In the present study, 15,108 daily records of milk yield (DMY), milk fat percentage (DFP), milk protein percentage (DPP), and milk somatic cell score (DSCS) in the first-two lactation periods of the Murciana-Granadina goats, collected from 2017 to 2024 in the southern part of Kerman province of Iran, were used. By applying latent variable modeling technique and confirmatory factor analysis a latent variable of daily milk production performance (DMP) in the first-two lactation periods of the Murciano-Granadina goat breed was constructed and evaluated statistically by four goodness of fit measures including standardized root mean square residual (SRMR), root mean square error of approximation (RMSEA), Tucker-Lewis Index (TLI), and comparative fit index (CFI). The values of CFI, TLI, RMSEA, and SRMR, were 0.97, 0.92, 0.05, and 0.02, respectively, implying the suitability of the confirmatory factor model proposed for DMP latent variable proposed in the present study. Standardized factor loadings of the traits used for describing DMP were statistically significant (P < 0.01) values of 0.24, -0.55, -0.54, and -0.37 for DMY, DFP, DPP, and DSCS, respectively. Therefore, it may be concluded that increase in milk yield would increase DMP. On the other hand, increases in milk fat percentage, milk protein percentage, and milk somatic cell score would decrease DMP. for constructing the latent variable of DMP in the Murciano-Granadina goats, emphasis should be on increasing milk quantity and decreasing milk composition traits including milk fat percentage, milk protein percentage, and milk somatic cell score. Pearson’s correlations among the investigated measurable traits were statistically significant (P<0.01), ranging from -0.14 (DMY-DFP and DMY-DSCS) to 0.30 (DFP-DPP). Pearson’s correlations between the DMP latent variable and the measurable milk yield and composition traits were also statistically significant (p<0.01) and were -0.76 (DFP-DMP), -0.75 (DPP-DMP), -0.52 (DSCS-DMP), and 0.33 (DMY-DMP). Overall, the study provides a statistical framework for describing DMP in contexts such as phenotypic evaluations, where a latent construct can capture the concept of milk yield and composition traits more effectively than individual traits alone.