4.6. VELMOD Results Using 500 km s-1 Smoothing
In our mock catalog tests, we found that using 500 km s-1 smoothing in the IRAS reconstruction resulted in ~ 20% overestimates of I (Section 3.1). However, because the mock catalog may not faithfully reproduce the dynamics of the real universe, it is useful to see how much I changes for the real data when the 500 km s-1-smoothed IRAS reconstructions are used. We carried out two such VELMOD runs, one with and one without the quadrupole. (We determined the quadrupole in the same way as for the 300 km s-1-smoothed reconstruction and found that the two differ little.) The resulting maximum likelihood estimates of I are listed in Table 2. The larger smoothing results in an increase in I, as expected. However, the 500 km s-1 result, I = 0.544 ± 0.071, is within 1 of our favored result obtained at 300 km s-1 smoothing. If we reduce this value by 20%, in accord with the bias seen in the mock catalogs, we obtain I = 0.45 ± 0.07, also within 1 of our preferred result. Our choice of a 300 km s-1 smoothing scale is thus unlikely to have led us seriously astray, even if the mock catalogs are imperfect guides.