A falling birth rate could lead to widespread social and economic problems. Like many countries in the post fertility transition, China is facing a precipitous decline in its population, setting the stage shortly for potential demographic and economic issues. According to a recent report by the Chinese Academy of Social Sciences [1
], China’s total fertility rate has been below 1.6 since 1996, which is lower than the replacement level fertility rate of 2.1. After a brief uptick from 2013 to 2016, the birth rate fell again in 2017, with 17.2 million babies born compared to 17.9 in 2016. Although the Chinese government has recognized the worrisome demographic trend and in 2015 abolished the one-child policy (replaced by a two-child policy) [2
], the overall number of births continued to drop. It has been estimated that the total Chinese population will present negative growth around 2028, though others believe it would come sooner or has already begun [1
Declining populations are creating many social problems in countries where this phenomenon is occurring [3
]. To address this issue, some scholars and practitioners have exerted efforts to understand the mechanism of fertility promotion globally [6
]. Smart technology drives healthcare and social development more than any other force [11
]. For instance, robot-assisted therapy has been developed in the care of older adults with dementia [14
]. Matarić et al., [15
] showed robot coaches effectively aid stroke patient rehabilitation by providing social and physical assistance. Recently, a trend for developing baby robots has emerged as a means of encouraging people’s willingness to become “parents.” It has been suggested that a baby robot affording the simulating experience of mothering and fathering is effective in influencing the perceptions regarding the significance of having a child and parenting behavior [16
]. The practice of developing baby robots, however, is implemented with both hopes and questions [17
]. Whether it improves the willingness of parenting or not is uncertain for empirical researchers.
The attempt of baby robots as infant simulators for fertility education can be traced to the early pilot program of virtual infant parenting [19
]. It was initially aimed to prevent teenage pregnancy in Australia and the US, nevertheless, controversial effects have been reported. For instance, a recent longitudinal experiment showed that the teenage girls who have access to an infant simulator were more likely to encourage childbearing than the control group [18
]. Besides, some studies revealed an insignificant difference between the student group with infant simulators and the group without the simulators [19
]. Nevertheless, the design-related factors, such as the visual appearance of infant simulators, have been much neglected in previous research investigating the consequent effects of virtual infant parenting. As relevant evidence revealed in robot research, visual appearance is influential in settings where it is preferred to have the user expect humanlike performance from the robot [20
]. Besides, more recent research suggests the phenomenon of “Uncanny Valley”(UV) in designing the visual appearance of humanoid robots [24
]. Its consistency in the application of baby robots is ambiguous.
Based on the well-defined theory of planned behavior in fertility research [25
], this study aims to investigate whether a baby robot performs as a prominent antecedent of fertility intention in China, and how this relates to its visual appearance. Our empirical observation as a design guideline can enrich the current literature on fertility promotion from an emerging perspective of baby robots.
MANOVA and Logit Regression were performed to analyze the data through SPSS 22.0. There were no missing or incomplete responses in the data set. Consistent with our predicted extent of human resemblance in the robot stimuli, manipulation check confirmed the extent of human resemblance gradually increased from No.1 robot to No.6 robot (Mean = 2.50, 2.57, 2.70, 3.53, 4.53, 7.00, respectively; SD = 1.78, 2.39, 2.20, 2.66, 3.11, 2.46, respectively). ANOVA analysis showed a significant difference in the human resemblance between No.1 and No.5/6 (both p < 0.05), suggesting experiment manipulation was successful.
As for the main analysis, a MANOVA analysis on fertility attitude, subjective norms and perceived behavioral control was conducted. Table 4
shows the means, SD (standard deviations), and Pearson correlations of different factors.
With respect to fertility attitude, one-way ANOVA results suggested a significant difference between different stimuli (Mean for No.1–No.6 robot = 6.80 vs 6.83 vs 4.90 vs 5.70 vs 7.03 vs 7.17, respectively; SD for No.1–No.6 robot = 2.06 vs 2.55 vs 2.92 vs 2.26 vs 1.77 vs 1.64, respectively; F (5, 174) = 4.863; p
<0.05; see Figure 4
and Table 5
). Consistent with the theory of UV [23
], people’s fertility intentions did not follow a linear trend. Instead, it dropped suddenly when the level of the robot’s human-resemblance increased to a certain level (No.3). Then, it gradually increased as the level of robot’s human-resemblance kept increasing (No.4/5/6). The post-hoc analysis showed people tended to have a significantly lower fertility attitude in No.3 robot compared to No.1/2/5/6 robot.
Moreover, different robot stimuli did not significantly influence people’s subjective norm (for family: F (5, 174) = 0.730, ns; and, for peers: F (5, 174) = 0.200, ns) and perceived behavioral control (for finance: F (5, 174) = 1.277, ns; for energy: F (5, 174) = 2.225, ns). Figure 5
and Figure 6
show people’s subjective norms and perceived behavioral control under different stimuli.
As for the fertility intention, a binary logistic regression was introduced to explore the factors that contribute to fertility intention. According to the theory of planned behavior, an empirical analysis was further developed to investigate the effect of fertility attitude, subjective norms (family and peers), and perceived behavioral control (finance and energy).
shows the result of this regression. The logistic regression was statistically significant (chi
-square (8) = 31.445; p
<0.05). The current model accounted for 21.5% of the variance in fertility intention with 68.9% identification accuracy. According to the result, fertility attitude was positively associated with fertility intention. While subjective norms from peers significantly contributed to fertility intention, subjective norms from family did not have a significant impact on fertility intention. Similarly, while perceived behavioral control of finance was positively related to fertility intention, behavioral control of energy only showed a marginal influence on fertility intention. In addition, age worked as a significantly negative indicator of fertility intention. Men showed somehow greater fertility intention compared with women however the effect was only marginal. Besides, no other significant effects were found.
Given the ambiguous relationship between the visual appearance of baby robots and people’s fertility intention, this paper introduced a behavioral experiment approach and tried to obtain a deeper understanding of the effect of a baby robot on people’s fertility behavior. Based on the decomposed theory of planned behavior, results show that the visual appearance of a baby robot could increase people’s fertility attitude. Under a temporal visual stimulation, the uncanny valley effects in baby robot design are associated with people’s fertility attitude, in which the degree of human-resemblance in a baby robot and fertility attitude followed a U-shape relationship. In other words, people tend to have a temporally higher fertility attitude when facing the baby robot with the most or the least human features, compared with the baby robot with the medium human features. However, the temporal visual effects of baby robots do not significantly contribute to the subjective norms (family and peers) and perceived behavioral control (finance and energy). This result in the current study is consistent with prior research: the temporal visual reaction towards stimuli would be more associated with individual cognition and attitude formation, rather than with long-term social cognition and control beliefs [66
]. Subjective norms are the cultural products of acceptable group conducts [68
], indeed, they are formed in the context of society and culture, representing informal knowledge that governs the conduct of members in the society [70
]. Perceived behavioral control is a subjective assessment of easiness and difficulty in the performance of behavior [57
], which is hardly determined by visual stimuli [55
Regarding the relationship between fertility attitude, subjective norms, perceived behavioral control, and fertility intention, results show that fertility attitude is a significant indicator of promoting fertility intention. This is consistent with the prior research [25
]. Besides, based on the decomposed theory of planned behavior [54
], this study subdivides the subjective norms and perceived behavioral control into two sub-constructs, family and peers, and finance and energy. Results indicate that, in the context of the Chinese population, subjective norms from peers is significantly associated with fertility attitude while subjective norms from family do not significantly influence fertility attitude. To specify, people who considered the opinions of their peers more seriously tend to have a lower fertility intention. The reason might lie in that the young generation in China might consider the responsibility to have a child as an old-fashioned idea [25
]. Similarly, fertility attitude was significantly associated with perceived behavioral control of finance while it was marginally related to perceived behavioral control of energy. People who had more monetary resources tended to have a stronger intention to have children, which is also consistent with previous literature [58
Moreover, age was also a reliable indicator of fertility intention: older people tended to give up having children more easily. Gender tends to work as a marginal factor for fertility intention: compared with men’s fertility intention, women showed a marginally weaker intention to have children. That might be caused by Chinese women’s increased education level and work participation [71
This study tended to have the following theoretical contributions. To begin with, although some prior research tried to address the antecedents of fertility intention and birth-intervention policies in the context of China [25
], they largely neglected to address this issue by introducing the latest technology applications, an emerging trend for adopting baby robots to encourage people’s fertility intention. By exploring the relationship of UV effect in baby robots and fertility intention, the current research examines an emerging way to increase people’s fertility attitude, thus promoting fertility intention at the end. Furthermore, previous research on the theory of planned behavior has treated the subjective norms and perceived behavioral control as a single construct. Based on the decomposed theory of planned behavior, the current study subdivided subjective norms and perceived behavioral control into sub-constructs and explored their influence on fertility intention in the context of China, trying to achieve a detailed picture of Chinese fertility behavior.
This study also has some limitations. First, this research explored the relationship between the uncanny valley and people’s fertility intentions, providing the preliminary guidelines for robot designers. The robot No.6 was adapted from a real infant picture which by far cannot be made at the current state of the robot industry. Thus, it might work as a confounding factor since people might be biased due to its realistic appearance. Future studies should try to use different actual social robots to validate the current result.
Second, there are many other features of a social robot that could be explored and work as promoting factors for people’s fertility intention, such as specific facial features [73
], expression [74
], and related other traits, which future study should try to explore.
Third, the main goal of this study was to explore the effect of the baby robot on fertility intention. Though this study recruited sufficient participants to take part in the experiment, the sample size might be relatively small when exploring the relationship between fertility attitude, subjective norms, perceived behavioral control, and fertility intention. Future studies should try to introduce a sizeable domestic survey to validate the conclusions in the current study.
Fourth, the measurement of behavioral control and fertility intention might be limited. The current study did not differentiate in the difference between the extensive and intensive margin of fertility, which might work as a cofounding factor. Further studies should try to examine the factors in detail and validate the current finding. Though the current study adapted the items from two perspectives, namely finance and physical energy [54
], there might be more emerging factors to influence people’s fertility ability, such as parenting skills [75
] and related attitudinal factors [76
], which could be included in future studies.