This article presents a general framework for representing constructs in organizational research, emphasizing the importance of construct depth (specificity versus generality) and dimensionality. It integrates this framework with methods of construct validation based on confirmatory factor analysis of multitrait-multimethod (MTMM) matrices. The authors apply these methods to the measurement of work values using the Work Aspect Preference Scale (WAPS). Results show that the WAPS performs better when used to represent relatively specific work values rather than more general ones. Further analyses supported the generalizability of the WAPS factor structure for men and women, although gender differences were found on structured means for several latent value dimensions. Rating scales are widely used in organizational research to operationalize theoretical constructs and make policy decisions. They are used in various areas, including task performance, job satisfaction, organizational citizenship, job characteristics, commitment, motivation, identification, and work values. A central concern with rating scales is construct validity, or the extent to which the measures capture the underlying construct. Various procedures are used in organizational research to establish construct validity, such as reliability estimation and the examination of relationships between focal measures and other relevant constructs. These procedures are integrated in the MTMM matrix approach, which involves examining internal consistency, convergence of multiple measures of the same construct, and distinctiveness of measures of different constructs. A key premise in construct validation is that constructs are represented at the appropriate depth. Construct depth refers to the specificity versus generality of a construct and its associated operationalizations. Specific constructs may refer to narrowly defined phenomena or fine-grained aspects of broader constructs, while general constructs typically correspond to global phenomena or combinations of specific constructs. The article discusses four models for representing constructs: total disaggregation, partial disaggregation, partial aggregation, and total aggregation. Each model has different implications for construct validation and the interpretation of the scale. The WAPS is used as an example to illustrate these models. The total disaggregation model is the most concrete representation, where each item is treated as a separate indicator. The partial disaggregation model reduces the number of parameters to be estimated and tends to decrease measurement error. The partial aggregation model is more abstract and focuses on facets or global levels. The total aggregation model is the most abstract, where a single global factor is hypothesized to account for variation in all measures. The article also discusses issues of generalizability, structured means, convergent and discriminant validity, and the use of MTMM matrices for construct validation. The authors conclude that a general approach to representing constructs in organizational research is essential for meaningful results, and that this approach should incorporate construct depth and dimensionality. The psychometric literature has not fully addressed construct depth, structured means, and the use of MTMM matrices within the context of hierarchical rating scales. The authors propose a framework that integrates construct breadth and depth, dimensionality, and construct validation within a single, unified framework for representing constructs in organizational research.This article presents a general framework for representing constructs in organizational research, emphasizing the importance of construct depth (specificity versus generality) and dimensionality. It integrates this framework with methods of construct validation based on confirmatory factor analysis of multitrait-multimethod (MTMM) matrices. The authors apply these methods to the measurement of work values using the Work Aspect Preference Scale (WAPS). Results show that the WAPS performs better when used to represent relatively specific work values rather than more general ones. Further analyses supported the generalizability of the WAPS factor structure for men and women, although gender differences were found on structured means for several latent value dimensions. Rating scales are widely used in organizational research to operationalize theoretical constructs and make policy decisions. They are used in various areas, including task performance, job satisfaction, organizational citizenship, job characteristics, commitment, motivation, identification, and work values. A central concern with rating scales is construct validity, or the extent to which the measures capture the underlying construct. Various procedures are used in organizational research to establish construct validity, such as reliability estimation and the examination of relationships between focal measures and other relevant constructs. These procedures are integrated in the MTMM matrix approach, which involves examining internal consistency, convergence of multiple measures of the same construct, and distinctiveness of measures of different constructs. A key premise in construct validation is that constructs are represented at the appropriate depth. Construct depth refers to the specificity versus generality of a construct and its associated operationalizations. Specific constructs may refer to narrowly defined phenomena or fine-grained aspects of broader constructs, while general constructs typically correspond to global phenomena or combinations of specific constructs. The article discusses four models for representing constructs: total disaggregation, partial disaggregation, partial aggregation, and total aggregation. Each model has different implications for construct validation and the interpretation of the scale. The WAPS is used as an example to illustrate these models. The total disaggregation model is the most concrete representation, where each item is treated as a separate indicator. The partial disaggregation model reduces the number of parameters to be estimated and tends to decrease measurement error. The partial aggregation model is more abstract and focuses on facets or global levels. The total aggregation model is the most abstract, where a single global factor is hypothesized to account for variation in all measures. The article also discusses issues of generalizability, structured means, convergent and discriminant validity, and the use of MTMM matrices for construct validation. The authors conclude that a general approach to representing constructs in organizational research is essential for meaningful results, and that this approach should incorporate construct depth and dimensionality. The psychometric literature has not fully addressed construct depth, structured means, and the use of MTMM matrices within the context of hierarchical rating scales. The authors propose a framework that integrates construct breadth and depth, dimensionality, and construct validation within a single, unified framework for representing constructs in organizational research.