Relational Coordination
Guidelines for Theory, Measurement and Analysis
Chapter 3: Measuring Relational Coordination (continued)
Unit of Observation and Unit of Analysis
The unit of observation for relational coordination is the individual participant in the work process, represented by the individual survey respondent. These individual respondents are then aggregated into a larger unit of analysis in order to construct a measure of relational coordination. That unit of analysis will depend on the hypothesis you are exploring. If you are studying an intervention that is expected to improve relational coordination of a particular work process, and the performance of that work process, your unit of analysis will be different periods of time, i.e. before and after the intervention has been implemented. If you are doing a cross-sectional study in which multiple sites that independently carry out the same work process are expected to have different levels of relational coordination, which are expected to result in different level of performance, your unit of analysis will be the site.
Focal Work Process or Individual Client?
Instead of asking relational coordination survey questions about a focal work process and perhaps a focal client population served by that work process, as seen in Appendices A, B and C, the relational coordination survey questions can be asked instead about individual clients. See Appendix D for an example of this alternative survey. With this alternative approach, one can construct a measure of relational coordination that is specific to individual clients, which is useful in organizations where different practices or interventions are being used for different clients. In this case, questions are asked about specific clients, rather than asking for general perceptions of typical patterns. Questions are asked about the respondents' specific interactions with other functions regarding a particular client. This introduces a greater potential for retrospective response error. To minimize that response error, it is desirable to survey participants as soon as possible after they have interacted with a particular client. The other challenge arises if the same participants are involved in providing service to multiple clients, thus requiring them to complete numerous surveys for the same study, one about each individual client, rather than a single survey about general patterns of interaction. Numerous surveys sent to a given participant about individual clients may be completed, but response rates are more challenging to achieve given the greater burden on the study participants.
One study that measured relational coordination for individual clients was reported in a paper called "Is the Doctor In? A Relational Approach to Job Design and the Coordination of Work". This was a one-hospital study in which some patients were cared for by physicians with the traditional job design, while other patients were cared for by physicians with a new "hospitalist" job design. It was hypothesized that the new physician job design would result in higher levels of relational coordination between physicians and other members of the care provider team, thus resulting in better risk-adjusted patient outcomes including shorter lengths of stay, lower total costs, fewer readmissions and lower mortality. Measuring relational coordination for individual patients enabled the assessment of this new job design that had been adopted for some patients and not others. xxxii
Why the Network Approach to Measuring Relational Coordination?
Relational coordination is measured based on a matrix, or network, methodology, in which each cross-functional tie is measured separately. Wouldn't it be much simpler to ask respondents for a global assessment each of these seven relational coordination dimensions, rather than creating a measure of relational coordination based on a matrix of specific cross-functional ties? Certainly. Indeed, a recent study in which researchers had access to only a few representatives of each organization, not nearly enough to enable a network measure of relational coordination, instead asked the relational coordination questions more generally about patterns of interaction in the organization as a whole. This study did find statistically significant relationships between the abridged measure of relational coordination and both psychological safety and learning from failures. xxxiii
However, the concept of relational coordination is more accurately captured as a network of ties. In the coordination of work, each tie potentially has a differential impact on performance, which would be lost in a more aggregated or global assessment of relational coordination. When respondents are asked to assess the quality of their communication and relationships with all functions globally, a particularly negative connection with one of the other groups could disproportionately influence the overall assessment. By asking respondents to evaluate separately their connections with each other function, the accuracy of measurement is enhanced.
The final and perhaps most compelling reason for a network measure of relational coordination is the ability to disaggregate the network into its component ties for the purpose of diagnosis and intervention. By measuring each cross-functional tie separately, the researcher reserves the possibility of doing a sensitivity analysis to learn which of the ties has the greatest impact on performance. The researcher can also diagnose for an individual site which ties are weakest. Cross-functional ties that have a significant impact on performance, and that are problematic for a particular site, should become a high priority for management attention in that site.
For example, in the study of physician job design reported above, the largest and most significant differences in relational coordination between the old and new physician job design were found in the ties between the physician and other members of the team, rather than among non-physician members of the team, with the biggest impact being on the physician/nurse tie. This type of finding, drilling down to the level of the dyad within the team, is only possible with a network measure. (Continued ... )