Lifetime totals from the measurement history currently retained in Airplay Live · updating live
Airplay Live should pass tests that do not depend on believing the score.
A real signal should repeat, known biases should disappear after correction, obvious differences should appear, and places where no difference exists should stay quiet.
Two Halves, One Answer
It could be noise. Noise would not repeat.
Split-half test · … tracks · plays split at random
Split-half agreement (r)
The Big 80s0.39
Box UK0.02
r = 0.39 across … tracks
Each dot is one track, scored independently from two non-overlapping halves of its plays. The Big 80s shows repeatable track-to-track differences across 617 tracks; Box UK shows almost none.Airplay Live live database · Random non-overlapping play halves
Off the Clock
The hour a track airs should not decide its score.
Schedule bias test · before and after ambient correction
Before correction, tune-outs partly follow the schedule. After ambient correction, that schedule link all but disappears: Dance UK’s drops from 0.38 to 0.00, Box UK’s from 0.15 to 0.00.Dance UK + Box UK · Raw vs. ambient-corrected measurement · Airplay Live live database
When It Says No
A meter that finds a difference everywhere finds nothing.
Genre-pair test · four stations · Appeal difference ±95 % CI
Two genre contrasts sit clearly away from zero, while two stay much closer to it. The meter does not produce similarly large differences across every comparison.4 genre-pair comparisons · 4 stations · 95% confidence intervals · Airplay Live live database
One Ballad Every Hour
A test where every radio person already knows the answer.
Off-format test · hourly ballads vs. catalog
Hourly ballads form a clearly lower score cluster: 34 qualifying ballads average about 8 points below the rest of the catalog.Dance stream · 548 catalog tracks · 34 ballads · Frozen archive · Schedule fixed before measurement No smoothing · End-of-hour ambient tune-out matched the rest of day · In-format genres scored alike
