From NYCT's Staff Summary on the service change:
"The resulting service plan provided a net increase in Astoria Line service, albeit with a decrease in frequency at local stations during the peak hour. Since approximately 43% of Astoria Line riders board at express stations, it was believed that the express/local operation would provide sufficient capacity on the N and W trains.
"In practice, since the service commenced in July 2001, the expected operational benefits of the express/local operation have not been realized, and the Astoria Line component of the service plan has proved problematic from a customer perspective. According to traffic checks conducted in August 2001, the ridership split between the N and W routes was not as even as expected, with N trains carrying 1.9 times as many passengers as W trains in the mornings, and 2.6 times as many in the evenings. As a result, a number of complaints were received from local riders, who experienced long waits and crowded conditions on the N, with lightly loaded W trains passing by on the express track.
The scheduled departures from Astoria and their arrivals at QBP are as follows: 7:49-7:59(N), 7:56-8:05(W), 7:59-8:09(N), 8:06-8:15(W), 8:09-8:20(N), 8:16-8:25(W), 8:19-8:29(N), 8:24-8:33(W), 8:27-8:37(N), 8:32-8:41(W), 8:35-8:45(N), 8:39-8:50(W), 8:46-8:57(N), 8:47-8:55(W).
The average interval from an N to a W is approximately 6 minutes; the average interval from a W to an N is approximately 3 minutes during the peak. There is no gain from using the W. It does not pass the previous N. There is no advantage for a passenger wait for the next W; his interest is best served by boarding the first train. Based on the differences in headways, 2/3's of the passengers would board a W and 1/3 an N. Thus one would expect that 0.66 x 43 = 29% of the passengers would use the W and the remaining 71% would use the N. The ratio of 71:29 is 2.4, which is close enough to the 1.9 ratio that the TA observed.
If the TA wanted to approach more equal loading a different scheduling stategy would have been required. Suppose every N departed 1 minute after the W, instead of 3. This would have increased the N to W headway to 8 minutes. The number of passengers on the W would have been 8/9 x 43% or 38%, with the remaining 62% on the N. The W running time was 1 minute less, so the 1 minute interval would have been a 2 minute interval at the merge before QBP.
The afternoon behavior is murkier because the W makes fewer stops in Manhattan. One would expect that it will have fewer passengers whether it is a local or an express on the Astoria line.
The fewer Manhattan destinations also means that changing the morning W to a local will not equalize completely loading.