Youve missed a couple of things.
First off, the diameter of the whiskers on the DB8 which increase end lengthening effects, and also the smaller width elements of the Winegard...which has the effect of decreasing the bandwidth of the design (meaning that swr and impedence rise at some points higher and lower to the resonant frequency closer to that resonance frequency.
EV: Sorry its taken so long to re-visit this.
I just went out and measured the diameters of the CM4228HD and DB8 "whiskers". I don't have a caliper, but with aid of a magnifying glass and my trusty old Engineering Ruler's 50 Units per (1") division scale, I observe 8/50'ths, or about 0.16" inch for both. I note that #6 AWG wire is 0.16202" diameter and thus both antennas appear to me to be using #6 AWG aluminum wire. So I don't understand what DB8 diameter aspect you are thinking of.
Also, I didn't quite understand the second part of this relative to the "width" of the Winegard elements. Unlike the CM4228HD and DB8, the Winegard 8800HD does not use wire. Instead it uses relatively flat rectangular elements, maybe about 20 gauge thickness aluminum, with (what I would call) a width
of about 1", and are as I indicate in my table, a length
of about 7 1/4 inches "long".
As such, the overall dipole length of the Winegard dipoles falls in between the CM4228HD's (longest) and DB8's (shortest). The primary effect of this, I would think, would be that the peak response of the Winegard would fall in between that of the CM4228HD and DB8, (all other factors being equal, which of course, they aren't).
However, the greater factor associated with the Winegard, at least to me, is the obvious difference that it is not a "fan dipole" ("bowtie") design. Rather its a two dimensional collinear array of common dipoles. From what I've read, the most significant aspect of that should be less bandwidth than that of the triangular fan/bowtie.
So, is the "width" you are discussing the "width" as I've defined it above, or something else?
If it is, then certainly I can accept that the characteristic impedance of the flat rectangular elements would be something quite different than that of triangular elements formed from wire. But, certainly it is the job of the balun/combiner design to be optimally matching characteristic antenna impedance to that of the transmission line, and thus optimally maximizing power transfer, right?
So after I pondered that I realized I still don't understand the point being made...