We present new observations of molecular anions with the Yebes 40 m and IRAM 30 m telescopes toward the cold, dense clouds TMC-1 CP, Lupus-1A, L1527, L483, L1495B, and L1544. We report the first detections of C3N− and C5N− in Lupus-1A as well as C4H− and C6H− in L483. In addition, we detected new lines of C6H− toward the six targeted sources, of C4H− toward TMC-1 CP, Lupus-1A, and L1527, and of C8H− and C3N− in TMC-1 CP. Excitation calculations using recently computed collision rate coefficients indicate that the lines of anions accessible to radiotelescopes run from subthermally excited to thermalized as the size of the anion increases, with the degree of departure from thermalization depending on the H2 volume density and the line frequency. We noticed that the collision rate coefficients available for the radical C6H are not sufficient to explain various observational facts, thereby calling for the collision data for this species to be revisited. The observations presented here, together with observational data from the literature, have been used to model the excitation of interstellar anions and to constrain their abundances. In general, the anion-to-neutral ratios derived here agree with the literature values, when available, within 50% (by a factor of two at most), except for the C4H−/C4H ratio, which shows higher differences due to a revision of the dipole moment of C4H. From the set of anion-to-neutral abundance ratios derived two conclusions can be drawn. First, the C6H−/C6H ratio shows a tentative trend whereby it increases with increasing H2 density, as we would expect on the basis of theoretical grounds. Second, the assertion that the higher the molecular size, the higher the anion-to-neutral ratio is incontestable; furthermore, this supports a formation mechanism based on radiative electron attachment. Nonetheless, the calculated rate coefficients for electron attachment to the medium size species C4H and C3N are probably too high and too low, respectively, by more than one order of magnitude.