We descriptively evaluated two published Vietnamese surname lists that were developed using different approaches. One of these lists (Lauderdale) was developed systematically using automated databases, while the other list (VCHPP) was developed using a community-based participatory approach [18
]. Interestingly, these two lists performed very similarly when applied in metropolitan Seattle. Specifically, the VCHPP list identified 4,283 potentially Vietnamese households, and 79% of the households with established ethnicity were Vietnamese; and the Lauderdale list identified 4,068 potentially Vietnamese households, and 80% of the households with established ethnicity were Vietnamese.
Mateos has recently summarized the potential limitations of using surname lists to identify ethnic population subgroups. Temporal differences in name distributions between a reference population (used to develop a list) and a target population can occur because of different migration waves and variations in geographic distribution patterns over time. Additionally, regional differences in the frequency distribution of names can occur between and within countries because of differential historic processes and migration flows [7
]. For example, the frequencies of Pakistani names occurring in the north and south east of England differ [10
]. However, the distribution of Vietnamese surnames in our recent Seattle study did not differ meaningfully from the distribution of Vietnamese surnames in the Lauderdale study that used a list of Vietnamese Americans who were born in North or South Vietnam before 1941 [18
The positive predictive value of any individual Vietnamese surname (the proportion of people with the name who are actually Vietnamese) depends on the proportion of Vietnamese in a particular geographic area [18
]. Our results indicate that some of the names on the VCHPP and/or Lauderdale lists also occur relatively frequently in other Asian ethnic groups. For example, less than one-half (45%) of the households in the original sample with the name Lam (which is also a relatively common Chinese name) were actually Vietnamese [9
]. If a study is being conducted in an area where the Asian population is predominantly Vietnamese, the positive predictive value of names such as Lam will be higher than if a study is being conducted in a geographic area where multiple Asian groups are represented.
In our study, the household surname and survey respondent surname could have been concordant because a husband and wife had the same commonly occurring name (e.g., Nguyen) prior to marriage, a woman had adopted the western custom of changing surname upon marriage, an adult daughter (with the same name as her father) completed the survey, or the household was headed by a woman. However, in 912 cases the household surname and respondent surname were discordant. Of the 912 respondents with a name that differed from the corresponding household name, only 81 (9%) had a name that is not on the VCHPP list or Lauderdale list. Further, none of these names occurred more than six times. This suggests that our combined VCHPP and Lauderdale list identifies a high proportion of Vietnamese Americans and includes all commonly occurring Vietnamese surnames.
Researchers rarely know whether persons with names that are on an ethnic surname list differ from persons of the same ethnicity with names that are not on the ethnic surname list, and few studies have addressed this issue. However, one study found that Korean Americans with the very common Korean surname of Kim did not differ demographically from all Korean Americans [22
]. Similarly, we found that the demographic characteristics of Vietnamese Americans with the very common Vietnamese surname of Nguyen, one of the top 15 Lauderdale names, one of the Lauderdale names, and one of the VCHPP names were very similar. This finding suggests that the demographic characteristics of Vietnamese Americans do not differ meaningfully by surname, and researchers could use a subset of commonly occurring Vietnamese surnames without introducing bias.
A recent analysis of 2000 Census data found that 94% of Americans with the surname of Nguyen are Vietnamese [23
]. In our study, only 84% of households that were identified using the surname of Nguyen were Vietnamese American. One limitation of our study is that some households probably moved between the time we conducted our sampling procedures and attempted to contact study households (the time lag ranged from one to 6 months). For example, a non-Vietnamese family may have replaced a Vietnamese family at a particular address. Therefore, it is likely that our study under-estimates the performance of Vietnamese names in identifying individuals of Vietnamese descent. Further, the interviewers were unable to establish the household ethnicity for 1,177 of the residential addresses in our original sample (because they were unable to access a secure complex or unable to contact a household after five attempts). However, there is no reason to believe that mobility (the frequency of address changes), residence in a secure apartment or condominium complex, and the likelihood of not being at home is related to Vietnamese surname. Finally, our study did not include households that have unlisted land phones or exclusively use cell phones. Younger people (<35 years of age) are more likely to use a cell phone as their primary telephone than older people (35 years of age or older) [24
New Contribution to the Literature
If accuracy (being sure of a person’s Vietnamese ethnicity) is the most important consideration in a study, researchers might want to consider using a subset of surnames from the VCHPP and Lauderdale lists. For example, they could use the following 12 commonly occurring Vietnamese names that are all associated with a high positive predictive value: Nguyen, Tran, Le, Pham, Huynh, Do, Bui, Truong, Hoang, Phan, Vo, and Vu. In contrast, if detecting the highest possible number and/or proportion of Vietnamese in a particular geographic area is the most important consideration in a study, researchers might want to use all the surnames that are included on the VCHPP and Lauderdale lists.