Attanucci (1974), in a survey of commuter carpooling in Boston, places the percentage between 25-45%, with co-workers representing another 50-70%. Kendall (1975), in a survey of Boston commuters found that 35% of carpools were intra-household. However, recent sources place the percentage of commuting involving family members much higher. Pisarski (2006) suggests the percentage is closer to 80% but provides no supporting evidence. Li (2007) in a survey of Texas carpoolers found that approximately 65% were family members, with co-workers representing another 30%. Morency (2007) using survey data from Montreal, Canada found that 82% of carpoolers were family members with another 9% representing co-workers.
Note that the Attanucci and Kendall surveys, while several decades earlier, were surveys of commuter carpoolers specifically, and found a much lower share of household members and a much higher share of co-workers. The Li and Morency surveys focused on all trip types and found much higher intra-household participation.
Intuitively, it is expected that commuting trips would involve smaller percentages of family members than trips for other purposes. Carpool statistics from MIT tend to support this; approximately 59 of 234 employees (25%) with registered carpool parking permits are family members. Note that this percentage is an understatement of true “fampools” at MIT, because employees that are dropped off at campus by a family member cannot be identified through parking permit registration. However, the statistics certainly suggest that “fampooling” for commuting trips is much lower than the 65-80% quoted elsewhere.
Beyond considerations of trip type, it is clear that the percentage of carpooling occurring between unknown participants is rare; Attanucci (1974) found 3%, Li (2007) found 3-5% and Morency (2007) found approximately 9%.