Empirical data from the Tokyo Metropolitan Expressway show how relations between traffic flow, density and speed evolve as the data are aggregated spatially and temporally. Considering larger geographic areas or longer time frames consistently reduces scatter associated with fundamental traffic flow relationships, according to quantitative results presented here. We argue that econometric results largely explain the somewhat counterintuitive finding. The relatively strong explanatory power of traffic flow relationships in aggregated data stems from correlations between observations of traffic on adjacent links and at adjacent points in time, as well as loop detector observational error. The data actually support a stronger claim that estimates of flow across a large area based on the aggregate fundamental diagram outperform estimates based on the summation of terms from linkspecific fundamental diagrams. This result again matches econometric findings.