You don't say which digital scale you have--but keep in mind that, unless it is a very high-test expensive one, running on a wall voltage supply, its accuracy is most likely expressed as plus or minus one tenth grain, at best, on the digital scale. Combine these tolerances on both scales, and the ES will be 0.4 gr.. That's above your current variation.
That's one factor, then. Then, even if the checkweights--what, 20 gr? 50 gr. are precise (and I believe they probably are), there can be variations due to dust and fingerprints on them. Finally, the only way you will get any sense of significant inaccuracy here is to measure ten or 20 loads at a time and then calculate the standard deviation at the 95% level, for both scales.
I suspect you may have forgotten about digital scales having a tolerance factor, too. Unless one heads over into precision lab equipment that costs much, much more than we spend for our reloading gear, I don't think you will get more accurate reading.
As for whether that perhaps 0.5-gr. 'actual' variation--at this level, it represents about 1 /1500th variation in charge weight--and, more appropriately, about 1 / 225 variation (0.3 gr.)--is an issue for your firearm, I can't tell you that. But, IMO, I'd trust the balance beam.
Run 10 charges out on the Lee, and see what the aggregate weight of all ten is. Do no excess powder handling--i.e., dumping from one tray to the next, etc. Weigh it on the digital scale. Repeat 10 times, and do the SD, etc.
Jim H.