A study on the reliability and validity of CWS in Chinese schizophrenia patients Authors: Xiaojing Zhang, Ying Jin, Dehua Meng, Jianhui Gao, Yuxuan Tao, Wenyue Jin, Nan Li, Liangliang Chen, Wenhua Yan Published: 2025 ABSTRACT Currently, the diagnosis and assessment of schizophrenia remain challenging. This study aimed to examine the reliability and validity of the Crisi Wartegg System (CWS) for assessing schizophrenia within the Chinese cultural context, with the goal of exploring its potential as an auxiliary evaluation tool. A total of 50 individuals diagnosed with schizophrenia and 50 healthy controls were recruited. All participants completed the Wartegg Drawing Test and the Minnesota Multiphasic Personality Inventory–2 (MMPI-2), and Positive and Negative Syndrome Scale (PANSS) scores were collected from patients with schizophrenia. Inter-rater reliability of the CWS was evaluated using the intraclass correlation coefficient (ICC). Structural validity was assessed by comparing CWS indices between the schizophrenia and healthy groups, as well as by examining correlations between CWS indices and MMPI-2 scales. Criterion validity was verified by analyzing the associations between CWS indices and PANSS scores. Results indicated that the CWS demonstrated moderate to high inter-rater reliability. Significant group differences were observed in EC+%, FQ+%, P%, and P+% scores, and several CWS indices were significantly correlated with MMPI-2 scales, including F (Infrequency) and Sc (Schizophrenia), supporting its structural validity. In conclusion, the findings provide evidence for the reliability and validity of the CWS in schizophrenia assessment. CWS indices not only differentiate individuals with schizophrenia from healthy controls but also show significant correlations with selected MMPI-2 scales, further confirming its structural validity. These results establish a foundation for the application of the CWS in cross-cultural clinical practice for schizophrenia.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12888-025-07742-y. Introduction Background Schizophrenia is a complex primary psychiatric disorder characterized by core symptoms such as hallucinations, delusions, disorganized thinking or speech, abnormal motor behavior (including catatonia), and negative symptoms. These manifestations severely impair patients’ ability to test reality, leading to a wide range of behavioral changes. The lifetime prevalence of schizophrenia is estimated to be between 0.3% and 0.7%, and the disorder is typically associated with a poor prognosis [1, 2]. Epidemiological data indicate that schizophrenia in China is associated with a high level of disability, with a monthly prevalence rate of 1.0%. However, limited public awareness of mental health disorders and widespread societal stigma toward individuals with psychiatric conditions have prevented many patients from seeking the treatment they need [3]. Between 2014 and 2019, the outpatient treatment rate for schizophrenia in China has been on a decline [4], and delays in treatment may exacerbate the progression of the disorder. Research has shown that individuals with schizophrenia have a suicide risk 23.8 times higher than those without the disorder, and suicides related to schizophrenia account for 9.7% of all suicide cases [5]. Given the disorder’s high disability rate and low recovery rate, which impose substantial burdens on families and society, accurate clinical assessment for early detection, timely treatment, and intervention is of critical importance [6]. Assessment of schizophrenia Nevertheless, the assessment of schizophrenia remains a challenging task. Its symptoms often overlap with those of other psychiatric disorders, such as mood disorders, obsessive-compulsive disorder, and post-traumatic stress disorder, making the differential diagnosis of schizophrenia more difficult compared to other mental illnesses [7]. In clinical practice, various scales have been developed to assess different symptom domains of schizophrenia. For positive and negative symptoms alone, commonly used instruments include the Scale for the Assessment of Positive Symptoms (SAPS), the Scale for the Assessment of Negative Symptoms (SANS), the Positive and Negative Syndrome Scale (PANSS), and the Brief Psychiatric Rating Scale (BPRS). In addition to positive and negative symptoms, clinical evaluation also requires separate assessments of thought disorder, functional impairment, depression, and suicidal ideation. While using multiple scales can improve diagnostic accuracy, the process is often cumbersome, time-consuming, and requires a high level of expertise from the evaluator [8]. For example, the PANSS, widely used at the Shanghai Mental Health Center, has been reported to have notable limitations in both content and structural validity, and typically requires 30–50 min to administer, imposing a significant burden in clinical settings [9]. Furthermore, clinical assessments often rely on a combination of observation, verbal reports, and inference. However, due to cognitive impairments in individuals with schizophrenia, self-reports may compromise the accuracy of the evaluation [10]. Given the current reliance on predominantly subjective reporting, some researchers have advocated for the incorporation of more objective methods, such as projective tests and medical examinations. into the diagnostic process for schizophrenia [11–13]. Application of projective tests in the assessment of schizophrenia Researches have indicated that, compared with self-report inventories, projective tests use semi-structured stimuli that can bypass the respondent’s defenses, revealing underlying motives or biases that the individual may be unaware of or unwilling to disclose [14]. In current clinical practice, clinicians often incorporate projective tests, in addition to self-report inventories, when assessing patients with schizophrenia [15]. Research has shown that the Rorschach Inkblot Test can not only distinguish individuals with schizophrenia from healthy controls but also that various Rorschach variables are significantly associated with both positive and negative symptoms [12, 16]. Similarly, studies have demonstrated that the Thematic Apperception Test (TAT) can differentiate schizophrenia patients from healthy individuals and can also distinguish between schizophrenia and dementia [13, 17]. Although these findings suggest that projective tests may provide valuable information for clinical assessment, there remains a lack of robust empirical evidence. Moreover, such tests are often time-consuming to administer and score, and their scoring systems are not sufficiently objective [18–20]. To address the limitations of traditional projective tests, such as lengthy administration times, complex scoring procedures, and high demands on examiners, Italian psychologist Alessandro Crisi developed a quantitative scoring system named the Crisi Wartegg System (CWS). The CWS is designed for the Wartegg Drawing Completion Test (WDCT), a semi-structured, graphic, performance-based personality assessment. The original version of the WDCT, created by Ehrig Wartegg, contained 16 boxes, but Crisi adopted an improved 8-box version for his more objective system. In addition to assessing personality structure across dimensions such as functional intelligence, thought organization, available energy, social skills, and emotional functioning, the CWS also includes several pathological indicators, schizophrenia among them, that can be applied in clinical evaluation [21]. The definitions and interpretations of the major CWS indices are provided in Supplementary Table S-1. Evidence suggests that the CWS demonstrates moderate to high inter-rater reliability, and that three of its indices, Evocative Character (EC), Affective Quality (AQ), and Form Quality (FQ), can distinguish between psychiatric and control groups [22]. However, the psychiatric group in that study included multiple mental disorders, schizophrenia among them, making it difficult to specifically highlight the CWS’s effectiveness in assessing schizophrenia. Furthermore, most existing CWS-related research has been conducted in Western countries such as Italy and the United States [22–25], with a lack of cross-cultural studies in China. Given these limitations, conducting a reliability and validity study of the CWS for schizophrenia assessment in the Chinese context holds significant practical value, as it could provide clinicians with a more efficient and accurate measurement tool for the diagnosis of schizophrenia. Selection of criterion measures Reliability and validity assessment of projective tests The reliability of psychological tests encompasses test–retest reliability, parallel-forms reliability, internal consistency, and inter-rater reliability [26]. Test–retest reliability reflects not only the properties of the test itself but also the influence of the time interval between administrations. Parallel-forms reliability, which primarily applies to text-based measures, is not suitable for projective tests [27]. Research indicates that internal consistency is more reflective of sample characteristics than the stability of the measurement tool [28]. Scholars have argued that tests requiring partially subjective scoring should report inter-rater reliability [26]. Therefore, the present study focuses on examining the inter-rater reliability of the CWS. The validity of psychological tests can be categorized into content validity, criterion-related validity, and construct validity [27]. Content validity is primarily applied in questionnaire-based tests, educational assessments, personnel selection, and physiological or behavioral observation measures. It is typically evaluated during the development, revision, or update of an assessment tool, as well as when the context of application changes [29]. Criterion-related validity includes concurrent validity and predictive validity, with concurrent validity being more widely applied in the validation of projective tests [27]. For example, in evaluating the concurrent validity of the “Person in the Rain” projective test, Lack used scores from Rorschach variables and the Wechsler Intelligence Scale for Children–Revised as criteria [30]. Similarly, Groth-Marnat and Roberts examined the concurrent validity of the House-Tree-Person Test and the Draw-a-Person Test by using the Coopersmith Self-Esteem Inventory and the Tennessee Self-Concept Scale as criteria [31]. Predictive validity is more complex, time-consuming, and costly to assess than concurrent validity. Nevertheless, some studies have attempted to establish the predictive validity of drawing-based tests [27]. For instance, one study examined the predictive validity of the Goodenough–Harris Draw-a-Person Test for children’s academic performance using multiple regression analysis [32]. Construct validity should be assessed through multiple approaches, including testing for differences between groups and comparing results with other measures [27]. For instance, some researchers evaluated the known-groups validity and convergent validity of the Roberts Apperception Test for Children (RATC) by comparing the differences between two distinct groups of children and adolescents, as well as by correlating RATC results with other scales [33]. Other researchers coded narratives from TAT using Cramer’s Defense Mechanisms Manual (DMM) and compared the results with the Defensive Functioning Scale (DFS) to assess the convergent validity between the DMM and the DFS [34]. Given that the primary aim of the present study is to explore whether the latent psychological constructs reflected by the CWS are intrinsically related to its theoretical framework and other relevant variables, construct validity, as a comprehensive type of validity, can provide the most compelling evidence to support the psychological interpretation and theoretical significance of the test results, and is therefore the most appropriate validity index for this study. Considering the characteristics of the CWS, the present study will adopt the intraclass correlation coefficient (ICC), as used in previous research [22], to report its inter-rater reliability. In terms of validity, this study will examine the construct validity of the CWS. Since the CWS is primarily designed to assess personality, it is essential to identify a criterion measure that is both suitable for the CWS and applicable to personality assessment in individuals with schizophrenia. Self-report personality inventories Empirical research has validated the Minnesota Multiphasic Personality Inventory–2 (MMPI-2) [35]. One study found that participants trained solely in deception techniques for the MMPI-2 were more successful at evading detection, whereas knowledge of schizophrenia symptoms did not help participants avoid detection by the MMPI-2 [36]. Further evidence has confirmed that the validity scales F, Fb, and Fp effectively distinguish genuine schizophrenia patients from simulators, highlighting the MMPI-2’s substantial clinical value in the differential diagnosis and assessment of schizophrenia [37, 38]. Using a Bayesian modeling approach to analyze MMPI-2 data, one study found distinct score patterns across various scales between individuals with schizophrenia and healthy controls [39]. Moreover, within the MMPI-2 clinical scales, D and Sc, as well as the content scales DEP and LSE, can differentiate schizophrenia from major depression, and the content scale DEP can also distinguish bipolar disorder from schizophrenia [40]. Although previous research has not identified personality inventory scales significantly correlated with the CWS, some published literature suggests that the MMPI-2 may be an appropriate criterion measure for the present study. Stephen Finn, the founder of Therapeutic Assessment (TA), found that CWS results were consistent with those from the MMPI-2, the Rorschach Inkblot Test, and the Adult Attachment Projective. This consistency led to the inclusion of the CWS as an assessment tool in TA [41]. Several related case studies have demonstrated consistency between the MMPI-2 and the CWS. For instance, one study reported that several evaluators jointly administered multiple assessments, including the MMPI-2, the Rorschach Inkblot Test, and the Wartegg Drawing Completion Test, to an adult client. The results indicated certain associations among the different measures [24]. Another case study demonstrated that, in the TA of a woman, the Minnesota Multiphasic Personality Inventory-2 Restructured Form (MMPI-2-R) indicated denial of distress and worry, while the CWS results suggested difficulty in effectively regulating emotions when experiencing intense affect. The two measures showed a degree of descriptive consistency [42]. While these case studies have pointed out similarities between the MMPI-2 and the CWS in their assessment dimensions, there remains a lack of quantitative research to substantiate these findings. Clinical assessment indicators Given the potential subjectivity of the MMPI-2, it is essential to incorporate objective clinical assessment indicators in the present study to obtain more comprehensive and multidimensional data. Identifying appropriate clinical assessment indicators is, therefore, of particular importance. Although the PANSS has certain limitations, as noted above, it remains the most widely used assessment tool in clinical evaluation and research on schizophrenia in China [43–47]. The PANSS consists of 30 items, divided into three dimensions: positive symptoms, negative symptoms, and general psychopathology. Scores are assigned by clinicians based on interviews and observations of patients’ responses and behaviors, helping clinicians understand patients’ symptom profiles and track changes in their condition [48]. The reliability and validity of the PANSS have been well established in clinical research. For example, a study conducted in 2014 assessed the negative symptom factor of the PANSS in evaluating negative symptoms among schizophrenia patients with prominent negative or disorganized symptoms, finding good test–retest reliability and validity [49]. In 2017, another study evaluated and compared several schizophrenia symptom rating scales, including the classic PANSS, and concluded that the PANSS is the “gold standard” for assessing schizophrenia symptoms, being highly useful in clinical research and providing objective measures of symptom severity and treatment response [50]. In summary, the present study adopts the MMPI-2 and PANSS as criterion measures for the CWS and proposes the following hypotheses: H1 The CWS demonstrates moderate to high inter-rater reliability. H2 The CWS demonstrates good construct validity, with significant differences between schizophrenia patients and healthy controls in its major indices, main indices of reference for CWS interpretive domains and specific pathological indices. H3 Scores on main indices of reference for CWS interpretive domains are significantly correlated with MMPI-2 scales reflecting schizophrenia-related characteristics. H4 Scores on the major indices, main indices of reference for CWS interpretive domains and specific pathological indices are significantly correlated with PANSS scores. Method Participants Patients with schizophrenia consisted of participants aged 18–60 years with 6–16 years of education. All participants met the ICD-10 diagnostic criteria for schizophrenia, were in a non-acute phase (PANSS total score ≤ 75). This threshold is commonly adopted in research to define clinical stability, as it generally reflects that core positive, negative, and general symptoms are of no more than mild severity [51, 52]. Additionally, all participants had maintained a stable dosage of antipsychotic medication for at least four weeks prior to enrollment. Written informed consent was obtained from each participant and their family members or legal guardians before the study commenced. Exclusion criteria for patients with schizophrenia included: withdrawal of informed consent by the participant or guardian; presence of suicidal ideation or impulsive aggressive behavior within two weeks prior to enrollment; severe physical illness or epilepsy; and substance abuse or dependence. The healthy group consisted of individuals aged 18–60 years with 6–16 years of education, good physical health, no history of psychiatric disorders, cardiovascular disease, or brain injury, and no history of substance abuse. Procedure The study procedure included obtaining an ethics approval letter, participant recruitment, test administration, clinical assessment, data collection and verification, CWS scoring, data analysis, and provision of research compensation. The specific steps were as follows: At a designated mental health center, 50 participants diagnosed with schizophrenia and currently in a stable phase were recruited by psychiatrists. Healthy controls, matched to patients with schizophrenia in age and educational level, were recruited through advertisements. Both groups first completed the CWS, followed by a 10-minute break, and then took the MMPI-2. The drawing materials we used for the CWS test were standardized drawing materials purchased from the Istituto Italiano Wartegg. Due to group differences, patients with schizophrenia required approximately 90–120 min to complete the assessments, whereas healthy controls required about 90 min. PANSS scores for the patient group were provided by psychiatrists to assess their mental status and for subsequent data analysis. Four professionally trained raters independently scored the drawings in a blind review. Each drawing was randomly assigned to two raters, and inter-rater reliability was calculated. For any scoring discrepancies, the final score was determined jointly by all four raters and the CWS’s creator, Alessandra Crisi. Data were entered into SPSS, and statistical analyses were conducted with MMPI-2 scale scores, chlorpromazine equivalents, and PANSS scores to generate the study results. Statistical analysis All statistical analyses were conducted using SPSS 26.0, with the following procedures: To examine differences in demographic characteristics between the two groups, independent-samples t-tests were used to compare age. Educational level was coded as follows: 1 = 6 years of education; 2 = 7–9 years of education; 3 = 10–12 years of education; 4 = 13–16 years of education. The Mann-Whitney U test was applied to compare differences in educational level between the two groups. Descriptive statistics were used to analyze age at onset, illness duration, and hospitalization history in patients with schizophrenia. To ensure the objectivity and reliability of the drawing scores, inter-rater reliability was assessed. This approach was selected to align with the methodology used in the research by Crisi, ensuring comparability [22]. Specifically, each drawing was randomly assigned to two of the four raters for scoring. The ICC was calculated using a one-way model. Reliability was classified based on the ICC values as poor (< 0.50), moderate (0.50–0.75), good (0.75–0.90), or excellent (> 0.90) [53]. For construct validity, the study aimed to examine the discriminative ability of the selected scales across different groups and the relationships among their underlying constructs. First, independent-samples t-tests were used to compare MMPI-2 scale scores between the two groups. Second, the Mann-Whitney U tests were applied to compare CWS index scores between healthy controls and patients with schizophrenia. In addition, Spearman correlation analyses were conducted to examine the relationships between CWS indices and MMPI-2 scales. To further verify the criterion validity of the CWS personality and schizophrenia indices, correlation analyses were performed to explore the associations between the CWS and PANSS scores. Results Demographic variables A total of 50 participants were recruited for each group (patients with schizophrenia and healthy controls). Independent-samples t-test results indicated no significant difference in age between patients with schizophrenia and healthy controls (p = 0.877). The Mann-Whitney U test results showed no significant difference in educational level between the two groups (p = 0.706). The mean age at onset, mean illness duration, and hospitalization history for patients with schizophrenia are presented in Table 1. Table 1Demographic information (N = 100)VariablePatients with Schizophrenia(N = 50)Healthy Controls (N = 50)t/Z p Age (M ± SD)45.040 ± 9.33144.760 ± 8.6930.1550.877Years of Education-0.3780.7066 years (%)17–9 years (%)111510–12 years (%)231613–16 years (%)1618Age at Onset22.860 ± 7.166-Illness Duration (years)22.180 ± 9.068-Hospitalization StatusLength of Stay/Discharge (months)Inpatient4729.819 ± 37.329-Discharged264.250 ± 65.407-Never Hospitalized1-- Table 1Demographic information (N = 100)VariablePatients with Schizophrenia(N = 50)Healthy Controls (N = 50)t/Z p Age (M ± SD)45.040 ± 9.33144.760 ± 8.6930.1550.877Years of Education-0.3780.7066 years (%)17–9 years (%)111510–12 years (%)231613–16 years (%)1618Age at Onset22.860 ± 7.166-Illness Duration (years)22.180 ± 9.068-Hospitalization StatusLength of Stay/Discharge (months)Inpatient4729.819 ± 37.329-Discharged264.250 ± 65.407-Never Hospitalized1-- Demographic information (N = 100) Inter-rater reliability In the present study, the ICC for Evocative Character (EC) scores across the boxes ranged from 0.758 to 0.880, while the ICC for Affective Quality (AQ) scores ranged from 0.774 to 0.920. The ICC for Form Quality (FQ) scores was slightly lower, ranging from 0.584 to 0.722. Detailed inter-rater reliability results for EC, AQ, and FQ in each box are presented in Table 2. Table 2Inter-Rater reliability for EC, AQ, and FQ in each CWS boxBOXECAQFQBOX 10.8800.9200.649BOX 20.8780.9040.673BOX 30.8640.8570.584BOX 40.8090.8330.623BOX 50.8280.9080.722BOX 60.7760.7740.694BOX 70.8260.8860.693BOX 80.7580.9300.603Note: EC = Evocative Character, AQ = Affective Quality, FQ = Form Quality Table 2Inter-Rater reliability for EC, AQ, and FQ in each CWS boxBOXECAQFQBOX 10.8800.9200.649BOX 20.8780.9040.673BOX 30.8640.8570.584BOX 40.8090.8330.623BOX 50.8280.9080.722BOX 60.7760.7740.694BOX 70.8260.8860.693BOX 80.7580.9300.603Note: EC = Evocative Character, AQ = Affective Quality, FQ = Form Quality Inter-Rater reliability for EC, AQ, and FQ in each CWS box Note: EC = Evocative Character, AQ = Affective Quality, FQ = Form Quality Construct validity Known-groups validity Major CWS indices After scoring and calculating each CWS drawing, a series of major indices were obtained. Since the Affective Stability Index (A/F) and the Adjusted Index of Inner Tension-2 (IIT-2 ADJ) are expressed as ratios, the difference between the left and right sides of the ratio was calculated for statistical analysis. Mann-Whitney U tests were used to compare differences in basic indices between the two groups in the CWS. Results showed that patients with schizophrenia had significantly lower scores than healthy controls on the percentage of Evocative Character (EC+%), the percentage of Form Quality (FQ+%), the percentage of Popular Responses (P%), the percentage of Popular Responses with Positive Form Quality (P+%), Adjusted Index of Inner Tension-1 (IIT-1 ADJ), Impulsivity Index (IM), and Impulsivity Index (AI). The difference between the left and right sides of the A/F ratio was significantly higher in patients with schizophrenia, whereas the difference between the left and right sides of the IIT-2 ADJ ratio was significantly lower than that of healthy controls (see Table 3). In terms of Percentage of Primary Contents, patients with schizophrenia had significantly lower scores on the percentage of Whole Animal Content (A%) (U = 957.000, p = 0.015, r = 0.243) and the percentage of Architecture Content (ARC%) (U = 1075.000, p = 0.007, r = 0.272), and significantly higher scores on the percentage of Abstract Content (ABS%) (U = 1075.000, p = 0.015, r = 0.243) compared with healthy controls. To control for the inflated risk of Type I errors from multiple correlations, the Benjamini-Hochberg procedure was used to control the False Discovery Rate (FDR) at 5%. Across the eight boxes, patients with schizophrenia obtained significantly fewer boxes coded as the most positive codes, Choice (C) or Positive Compensation (PC), and significantly more boxes coded as the most negative codes, Negative Compensation (NC) or Delay (D), than healthy controls, while there was no significant difference between the two groups in the number of boxes coded as Ambivalent Choice (AC) or Ambivalent Delay (AD) (see Table 4). Table 3Non-Parametric test of differences in major CWS indices between patients with schizophrenia and healthy controls (N1 = 50, N2 = 50)CodeMean RankMann-Whitney U p r PatientsHealthy ControlswithN = 50SchizophreniaN = 50EC+%36.16064.840533.000 0.000 0.497AQ+%48.28052.7201139.0000.4390.077FQ+%37.60063.400605.000 0.000 0.452A/F59.81041.190784.500 0.001 0.325P%38.55062.450652.500 0.000 0.416P+%43.18057.820884.000 0.004 0.287O%53.58047.4201096.0000.1130.159O+%53.16047.8401117.0000.1700.137IIT-1 ADJ40.17060.830733.500 0.000 0.357IM41.73059.270811.500 0.001 0.320AI39.87061.130718.500 0.000 0.372IIT-2 ADJ36.58064.420554.000 0.000 0.481Note: EC+% = Percentage of Evocative Character, AQ+%= Percentage of Affective Quality, FQ+%= Percentage of Form Quality, A/F = Affective Stability Index, P%=Percentage of Popular responses, P+%=Percentage of Popular responses with positive Form Quality, O%=Percentage of Original responses, O+%=Percentage of Original responses with positive Form Quality, IIT-1 ADJ = Adjusted Index of Inner Tension-1, IM = Impulsivity Index, AI = Impulsivity Index, IIT-2 ADJ = Adjusted Index of Inner Tension-2 Table 3Non-Parametric test of differences in major CWS indices between patients with schizophrenia and healthy controls (N1 = 50, N2 = 50)CodeMean RankMann-Whitney U p r PatientsHealthy ControlswithN = 50SchizophreniaN = 50EC+%36.16064.840533.000 0.000 0.497AQ+%48.28052.7201139.0000.4390.077FQ+%37.60063.400605.000 0.000 0.452A/F59.81041.190784.500 0.001 0.325P%38.55062.450652.500 0.000 0.416P+%43.18057.820884.000 0.004 0.287O%53.58047.4201096.0000.1130.159O+%53.16047.8401117.0000.1700.137IIT-1 ADJ40.17060.830733.500 0.000 0.357IM41.73059.270811.500 0.001 0.320AI39.87061.130718.500 0.000 0.372IIT-2 ADJ36.58064.420554.000 0.000 0.481Note: EC+% = Percentage of Evocative Character, AQ+%= Percentage of Affective Quality, FQ+%= Percentage of Form Quality, A/F = Affective Stability Index, P%=Percentage of Popular responses, P+%=Percentage of Popular responses with positive Form Quality, O%=Percentage of Original responses, O+%=Percentage of Original responses with positive Form Quality, IIT-1 ADJ = Adjusted Index of Inner Tension-1, IM = Impulsivity Index, AI = Impulsivity Index, IIT-2 ADJ = Adjusted Index of Inner Tension-2 Non-Parametric test of differences in major CWS indices between patients with schizophrenia and healthy controls (N1 = 50, N2 = 50) Note: EC+% = Percentage of Evocative Character, AQ+%= Percentage of Affective Quality, FQ+%= Percentage of Form Quality, A/F = Affective Stability Index, P%=Percentage of Popular responses, P+%=Percentage of Popular responses with positive Form Quality, O%=Percentage of Original responses, O+%=Percentage of Original responses with positive Form Quality, IIT-1 ADJ = Adjusted Index of Inner Tension-1, IM = Impulsivity Index, AI = Impulsivity Index, IIT-2 ADJ = Adjusted Index of Inner Tension-2 Table 4Non-Parametric test of differences in CWS code indices between the patients with schizophrenia and healthy controls (N1 = 50, N2 = 50)CodeMean RankMann-Whitney U p r PatientsHealthy ControlswithN = 50SchizophreniaN = 50C41.31059.690790.500 0.001 0.327AC53.33047.6701108.5000.3000.104NC57.55043.450897.500 0.010 0.259PC37.81063.190615.500 0.000 0.453AD52.98048.0201126.0000.3630.091D61.74039.260688.000 0.000 0.409Note: C = Choice, AC = Ambivalent Choice, NC = Negative Compensation,PC = Positive Compensation, AD = Ambivalent Delay, D = Delay Table 4Non-Parametric test of differences in CWS code indices between the patients with schizophrenia and healthy controls (N1 = 50, N2 = 50)CodeMean RankMann-Whitney U p r PatientsHealthy ControlswithN = 50SchizophreniaN = 50C41.31059.690790.500 0.001 0.327AC53.33047.6701108.5000.3000.104NC57.55043.450897.500 0.010 0.259PC37.81063.190615.500 0.000 0.453AD52.98048.0201126.0000.3630.091D61.74039.260688.000 0.000 0.409Note: C = Choice, AC = Ambivalent Choice, NC = Negative Compensation,PC = Positive Compensation, AD = Ambivalent Delay, D = Delay Non-Parametric test of differences in CWS code indices between the patients with schizophrenia and healthy controls (N1 = 50, N2 = 50) Note: C = Choice, AC = Ambivalent Choice, NC = Negative Compensation, PC = Positive Compensation, AD = Ambivalent Delay, D = Delay Main indices of reference for CWS interpretive domains Based on the importance of the main indices of reference for CWS interpretive domains, each index was weighted on a scale of 1–3. The weighted scores were entered into SPSS and converted into T-scores to examine group differences across the interpretive domains. The results are presented in Table 5. In the domain of Functional Intelligence, patients with schizophrenia scored significantly lower than healthy controls on High Intelligence and Medium Intelligence, while scoring significantly higher on Low Intelligence. In addition, patients with schizophrenia scored significantly lower on Abstract Intelligence. In the domain of Organization of Thought Processes, patients with schizophrenia scored significantly lower on Intact Reality Testing, while scoring significantly higher on Inadequate Reality Testing, Disturbances in Thought Processes, and Disturbances in Thought Content. Furthermore, patients with schizophrenia scored significantly lower on Adherence to Conventional Thinking and significantly higher on Oppositional Style. In the Adequate and Functional Energy dimension, patients with schizophrenia scored significantly lower than healthy controls. In the domain of Emotional Functioning, patients with schizophrenia scored significantly lower on Normative Values Indicating Overall Maturity and Stability. In the domain of Social Skills, there were no significant differences between patients with schizophrenia and healthy controls in Adequate Socialization, Difficulties (Detachment and Isolation), or Difficulties (Dependence). Table 5Non-Parametric test of differences in scores on main indices of reference for CWS interpretive domains between patients with schizophrenia and healthy controls (N1 = 50, N2 = 50)CodeMean RankMann-Whitney U p r PatientsHealthy ControlswithN = 50SchizophreniaN = 50High Intelligence39.75061.250712.500 0.000 0.378Medium Intelligence35.70065.300510.000 0.000 0.513Low Intelligence64.77036.230536.500 0.000 0.493Practical Intelligence54.23046.7701063.5000.1950.130Abstract Intelligence42.54058.460852.000 0.005 0.282Highly Intellectualized Intelligence47.62053.3801106.0000.2920.105Intact Reality Testing34.99066.010474.500 0.000 0.536Inadequate Reality Testing65.34035.660508.000 0.000 0.513Disturbances in Thought Processes66.99034.010425.500 0.000 0.571Disturbances in Thought Content66.20034.800465.000 0.000 0.547Adherence to Conventional Thinking41.58059.420804.000 0.002 0.309Conformist Style48.05052.9501127.5000.3920.086Oppositional Style60.02040.980774.000 0.001 0.335Adequate and Functional Energy43.76057.240913.000 0.019 0.234Adequate Socialization45.46055.540998.0000.0800.175Difficulties (Detachment and Isolation)53.48047.5201101.0000.3010.103Difficulties (Dependence)48.88052.1201169.0000.5750.056Normative Values Indicating Overall Maturity and Stability41.26059.740788.000 0.001 0.319 Table 5Non-Parametric test of differences in scores on main indices of reference for CWS interpretive domains between patients with schizophrenia and healthy controls (N1 = 50, N2 = 50)CodeMean RankMann-Whitney U p r PatientsHealthy ControlswithN = 50SchizophreniaN = 50High Intelligence39.75061.250712.500 0.000 0.378Medium Intelligence35.70065.300510.000 0.000 0.513Low Intelligence64.77036.230536.500 0.000 0.493Practical Intelligence54.23046.7701063.5000.1950.130Abstract Intelligence42.54058.460852.000 0.005 0.282Highly Intellectualized Intelligence47.62053.3801106.0000.2920.105Intact Reality Testing34.99066.010474.500 0.000 0.536Inadequate Reality Testing65.34035.660508.000 0.000 0.513Disturbances in Thought Processes66.99034.010425.500 0.000 0.571Disturbances in Thought Content66.20034.800465.000 0.000 0.547Adherence to Conventional Thinking41.58059.420804.000 0.002 0.309Conformist Style48.05052.9501127.5000.3920.086Oppositional Style60.02040.980774.000 0.001 0.335Adequate and Functional Energy43.76057.240913.000 0.019 0.234Adequate Socialization45.46055.540998.0000.0800.175Difficulties (Detachment and Isolation)53.48047.5201101.0000.3010.103Difficulties (Dependence)48.88052.1201169.0000.5750.056Normative Values Indicating Overall Maturity and Stability41.26059.740788.000 0.001 0.319 Non-Parametric test of differences in scores on main indices of reference for CWS interpretive domains between patients with schizophrenia and healthy controls (N1 = 50, N2 = 50) Convergent validity First, for the MMPI-2 analysis, the study began with 50 protocols from the patient group. To ensure data quality, all protocols were screened for validity based on the criteria in the MMPI-2 Chinese Manual [54]. Specifically, protocols were excluded if scores on the validity scales fell outside the acceptable ranges (i.e., VRIN: T < 30 or ≥ 80; TRIN: T ≥ 80; L: T ≥ 80; F: T ≥ 91 without a valid VRIN; ICH raw score ≥ 10). This rigorous screening led to the exclusion of 9 invalid protocols, resulting in a final sample of 41 valid datasets for analysis. Second, some MMPI-2 scales were extracted as key scales for Spearman correlation analysis with key major CWS indices.The selection of MMPI-2 scales was guided by prior research on the psychopathology of schizophrenia, focusing on scales that consistently differentiate patients from healthy controls [37, 38, 40]. It was confirmed that these scales were also significantly elevated in our patient group compared to controls (see Table S-2). Finally, the statistical results were imported into MATLAB R2021a to generate a heatmap of Spearman correlation coefficients between the key major CWS indices and the key MMPI-2 scales in patients with schizophrenia (see Figure S1). The results indicated that several key major CWS indices were significantly correlated with multiple key MMPI-2 scales. P% was significantly negatively correlated with seven MMPI-2 subscales, F, Fb, Fp, Pt and Sc, making it the major CWS indices with the largest number of significant correlations with MMPI-2 scales. P+% was significantly negatively correlated with the F and Sc scales. EC+% was negatively correlated with Sc scale, whereas the percentage of FQ+% was significantly negatively correlated with the F and Sc scales. IIT-2 ADJ showed a significant negative correlation only with the Sc scale, and A% showed a significant negative correlation only with the Pd scale. To further examine the convergent validity of the main indices of reference for CWS interpretive domains across interpretive domains, Spearman correlation analyses were conducted between the key indices in each CWS interpretive domain and the key MMPI-2 scales in patients with schizophrenia. The results showed that in the domains of Functional Intelligence, Organization of Thought Processes, and Available Energy, several indices were significantly correlated with the F and Sc scales. In addition, High Intelligence and Adequate and Functional Energy were significantly negatively correlated with Fb and Fp, while Adequate and Functional Energy was also significantly negatively correlated with BIZ and TRT. Oppositional Style was significantly negatively correlated with Pd (see Table S-3). To control for the inflated risk of Type I errors from multiple correlations, the Benjamini-Hochberg procedure was used to control the False Discovery Rate (FDR) at 5%. Criterion validity Spearman correlation analyses were conducted between the key major CWS indices and the PANSS scale scores. The results showed that P% was significantly negatively correlated with the Negative Scale, General Psychopathology Scale, and PANSS total score (see Table S-4). None of the content indices were significantly correlated with the PANSS scores. Within the key main indices of reference for CWS interpretive domains, the High Intelligence index showed a negative correlation trend with the Negative Scale (r = − 0.315, p = 0.045). Both Available Energy (r = 0.386, p = 0.013) and Emotional Functioning (r = 0.329, p = 0.035) were significantly positively correlated with the PANSS Positive Scale. To enable a more targeted comparison between the CWS and PANSS, the schizophrenia index from the specific pathological indices was introduced, as it is more specific than the main indices of reference for CWS interpretive domains. Similar to the weighting process applied to the main indices of reference for CWS interpretive domains, four researchers discussed and assigned weights of 1–3 points to each criterion of the schizophrenia index. The resulting total score was used as the score for the schizophrenia index. An Mann-Whitney U test comparing the schizophrenia index scores between patients with schizophrenia and healthy controls showed a significant difference (U = 539.000, p = 0.000). After standardizing the schizophrenia index scores into z-scores, Spearman correlation analyses were conducted with the scores of the Positive Scale, Negative Scale, General Psychopathology Scale, and total score. The results indicated that the schizophrenia index was not significantly correlated with the Positive Scale (r = − 0.114 p = 0.432), Negative Scale (r = 0.117, p = 0.417), General Psychopathology Scale (r = 0.068, p = 0.638), or total score (r = 0.005, p = 0.974). Discussion Inter-rater reliability First, in the present study, the CWS demonstrated moderate to high inter-rater reliability, which is consistent with the results reported in previous research [22]. Similar to the findings of Crisi’s study, the ICC for FQ in this study was the lowest among the three scoring categories. This may be related to the scoring criteria for FQ, which assess whether the drawing is clear and easily understandable, an aspect closely linked to the participant’s level of cognitive functioning [21]. One of the core symptoms of schizophrenia is disorganized thinking [2], which significantly impacts cognitive functioning and may, in turn, affect the clarity of the drawings, thus complicating the scoring of FQ.Second, although the CWS was developed with reference to the Rorschach [21], a key difference is that the Rorschach has developed an FQ scoring table over many years. Research has shown that without such a table, raters’ evaluations of FQ tend to rely more on personal judgment, resulting in lower reliability [55]. In the CWS, no similar FQ table exists, and the subjective nature of scoring may reduce inter-rater consistency.Finally, researchers from Muller’s team have noted that ICC values are influenced by the overall variance of the sample, meaning that the calculated ICC cannot be interpreted in isolation [56]. From this perspective, even though the FQ consistency for some boxes did not meet the traditional standard, it can still be concluded that, in the context of this study, considering the CWS is a projective test and lacks an FQ scoring table similar to that of the Rorschach, the CWS demonstrates moderate to high inter-rater reliability, reflecting the objectivity of its scoring. Construct validity Known-groups validity Major CWS indices In terms of construct validity, the present study found significant differences between schizophrenia patients and healthy controls in several major CWS indices. First, when EC+% is below the normative value, it indicates that the individual ignores or does not understand the evocative character within the box, reflecting cognitive deficits, rigid adaptation, and a lack of flexibility, thereby confirming the utility of this index for assessing reality testing ability [21]. Second, lower FQ+% compared with the healthy group may be related to the disorganized and idiosyncratic drawing style of individuals with schizophrenia, which is a typical manifestation of disorganized thinking [57]. Third, a smaller difference between the left and right values of the A/F ratio reflects the immaturity, dependency, and impulsivity of individuals with schizophrenia [22]. This finding is consistent with previous research showing that schizophrenia patients, due to delayed or arrested neurodevelopment, experience cognitive deficits and functional impairments that lead to dependency on others and increased impulsivity [58–60]. Fourth, lower P% and P+% scores may result from the lower overlap between the drawing content of schizophrenia patients and that of healthy individuals, as well as from the uniqueness of their drawings, an expression of symptoms such as disorganized thinking and delusions in their artwork [57]. This index suggests that the thinking of individuals with schizophrenia diverges from mainstream thought and may indicate symptoms of social functioning impairment or intellectual deterioration [21]. Fifth, lower IIT-1 ADJ and AI in patients with schizophrenia may be due to their drawings having smaller average sizes and contour lengths compared with those of the healthy group [61]. Overly simple drawing content may result in fewer specific Special Scores (SS). Although lower scores on this index reflect higher self-confidence and lower levels of tension and anxiety [21], this may be attributable to affective flattening, one of the typical negative symptoms of schizophrenia [2], or to the use of antipsychotic medications that reduce anxiety [62]. Finally, the significantly smaller difference between the left and right values of the IIT-2 ADJ ratio in patients with schizophrenia suggests a lower level of personality integration [21]. Otto Kernberg classified levels of psychological organization into high, medium, and low; the pathological manifestations at the low level include pre-psychotic personality organization, in which personality integration is insufficient [63]. Individuals with schizophrenia are precisely at this lower level of psychological organization. The current understanding of the pathophysiology of schizophrenia involves multiple hypotheses. Among these, the neurodevelopmental hypothesis posits that individuals with schizophrenia experience disruptions in macro- and micro-connectivity of the prefrontal, temporal, and parietal cortices, as well as the hippocampus, due to congenital or acquired influences [64, 65]. These disruptions may lead to deficits in executive and abstract functions, which could manifest in the CWS test as difficulties in capturing the “Evocative Character” of each CWS square and in the abstract nature of the drawn content. This is in line with previous research findings that the paintings of schizophrenia patients, especially those with thought disorders, tend to be geometric and abstract [66]. Among the remaining major CWS indices, in terms of content percentages, patients with schizophrenia produced more abstract drawings (i.e., representations of concepts or emotions like “happiness”) and fewer animal and architectural drawings compared with the healthy group [21]. This finding further supports previous research showing that the artwork of individuals with schizophrenia tends to be repetitive and abstract [67]. Differences in the statistical results of box codes indicated that schizophrenia patients have lower levels of personality integration and greater internal conflict [21], which is consistent with Otto Kernberg’s concept of low-level psychological organization in pre-psychotic personality and with earlier findings on ambivalence in individuals with schizophrenia [63, 68]. In summary, the major CWS indices demonstrate strong effectiveness in assessing multiple characteristics or symptoms of schizophrenia, including inadequate reality testing, disorganized thinking, and affective flattening. Main indices of reference for CWS interpretive domains In the Functional Intelligence domain, the results showed that patients with schizophrenia had significantly lower intelligence than the healthy group, reflecting the widespread presence of intellectual deficits in this population [69]. Among these, the abstract thinking ability of schizophrenia patients was significantly lower than that of the healthy group. Although individuals with schizophrenia may sometimes produce seemingly “abstract” responses, for example, using particularly simple symbols to represent something in the CWS, such responses are in fact manifestations of illogical thinking [70]. Deficits in Functional Intelligence among schizophrenia patients are closely tied to deficits in Organization of Thought Processes. Indicators in this domain showed that schizophrenia patients had impaired reality testing and exhibited disorganized thinking caused by organic or physiological factors, along with thought disorders such as poverty of speech, neologisms, and illogicality [21]. These findings not only reflect organic brain differences between individuals with schizophrenia and healthy controls but also highlight the relationship between psychotic symptoms, such as hallucinations and delusions, and performance on the test [71–74]. In the Available Energy domain, schizophrenia patients had less functional energy, possibly due to functional impairments spanning multiple domains that cannot be fully rehabilitated through treatment [75]. In the Social Skills domain, no significant differences were found between patients with schizophrenia and the healthy group across the indices. This may be related to cross-cultural findings indicating a higher prevalence of insecure attachment among Chinese individuals [76], suggesting that the CWS may have limited sensitivity in detecting social skills deficits in schizophrenia patients within this specific cultural context. This result may also be influenced by statistical bias due to the relatively small sample size. In the Emotional Functioning domain, schizophrenia patients exhibited lower emotional maturity and stability compared with the healthy group, supporting previous findings that individuals with schizophrenia have deficits in emotional perception and expression [77]. In summary, the main indices of reference for CWS interpretive domains can effectively assess characteristics of schizophrenia patients such as functional intelligence, organization of thought processes, and available energy, and can distinguish them from healthy individuals. Convergent validity Analysis of the correlations between the key major CWS indices and the key MMPI-2 scales revealed that EC+%, FQ+%, P%, P+%, IIT-2 ADJ and A% were all significantly negatively correlated with one or more key MMPI-2 scales. Among these, several scales showed significant negative correlations with the F scale, which reflects feelings of unhappiness and discomfort in respondents, and with the Sc scale, which can predict thought processes similar to those in schizophrenia, anhedonia, deficits in social skills, and difficulties in daily functioning [78]. On the one hand, these correlations are consistent with findings from studies in China showing that individuals with schizophrenia tend to obtain higher scores on the F and Sc scales [79, 80], supporting the relevance of these major CWS indices to certain personality characteristics of schizophrenia. On the other hand, these major CWS indices reflect the influence of symptoms such as delusions and hallucinations on the size, uniqueness, and complexity of drawings [57, 61]. Their correlations with the MMPI-2 validate the convergent validity of these indices for assessing reality testing ability, environmental adaptability, judgment, adherence to mainstream thinking, interpersonal relationships, and emotional stability. The negative correlation between A% and Pd scores is inconsistent with the description in the CWS manual, which associates higher A% values in psychiatric patients with psychopathology. This discrepancy may be attributable to a bias stemming from our small sample size or the fact that schizophrenia, while a form of psychosis, may not be representative of psychiatric patients more broadly [21]. For the main indices of reference for CWS interpretive domains, High Intelligence, Medium Intelligence, and Low Intelligence were all significantly correlated with multiple key MMPI-2 scales, each reflecting characteristics of cognitive impairment, disorganized thinking, and poor social adaptation in schizophrenia. This provides strong evidence for the construct validity of the CWS Functional Intelligence indices and also reflects the impact of functional intelligence, affected by psychotic symptoms such as disorganized thinking, on personality [21, 35]. In the Organization of Thought Processes domain, scores for Intact Reality Testing and Inadequate Reality Testing were all significantly correlated with the F scale, which reflects the severity of psychological distress. This supports the construct validity of these two main indices of reference for CWS interpretive domains, indicating that the CWS can effectively capture the core impairments in thought organization and reality perception in schizophrenia. Such impairments may contribute to the distinctive response patterns of schizophrenia patients in the CWS compared with healthy individuals, thereby showing strong associations with the F scale [21, 35]. However, the Disturbances in Thought Processes and Disturbances in Thought Content index did not show any significant correlations with the key MMPI-2 scales. Thought disorders in schizophrenia include delusion; intrinsic thinking disturbance; formal thought disorder; and deficient real-world knowledge, but not all schizophrenia patients display intrinsic thought disturbances [81]. Although both the CWS and MMPI-2 can reflect thought disorders to some extent, the dimensions they measure may not fully overlap. The Adherence to Conventional Thinking index was significantly negatively correlated with the F and Sc scales, while the Oppositional Style index was significantly positively correlated with the Pd and Sc scales. The two scales of the MMPI-2 also align precisely with the meaning of the CWS’s Oppositional Style index [21, 35]. These correlations reflect the links between these two indices and psychological distress or schizophrenia-related characteristics, validating their construct validity in assessing thought patterns in schizophrenia. The Adequate and Functional Energy dimension was significantly negatively correlated with multiple scales, including F, Fb, Fp, Sc, BIZ and TRT revealing its association with psychological distress and dysfunction [35] and confirming its convergent validity. Overall, both the major CWS indices and the main indices of reference for CWS interpretive domains showed significant correlations with MMPI-2 scales such as F and Sc, which reflect personality characteristics of schizophrenia, thereby validating the convergent validity of these indices. Criterion validity The statistical results for criterion validity confirmed that P%, and SIG% in the CWS have criterion validity for assessing negative symptoms in schizophrenia, as well as a broad range of psychopathological manifestations beyond positive and negative symptoms. Lower P% and P+% suggest that the thinking of individuals with schizophrenia diverges from mainstream thought and that they may present with symptoms of social functioning impairment or intellectual deterioration [21]. This is similar to the symptom descriptions in the PANSS Negative Scale, such as impaired emotional communication, passive or apathetic social withdrawal, difficulty in abstract thinking, and stereotyped thinking, as well as in the General Psychopathology Scale, which includes uncooperativeness, unusual thought content, poor judgment and insight, and active social avoidance [82]. Higher SIG% suggests possible cognitive deficits [21], which may correspond to symptoms measured by the General Psychopathology Scale of the PANSS, such as unusual thought content [82]. No other major CWS indices or main indices of reference for CWS interpretive domains showed significant correlations with the PANSS scales. Several factors may explain the non-significant correlation found between the CWS schizophrenia index and the PANSS scores. First, it is important to note that the CWS schizophrenia index is not a direct measure of symptoms but is derived from a constellation of underlying personality indicators. In contrast, the PANSS is the most commonly used clinical tool for assessing individuals with schizophrenia and evaluates symptom-level content [82]. The PANSS itself has been a subject of psychometric debate, with studies suggesting that instabilities in its structure and items could obscure its relationship with other measures [83]. Second, the modest sample size of our study may have had insufficient statistical power to detect a true, albeit potentially weak, relationship. Finally, no prior literature has validated the CWS schizophrenia index against an external clinical scale. Therefore, the analysis represents a preliminary, exploratory investigation into this index’s clinical utility, and this null finding itself is a valuable contribution that underscores the need for further research. But patients with schizophrenia and healthy controls showed significant differences in their scores on this index, also confirming the construct validity of this specific pathological index. This suggests that, although the index may have some potential for assessing schizophrenia, the symptoms it evaluates differ from those assessed by the PANSS, leading to the lack of significant correlations between the two. In summary, this study confirmed the criterion validity of several major CWS indices. Given the differences between personality measures and symptom assessment scales, future research should continue to explore more suitable criterion measures for such studies. Advantages In previous research, projective tests have often been considered to involve subjective scoring, with low inter-rater reliability [18, 84]. A key breakthrough of the present study is the use of a quantitative research approach to verify the inter-rater reliability of the CWS, with results closely aligning with those reported by Crisi [22], thereby providing empirical support for the objectivity of this scoring system. High reliability is the foundation for validity exploration. This study first established the objectivity of the CWS, showing that ratings by trained scorers are reliable and trustworthy, an achievement that is particularly valuable in the field of projective tests, which typically rely heavily on subjective interpretation. This provides a solid foundation for subsequent validation of construct validity and criterion validity, making the validity results more credible. Second, although the WDCT is widely used as a projective test, it lacks sufficient theoretical research, and previous studies have been conducted under different cultural contexts and scoring systems, making it difficult to quantitatively investigate and validate its underlying theoretical structure [85]. The CWS, as an objective and quantifiable scoring system, offers the possibility for such research [21, 22]. Building on previous studies, the present research verified the construct validity of the CWS, indicating that what the CWS measures is not merely a collection of disparate indices but rather a set of indices that can aggregate into latent structures with psychological significance. This provides strong empirical support for the theoretical model of the CWS, demonstrating that it is not a “black box” but a clearly structured assessment tool with a solid theoretical foundation. The present study also revealed its relationship with the MMPI-2, providing quantitative evidence to support previous case studies on the relationship between the CWS and the MMPI-2 [24, 41, 42]. In addition, the clinical utility of the CWS is immediately evident in its practical robustness. In this study, all participants provided a valid CWS protocol — in part because the standardized procedure allows for ‘additional instructions’ when necessary — a stark contrast to the high invalid rate for the MMPI-2. Finally, although the present study found that the CWS schizophrenia index was not significantly correlated with the PANSS and its criterion validity has yet to be established, it is well recognized in clinical practice that the assessment of schizophrenia cannot be accomplished using a single scale alone [8]. The finding in this study that the index has the potential to distinguish between individuals with schizophrenia and healthy controls offers a new possibility for assessing schizophrenia symptoms in clinical settings. Limitations This study has several limitations. First, regarding sample representativeness, the age range and geographical regions of the recruited schizophrenia patients did not encompass the entire schizophrenia population, and the sample size was relatively small. This was largely attributable to the study’s stringent inclusion criteria, which were implemented to ensure sample homogeneity but consequently limited the recruitment pool, particularly within the study’s limited timeframe. Consequently, the study was unable to conduct a comparison by further subdividing the patients with schizophrenia. The study was also limited to correlational analyses, as more advanced models like moderation would lack the statistical power required for reliable conclusions. Second, the study employed a cross-sectional design, which cannot establish causal relationships. Longitudinal studies are needed to investigate the potential of the CWS in predicting schizophrenia outcomes and its utility in clinical assessment over time. Third, the current study only tested the effectiveness of the CWS in differentiating between healthy individuals and those with schizophrenia, without exploring its capacity to distinguish between various psychiatric disorders. Therefore, this study cannot serve as evidence for the effectiveness of CWS in assessing schizophrenia. Further studies are necessary to validate the CWS’s utility across a broader spectrum of psychiatric conditions. Fourth, the test-retest reliability of the CWS was not assessed in this study. The nature of the CWS as a projective instrument necessitates a lengthy re-administration interval, typically six months or more, to minimize potential practice and memory effects. Such a longitudinal assessment was not feasible within the timeframe and logistical constraints of the current research. Fifth, potential sources of error include variations in test administration (e.g., different administrators and settings) and the inherent subjectivity in scoring projective measures (particularly for indices such as FQ). Although we demonstrated moderate to high inter-rater reliability, these factors could have influenced the results. Sixth, the existing literature on the CWS in schizophrenia is largely confined to the work of Crisi’s group [22]. This scarcity of comparative data prevents a systematic comparison with previous studies and an exploration of potential cross-cultural variations. Finally, as the first study in China to assess the reliability and validity of the CWS, future research should include replication studies and examine cross-cultural factors. Addressing these limitations will provide a more comprehensive understanding of the clinical value and applicability of the CWS across different contexts and populations. Conclusion In summary, the results of this study confirmed that the CWS demonstrates moderate to high inter-rater reliability and construct validity. The schizophrenia index showed significant group differences between patients with schizophrenia and healthy controls, suggesting its potential for distinguishing between these populations. At the personality level, several key CWS indices were significantly correlated with key MMPI-2 scales, supporting its good convergent validity in personality assessment. However, when PANSS scores were used as the criterion, the correlations with the CWS were relatively weak, and its criterion validity remains to be further verified in future studies. As the first CWS study conducted with a clinical population in China, this research not only established the reliability of CWS scoring but also validated its construct validity in a cross-cultural context and revealed its relationship with the MMPI-2 through quantitative analysis. In addition, the study found that the schizophrenia index has potential for distinguishing between individuals with schizophrenia and healthy controls. These findings provide new possibilities for the clinical assessment of schizophrenia in the future. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2