It's hard to see how this claim could be substantiated given that we don't know how much existed in the first place.
In a 2013 article Lost writings of Latin literature, Peter Knox (Professor of Classics at the University of Colorado) and J.C. McKeown, (Professor of Classics at the University of Wisconsin) observe that:
No doubt the Roman world produced many literary masterpieces of which
we are completely unaware.
Similarly, the Jstor article The Lost Parts of Latin Literature state that
The total number of writers regarding whom any notice has been
preserved to us is 772, so far as recorded in the pages of Schanz and
Teuffel. How many more actually figured in the course of Roman
literary history we have no means of knowing, or even of guessing with
a fair chance of coming near the truth.
Granted, this article dates from 1905 and finds have been made since, but our knowledge remains patchy.
The original source: J. A. Willis
The author of the source cited by Trompf would appear to be the same J. A. Willis listed as a former Professor of Classics and Ancient History at the University of Western Australia (1962 - 1988).
I have found a few citations of his articles, but none of them are the one cited by Trompf. None of the sources cited in this posted, most of which post-date Willis' 1968 article, mention the estimate of 94%. Also, as with the OP, extensive searching has turned up no further sources which cite Willis' figure.
Although not stated, the 94% probably refers to titles / works. If so, Willis' estimate of 94% lost may be based on known works.
So what estimates are there?
Few sources cite any, but this article Loss of Information from Late Antiquity to the Thirteenth Century states that (for classical literature):
Estimates of the percentage of classical literature that is thought to
have survived to the present vary; one widely used estimate is only
This article cites Scribes and Scholars: A Guide to the Transmission of Greek and Latin Literature by L. G. Reynolds and N. G. Wilson, but I can find no mention in this book for this estimate (I have a copy), so the above '10% survived' should be taken with a very large pinch of salt.
The article Reference for the claim that only 1% of ancient literature survives gives us a different figure and, as with the above 10%, is not specifically for Latin literature.
Other sources I have checked do not give percentages. Michael von Albrecht in A History of Roman Literature (1997) says:
Only a small portion of Roman literature has come down to us, and we
should never forget how much has been lost.
The Oxford Anthology of Roman Literature (by Knox & McKeown, see above) is even less precise:
It is....hardly surprising that the survival of literary works has
been haphazard at best.
The Cambridge History of Classical Literature, vol. 2 Latin Literature makes similarly vague statements in relation to different authors and genres.
Perhaps more useful (and productive) than trying to guess at a percentage is to look at the authors:
for Latin, we have the names of 772 classical authors. Of these, not
a word survives from 276 of them. We have fragments ranching from an
aphorism to several pages of 352 of the authors. Of the remaining
144, we possess at least one of their works but rarely all of them.