In this work it was determined if there exist delays between the emissions in different bands in 16 Fermi/LAT detected blazars, via the cross-correlation function (CCF). The bands we are working with are: gamma-rays (0.1-300 GeV), optical V band, near-infrared J, H and Ks bands (observation frame) and emission lines (object rest frame). We also want to investigate the performance of distinct methods used to obtain the cross-correlation function under different conditions, and we pretend to point out their flaws and strengths, so that in the future we can make the correct decision about what method to use, depending on the specific characteristics of the data available. The cross-correlation analysis was carried out by three different methods discussed previously in the literature, the interpolation method (Gaskell & Sparke, 1986), the discrete cross-correlation function (Edelson & Krolik, 1988) and the Z-transformed discrete correlation function (Alexander, 1997). The idea behind using more than one method is to increase the sturdiness of the results obtained. Nonetheless, some modifications were made with respect to the original methods due to some problems present in these, mainly to correct for the nonlinearity of the AGN light curves. Our results show that for all objects, the three NIR bands vary simultaneously. For four objects of our sample (3C 273, Mrk 501, PKS 2155-304, and PMN J0808-0751), no correlation is found between any of the bands available in this study. For the remaining twelve, a correlation was found between the V band and the NIR bands; in most of them indicating that the V band and the NIR bands vary simultaneously or with small delay. For just 3 objects (3C 454.3, PKS 0235+164, and PKS 1510-089) was found a correlation between the gamma-ray emission and the V band; for the same three objects is found a correlation between gamma-ray emission and the NIR bands. In this work, we arrived to the conclusion that due to the different flaws existent in the three methods to compute the CCF, there is not a single method better than the others; and if we want better robustness in the results it is necessary to use more than one method.