View description.

This presents a graphical form of performance comparison of plain mean and Volz-Heckathorn (VH) estimators for respondent-driven samples on simulated networks. The x axis represents a network topology sensitivity constant, and y axis represents the number of coupons used in simulated RDS processes. Every component of the plot consists of upper and lower set of 3 half circles. Large half circles correspond to absolute difference in average mean squared error (MSE), left small half circles correspond to normalized absolute difference in squared bias, and right small half circles correspond to absolute difference in variance. All differences are between the VH and plain mean estimator. Yellow background colour means positive sign, and blue colour means negative sign. Upper sets of half circles correspond to exact degree reporting, and lower sets of half circles correspond to random degree reporting. The pairs of numbers separated by commas mean the average number of times the RDS process had to restart, and the average number of waves needed to collect the sample (sample size is 30 for networks of sizes 100 and 300, and 300 for networks of sizes 1000 and 3000).

Functionality description:

- Network topology is set by a group of 3 radio buttons with values
*homophily*,*inv. homophily*and*rich-gets-richer*. - Seed selection and participant referral function can be selected with radio buttons below the network topology selection controls.
- Network size can be set using the group of radio buttons with values
*size 100, size 300, size 1000*and*size 3000*.

This is a supplementary publication visualization 1 that has been made available along with supplementary publication visualization 2.

Tags: estimation, homophily, Joe, Processing, RDS, visualization