Within the context of the Triad’s second-generation data quality model, "data quality" is assessed according to the data’s ability to support confident decision-making. "Decision quality data" are defined within the Triad as data that can support the decision to be made at the desired level of confidence. Decision quality data sets often are comprised of data from different sources (e.g., a smaller number of analyses from standard fixed laboratory methods supplemented with a larger number of real-time measurements) so that all aspects of data uncertainty (sampling, analytical, and relational) are managed. The term "collaborative data" is used in the context of the Triad to express this.

Following the same line of reasoning, "screening quality data" are data that provide some useful information for the decision to be made, but taken alone are not sufficient to make the decision at the desired level of confidence. Screening quality data are data of known quality, however, there is either excessive analytical, sampling, or relational uncertainty present with respect to the decision to which the data are being applied. Excessive analytical uncertainty can include detection limits that are too high, or analytical bias or imprecision that is too great. Excessive sampling uncertainty occurs when there are insufficient data points to support conclusions about important population parameters, such as whether the mean concentration is below cleanup guidelines, whether hot spot concerns are absent, or the actual boundaries of spatially patterned contamination. Excessive relational uncertainty exists when the interpretation of measurement results is ambiguous with respect to decision-making. Both standard fixed laboratory analyses and field deployable technologies can produce decision quality data or screening quality data, depending on the level of analytical, sampling, and relational uncertainty present with respect to the decision.

Generally, screening quality data will be used in conjunction with other data with the goal that the combined information will manage sampling, analytical, and relational uncertainties (i.e., the overall uncertainty) in the site data and the CSM constructed from that data. For sampling uncertainty, once there is some understanding of what the target analyte list might be, less expensive methods could be used to map the area for those analytes. Higher analytical quality data with lower reporting limits or less bias may be needed to control the analytical uncertainty. Specific areas may need to be sampled and analyzed with higher quality methods where relational uncertainty is unacceptable. The outcome is a collaborative data set that represents decision quality data and confidence in the CSM. Within the Triad approach, decision quality data are used ultimately to support decisions that need to be made. Less expensive, real-time measurement technologies, if used alone, often produce screening analytical quality. Expensive fixed lab methods, if used alone, usually produce screening sampling quality or screening data representativeness.