The thing to keep in mind when looking at measurements by voice coil is the they are taken in a speaker enclosure, not IB like a company would show. In the case of this driver, it was a 8x17" baffle. This means baffle step is a factor. If you don't know about baffle step, it is essentially a 6dB loss at low frequencies because after the wavelengths get longer than the enclosure is wide, the energy starts to eminate backwards as well as forward. So, when comparing the voice coil measurement to the manufacture's you have to assume an additional 6dB that gradually decreases as frequency rises. In the case of this driver, it fill in quite nicely to the desired response. Additionally there is some diffraction that will mess up the response as well. Namely at 800hz, that small bump, and some other stuff higher in frequency that is being masked by smoothing (1/6th, I hate that they do this) and breakup.
So overall, I'd say that the manufactures measurement is quite close to what vance measured and I would actually trust it more in this case due to the lack of smoothing used. Regardless, the response in your car will vary greatly due to the environment it is in, but as a sanity check, the supplied measurements are good.
Most car audio products I've seen inflate sensitivity greatly, either measuring it via 2.83V (not lying, just misleading IMO) or just choosing a random spot in the response (lying IMO). It should truly be done based on 1watt and based on the t/s parameters. If you want the best idea of the actual sensitivity of a driver (compared to other drivers, not real world), using a program like winisd or hornresp is your best bet. Low frequencies don't vary from driver to driver as high frequencies do. If a manufacturer posted spec varies from what a program says below about 400hz, then the driver has some sort of resonance, or the company is lying about something.