1960: The first mail-out census forms were used in 1960. Census questionnaires were mailed to every household and residents were asked to complete the forms and hold them for collection by an enumerator. When the enumerators collected the initial forms they left a second form with additional questions at 25% of the households. These forms were to be mailed back and checked for accuracy. Phone or in-person interviews were conducted to complete any unanswered questions. 
1970: Census forms were again mailed out to all households (20% receiving a long form). “In larger metropolitan areas and some adjacent counties (approximately 60 percent of the United States’ population), households were asked to complete and return the questionnaire by mail on April 1, 1970 (resulting in an 87 percent mail back response rate).” As in the 1960 mail-back sample, these forms were reviewed for consistency and completeness. Follow-ups were done when necessary. Census enumerators picked up the questionnaires from the remaining 40% of the population. 
1980: The mail-out/mail-back process in 1970 met with such success that the 1980 census was conducted almost entirely by mail, covering greater than 95% of the population. 
For genealogists, the census provides invaluable information amid two persistent problems: missing families (the undercount) and misspelled names. Fortunately, both of those issues should become less problematic for future family historians. Misspellings don't matter for federal apportionment of funds, but luckily, with the advent of self-enumeration forms this problem will disappear on its own. The government does, however, care about the problem of the undercount and worked hard to address it in advance of the 1970 census. I haven't found statistics for how many households were thought to be undercounted and hence missing from the census. That makes it difficult to say how effective their measures were.
So what was curb stoning? Since the inception of the census “ a small percentage of enumerators completed questionnaires. . . for an individual or multiple households from the curb, without actually conducting an interview or checking the accuracy of their 'guesses.' This practice was motivated, in part, by the requirement to meet quotas or payment for work done on a 'piece-of-work' basis.”  Tighter controls on the enumeration process helped deal with the problem of “curb stoning” during the transition to the current census-by-mail system. Could this be why I haven't succeeded in locating a few of my people in the census? It's possible, but statistically unlikely. I guess I'll just keep hunting.
Fun fact: Michigan ranked 7th in population in 1940 with about 5.2 million inhabitants. For those of you who want to know, the top six were: New York, Pennsylvania, Illinois, Ohio, California and Texas.
As of April 27th, the indexing of Michigan at Family Search was 7% complete (last night I did a batch from Cass County). If we all do a little indexing we can knock out the 149,720 images for Michigan in . . . well. . . every little bit helps.