This study introduces new methods of quantifying and evaluating the human c
hewing cycle. These methods were validated on a sample of 26 young adults (
11 women and 15 men) between 20-35 years of age. Movements of the mandibula
r central incisors were recorded (100 Hz) using an optoelectric computer sy
stem while the participants chewed gum. A subsample of 10 cycles was automa
tically selected, based on multiple objective criteria to ensure representa
tive cycles for each individual. Once representative cycles had been identi
fied, multilevel statistical models were used to evaluate and describe the
sample's kinematic patterns. The multilevel procedures allow for missing ob
servations, they do not assume equal intervals. and variation can be partit
ioned hierarchically. Two-level models showed significantly shorter cycle d
uration for males (835 msec) than females (973 msec). Inferior-superior (IS
) cycle range was 2.6 mm larger and maximum LS velocity was 19.6 mm/sec fas
ter in males than females. There were no significant differences in medial-
lateral (ML) and anteroposterior (AP) excursive ranges or velocities. With
the exception of cycle duration and ML ranges of motion, random variation w
as three to five times larger between individuals than between cycles. The
three-level models showed that eighth-order polynomials were necessary to d
escribe IS, AP, and ML chewing movements of the entire cycle. The models id
entified highly significant sex differences in cycle kinematics (excursions
, velocities, accelerations, etc.) for each aspect of movement (AP, IS, and
ML). It is concluded that this;approach provides several important advanta
ges over existing methods, including (a) its objectivity, (b) a more comple
te description of kinematic patterns, (c) a hierarchical description of var
iation, and (d) its ability to test hypotheses statistically, (C) 2000 Else
vier Science Ltd. All rights reserved.