Yes, it is accurate in a broad/general sense if done correctly, and in fact people do this all the time subconsciously with a great degree of accuracy. The problem is when it is discussed in a manner in which people are not at all willing to be open to the fact that some correlations do in fact indicate causations (a great majority of the Ti using population, especially when backed by Fe SJW emotionality aka adherence to the church of peer review).
Even if they do agree that there is in fact some fundamental psychological or personality trait correlated with a particular physical feature, since they are so strongly opposed to the possibility of phrenology existing (due to the fact that if phrenology does exist, it would decimate the framework their entire worldview is precariously sitting on, backed by group consensus), it would be impossible for them to connect the dots as to what the correlation actually indicates.
Another factor is, in my opinion, only a minority of the population actually catalogues facial recognition data to the degree that phrenologists do which enables them to draw any sort of constructive possibilities with the data (a lot of the time, they do, subconsciously, but are unable to demonstrate exactly what it is their brain is informing them of in the personality of the subject they are observing consciously, based only on facial features and expression alone). I would also argue that those who excel at phrenology are frequently phrenological outsiders themselves, for a myriad of reasons which would be a little too tiresome to expand on this post.
I think a better question would be...would you be comfortable if phrenology was real, and there were phrenologists that existed right now that in fact could tell if you liked apples simply from the shape of your jaw?