How is SVV (β) defined? It seems to me that the "angle of deviation" (Δ) would be what is considered the SVV. The study which provides the normative data used for the confidence interval defines SVV as "the set angle, as measured with reference to the true vertical", but this almost seems like the definition of "angle of deviation".
Answer: I believe the SVV (β) is defined simply as the angle at which the viewer sets the luminous line. For example, if the viewer set the line at 0° then it would be parallel with gravity. If they set the line -15° then it would be tilted to their left by 15 degrees and so on.
So the question then becomes how is 0°, -15° etc… defined? This is defined by a sensor in the instrument. Without going into the technicalities, there is a self-aligning transducer in the instrument which defines vertical using gravity. (the same principle as a plumb line.)
So referring to the article by Schoenfeld and Clarke (2011), which is the one I believe you’re quoting from in your question, I guess what they mean by the phrase “with reference to the true vertical” is at what angle did the viewer set the line relative to the true vertical as measured by the self-aligning transducer. They aren’t referring to the "angle of deviation" (Δ) with this phrase, so far as I can tell.
What appears somewhat counterintuitive at first, and is perhaps the source of your puzzlement, is that the instrument displays β as a value relative to the head tilt angle (α). However, the head tilt angle must be defined by reference to the self-aligning transducer (in addition to the luminous line), so these two values are related by the same reference.
References and caveats
Schönfeld U., Clarke A. H. – 2011: A Clinical Study of the Subjective Visual Vertical during Unilateral Centrifugation and Static Tilt; Acta Otolaryngol 131(10):1040-50