if potential visitors have an aversion to the newer TLDs.
It is a very subjective question.
I would like to offer a more nuanced approach than the answers already given.
Chicken-and-egg
This is a classical chicken and egg problem: people are afraid of new TLDs (please do not say "extension") and/or do not use them because no one (or not enough critical mass) uses them... and no one uses them (or are afraid to use them) because people are afraid to use them.
First movers are at a disadvantage... except of course they get the opportunity to register domain names that may be long unavailable in legacy TLDs.
Do note however that the ICANN new TLD program started to be alive in 2012,
with new TLDs added at end of 2013. This is now old in term of Internet time. So the 2016 study may or may not be still relevant today.
TLD relevance
How often does it happen that you need to spell out your domain name? I guess not so much. It is true that there is the empirical radio test (see https://www.domainsherpa.com/domain-name-dictionary/radio-test/ and https://www.namecheap.com/blog/domain-name-radio-test/) and many cases of wrong choices of domain names that can be read differently (see some lists at https://www.boredpanda.com/worst-domain-names/ or https://www.shoutmeloud.com/worse-funny-domain-names-websites.html or https://funnyshit.com.au/bad-domain-names.html)
This clearly shows I think that you already can go very wrong even in the perimeter of a single TLD and even a legacy well-known one.
But, do you really yell out your domain name for people to come on your website? I guess not, they probably come because of some hyperlink somewhere, or some ad that you bought. So in all these cases people just click on a link, I doubt most of them even look at the URL, and in which case I say that the TLD does not matter so much.
TLD acceptance
There is also a little negative side effects of new gTLDs but in fact it happened even before the new round of 2012: there is a lot of broken software out there that is built on false assumptions such as "all gTLDs are 3 characters" or "no digits in a TLD", etc.
Often there is even a list of TLD, and not being carefully maintained it won't of course know about new gTLDs and hence you get rejections in some online forms when asking for your email or such.
Note that this bad fate is also shared by IDNs, that is domain names using other characters than from the ASCII repertoire and people then speaking about "special" characters and all difficulties they bring.
For more information on these subjects, there is a working group inside ICANN devoted to it: https://uasg.tech/
New registries
The other question one may want to ask before choosing the domain name is: can I trust the registry?
Because, in new gTLDs, some registries did already disappear but mostly brands. Have a look at this page for a list of terminations that happened: https://www.icann.org/resources/pages/terminated-registry-agreements-2016-03-02-en
One may argue that .COM being as big as it is will never fail. Then of course the history has plenty examples where even the biggest companies fail in some way, without being backed up by the government or equivalent and hence disappear.
So, like any non-trivial choice, this one is never clear cut in neither side, but it is a point to take into account (whose solution could be for a big business to have multiple domains registered in multiple TLDs, and while using normally just one, having everything in place to switch to another one if needed; it may not be because of registry termination, it could as well be because of technical problems, etc.).
For new gTLDs, note that ICANN has a whole program in place where EBEROs (Emergency Back-End Registry Operators) are on standby to save any failing registry and transition DB content so that the domains registered under the TLD continue to work, hence not harming current registrants and services built on top of the TLD.