An inherent problem in measuring the influence of expert reviews on the demand for experience goods is that a correlation between good reviews and high demand may be spurious, induced by an underlying correlation with unobservable quality signals. Using the timing of the reviews by two popular movie critics, Siskel and Ebert, relative to opening weekend box office revenue, we apply a difference-in-differences approach to circumvent the problem of spurious correlation. After purging the spurious correlation, the measured influence effect is smaller though still detectable. Positive reviews have a particularly large influence on the demand for dramas and narrowly-released movies.