Table 1. Details of the data analysis.

Step	Description
1) Initial concept map and coding	a. Four coders (MJL, KAB, LL, XG) were each assigned a random selection of 5% of the data* b. Four coders independently developed:  i. Visual concept maps to describe the content of their subset  ii. Qualitative codes describing the content of the calls related to why the complainant was concerned about rats and what they wanted to occur because of their complaintc. Coders compared and contrasted maps and codesd. Coders collaboratively built a single concept map and set of codes for the entire 20% random selection of data
2) Coder calibration	a. The four coders independently mapped and coded the same 50 complaints b. Compared maps and codes c. Discussed discrepancies and emergent information d. Came to a consensus on each discrepancy before proceeding
3) Coding and concept mapping	a. Each of the four coders was randomly assigned 25% of the overall dataset b. Each coder independently:  i. Qualitatively coded their subset of the data   • Categorized each complaint by the co-developed qualitative codes   • The initial set of codes was not changed throughout the coding process to facilitate consistency among coders and enumeration across the entire dataset  ii. Refined the initial collaborative concept map to account for new information not captured by the qualitative codes developed in step 1   • Individual concept maps were built through the addition of new concepts, restructuring, and reorganization as new information emerged from their respective datasets   • Memoing was used to track reasoning behind changes   • Maps were continually updated until no new call triggered the addition of new information to the map and the content of every call could be categorized within the framework
4) Collaborative concept mapping	a. Coders compared and contrasted concept maps b. Discussed and agreed upon discrepancies in maps c. Organized concepts into higher level themes d. Collaboratively built a visual thematic framework to describe the content of the overall dataset
5) Coding summaries	a. Descriptive statistics were used to assess the proportion of each code in the overall dataset b. Summaries were written for each code and are described in the Results section

^*All random selection occurred within categories of complaints predefined by the city. The city categorized calls into 29 categories (i.e., property use complaint, fire safety hazard, etc), such that for step 1 in this table, 5% of the codes were randomly selected from each of these 29 categories to give each coder 5% of the overall number of complaints.