I second @Richard's comment that it's a poor definition. Here is some evidence from other textbooks than AP Barron's.
In Steven G. Laitz's The Complete Musician (which uses the "phrase model" mentioned by @Richard), retrogression is defined as follows:
a backward motion [such as] from D to PD [dominant to predominant] is called a retrogression.1
This is used to introduce the concept of back-relating dominants, in which a dominant chord serves to expand the preceding tonic harmony but does not itself resolve to the tonic. The example given is from Bach's Prelude in Eb Major from The Well-Tempered Clavier, Book 2 (BWV 876), excerpted here:
X: 1
T: Prelude in Eb Major, BWV 876
T: from The Well-Tempered Clavier, Book II
C: Bach
M: 9/8
K: Eb
L: 1/8
%%score (V1 V2) | V3
[V:V1 stem=up] B3 z3 z Ge | {e}d3 z3 z FG | ACE
[V:V2 stem=down] G3 z3 z G2 | F3 z3 y3 | y3
[V:V3 clef=bass] z ED EB,G, E, zz | z B,=A, B,F,D, B,, zz | F,zz
w: I | V7 | ii7
w: Tonic | BRD | PD
Aldwell and Schachter, in Harmony and Voice Leading, devotes a brief section to back-relating dominants, but nowhere uses the term "retrogression."2 The Oxford Companion to Music and The Norton/Grove Concise Encyclopedia of Music also do not have entries for "retrogression."
1 Steven G. Laitz, The Complete Musician, 2nd ed. (2008, Oxford University Press), pages 452-53.
2 Edward Aldwell and Carl Schachter, Harmony and Voice Leading, 2nd ed. (1989, Harcourt Brace Jovanovich), pages 146-47.
V I
makes perfect sense. But, as Richard said, the full definition might tell more.