We analyze the LSND , KARMEN and MiniBooNE data on short-baseline \bar { \nu } _ { \mu } \to \bar { \nu } _ { e } oscillations and the data on short-baseline \bar { \nu } _ { e } disappearance obtained in the Bugey-3 and CHOOZ reactor experiments in the framework of 3+1 antineutrino mixing , taking into account the MINOS observation of long-baseline \bar { \nu } _ { \mu } disappearance and the KamLAND observation of very-long-baseline \bar { \nu } _ { e } disappearance .
We show that the fit of the data implies that the short-baseline disappearance of \bar { \nu } _ { \mu } is relatively large .
We obtain a prediction of an effective amplitude \sin ^ { 2 } 2 \vartheta _ { \mu \mu } \gtrsim 0.1 for short-baseline \bar { \nu } _ { \mu } disappearance generated by 0.2 \lesssim \Delta { m } ^ { 2 } \lesssim 1 \text { eV } ^ { 2 } , which could be measured in future experiments .