The Eggcorn Database already has skew for skewer, and the Eggcorn Forum has the reverse substitution, beginning with Ken Lakritz’s contribution from 2005, with the text:
Ben Zimmer contributed the eggcorn ‘skew’ for ‘skewer’ to the database. This converse eggcorn is also fairly common. To skewer is to attack or mock, so when an opinion or a mathematical distribution is ‘skewered’ rather than skewed, the asymmetry or partiality of the opinion or distribution is being held up as evidence of attack or tampering. Examples:…
From my own files:
As a high school teacher, I’ve seen the same glazed looks on my students during these tests. They understand that they are tools for adults, not them, and so they don’t try. The few that do on the first day of testing are burned out by the third. Any basic statistic class — or even just common sense — would tell you that this skewers the data. (Shumit DasGupta, “Perspective” on KQED’s Morning Edition, 8/27/10 (link))
This creates some problems in comparison since in some countries, housing costs are calculated in the basic unemployment benefit (e.g. France), but the explanatory tables to the OECD data don’t suggest this skewers the data by a significant amount. (Michael Taft, “Open Letter to Sarah Carey”, Irish Left Review 3/23/09 (link))
Plus a few dozen more.
I’m not at all sure that eggcorn is the right category here. To me, it looks a lot like flounder for founder and similar substitutions: some kind of malapropism, probably a classical malapropism, based on similarity in sound and overlap in meaning.
Footnote: there are literal uses of “skewer the data”, as in:
Ms. [Susan] Faludi is at her best in debunking the studies, experts and trend stories that made their way into our collection of “common wisdom” in the last decade. She skewers the data, the data collectors and the data purveyors. (Ellen Goodman, “The ‘Man Shortage’ and Other Big Lies”, NYT, 10/27/91 (link))