*3.3. Data Collection*

We conducted on-site trials of smart safety seats in shopping malls and auto 4S stores to understand consumers' feelings better. Consumers were required to fill out questionnaires after the trials. We have included a detailed introduction to the smart safety seat in the first section of the questionnaire for users who cannot enter the place to try out the smart safety seat. Consumers can clearly understand the smart safety seat's function, appearance, and characteristics thanks to these introductions. Respondents were required to complete the questionnaire.

Respondents were asked to complete the questionnaire online through Wechat or a browser. Professional research institutions that set the same IP address cannot submit the questionnaire repeatedly, and the same Wechat can only participate in the questionnaire once to improve the validity of the data. Wechat accounts and suspicious IP addresses have been blocked. All respondents were informed that the information was private and would not be publicized. They voluntarily completed the questionnaire and were rewarded with 2 yuan (CNY) in exchange. We collected 1152 questionnaires between March and July 2022. We rescreened the data, removing questionnaires with the same score for all options, and obtained 1057 valid responses.

Table 1 provides an overview of the respondents' age, gender, educational background, household income, and frequency of trips with children. Of the respondents, 38.7% (409) were female, and 61.3% (648) were male, both of whom were parents of existing children. The majority, about 77.6%, were reported to be between the ages of 26 and 45. More than 90% had higher education. The largest proportion of households with an annual household income of RMB 100,000–300,000 yuan was 46.2%.


**Table 1.** Demographic profile of respondents (N = 1057).

#### *3.4. Data Analysis*

Using Smart PLS 2.0, partial least squares structural equation modeling was performed. Smart PLS's strengths lie in its flexibility [113]; it can be applied to the analysis of non-normal data or studies with small sample sizes, as well as to the analysis of more sophisticated multi-order latent variable models and the exploration of novel models [114,115]. The model is complicated since it is a second-order model with seven latent variables, and the data used in this study are not strictly normally distributed. This exploratory model has not been the subject of any prior research. The aforementioned considerations led to the choice of the partial least squares structural equation modeling method for this investigation.

The useful data were examined in three stages; the first two steps are analyzed according to the two-step method proposed by Anderson and Gerbing [116].

Step 1: Descriptive statistics were run on the population data. Regarding the scale, the correlation between latent variables and observable variables was used to assess measurement models. Reliability tests and validity tests were run on the data. The validity tests were further broken down into convergent and discriminant validity tests.

Step 2: Examine the structural equation model, which comprises the paths that latent variables take to interact with each other. Pay special attention to the regression weight and significance level, as well as the amount of variance that these latent variables explain. Validity assessment of structural models using the blindfolding procedure was conducted. Step 3: Out-of-sample prediction [117,118].
