Objective: Eating disorder (ED) research is increasingly focusing on the pr
oximal antecedents to disordered eating behavior. Such antecedents may incl
ude cognitions, environmental stimuli, social interactions, and affective s
tates. Current ED theories suggest that the relationships between anteceden
ts and eating behavior may be complex, including interaction associations,
time-lagged effects, and associations that persist only for brief periods o
f time. Similarly, these theories often include the consequences of behavio
r-influencing variables of interest (e.g., short-term reductions in negativ
e affect). Careful examination of such theories, however, has been limited
by a reliance on data collection methods not appropriate for testing these
effects. Method: This study examines alternative methods for data collectio
n and analysis that overcome previously noted limitations, using data colle
cted in several studies with eating-disordered participants. Results: The d
evelopment of a technique called ecological momentary assessment (EMA) allo
ws the ongoing study of behavior in its natural context and reduces biases
associated with retrospective recall. The development of technology that al
lows the sophisticated collection and storage of such data (e.g., palm-top
computers), along with statistical procedures for analyzing hierarchically
nested, repeated measures data, allow precise testing of complex theoretica
l models. Discussion: We demonstrate several important features of this res
earch: (I) patients are willing and able to engage in EMA studies, (2) data
not possible to collect using other designs are obtainable, (3) complex th
eoretical models can be evaluated using these data and appropriate statisti
cal methods, and (4) the collection and analysis of EMA data present unique
difficulties to ED researchers. Finally, we endorse and provide recommenda
tions for the use of EMA in future ED research and practice. (C) 2001 by Jo
hn Wiley & Sons, Inc.