This work is a study on scales used in the sociology of education as important factors in student and school performance. First, we employ nonparametric item response theory to analyze the properties of three one-dimensional scales obtained from the PISA- 2000 data base, for the theoretical constructs, social capital, cultural capital and economic capital, of students from Peru. Then, we use confirmatory factor analysis as a way to examine construct dimensionality and to assess model fit. Last we use these scales in a two level model to investigate school effects: we use the three scales of capital, as measured by both methods, as controls at the student level and at the school level to investigate the school contribution to student performance. Our results has shown that we can obtain a fairly strong scale for economic capital, with a high scalability coefficient (H), which implies that the respondents can be well ordered with respect to the latent trait. The construct social capital presented an intermediate scalability and the construct cultural capital presented a low scalability. The multilevel analysis has shown that there remains a significant school effect, even after controlling students achievement by economic capital and cultural capital at student level and average cultural capital at school level. Among the factors related to the resulting differences in school achievement we found a significant effect due to the variables educational resources, school infrastructure, teacher's motivation and lack of teachers; and the first one educational resources is the one that accounts for the largest difference.