I have three electronic scales that together cover the range of masses I need to measure. I also have NIST-traceable calibration weights for each. The 'linearity' calibration procedure for these scales requires weights at or near either end of the range (e.g. 0 grams, no weight, and 200 grams for a 200 g scale) plus a weight in the middle (100 g for this example).
Without your own calibration weights that weren't supplied by the manufacturer of the scale you have no idea the actual accuracy. This is no different than for a, say, micrometer. All you know with a micrometer is 0 is 0, because you adjusted the spindle to make it so. If you also have a calibrated 1" gauge block you now know it is correct at 0 and 1" (assuming it is), but you don't know about the accuracy in between. That's why micrometer calibration checking sets contain several blocks, to test the micrometer mid-range and at several orientations of the spindle. If a micrometer passes all those tests it's still possible it could be off at some other point in the travel, but the odds are small.
With an unknown electronic scale, irrespective of how many digits the readout has, you won't have any idea of its accuracy without at least two weights calibrated by someone other than the scale's manufacturer.