Another point is that the article doesn’t specify what type of audit it’s using for comparison. Audits range from a simple correspondence audit, where the IRS simply sends a taxpayer a letter requesting substantiation on an item that was reflected on the taxpayer’s return, all the way up to a full-blown request for copies of all underlying data and documentation that justifies all material positions taken on the return.
For example, if one is claiming a foreign tax credit on income from services (e.g., a person who works both in the US and outside the US as an independent contractor), one may be subjected to a day-count audit, where one has to prove that one actually spent the number of days working outside the US that we’re used to compute the amount of the foreign tax credit claimed (assuming that one did not qualify for, or did not elect to take, the foreign earned income exclusion). A day-count audit is much more involved and “painful” than a simple correspondence audit asking one to provide copies of certain documents. Particularly if the focus is on proving workdays, not just days of physical presence outside the US.
So, given that there is a wide degree of different types of audits, it would not surprise me in the least if the claimed disparity was because most of the lower-income audits were simple correspondence audits, while the high-income audits were full blown actual audits requiring a review of underlying books and records, and substantiation of amounts claimed in various financial statements.