I apologise in advance to folks in the humanities for my skepticism, but when you're quantifying something like "empathy" which is not well defined and canot be measured directly, then a quantitative difference on its own doesn't mean much. That 40% shift may be because the questionnaire itself changed over time, or because it didn't change enough with the times, or any other factor that is has nothing to do with the actual "empathy" of the subjects.
Edit: I was once given a personality test by a psychologist [for reasons I'm not at liberty to disclose, like]. One of the questions was along the lines of "a holy man can heal a patient by laying on hands: true or false?". At the time, I had no idea what that even meant. "Laying on hands"? What, like a D&D Paladin? Years later I learned that some evangelical sects in the US practice this as some kind of prayer or ritual. The personality test I took was most likely drafted originally in the US and there were many more questions that I found very confusing, like that one. So, if I had no idea what many of the questions meant, what exactly was that questionnaire supposed to be measuring? Yet I believe it was some standardised psychological test given to people around the globe.