Norm Eisen reviewed thousands of New York cases charging the falsification of business records cases. While he concluded that, in his opinion, jail time was "likely" for Trump, his data showed that in nearly 90% of cases charging someone with falsifying business records, there was no jail time imposed. A subsequent review that cites Eisen's survey tried to massage the data further still found that over 80% of such cases in which there was a conviction or guilty plea, no jail time was imposed. https://www.justsecurity.org/97186/trump-sentencing-cases-survey/
It is important to note that the authors of that survey sought to show via 26 examples (out of a universe of over 300) that "comparable" cases to Trump's resulted in jail time. That may be what you are referring to. Apart from the limited nature of the examples -- which appear to have been chosen to support a specific outcome, a close review of the examples shows that they are not necessarily comparable because, for example, the defendants were, in fact, charged with other felonies, including felonies with more severe penalties that that applicable to a falsification of records charge, even if they weren't convicted of those charges under a plea deal or were convicted of those charges in a separate case. Moreover, as any experienced criminal trial lawyer knows, sentencing is, at its core, a fairly subjective process and no attempts to compare one case to another can possibly take into account the subjective elements that may have impacted whether a defendant was sentenced to jail time.