For the list, I relied on a blend of factors, including contextual DIs, taking into account time and place, while relying on publications, videos, and still photography from various sources. For example, a highly visible tornado that occurred during broad daylight on the Great Plains yet still produced a relatively high death toll and/or disproportionately severe injuries is automatically a candidate for “most violent,” provided other context is available and confirmed via multiple, reputable sources. The 1936 Tupelo (F5) tornado can also be considered quite violent, despite occurring after dark and striking poor African-American communities in the segregationist Deep South, because, among other things, it mainly impacted single-family homes outside the business district yet killed more people than the slower-moving Gainesville, GA, (F4) tornado, which struck densely inhabited, multi-story, downtown structures a day later, at the start of the morning workday, in fact.
The Chapman, KS, tornado of 25 May 2016 could conceivably belong to this list, along with the Rocksprings, TX, tornado of 12 April 1927, the massive F5 tornado family in KS on 7 May 1927, and the Tianjin tornado of 29 August 1969. The Chapman tornado, as mentioned, produced very intense ground scouring and damage to sizeable vehicles, while the Rocksprings and KS tornadoes in 1927 apparently reduced numerous, large, mature trees to debarked stubs, while leaving behind very little debris on empty homesites. The Tianjin tornado also debarked and denuded entire trees and snapped off steel-and-concrete rebars in a densely populated industrial zone. As far as I know, only the Brandenburg, Parkersburg, and Joplin tornadoes come remotely close to this level of damage, given that these tornadoes either toppled or snapped poured concrete basement walls. The Bakersfield Valley, Hackleburg, and Smithfield tornadoes did manage to break apart concrete surfaces and/or foundations in general, as far as I know.
I also put the Atkins–Clinton–Zion tornado on the list because the damage to the boat factory is exceptional, matched only by similar industrial damage in the Jackson–Forkville, San Justo, Guin, Niles–Wheatland, Edmonton, and Roanoke tornadoes. (Tianjin itself is on another level.) Additionally, the Atkins–Clinton–Zion tornado also snapped and partly debarked numerous, mature hardwoods only a few feet above ground level, while reducing large structures north of Clinton to bare slabs and trailers to chassis, with practically no debris left on foundations in some areas. I recall an old thread on American Weather that mentioned the slabbed foundations being visible in an aerial that unfortunately has long since vanished from the Web. If any tornado in February deserves to be rated EF5, it’s probably the Clinton tornado from Super Tuesday (2008), given that both Significant Tornadoes and NWS JAN only found F4 damage in LA and MS from the “official” F5 of 21 February 1971. That F5 should be downgraded and replaced by Clinton, in my view.
The Udall, Hudsonville, Lake Pleasant/Coldwater Lake #1, and Lebanon–Sheridan tornadoes each produced extensive ground scouring and finely granulated debris, whereas the other “potential” F5s from Palm Sunday (1965) did not leave behind those indicators. The first Coldwater Lake tornado on Palm Sunday apparently produced some very intense damage to vehicles as well. This is the same long-tracked tornado that became a massive wedge over southern MI and produced a wind gust of 131 knots (151 mph) at Tecumseh. The second long-tracker that followed it was apparently not as intense as the first and caused somewhat fewer casualties, according to Significant Tornadoes. The Prague–Iron Post–Sapulpa tornado, according to my recollection, apparently removed half a foot of topsoil from hillsides, “stripped” them of “all vegetation,” and tossed oil tanks an unspecified distance over rural areas of the Cross Timbers in northeastern OK, per an old post that I recall by the SPC’s Richard Thompson on American Weather, long since taken down.