Putting differences in test barrels and powder lots aside that can see such variations, there is another factor dependent on how the data compiler lists loads. That is type a) where starting and maximum charge weights are shown, or type b) Hornady, Sierra, Speer where MVs are shown in columns, usually in 100 fps steps, sometimes in 50 fps for smaller cartridges, and the appropriate charge weight is shown underneath.
Both sets of providers load, shoot and measure pressures in a relatively small range of charge weight steps, this fed directly into a laptop or PC these days. The top charge will normally be be below peak allowed pressures. A charge weight v pressure graph is then drawn with the line falling between the points scatter (at one time manually now done by the software) which allows the maximum charge to be extrapolated to the point where the graph line says it produces maximum SAAMI pressure. (Or, of course whatever pressure level the powder / bullet company sets as its maximum which may be below SAAMI to give some safety factor.)
If a Type a) presentation, that value is quoted as is. In the case of Type b) (MV columns), the software produces a set of calculated charge weights that match those MVs used in the table. Maximum charge weight = that for the highest MV available within allowable pressures. If you look at these tables, the steps are usually constant give or take small variations. So in the case of Sierra (Edition V) data for 308 with the 168 SMK/H4895, it is in 1.3gn steps = 100 fps and the maximum quoted is 39.9gn for 2,600 fps MV, the next step would likely be 41.2gn for 2,700 fps. Now it may be that 39.9 exactly produces Sierra's chosen PMax, but the odds are against that. There is an equal chance that the graph says it is 41.1gn producing an estimated 2,692 fps (each tenth of a grain variation = 7.7 fps off the graph). Sierra can't quote that value as it doesn't match the fixed speed and it can't quote the marginally higher one that does as it'll be 'over-pressure'.
In this case, that still doesn't explain fully why the Sierra load is apparently low. I find the 1.3 gn steps surprisingly low after many, many years of loading 308 Win. A full grain change usually changes MV by ~60 fps in this cartridge / bullet weight in a 30-inch barrel, likely a little less in a shorter one, that is ~ 6 fps change per 0.1gn powder. One would therefore expect 1.6-1.7gn steps to be needed to achieve a 100 fps MV change.
There is another factor too that can affect loads manual maximum charges but which doesn't apply here as both Sierra and Hogdon quote charges for the 168gn MK on its own (in Sierra's Edition V anyway). Where there are lots of bullet models in a given weight (or a small weight range), the bullet manufacturer may lump them all together. Most will likely produce very similar pressures, but there may be one or two that produce significantly higher pressures than the others. For safety, that model or those models must be used by the data-set compiler thereby depressing the allowable maximum safe loads for other models that produce less pressure. The latest (10th edition) Hornady manual has tables for 10 different bullets in some weights in 0.308 calibre for instance.