Validation of the Chinese version of the psychological safety in high-fidelity simulation scale: a culturally compatible methodological study Authors: Yue Yi Li, Mio Leng Au, Lai Kun Tong, Im Ha Fong, Xing Hua Ren, Liying Cheng Published: 2025 ABSTRACT BackgroundCurrently, there is a lack of a specialized scale in China for assessing the psychological safety of nursing students engaged in High-fidelity Simulation (HFS). The aim of this study was to translate the Psychological Safety in High-Fidelity Simulation Scale (PSHFSS) into Chinese and evaluate its psychometric properties.MethodsThe translation and cross-cultural adaptation strictly followed the TRAPD (Translation, Review, Adjudication, Pretesting, and Documentation) protocol. A cross-sectional survey was conducted using the Chinese version of the PSHFSS (CV-PSHFSS) with convenience sampling of nursing students. Psychometric evaluation of the CV-PSHFSS was based on Classical Test Theory and Item Response Theory.ResultsA total of 522 participants was analyzed, with item-level content validity index (CVI) ranging from 0.8 to 1.0 and a scale-level CVI of 0.96. Confirmatory factor analysis yielded a goodness-of-fit test of 2.5, root mean square error of approximation of 0.054, the goodness-of-fit index of 0.963, comparative fit index of 0.982, and incremental fit index of 0.982. Cronbach’s α coefficients for the dimensions of dealing with uncertainty, being exposed, being unsupported, interpersonal risk, and the overall scale were 0.820, 0.885, 0.888, 0.896, and 0.927, respectively. Meanwhile, the McDonald’s omega coefficients for the same dimensions—dealing with uncertainty, being exposed, being unsupported, interpersonal risk, and the overall scale—were 0.821, 0.882, 0.890, 0.897, and 0.925, respectively. The intraclass correlation coefficient (ICC) of the CV-PSHFSS was 0.925, and that of the items ranged from 0.715 to 0.855. Infit mean-square (MNSQ) values ranged from 0.69 to 1.24, and outfit MnSQ values ranged from 0.69 to 1.37. Point-measure correlations were between 0.59 and 0.77. Item difficulty measures ranged from − 0.55 to 0.85 logit, while individual psychological safety ranged from − 5.15 to 4.95 logit. Differential items functioning contrast ranged from − 0.24 to 0.28 logits, with all p-values exceeding 0.05.ConclusionsThe CV-PSHFSS has satisfactory reliability and validity among Chinese nursing students. Nursing educators are encouraged to use this scale to assess students’ psychological safety in HFS, thereby improving learning outcomes and engagement.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12912-025-04203-5. Background High-fidelity simulation (HFS) has been widely applied in nursing education due to its effectiveness in enhancing nursing students’ knowledge, skills, critical thinking, collaboration, and caring capabilities [1–3]. However, the educational effectiveness of HFS is influenced by various factors, including simulation design, facilitators, and the individual characteristics of students [4]. Psychological safety is one of students’ individual characteristics. It refers to individuals’ psychological state of feeling safe and able to express their thoughts, perceptions, and opinions without fear of retribution or embarrassment [5, 6]. In the context of HFS, psychological safety is the sense of security and confidence that allows participants to express their feelings, perceptions, and opinions without feeling judged, embarrassed, or reprimanded [7, 8]. According to the NLN Jeffries Simulation Theory, students with psychological safety in HFS are more actively engaged in scenarios, communicate effectively with their peers and facilitators, and are willing to take risks without fearing negative consequences [4, 9, 10]. Furthermore, psychological safety encourages students to think beyond task completion and engage in critical thinking and ongoing reflection during simulations, and fosters a dynamic learning process, ultimately improving student experiences and outcomes [4, 9, 11]. Conversely, students who lack psychological safety may experience anxiety and pressure, disrupting their memory and leading to physical and mental fatigue, which negatively impacts their learning outcomes [12, 13]. Therefore, the psychological safety of students in nursing simulations can thus significantly affect their learning outcomes. In recent years, many researchers have focused on exploring effective measures to ensure psychological safety for students in HFS, aiming to better enhance their learning outcomes [14–16]. Nevertheless, it is important to note that before evaluating which measures are more beneficial for improving nursing students’ psychological safety, a dedicated scale for assessing their psychological safety in HFS must be developed [17]. Such a scale would enable an objective evaluation of the effectiveness of various measures. Unfortunately, there is currently no scale in China specifically designed to assess the psychological safety levels of nursing students in HFS, which somewhat limits regional comparisons. Therefore, the development of an appropriate scale is essential for accurately determining the psychological safety of nursing students in HFS. The Psychological Safety in High-Fidelity Simulation Scale (PSHFSS) was developed by Korean scholars to evaluate the psychological safety of nursing students during HFS [17]. The items and dimensions of the initial PSHFSS were formed based on a comprehensive literature review and a focus group interview regarding the experience of psychological safety in simulation-based learning among nursing students [16, 17]. After conducting exploratory factor analysis, 14 items were retained across four dimensions: dealing with uncertainty, being exposed, feeling unsupported, and interpersonal risks (Supplementary Table 1) [16, 17]. A strong correlation exists between these dimensions and the four factors identified in previous scholarly conceptual assessments of psychological safety in healthcare: structural factors, system factors, interpersonal factors, and individual factors [16–18]. Moreover, the PSHFSS scores show statistically significant positive correlations with the Academic Safety Assessment tool and the Educational Satisfaction Scale in Simulation, with a Cronbach’s coefficient of 0.91 [17]. Due to the robust structure and the strong reliability and validity of the PSHFSS,, it effectively assesses the psychological safety of nursing students in HFS, which has led to its translation into other languages and widespread use [14, 19–22]. To better assess the psychological safety of nursing students in HFS and to achieve objective comparisons between regions, the aim of this study was to translate the Korean version of the PSHFSS into Chinese and evaluate the psychological characteristics of the Chinese version of the PSHFSS (CV-PSHFSS). Given that Classical Test Theory (CTT) is primarily used for straightforward assessments of test reliability and validity, focusing on overall test performance, while Item Response Theory (IRT) offers a nuanced understanding of item characteristics and individual differences, the psychological characteristics of the CV-PSHFSS were evaluated using both CTT and IRT. Methods The STROBE statement was utilized to guide the reporting of this cross-sectional study [23]. A checklist based on the STROBE statement is provided in related material. Two phases of this study were conducted: (1) translation and cross-cultural adaptation, and (2) psychometric properties test. Phase 1: Translation and cross-cultural adaptation The translation process rigorously followed the team’s translation protocol TRAPD (Translation, Review, Adjudication, Pretesting and Documentation) to maintain semantic consistency with the PSHFSS [24]. The process included the following five steps: (1) Translation: The Korean version of the PSHFSS was independently translated into Chinese by two bilingual experts residing in China. Both translators hold doctoral degrees and have lived and studied in South Korea for over five years. One translator is a nursing educator with over 20 years of experience in nursing education, while the other specializes in Korean translation and has 15 years of teaching experience in China. The translators have both received training in questionnaire translation and have been thoroughly briefed on this project’s characteristics; (2) Review: A panel consisting of two translators, as well as specialists in nursing, education, psychology, sociology, epidemiology, and statistics conducted a comprehensive review to identify differences between the two translated versions and the original version. This process ultimately resulted in the creation of an initial version of the CV-PSHFSS; (3) Adjudication: An adjudicator, well-versed in the translation project, methodologies, psychological safety, and HFS, but not directly involved in the translation process, worked alongside a nursing expert to finalize the consensus version of the CV-PSHFSS; (4) Pretesting: Ten nursing students were invited to complete the consensus version of the CV-PSHFSS and participated in cognitive interviews to identify any expressions that were difficult to understand; (5) Documentation: Documentation is included throughout the entire TRAPD process, comprising the different translation versions, discussions between the two translators and experts, the consensus reached by the adjudicator and nursing expert, as well as feedback from nursing students during the pretesting phase. Phase 2: Psychometric properties test Settings Due to the relatively high cost of high-fidelity simulation equipment, not all institutions have access to these resources [25], leading to an uneven distribution of qualified nursing student participants in the study. Therefore, it is essential to take these economic disparities into account when selecting sample sources. Based on economic development levels, mainland China can be divided into four regions: Northeast, East, Central, and West. In 2024, the gross domestic product (GDP) is 13,490.835 billion yuan, with the Northeast, East, Central, and West regions accounting for 4.7%, 52.4%, 25.5%, and 17.3% of the national GDP, respectively [26]. Additionally, considering the economic differences between the Special Administrative Regions and mainland China, this study has selected Macao, one of the Special Administrative Regions, as a source for the samples [27]. To ensure the sample’s representativeness, dedicated coordinators were appointed in the aforementioned regions to promote the research and facilitate the collection of samples. Participants Convenience sampling was employed in the present study. To validate the reliability and validity of the CV-PSHFSS, students were included based on the following criteria: (1) they must be Chinese nursing students, (2) aged 18 years or older, (3) have previously participated in HFS, and (4) voluntarily agree to participate. Those who were unable to understand or produce written Chinese were excluded from the study. As CTT and IRT will be utilized to evaluate the psychometric properties, a sample size of 400 students is believed to be adequate [28]. In validating content validity, ten experts from various fields were invited to enhance the CV-PSHFSS by aligning it with Chinese expressions and cultural practices, utilizing their professional expertise and experience. Each expert possesses a doctorate and specializes in nursing, education, medicine, psychology, social sciences, statistics, or epidemiology. Data collection Data collection for this study was conducted using Wenjuanxing (https://www.wjx.cn/), a prominent online questionnaire platform in China. The research team began by developing promotional materials, including posters with QR codes that directed participants to the electronic questionnaire. Research coordinators in each region were tasked with disseminating study information through various social media channels to maximize outreach. To maintain data integrity, responses were limited to one per IP address. The data collection period spanned from June 6 to July 28, 2025. Following the completion of data collection, the research team conducted a comprehensive review and cleaning of the data, which included the removal of incomplete or inconsistent responses to guarantee the accuracy and reliability of the final dataset. Statistical analysis The data were analyzed using SPSS 29.0, AMOS 22.0, and Winsteps 3.74.0. Descriptive statistics were applied to assess the baseline characteristics, with a two-tailed p-value of less than 0.05 marking statistical significance [29]. CTT analysis CTT analysis CTT analysis The analysis of validity encompassed both content and construct validity. Content validity was evaluated using the Content Validity Index (CVI). This assessment included both item-level CVI (I-CVI) for individual item and scale-level CVI (S-CVI) for the overall scale. The I-CVI is determined by the proportion of experts who rated its relevance as 3 or higher on a 4-point Likert scale (1 = not relevant, 2 = somewhat relevant, 3 = quite relevant, and 4 = highly relevant), while the S-CVI is calculated as the average of the I-CVI values [30]. Those with an I-CVI greater than 0.7 and an S-CVI greater than 0.8 were considered valid [30]. Since the original scale has clear dimensions, this study employed confirmatory factor analyses (CFA) to test the construct validity of the four dimensions [17, 31]. In CFA, model modifications were guided by statistical modification indices as well as theoretical justification. The reasonable criteria of indices were as followings: a goodness-of-fit test (CMIN/DF) < 3.0, root mean square error of approximation (RMSEA) < 0.08, the goodness-of-fit index (GFI), comparative fit index (CFI), and incremental fit index (IFI) > 0.95 [32, 33]. The reliability of the CV-PSHFSS was assessed through the analysis of its internal consistency and external reliability. Internal consistency was evaluated using Cronbach’s alpha and McDonald’s omega (Ω), with values exceeding 0.7 deemed acceptable for establishing internal consistency [34]. To assess external reliability, a test-retest was carried out with thirty participants two weeks after the initial assessment. The conditions for the second test were intentionally structured to closely resemble those of the first assessment. Specifically, participants completed the CV-PSHFSS immediately prior to their second high-fidelity simulation exercise. Both simulation exercises were designed to maintain a similar level of difficulty, which facilitated a fair comparison of psychological safety levels across the two assessments. The intraclass correlation coefficient (ICC) was calculated to evaluate external reliability, with a minimum acceptable value set at 0.70 [30]. (2)IRT analysis (2)IRT analysis IRT analysis In this study, the Rasch Rating Scale Model (RRSM) of IRT was utilized to estimate both item and person parameters. Prior to applying the Rasch model, it is essential to confirm that the assumption of unidimensionality is satisfied. Unidimensionality refers to the requirement that each item measures a single concept or construct [35]. To assess this assumption, principal component analysis (PCA) of the Rasch residuals was conducted [35]. For unidimensionality to be established, the measurement dimension must account for at least 40% of the variance, and the unexplained variance associated with the eigenvalue of the first contrast must remain below 3.0 [36]. In order to determine the fit of the items to the RRSM, infit mean-square (MNSQ) and outfit MNSQ values were calculated, with acceptable ranges defined from 0.6 to 1.4 [35]. Item polarity is evaluated using point-measure correlation (PTMEA CORR.), which analyzes how an individual item relates to the total score of the underlying construct [37]. A point-measure correlation ranging from 0.4 to 0.8 is deemed acceptable [37]. To verify that items effectively differentiate between varying levels of individual performance and that respondents can perceive calibration differences among items, separation indices were computed for both items and individuals. An item and person separation index greater than 2, along with a separation reliability coefficient exceeding 0.7, is typically regarded as sufficient [38]. To visualize item characteristics and individual measurements, a Wright map was used to illustrate how all items ranked on a unified scale. Moreover, differential item functioning (DIF) was used to evaluate measurement invariance across genders, usually indicated by a DIF greater than 0.5 logits and a p-value lower than 0.05 [39]. Ethical considerations Permission to translate the PSHFSS was granted by the original author [17]. The ethical considerations of this study were approved by the Academic Ethics Committee of the School of Education at City University of Macau (Approval No. SOE-09-03-2425-06-03-DEDCF). This research was conducted in accordance with the principles set forth in the Helsinki Declaration (World Medical Association Declaration of Helsinki). Upon accessing the Wenjuanxing platform (https://www.wjx.cn/), students were required to read the informed consent form and click the “Agree” button to participate in the survey. The survey was conducted anonymously, ensuring that all collected information was securely stored and used solely for research purposes. Participants maintained the right to withdraw from the study at any time without facing any negative repercussions. Results Demographic characteristics of participants A total of 522 participants were included in the analysis, with their demographic characteristics summarized in Table 1. The participants’ ages ranged from 18 to 60 years, with a mean age of 23.5 ± 4.9 years. Female participants comprised 78.7% of the sample, and 83.9% were currently pursuing a bachelor’s degree. Table 1Participants’ characteristics (N = 522)Variables N %Gender Female41178.7 Male11121.3Current educational pursuit Associate’s degree458.6 Bachelor’s degree43883.9 Master’s degree346.5 Doctoral degree51.0Region Northeast China366.9 East China19337.0 Central China6813.0 West China499.4 Macao SAR17633.7Note. SAR = Special Administrative Region Table 1Participants’ characteristics (N = 522)Variables N %Gender Female41178.7 Male11121.3Current educational pursuit Associate’s degree458.6 Bachelor’s degree43883.9 Master’s degree346.5 Doctoral degree51.0Region Northeast China366.9 East China19337.0 Central China6813.0 West China499.4 Macao SAR17633.7Note. SAR = Special Administrative Region Participants’ characteristics (N = 522) Note. SAR = Special Administrative Region CTT analysis The I-CVI of the CV-PSHFSS ranged from 0.8 to 1.0, and the S-CVI was 0.96, as shown in Supplementary Table 2. All the I-CVIs were greater than 0.80 and the S-CVI was greater than 0.9, suggesting that the content validity of the CV-PSHFSS was good. The construct validity was evaluated by CFA. The results of CFA are shown in Fig. 1. The initial model was modified according to the modification indices and theoretical justification. In the revised model, the CMINDF was 2.5, RMSEA was 0.054, the GFI was 0.963, the CFI was 0.982, and the IFI was 0.982, indicating that the model fit the data well. Fig. 1The confirmatory factor model of the CV-PSHFSS (N = 522) The confirmatory factor model of the CV-PSHFSS (N = 522) Cronbach’s α coefficients for the dimensions of dealing with uncertainty, being exposed, being unsupported, interpersonal risk, and the overall scale were 0.820, 0.885, 0.888, 0.896, and 0.927, respectively. Meanwhile, the McDonald’s omega coefficients for the same dimensions—dealing with uncertainty, being exposed, being unsupported, interpersonal risk, and the overall scale—were 0.821, 0.882, 0.890, 0.897, and 0.925, respectively. Thirty participants were invited to complete the scale after two weeks. The ICC of the CV-PSHFSS was 0.925, and that of the items ranged from 0.715 to 0.855, as shown in Supplementary Table 2. IRT analysis Based on the PCA of the standardized residuals, the dimension identified by the RRSM accounted for 50.1% of the total variance. Additionally, the unexplained variance linked to the eigenvalue of the first contrast was found to be 2.8. This finding supports the assumption of unidimensionality of the construction under investigation. Table 2 presents the item statistics for each of the 14 items, which include item difficulty, standard error of measurement, fit statistics, and point-measure correlation. The infit MnSQ values ranged from 0.69 to 1.24, while the outfit MnSQ values ranged from 0.69 to 1.37. The minimum point-measure correlation value observed was 0.59, while the maximum value reached 0.77. Moreover, the CV-PSHFSS demonstrated a person reliability coefficient of 0.90 and a person separation index of 2.96, suggesting that participant scores could be reliably estimated and categorized into at least four strata [(2.96*4 + 1)/3]. The item reliability coefficient was 0.98, with an item separation index of 6.48, indicating that these items provided precise measurements and could be effectively distinguished into at least eight strata [6.48*4 + 1)/3] of psychological safety. Table 2Item difficulty, standard error, fit statistic, and point-measure correlationItem numberItem difficultyStandard errorInfit MnSQOutfit MnSQPTMEA CORR.T1-0.160.061.241.290.61T20.850.061.231.210.65T3-0.440.061.241.370.59T40.330.061.221.210.66T50.540.061.251.250.68T60.320.060.880.870.74T70.020.060.690.690.77T8-0.050.060.700.710.77T90.270.060.880.860.75T10-0.550.060.900.930.69T11-0.070.060.940.960.72T12-0.440.060.991.000.66T13-0.240.060.900.910.69T14-0.370.060.900.910.71Note: MNSQ = infit mean-square; PTMEA CORR.= point-measure correlation Table 2Item difficulty, standard error, fit statistic, and point-measure correlationItem numberItem difficultyStandard errorInfit MnSQOutfit MnSQPTMEA CORR.T1-0.160.061.241.290.61T20.850.061.231.210.65T3-0.440.061.241.370.59T40.330.061.221.210.66T50.540.061.251.250.68T60.320.060.880.870.74T70.020.060.690.690.77T8-0.050.060.700.710.77T90.270.060.880.860.75T10-0.550.060.900.930.69T11-0.070.060.940.960.72T12-0.440.060.991.000.66T13-0.240.060.900.910.69T14-0.370.060.900.910.71Note: MNSQ = infit mean-square; PTMEA CORR.= point-measure correlation Item difficulty, standard error, fit statistic, and point-measure correlation Note: MNSQ = infit mean-square; PTMEA CORR.= point-measure correlation Figure 2 illustrates a Wright map that displays the 14 test items arranged according to participant ability. This map offers a visual representation of the distribution of item difficulty and participant ability, both represented on the same interval logit scale. The right side of the Wright Map shows the distribution of items organized by their difficulty, with the easiest item (T10) at the bottom and the most challenging item (T2) at the top. The difficulty measures for the items ranged from − 0.55 to 0.85 logits, resulting in a spread of 1.4 logits. On the left side of the map, individuals are positioned according to their psychological safety levels. Nursing students exhibiting lower psychological safety are found at the bottom, while those with higher psychological safety are at the top. Individual psychological safety estimates varied from − 5.15 to 4.95 logits, yielding a spread of 10.1 logits. The average item difficulty for the test is 0.00 logits, whereas the mean psychological safety levels is approximately 0.42 logits higher. Fig. 2Wright map of the CV-PSHFSS Wright map of the CV-PSHFSS The DIF analysis assesses the equivalence of constructs across different groups. Figure 3 presents a comparison of item DIF against the overall baseline item difficulty, categorized by gender. The DIF contrasts for all items ranged from − 0.24 to 0.28 logits, with all p-values exceeding 0.05, as detailed in Supplementary Table 2. This indicates that there is no significant differential item functioning (DIF) related to gender. Fig. 3DIF across genders for the CV-PSHFSS DIF across genders for the CV-PSHFSS Discussion In this study, the PSHFSS was translated into Chinese, and its psychometric properties were evaluated among Chinese nursing students using CTT and IRT. The findings indicate that the Chinese version of the PSHFSS exhibits satisfactory reliability and validity in this population. Nursing educators can apply this scale to assess the psychological safety of students in high-fidelity simulations, thereby enhancing their learning outcomes and engagement. Adherence to the TRAPD protocol for effective translation and adaptation In recent years, the TRAPD protocol has been recognized as a valuable and increasingly utilized methodology for the translation and adaptation of psychological scales [24]. This is not only due to the clear guidelines provided for each step within the TRAPD protocol, but also because it emphasizes the inclusion of diverse expertise during the translation and review processes, effectively addressing the shortcomings of traditional translation methods [24]. This study rigorously adhered to the TRAPD protocol throughout the translation and cross-cultural adaptation processes. Two doctoral-level translators with over five years of experience in South Korea are involved in the translation process. One is a nursing educator with 20 years of experience, while the other specializes in Korean translation and has 15 years of teaching experience in China. The review process involved ten experts from diverse fields, including nursing, education, psychology, sociology, epidemiology, and statistics. In the translation process, the results of the two translators were largely consistent, with only minor discrepancies emerging during the cross-cultural adaptation phase. Overall, these results underscore the importance of insights from experts across various disciplines during the translation and cultural adaptation processes, ensuring both the accuracy and appropriateness of the final translation. Content and construct validity, alongside reliability of the CV-PSHFSS In the CTT analysis, all I-CVIs exceeded 0.80, and the S-CVI surpassed 0.9, indicating that the content validity of the CV-PSHFSS is robust. These findings reflect a strong consensus among experts regarding the relevance and appropriateness of each item on the scale [40]. They believe that the scale effectively measures the intended constructions, highlighting its significant implications for future practical applications. In exploring the construct validity of the scale, this study only performed CFA and did not conduct EFA. This decision was primarily based on the fact that the original scale was developed from a clearly defined four-dimensional structure established through prior qualitative interviews and a systematic literature review [16, 17] While EFA focuses on exploring the factor structure of a scale, CFA is concerned with validating the fit of that structure [31]. Therefore, EFA for the CV-PSHFSS is not considered at this time. In the CFA, the fit indices of the initial model did not meet the desired criteria, so modifications were made to the initial model based on the modification indices and theoretical justification. According to previous research, psychological safety is not only affected by cognitive factors but also strongly influenced by emotional factors [16]. When participants feel anxious or uneasy in a high-fidelity simulation environment, these emotions can strengthen their connections to one another [16]. In the CV-PSHFSS, all 14 items mention anxiety and unease; thus, residual relationships were established between them. The revised four-dimensional structure model exhibited favorable fit indices, including CMIN/DF, RMSEA, GFI, CFI, and IFI, all of which met the established standards. This indicates that the model fits the data well, and the CV-PSHFSS retains a structure consistent with the original scale. In terms of reliability, the Cronbach’s α coefficients and ICCs for the four dimensions and the overall scale exceeded 0.7, indicating satisfactory internal consistency and external reliability of the CV-PSHFSS [41]. These findings highlight the CV-PSHFSS as a reliable tool for assessing the psychological safety of nursing students over time. The robust reliability metrics not only ensure its applicability but also ensure its accuracy for capturing psychological safety in this population. However, it is noteworthy that the Cronbach’s α coefficient for the CV-PSHFSS is slightly higher than the original version’s 0.91 and the Japanese version’s 0.906 [17, 22]. Several factors may contribute to this result. First, the scale’s cultural adaptation may more effectively capture the concept of psychological safety within the Chinese context, leading to more consistent responses from participants [42, 43]. Second, the language used in the translation is likely to resonate better with the Chinese nursing students, thereby reducing potential misunderstandings [44]. Regarding the ICC, the original scale does not provide relevant information [17]. The CV-PSHFSS has an ICC of 0.925, which is slightly higher than the Japanese version’s 0.889 [22]. However, since the Japanese version does not specify the number of nursing students involved in the retest and the interval time for the retest, further comparisons cannot be made. Item characteristics and respondent interactions of the CV-PSHFSS In addition to employing CTT to assess the psychological properties of the scale, IRT was utilized to gain a more comprehensive understanding of the distribution of survey items and the variability in respondents’ behaviors [45]. This combined approach allows for a thorough examination of both item characteristics and respondent interactions, enhancing the overall analysis [45]. In the IRT analysis, the infit and outfit MnSQ values for all items fell within the range of 0.6 to 1.4, while the point-measure correlation values ranged from 0.4 to 0.8. These findings indicate that the items demonstrate a stable performance in measuring psychological safety in HFS [37]. The point-measure correlation values also indicate the effectiveness of the items, reflecting a moderate to strong relationship between individual responses and the total score [46]. As a result, the items contribute significantly to the comprehensive measurement of the construction, therefore enhancing the validity of the scale. In the Wright map, both item difficulty and respondent abilities are represented on the same logit scale [47]. This map aids in assessing whether the test items cover a range of difficulty that aligns with the psychological safety levels of nursing students. The distribution of respondents spans approximately 10.1 logits, allowing for the categorization into at least four distinct strata. In contrast, the test items demonstrate a span of 1.4 logits, effectively distinguishing at least eight strata of psychological safety. Notably, the mean ability of respondents is approximately 0.42 logits higher than the mean difficulty of the items. This discrepancy may result from the inclusion of nursing students with graduate-level education in this study. A proportion of participants demonstrate psychological safety levels exceeding those of the most difficult item (T2), this somewhat restricts the scale’s ability to differentiate among individuals with high levels of psychological safety. Additionally, some items (i.e., items 7, 8, and 11) were found to be at the same difficulty level, indicating redundancy. Future research may necessitate the introduction of more challenging items while removing redundant ones to effectively assess the psychological safety levels of nursing students [38]. Overall, despite the higher mean ability, the range between − 0.55 logits and + 0.85 logits encompasses all items, with approximately 60% of respondents also falling within this range. This congruence indicates a commendable degree of targeting, suggesting that the items are well-calibrated to the population being assessed. DIF analysis offers valuable insights into item performance across different groups, thus ensuring the scale’s validity [48]. Because gender may influence individuals’ experiences, perceptions, and responses in simulation environments [49], a DIF analysis was performed to investigate whether the items on the CV-PSHFSS function differently based on gender. The absence of significant differences in psychological safety scores between male and female students suggests that the scale is gender neutral. This finding enhances its applicability in various educational contexts and underscores the importance of investigating other potential influencing factors. Limitations There were some limitations in this study. Firstly, while the economic disparities across different regions of China were considered during sample collection, efforts were made to engage local representatives for promoting the research and gathering samples to ensure equal representation. Nevertheless, significant differences in sample sizes across regions remained. This discrepancy is primarily attributed to the high cost of high-fidelity simulation equipment, which limits the ability of economically disadvantaged areas to conduct such simulations [25], thereby restricting sample inclusion. Secondly, the study’s sample consists mostly of female bachelor’s degree students, which may limit its applicability to male nurses and other educational levels. Meanwhile, since sample sources have diverse educational levels, DIF cannot be used to analyze measurement invariance among these levels. Lastly, despite employing CTT and IRT for the translation and validation of the CV-PSHFSS, a criterion validity test was not conducted due to the absence of a gold standard for evaluating psychological safety among nursing students. Future research should emphasize the use of more diverse and representative sampling methods, such as stratified random sampling, to enhance the external validity of the findings. Furthermore, integrating a broader range of indicators to assess the psychometric properties of the scale would aid in identifying and correcting potential biases, thereby improving measurement accuracy and applicability across various cultural contexts. Conclusions This study successfully translated the PSHFSS into Chinese and rigorously evaluated its psychometric properties among Chinese nursing students using CTT and IRT. The results demonstrate that the Chinese version of the PSHFSS possesses satisfactory reliability and validity among Chinese nursing students. Therefore, nursing educators are encouraged to utilize this scale to assess the psychological safety of students in HFS, thereby enhancing their learning outcomes and engagement. Future research should utilize diverse sampling methods and incorporate a broader range of indicators to enhance the psychometric properties of the scale, improving its accuracy and applicability across cultural contexts. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1