"The results could indicate a more moderate degree of protection of 15-20%" to the wearer.
"The findings are consistent with previous research. Health experts have long said a mask provides only limited protection for the person wearing it, but can dramatically reduce the risk to others if the wearer is infected, even when showing no symptoms. Preventing the spread to others is known as source control."
Now consider the following: if the "current" R number (the average of how many people are further infected by one infected person) in some environment is 1.2 that still results in the number of infected per day exponentially(!) increasing. Example: after the sequence of 10 infections with R 1.2 there are 6 infected, after the sequence of 20 -- 38, after the sequence of 30 -- 237 and after the sequence of 100 -- 82 million people infected, starting from just one. It is estimated that it takes just a few days for one sequence. Taking that number to be around 3.6 (it doesn't have to be that number, it's just to take some convenient number as an illustration) means that 100 sequences can be imagined as one year. So 82 million in one year.
Now if every wearer is protected only 20%, it can appear to the wearer who doesn't understand math that he should not wear it. However, reducing current R number from 1.2 to 1 means that the number of infected per day doesn't increase at all but remains constant. Example: starting with 1 infected, after the sequence of 100 there are just 100 infected. Using the same convenient number, after one year, still just 100 infected.
That, in turn, means that if everybody wears it, everybody wins: allowing exponential increase of infected will result in everybody becoming infected, whereas the constant number of daily infected could guarantee that most don't become infected before they can be vaccinated.
Not to mention that if the number of infected per day can be kept constant some other kinds of measures (lockdowns with closures, bad for economy) could be completely avoided, especially if it is achieved at some low level. Which means that it can be of immense benefit to the wearer to wear the mask as much as possible even if he increased his own direct protection in some specific situation for mere 20%! Moreover, the benefits are dependent on most others also wearing the mask! If he's doing that alone, than it's in vain -- he alone can't change R for the whole population: his highest interest is to convince others to do the same.
Being willing to understand math and to read carefully helps.