Even under these conditions of self-awareness, project analyses show that discards generally adhere to the same patterns found in garbage collected anonymously at the census tract level (Ritenbaugh and Harrison 1984).

Although independent of informant-based distortions, refuse analysis is susceptible to other forms of bias. The most obvious one is garbage disposals, and the Garbage Project has conducted studies, not unlike those of ethnoarchaeologists, to develop correction factors for ground up food (Rathje and McCarthy 1977). Other biases include people who carry recyclables to drop-off buy-back centers and the fact that behavior can be characterized only at the household level, not on the individual level.

Overall, the advantages of garbage sorting as both an alternative and a quantitative measure of behavior outweigh its limitations, and the first pattern identified—that self-reports differ from refuse records—has opened up a broad new research area.

The second conclusion drawn from refuse analysis is that there is a clear patterning in the differences between what people report they do and what they actually do. This conclusion was drawn from a number of Garbage Project studies that were designed to verify consumer responses to various kinds of diet questionnaires by comparing self-reports about food use against packaging and food debris in fresh refuse. One specific self-report/refuse pattern the Garbage Project has documented is “the good provider syndrome”: a female adult reporting for a household as a whole has a tendency to overreport everything the household uses by 10 to 30 percent or more. Another pattern is “the surrogate syndrome”: to find out how much alcohol is consumed by household members, do not ask a drinker; drinkers consistently underreport their alcohol consumption by 40 to 60 percent. Instead, ask a nondrinker, as people in this category report accurately what drinkers drink (Dobyns and Rathje 1987; Johnstone and Rathje 1986). The second conclusion is again no real surprise.

Unlike the first two, the third conclusion was full of surprises. The differences between respondent reports and the material remains in refuse frequently indicate directly opposed behaviors; to be more specific, respondents normally report rational behaviors while their actual behaviors often appear irrational. One of the best examples of this kind of counterintuitive relationship between self-reports and refuse occurred during the highly publicized “beef shortage” in the spring of 1973. At that time, when consumers were complaining bitterly about high prices and the erratic availability of beef, the Garbage Project was recording the highest rate of edible beef waste it has ever documented (Rathje and McCarthy 1977).

Several other instances of this kind of counterintuitive report/refuse pattern have been documented. In 1977, the Garbage Project gave “meat fat” its own separate category. Using a long-term database, the Garbage Project determined that in 1987 people began cutting off and discarding much larger than normal quantities of the separable fat on fresh cuts of red meat and, at the same time, they also bought less fresh red meat. Both actions seemed to be responses to a 1986 National Academy of Science study that was widely reported in the media and which identified fat from red meat as a cancer risk factor. There was just one problem. The consumers studied replaced the fresh red meat in their diet with processed red meat—salami, bologna, sausage, hotdogs, etc., which contained large quantities of hidden fat—so the level of fat intake in the diet did not fall; instead, it stayed the same or increased (Rathje and Ho 1987).

A third case involved household hazardous wastes (see Rathje, Wilson, Hughes, and Herndon 1987). In 1986, Marin County in California sponsored a Toxics Away! Day to collect household hazardous wastes such as used motor oil and unused pesticides. The Garbage Project recorded residential refuse two months after the collection day and compared it to household discards sorted before the collection day. The results were completely unexpected: there were nearly two times more potentially hazardous wastes recorded in the refuse after the collection day than there had been before. The data clearly demonstrated that all of the increase in hazardous wastes was owing to the discard of large quantities of items from only a few households