Technical Abstract: The explosion of genomic data is revolutionizing plant breeding and genetics research, but in most applications good phenotypic data are crucial for making efficient use of genomic data. While new technologies have considerably improved accuracy and throughput of genotyping, relative slowness of methodological advance makes phenotyping the major bottleneck for genomic studies (e.g. linkage mapping, association mapping, genomic selection, marker-assisted selection, etc.). In other applications, good phenotypic data are of interest in themselves, e.g. for assessing the value of accessions in genebanks or breeding programs. Combining data from different trials becomes complicated if some of the evaluations were performed on ordinal scales. This paper discusses methods that can be used to combine data from ordinal rating scales into a single dataset.