| Invariant mass distribution
||Dielectron invariant mass spectrum, where red dots with error
bars are data and blue shadeds are other Standard Model
background and green ones are QCD background. White region is
Drell-Yan background. Background contributions are added. Left
plots show the mass region from 50 to 500 GeV/c2 in a
log scale and right plots show high mass region above 150
GeV/c2 in a linear scale.
EPS or GIF.
||(data-prediction)/uncertainty as function of dielectron mass.
EPS or GIF.
| Signal Significance
||We search for deviations from the Standard Model prediction
in the dielectron invariant mass spectrum by estimating the
probability of any excess originating only from fluctuations of
standard model backgrounds as follows. We scan the spectrum
between 150 and 500 GeV/c2 in 5 GeV/c2
steps. At each step we fit the distribution to a background-only
hypothesis and a background + narrow resonance hypothesis. For
the background + narrow resonance hypothesis, we add to the
background prediction a gaussian distribution centered on the
step whose width is fixed to the known detector resolution. We
then compute the difference in &chi2
(&Delta&chi2) between the two hypotheses as a figure
of merit to quantify the significance of any excess (shown in the
plot). Finally, we estimate the probability to observe the
maximum value of &Delta&chi2 we observe in the data
anywhere in the spectrum from 150 to 500 GeV/c2 under
the background-only hypothesis using background-only
pseudo-experiments. The probability of observing an excess as
significant or more than that observed in data estimated with
this method is 1.7%.
| Event Display
||Typical dielectron event display.
COT View (EPS or GIF).
CAL View (EPS or GIF).
| Other Studies
using the dielectron channel and slightly
more integrated luminosity did not report a similar excess but
the two analyses are statistically consistent with each other at
the 7% level. Another study using dimuon pairs is in progress.
||The data size will be increased by about a factor of 2
before the planned shutdown in the summer of 2007. The data will be
analyzed through the same procedure and reported.