We use a moving average forecast method to develop forecasts of sales/demand. Based on an analysis of outliers in our sales data, we surmise that most outliers were caused by random events. Should we go 'long' or 'short' (take more or fewer periods) in our determination of the # of periods to use for computing our moving average forecasts? O Go 'short' - Use fewer periods since we wish to stay include outliers while developing our forecasts Doesn't make a difference since the outliers are random in occurence. Go 'long' - Use more periods since we do not want to include outliers while developing our forecasts