The bandwidth value used to select a peer by population criteria is the historical average bandwidth of the peer. The bandwidth value displayed in the test results for each data point is the exact bandwidth when the test is executed.
Since the instant bandwidth of a peer may vary and may not necessarily be the same as the historical average, you will see test results in the chart from all bandwidths that have been reported. If the instant bandwidth is lower than the population criteria, you will see data points with lower bandwidth in the chart.
For the minimum bandwidth expected for a high broadband peer, the peer should report the same instant bandwidth as configured for at least 90% of the times. In other words, you may see all bandwidths listed under the Filters, but when you chart the non-selected bandwidth range(s), there should only displays sporadic data points.
There is no minimum bandwidth expected for a high broadband peer.
Since the peer is installed on the real user's machine, even if the machine has high broadband, the bandwidth assigned to the peer could be low when the test is run.
If a peer consistently reports lower bandwidth for your test, it could be a peer issue. In that case, we suggest you blacklisting that peer so your test will not be executed by it. For details, see Creating a Peer Population Blacklist in the documentation.