After recent results from solar neutrino experiments and KamLAND we can definitely say that neutrinos from SN1987A underwent flavor conversion , and the conversion effects must be taken into account in the analysis of the data .
Assuming the normal mass hierarchy of neutrinos we calculate the permutation factors p for the Kamiokande-2 , IMB and Baksan detectors .
The conversion inside the star leads to p = 0.28 - 0.32 ; the oscillations in the matter of the Earth give partial ( and different for different detectors ) regeneration of the original \bar { \nu } _ { e } signal , reducing this factor down to 0.15 - 0.20 ( at E = 40 MeV ) .
We study in details the influence of conversion on the observed signal depending on the parameters of the original neutrino spectra .
For a given set of these parameters , the conversion could lead to an increase of the average energy of the observed events up to 50 % and of the number of events by a factor of 2 at Kamiokande-2 and by a factor of 3 - 5 at IMB .
Inversely , we find that neglecting the conversion effects can lead up to 50 % error in the determination of the average energy of the original \bar { \nu } _ { e } spectrum and about 50 % error in the original luminosity .
Comparing our calculations with experimental data we conclude that the Kamiokande-2 data alone do not favor strong conversion effect , which testifies for small difference of the original \bar { \nu } _ { e } and \bar { \nu } _ { \mu } spectra .
In contrast , the combined analysis of the Kamiokande and IMB results slightly favors strong conversion effects ( that is , large difference of the original spectra ) .
In comparison with the no oscillation case , the latter requires lower average energy and higher luminosity of the original \bar { \nu } _ { e } flux .