We investigate the relationship between the \hbox { Mg { \sc ii } } \lambda 2798 emission-line and the 3000 Å continuum variations using a sample of 68 intermediate-redshift ( z \sim 0.65 - 1.50 ) broad-line quasars spanning a bolometric luminosity range of 44.49 erg s ^ { -1 } \leqlog L _ { bol } \leq 46.31 erg s ^ { -1 } ( Eddington ratio from \sim 0.026 to 0.862 ) .
This sample is constructed from SDSS-DR7Q and BOSS-DR12Q , each with at least 2 spectroscopic epochs in SDSS-I/II/III surveys .
Additionally , we adopt the following signal-to-noise ratio ( S/N ) selection criteria : a ) for Mg ii and the 3000 Å continuum , S/N \geq 10 ; b ) for narrow lines , S/N \geq 5 .
All our quasar spectra are recalibrated based on the assumption of constant narrow emission-line fluxes .
In an analysis of spectrum-to-spectrum variations , we find a fairly close correlation ( Spearman \rho = 0.593 ) between the variations in broad Mg ii and in the continuum .
This is consistent with the idea that Mg ii is varying in response to the continuum emission variations .
Adopting the modified weighted least squares regression method , we statistically constrain the slopes ( i.e. , the responsivity \alpha of the broad Mg ii ) between the variations in both components for the sources in different luminosity bins after eliminating intrinsic biases introduced by the rescaling process itself .
It is shown that the responsivity is quite small ( average \bar { \alpha } \approx 0.464 ) and anti-correlates with the quasar luminosity .
Our results indicate that high signal-to-noise flux measurements are required to robustly detect the intrinsic variability and the time lag of Mg ii line .