that the files in them will most likely have to be extracted at some point. If the extraction cannot be guaranteed to produce correct results then the whole program is absolutely pointless. Actually, worse than pointless - it causes data loss and hence damage.