Historical changes in baby names in China

Dear Dr. Ogihara,

Thank you for proposing these suggestions to our article. I agree that answering these three questions could help clarify the implications of our findings and facilitate the understanding of cultural changes in China. As the first author of this article (Bao et al., 2021), I am responsible for providing a reply to address your concerns. Overall, it is important to note that no research is perfect without limitations. While there are studies using more comprehensive data on baby names to test cultural changes in the United States (Twenge et al., 2010, 2016) and France (Mignot, 2022), the complete annual baby names data for the other countries (e.g., the United Kingdom, Japan, and China) are difficult if not impossible to access. Thus, the 2005 China’s 1% Population Census data used in our study (Bao et al., 2021) is still the most qualified dataset to our knowledge. Below is my detailed point-by-point rebuttal.

1. Were the samples of older birth cohorts truly representative?

Let me first reiterate the data source as described in Bao et al. (2021):
To obtain a nationally representative sample of Chinese names covering a long period, we accessed data from the 2005 China’s 1% Population Census (National Bureau of Statistics [NBS] of China, 2005). The 2005 China Census was conducted using a three-stage stratified cluster sampling method, with respondents randomly selected from each of 340 prefectural-level cities or regions in China. Our sample was a random subset (N = 2,585,481) drawn by the NBS, which had been widely used in previous economic and population research.

Notably, the 2005 China Census was conducted by the National Bureau of Statistics of China, and the random subset we used was also drawn by the NBS. Given the three-stage stratified cluster sampling method for the census, it is reasonable to describe the 2005 China Census as a “nationally representative sample” in terms of its high representativeness of the Chinese population alive in 2005. Indeed, we highlighted its representativeness of the Chinese population but did not overstate its representativeness of any specific birth cohort.

More importantly, four points deserve special consideration, rendering the doubt about the representativeness of older birth cohorts unwarranted or less necessary.

First, although the 2005 China Census is a “one-shot cross-sectional survey” rather than a “cross-temporal survey”, this would not necessarily bias the research on names. The main reason for research using cross-temporal surveys (e.g., cross-temporal meta-analysis) to study psychological changes is to disentangle cohort effects from age effects in participants’ self-report responses. However, names are not a self-report measure and thus unlikely to be confounded with age effects. While a small proportion of people may change their names over the lifetime, most people retain their birth names throughout their lives. Therefore, it is not much problematic to use cross-sectional name data.

Second, it should be clarified that the Study 2 in Cai et al. (2018) was also based on cross-sectional (but not cross-temporal) name data, because “the name database in the Chinese Public Security Department” (Cai et al., 2018, p. 3) only included people alive in that year but did not include the deceased population. Chinese who have died are usually removed from the household registration records in the Chinese Public Security Department. Therefore, the large sample of 2 million names with at least 2,000 names per year (Bao et al., 2021) was certainly more representative, reliable, and valid than the small sample of 600 names with only 10 random names per year (Cai et al., 2018).

Third, Ogihara mentioned that “the average life expectancy in China in 2005 was approximately 73 years”, which might introduce selection bias due to the health and wealth of older people who managed to live longer. The rationale for this argument is probably that name uniqueness can dramatically affect, or at least be strongly associated with, one’s (physical) longevity and (economical) prosperity, which actually lacks empirical support. Moreover, it is critical to think about what “reality” is when speaking of older generations. One important factor that Ogihara did not mention, but which has caused the unexpected deaths of Chinese born before 1945, is war. In the Chinese People’s War of Resistance Against Japanese Aggression (1931~1945), tens of millions of Chinese people were killed. How do we account for the possible impact of the war on the representativeness of the older Chinese? Is including or excluding the Chinese who died in the war more likely to reflect the reality?

Finally, and reciprocally, a critical scholar may also ask, for instance, whether the name samples used in Ogihara et al. (2015) were truly representative. Specifically, Ogihara et al. (2015) retrieved Japanese names from two companies: one was from the “Benesse Corporation”, which collected “baby names from customers who bought their products”; the other was from the “Meiji Yasuda Life Insurance Company”, which collected “baby names from people holding insurance policies with the company”. In both cases, the problem of unrepresentativeness seemed much more serious because the two samples were likely drawn from wealthier people who had money to buy these products and insurance. It would also be important to address this systematic selection bias in the samples of Japanese names.

Nevertheless, it is safer to claim that the 2005 China Census is at least representative of Chinese born between 1949 and 2005, and that the results before 1949 are exploratory or preliminary but still informative for understanding Chinese cultural changes over a longer time span. We hope that future research can access more comprehensive data on baby names in China.

2. Did unique names increase only after the 1970s?

Simply put, yes, but the shifting pattern of name uniqueness varies across regions within China, which warrants future research and explanations.

In both articles (Bao et al., 2021; Cai et al., 2018), despite the huge difference in sample sizes, we consistently found that name uniqueness in China began increasing after the 1970s. This is possibly a result of the China’s Reform and Opening-Up Policy since 1978. Therefore, the statement that “the study did not replicate the previous finding” is unwarranted and incorrect. We did replicate the previous finding by using two similar indices of character uniqueness (in the third point below, I will explain why the name length indices were different from character uniqueness and improper to be used to indicate name uniqueness).

Nonetheless, this is only an overall result at the national level. My unpublished findings, which are also based on the 2005 China Census, indicate that the patterns of historical changes in name-character uniqueness actually varied across different regions within mainland China. While there was a continuous sharp increase in name uniqueness after the 1970s (consistently since 1978) in all regions of mainland China, I found that from 1920 to 1978, name uniqueness (1) increased steadily in Northeast China, (2) remained almost unchanged in North and Southwest China, and (3) decreased in South, Central, East, and Northwest China.

I will refrain from reporting and explaining more here, as these results remain unpublished. But a more comprehensive understanding of Chinese cultural changes requires distinguishing between different regions within China, since China is a country spanning about 9.6 million square kilometers and there are substantial within-culture variations in China.

3. How are the historical changes in average name length interpreted?

The changes in the four indices of name length were not sufficiently interpreted in our article (Bao et al., 2021). Here I provide further probable explanations for the increased proportion of one-character given name and the decreased average name length (note that these two indices should be understood as reverse because the average name length was largely determined by the relative proportions of one-character and two-character names).

The main reason for this pattern may be the joint effect of the China’s One-Child Policy (1979~2015) and the declined use of “generation names” in China (probably since the 1960s), both of which are factors unrelated to the need for uniqueness, making the indices based on name length less representative of cultural uniqueness and individualism. Take my own family as an illustrative example: my grandparents have three children born in the 1960s and 1970s, and all of them have one-character given names (i.e., Yue, Tai, and Dai). This actually deviates from the traditional naming practice in China of using “generation characters” in given names. We have also mentioned this in our article (Bao et al., 2021, p. 2): “Traditionally, the majority of Chinese given names consist of two characters: one represents the generation in a family and the other denotes the unique identity of a person (Zhu and Millward, 1987; see also Figure 1).” Meanwhile, the China’s One-Child Policy also matters. It is likely that families with only one child no longer need to use generation names for multiple children, which may explain why the proportion of one-character names reached its highest level in the 1980s. In addition, the reason why the proportion of three-character names increased but the proportion of one-character names decreased from 1990 to 2005 may be that people began to be aware of the high probability of name duplication if their names only had one character, and thus they pursued two-character or even three-character baby names. Indeed, this is a plausible explanation based on my observations and informal interviews of my family members and peers.

All of these plausible interpretations, however, require more rigorous empirical examination. We hope that future research can empirically test the impact of political policies on naming practices in China. Nonetheless, because name length indices are more susceptible to such social or political impacts, they are less appropriate to indicate cultural emphasis on uniqueness or individualism. Therefore, we suggest that name-character uniqueness is the most appropriate index for research on Chinese cultural change.

I hope that the above responses have largely addressed your concerns and can enhance the understanding of historical changes in baby names and related cultural phenomena in China.

Best regards,
Han-Wu-Shuang Bao