Of course, especially once study result make it into mass media all statistical meaning of the words used is completely lost and reinterpreted from the point of view of normal-life usage.
When you get a statistical "explanation", for example in a linear model, factor X explains 50% of the observed variation, you never know if that is actually the factor - or if it's random, or if it's confounded by an unknown.
"explains" in statistics is meant to be used within the context of your model equations and numbers.
Even if you get causation by experimenting and controlling the variables and don't just have correlation, who would be satisfied with a statistical "explanation"? You still won't know if there isn't a common factor that you just don't know about yet.
Same with statistical significance: It is useless unless someone who actually knows the field checks how/if the model you have relates to the real world. It's not enough to show that everything works fine within your model - you have to know if you have the right model, which requires knowledge outside of statistics. It's just a tool (statistics), by itself it means as much as naked numbers.
Statistics can tell you that there is a connection. It can't tell you how (it works).