The other way around. By far.
A balance is the most accurate and precise means of measuring the mass of normal objects. If you are talking about measuring the mass of an electron or a planet, then that's a different story. Even the international standard for electrical current uses a balance.
Don't be fooled by the "accuracy" claims of the scale vendors - those claims are written up by marketing types that don't have the slightest clue what accuracy means and, frankly, they don't care.
I can easily design a digital scale that has a RESOLUTION of 0.001 lb. All this means is that the display reads out in those increments. If I were to design such a scale, it would invariably be advertised as having an accuracy of 0.001 lb and have a correspondingly high price tag. Never mind that if you placed a 100 pound object on the scale it might read 110.352 lb and if you took that object off the scale and then set it back on it might now read 114.791 lb.
Marketing people do not care about truth in advertising. Don't believe me? Check out the air compressors at your local hardware store. They will advertise, in big numbers right on the side of the unit, that it is a 4 HP or 5 HP or even 6 HP compressor that runs off of a 120V 15A outlet. The only problem is that the motors on those compressors are almost invariably 1.5 HP motors since the maximum amount of power that can be delivered by such an outlet is less than 2.5 HP before tripping the breaker and even a 2 HP motor will generally trip the breaker as it starts under load.
By I digress - back to scales.
A spring scale - and a digital scale is nothing but a spring scale with a digital readout - suffer from a host of influences that affect their accuracy (how close the the displayed reading is to the correct reading) and precision (how repeatable a measurement is without respect to accuracy).
A spring scale depends on the deformation of some type of spring. The readout - be it mechanical or digital - then somehow responds to this deformation. The deformations are very minute, a tiny fraction of an inch in a mechanical spring scale and not much more than the width of a few dozen hairs in the case of many digital scales. Sensing such small deformations accurately and precisely and CHEAPLY is not easy. Furthermore, there are lots of other factors that can also cause deformations such as how the weight is distributed on the platform and the temperature.
Furthermore, spring scales are not intrisically linear. By this I mean that if you place a ten pound weight on the scale and note the reading, then add a one pound weight and note then increase, this is probably not going to be the same increase that you would observe by starting with a one hundred pound weight and then adding that same one pound additional mass. This is particularly true of digital scales because there are inherent nonlinearities in the sensing circuit and, while those nonlinearities can be greatly reduced with additional strain gages and circuitry, this is never done in a consumer scale because it is cost prohibitive.
For this reason, a mechanical dial spring scale is almost always more accurate and more precise than a consumer quality digital scale. Now, if we are talking about the much higher end scales that have a fully populated and compensated Wheatstone Bridge as the sensing unit, that's a different matter. Those can be quite repeatable though they must still be calibrated fairly frequently.
Spring scales also inhibit significant hysteresis. This means that if you apply the load going up (stepping on the scale slowly and smoothly) that you will get a different reading than if you apply the load going down (step on the scale holding a gallon jug of water and then smoothly set the jug on the counter).
So when a scale claims that is it "accurate up to 400 pounds" all that means is that it is capable of displaying a result up to 400 pounds. Likewise, if it says that it is "accurate to 0.1 pounds" all that means is that it can display results in 0.1 pound increments.
Beam balances, on the other hand, are virtually immune to all of these influences by the very physics of the way they operate. Furthermore, they basically never need calibrating (beyond re-zeroing) because the calibration is due solely to the location of the pivot points and not upon the spring constant of some piece of material that changes with temperature and over time.
To give you an example, I have a Health-O-Meter Model 200S Medical Beam Balance that is marked up to 350 lb in 0.25 lb increments. And this is what the marketers tout as the "accuracy" numbers for the scale. In point of fact, the accuracy and precision on it are closer to (and probably better than) 0.125 lb.
A point that really illustrates this - as well as demonstrating the superb linearity of the scale - is that I needed it to measure up to 380# so I added a homemade counterweight. I sized the counterweight by placing three 10 pound weights on the scale and, with the scale set to zero, adjusted the number of washers on my counterweight until it balanced After that, I just weighed myself and added 30 pounds to whatever the result was.
When I finally got down to under 345 pounds, the scale was reading 314.75 (for a total of 344.75) and I took the counterweight off and took a new reading. It read right on 344.75 pounds. So a 30 pound difference at 0 and a 30 pound difference at ~350 lb were indistinquishable. Not bad at all. In order to achieve that kind of linearity, the absolute accuracy also has to be better than 0.25 pound.
BTW: You can buy a counterweight for that scale that extends the range up to 450#. The counterweight is currently about $50 but you can extend the range (probably up to about 600# - the limit is when you start physically damaging the scale platform and that is almost surely above the 700# level) with a nail (or some string) and a bunch of five cent washers.
I know that was long, but hopefully it was worth it.
If you are interested in such a scale, you definitely have to shop around for the best price.
For instance,
www.health-o-meter.balances.com has the scale that I use for $239.95 and claim that the MSRP is $320.
www.bodytrends.com goes one better and sells it for #336.49.
Yet
www.drugstore.com has the EXACT same scale for $189.99. A shorter version of this scale is $119.99 on this same site.
I bought mine at Best about six years ago for $89.00 and I have a hard time believing that inflation has more than doubled its price, so you may well find it at Sam's Club or Wal-Mart for under $150.