We measure the fraction of galaxies undergoing disk-disk major mergers (f_m) at intermediate redshifts (0.35 <= z < 0.85) by studying the asymmetry index A of galaxy images. Results are provided for B- and Ks-band absolute magnitude selected samples from the Groth strip in the GOYA photometric survey. Three sources of systematic error are carefully addressed and quantified. The effects of the large errors in the photometric redshifts and asymmetry indices are corrected with maximum likelihood techniques. Biases linked to the redshift degradation of the morphological information in the images are treated by measuring asymmetries on images artificially redshifted to a reference redshift of z_d = 0.75. Morphological K-corrections are further constrained by staying within redshifts where the images sample redward of 4000 A. We find: (i) our data allow for a robust merger fraction to be provided for a single redshift bin, which we center at z=0.6. (ii) Merger fractions at that z have lower values than previous determinations: f_m = 0.045+0.014-0.011 for M_B <= -20 galaxies, and f_m = 0.031+0.013-0.009 for M_Ks <= -23.5 galaxies. And, (iii) failure to address the effects of the large observational errors leads to overestimating f_m by factors of 10%-60%.
Combining our results with those on other B-band selected samples, and parameterizing the merger fraction evolution as f_m(z) = f_m(0)(1+z)^m, we obtain that m = 2.9 +- 0.8, and f_m(0) = 0.012 +- 0.004. For an assumed merger time-scale between 0.35-0.6 Gyr, these values imply that only 20%-35% of present day M_B <= -20 galaxies have undergone a disk-disk major merger since z ~ 1.
Assuming a Ks-band mass-to-light ratio not varying with luminosity, we infer that the merger rate of galaxies with stellar mass M >= 3.5x10^10 M_Sun is R_m = (1.6+0.9-0.6)x10^-4 Mpc^-3 Gyr^-1 at z=0.6. When we compare with previous studies at similar redshifts, we find that the merger rate decreases when mass increases.