Evaluating career self-awareness instruments for victims of violence using the Rasch model

Abstract

Background

Victims of violence have trauma that often affects career development. A tool is needed to measure a person’s self-perception of self and work environment. The main problems in instrument measurement are items that are not independent of what is measured, instrument tests that are not linear and do not have significant value accuracy, lack of empirical coherence of responses, items, and constructs, and missing data errors. the purpose of solving these problems requires in-depth analysis using the Rasch model.

Methods

This research examines the validity and reliability of career self-awareness instruments with victims of violence. The research was conducted with 50 victims of violence in Bandung City. A quantitative approach with a cross-sectional study design was used. Data were analyzed using the Rasch model and Winsteps 3.73 application, Rasch model procedures and criteria are unidimensionality, person instrument analysis and item reliability, person item map, item difficulty, item fit, rating scale diagnostics, and item bias detection.

Results

The significance of the research findings is assessed based on unidimensionality in the “good” category, Analysis based on Andrich’s threshold shows the score increased to the appropriate level, Cronbach alpha value represents the interaction between person and item in the good category, and the person reliability value is 0.89 while the item reliability is 0.98.

Conclusion

Based on these results, a career self-awareness instrument can be used to assess and measure the level of career self-awareness of victims of violence.

Introduction

The career self-awareness instrument (CSAI) was developed to complement the shortcomings of existing self-awareness scales that ignore the empirical coherence of responses, items, construct validity and even confidence for missing responses, as well as raw scores do not show individual abilities and the difficulty level of questions/statements (not linear). To optimize the results of the analysis using the Rasch model which has an interactive process that is carried out repeatedly until the researcher finds the optimal composition of the instrument developed because all criteria are met.

Self-awareness is an essential aspect of achieving individual goals, including career development. Someone who has solid self-awareness can recognize the obstacles they face in achieving their goals and improve their choice of strategies to solve problems [1,2,3]. In a career context, individuals with goals would be able to determine positive actions for themselves in career development to achieve career success [4]. Savickas’ career construction theory states that career success requires an individual’s efforts to meet his or her expectations and the expectations of others [4]. One of these efforts is to develop a deep self-understanding and understanding of the world of work and responses to stimuli from the environment, cognitive phenomena, specific events [5, 6] and better recognize their own goals [5, 7, 8]. Therefore, highly self-aware individuals might compare ideal conditions with actual conditions in the struggle to achieve their goals [3].

Many studies have been conducted on self-awareness instruments, but none have been specifically related to individual career development. One of the first attempts to develop a scale to measure self-consciousness was made by Fenigstein, Scheier, and Buss: The Self-Consciousness Scale (SCS) [6]. The SCS instrument showed that self-consciousness is stable enough to be considered a personality trait [9]. Other limitations The scale used includes 3 alternative choices, whereas to analyze behavior at least 4 alternative choices.The SCS consists of three subscales: private, public, and social anxiety [6, 10, 11]. There are several versions of the SCS [12, 13] and the SCS has been translated into several languages [14]. In 1990, Trapnell and Campbell reassessed the psychometric properties of the SCS and showed that the personal self-awareness subscale includes two distinct constructs, self-reflection and self-contemplation [15].

Whereas the Situational Self-Awareness Scale (SSAS) instrument [16]. has limitations that only focus on environments that highlight unique characteristics (e.g., being the only male in a group of females) thus leading to individuation [15]. in addition the instrument only encourages similarities in behavior, appearance, and values (e.g., soldiers) [17, 18]. and awareness and focus on cognitive schemas where self-perception outcomes vary based on context or relationships [19].

The new scale is necessary because the existing scales in the literature are too limited. They do not focus on the complexity of the self-awareness process and only partially utilize the dimensions of self-awareness that operate in the context of career development. Therefore, each of the above scales has desirable but incomplete aspects. For example, Fenigstein, Scheier, and Buss’ scale considers the ideas of self-reflection and attention to inner thoughts and feelings [6, 20, 21, 22]. Career awareness, which is one of the sub-concepts, is the degree of individual knowledge and understanding of one’s setting and opportunities associated with occupational fields or work groups to prepare for roles in the professional world [23,24,25,26].

An instrument must be tested to ensure that it is valid and reliable with the actual situation, a valid, consistent, and accurate instrument is needed that provides reliable research data [27]. If the instrument is tested based on classical model analysis, it is feared that the raw data based on the rating response cannot represent the original characteristics of quantitative data which is a continuum, a higher accuracy ratio on the principle of probability [28, 29]. so the main reason for using the Rasch model rationale is consistent with existing research data [30].

The study highlights five essential parts in the analysis with the Rasch model, including the calibration and item estimation capabilities, the item characteristics in the parameter model, the item information functions and the instrument, the interaction maps between items and respondents, and the fit between items and respondents. These advantages make the results of the statistical analysis of the Rasch model in the conducted research more accurate and, more importantly, provide standard error values for the instruments used. Thus, it could increase the accuracy of the calculation [27, 31]. Furthermore, Sumintono & Widhiarso explained that testing respondents (individuals) and items simultaneously showed that quantitative research could be conducted by scientific and social science disciplines usually identified with qualitative research [31]. In this study, victims of violence were included as study participants. Researchers concluded that victims of violence had low career self-awareness based on their analysis of the literature. Several studies provide evidence of self-destructive behavior after victimization [32]. Green said concluded that feelings of worthlessness, crime, and self-loathing in individuals who experience violence are the aftermath of the assault, rejection, and parental scapegoating from self-destructive behavior [32].

In women, victimization has a greater impact on various cognitive aspects that can affect work efficiency, sleep disturbances, and the ability to apply social and organizational rules [33]. Male victims of violence have difficulty understanding emotions, thoughts, and expressions that should be shown [32, 33]. The conditions described implicitly illustrate that victims of violence have low career self-awareness. Based on the analysis of the previous findings, a specific instrument is needed to describe issues related to career self-awareness for victims of violence. The career self-awareness instrument (CSAI) was developed to complement the shortcomings of existing self-awareness scales including strengths, weaknesses, interests, values, and career development and needs to be tested for validity and reliability to measure the career self-awareness of tested victims of violence. Instrument testing using the Rasch model is an important part of an effort to accurately analyze data both in terms of items and persons.

Methodology

This study uses a quantitative approach with a cross-sectional survey research design that measures attitudes and provides information quickly [34, 35] and collecting data at one specific time [35]. The demographic data of the participants in this study are shown in Table 1.

Participants

The total number of respondents who met the requirements and were willing to voluntarily complete the instrument through three social media platforms, WhatsApp, Instagram, and Facebook, was 50 people.

The population size has a value of more samples in 50 sample but the sample size in Rasch modelling can provide a solution that can be analysed through a linear scale, namely the logarithm odds unit. Thus item calibration is stable within ± 1 logit with 95% confidence level, sample range 16–26, hence sample size is feasible 30; item calibration is stable within ± 1 logit with 99% confidence level, sample range 27–61, hence sample size is feasible 50; item calibration is stable ± 0.5 logit with 95% confidence level, sample range 64–144, hence sample size is feasible 100; and item calibration is stable within ± 0.5 logit with 99% confidence level, sample range 10–243, hence sample size is 150.

Purposive sampling of those willing to participate as respondents. Data collection was carried out in 2021 in the Covid-19 (feb until July 2021) situation so that it was carried out online using a questionnaire made with Google Form media, the prerequisites for respondents to participate in the study as follows: (1) aged between 15 and 25 years old; (2) having experienced violence (physical, psychological, or sexual); and (3) filling out the questionnaire freely. These conditions were made to screen respondents who met the criteria for this study.

The participants were informed of the overall purpose of the study, and they were assured that the information they provided would be kept confidential. All individuals freely contributed to the study, with no incentives or rewards from the researchers.

The questionnaire was first distributed through two WhatsApp groups owned by the researcher, the number of participants in the two groups was the first group of 84 people, the second group was 250 people, from this distribution 5 respondents were obtained according to the researcher’s criteria.

The second questionnaire was sent through the researcher’s Instagram account, with the help of 30 fellow researchers. Researchers and 30 co-researchers have 15,000 Instagram followers. The distribution of the questionnaire resulted in 42 respondents. The questionnaire was not sent through the Facebook page of one of the researcher’s colleagues. The account currently has 100 friends. The results of the questionnaire distribution showed that three participants. Table 1 demonstrates the data of the participants:

Table 1 Demographic characteristics of the respondents

Career self-awareness instrument development

Here are the details of the instrument development, Table 2.

Table 2 Stage career self-awareness instrument development

The final stage of instrument development was to conduct a readability test on respondents with similar characteristics, a total of three people who were not used as a population or sample in the study. The basis for selecting a Likert scale is to examine individual attitudes in detail and accurately based on individual conditions that are generalized in the form of statements at different levels. decision of 25 items based on the results of the Rasch model analysis that has been carried out. The results of the readability test were carried out with the aim that the instrument items could be easily understood according to the conditions, the items minimized incomplete sentences, and the items formed a systematic answering pattern based on statements experienced in real life.

Data analysis procedure

The following describes the steps taken to prepare the instrument: (1) SCS and CAI provided a framework for constructing the instrument; (2) Items Development; (3) Assessment of validity; (4) Test preparation; (5) Psychometric quality estimation; and (6) Instrument publication. The relationship between theoretical constructs (SCS and CAI) and the Rasch model based on the theory of career awareness is a development process that is formed through the process of interaction between humans and their environment, the aspect of achievement can change and build individuals who have optimal career awareness, the accuracy of the level can be seen from the analysis of the Rasch model rating scale, and to find out valid items and the right people to be used as samples can be studied based on statistical conclusions in the Rasch model. The resulting test results were then processed using Rasch Modeling and the Winstep application. The raw score was processed in the Rasch model, and finally, the construct map information was completed. The analysis of the items used is calibrated simultaneously: the measurement scale, the respondents (person), and the items; thus, the desired data could be obtained [28]. Analysis with the Rasch model produces a statistical analysis of suitability which provides information to researchers on whether the data obtained ideally show that individuals with high ability give response patterns to items that correspond to the level of difficulty [27]. The purpose of the Rasch model is to meet the objective measurement criteria so that the instrument has the same quality as science in the social sciences [28].

The steps of the Rasch model analysis

• 1. The analysis process using the Rasch model begins with a one-dimensional measurement to test the accuracy of the items with the model.

• 2. The measurement of item bias. This measurement was done because each item could be biased, especially if an item favors a person with certain characteristics. Meanwhile, individuals with oppositional characteristics are disadvantaged [28].

• 3. Rating scale diagnostic.

• 4. The individual measurements. In addition to measuring individual abilities more accurately, the Rasch model could also determine the accuracy of abilities based on given response patterns [28].

• 5. The Cronbach Alpha value shows the interaction between person and item [31].

Results

Measurement scales

Unidimensional

The analysis of unidimensionality determines how many attributes or dimensions are measured by the instrument. This analysis used Output Table 2 to observe the Raw variance explained by measures and Unexplained variance in the 1st to fifth contrast. According to Sumintono & Widhiarso, the unidimensionality measurement could be proven if the Raw variance explained by measures ≥ 20% with a note that the general criteria for interpretation are in an acceptable category if the variance is around 20–40%, good category if it is 40–60%, excellent if it is above 60%, and if Unexplained variance in first to fifth contrast of residuals is < 15% each [31]. The unidimensional are presented in Table 3 as follows.

Table 3 Unidimensional

The results of data analysis showed that the Raw variance explained by measures is 32.1%, which means it is in the acceptable category. Meanwhile, the Unexplained variance in the 1st to fifth contrast of residuals is 12.0%; 9.6%; 7.1%; 5.1%; and 4.1%. It appears that each variance is less than 15%. Therefore, the constructed instrument used measured one variable, career self-awareness.

Instrument analysis

The information is presented in the output Table 3 Summary Statistics is used in the instrument analysis. In detail, the analysis of the instrument is given in Table 3.

Table 4 Summary statistics

Person measure showed the average score of all participants working on each data instrument item for disclosing career self-awareness. The average person value, greater than the average item (where the average item is 0.00 logit), indicates that the participants’ ability is generally greater than the difficulty level of each item in the instrument.

The Cronbach Alpha value, which represents the interaction between the person and the items as a whole, is 0.89, which is in the very good category. Based on the criteria presented by Sumintono & Widhiarso, the person reliability value is 0.63, which is in the low category, while the item reliability is 0.92, which is considered very good [31].

Other data in Table 3 that could be used are INFIT MNSQ and OUTFIT MNSQ in both the Person and Item tables. Based on the Person Table, it is known that the average Infit Mean-square and Outfit Mean-square values are 1.02 and 1.00, respectively. Meanwhile, based on the Item table, it is known that the average Infit Mean-square and Outfit Mean-square values are 1.01 and 1.00, respectively. If it is closer to number 1, the criterion is better because the ideal value is 1 [28]. Therefore, the average person and item are close to the ideal criteria.

Meanwhile, related to Infit Z-standard and Outfit Z-standard, the average values for the person are − 0.06 and − 0.09, respectively. Meanwhile, each item’s Infit Z-standard and Outfit Z-standard values are 0.03 and 0.01. The ideal Z-standard value is 0; the closer to 0, the better [31]. Thus, the quality of the person and item is good.

From the output of Table 4 it is known that the person separation is 1.30, and for items is 3.38. The greater the separation value, the better the quality of the person and instrument. The separation value is calculated more accurately through the formula: H= {(4 x separation) + 1}/3. Therefore, the person separation value is 2.07, rounded down to 2, while the item separation is 4.84, rounded up to 5. This result means that research participants have a variety of abilities that could be categorized into two groups. Meanwhile, the difficulty level items were spread into five groups starting from the easiest to the most difficult.

Participants

Fig. 1

Wright Map Analysis (Person-Item Map)

Referring to Output Fig. 1. Wright Map, it is known that the Wright Map Analysis Career Self-Awareness on victims of violence spreads in the range of 0 to 3 logit.

Their Career Self-Awareness management positions are mostly between 0SD and + 1SD. Nevertheless, some of them have abilities that are outliers, which means extremely high. The logit average of the Wright Map Analysis Career Self-Awareness on victims of violence is + 1.37, above 0.00, the average logit item, representing the average of respondents above the average standard difficulty level item.

Meanwhile, the difficulty level item map spreads from − 1 to 1 logit. The position of the difficulty level of 24 items is between 0SD and + 1SD, while one item, number 1, is below − 1SD. Therefore, item number 1 has an outlier of item difficulty level. The average standard item difficulty level is below the level of the respondents’ ability. Thus, the Career Self-Awareness instrument items were easily approved by the respondents.

Items analysis

The analysis of this item includes the difficulty level item (item measure), the suitability level item (item fit), and the item bias detection.

Difficulty level item

To find out the difficulty level of the items, it could be analyzed from Table 1 Item Measure Order. From the table, it is known that the SD value is 0.61. If the SD value is combined with the average logit value, the item difficulty level could be grouped into the very difficult category (greater + 1 SD), the difficult category (0.0 logit + 1 SD), the easy category (0.0 logit − 1 SD), and very easy category (less than − 1 SD). Thus, the value limit for the very difficult category is more than 0.61, the difficult category is 0.00 to 0.61, the easy category is -0.61 to less than 0.00, and the very easy category is less than − 0.61. The following is Table 5, relating the difficulty level item in detail.

Table 5 Difficulty level item

By looking at the logit value of each item in Table 5 difficulty level item, sequentially based on the level of difficulty (from the most difficult item to the easiest), it is known that five items fall into the very difficult category, namely item numbers 2, 16, 15, 19, and 21. The difficult category has eight items numbered 12, 18, 8, 20, 24, 6, 10, and 11. The easy category has eight items; they are 13, 14, 23, 22, 5, 25, 3, and 7. At the same time, the very easy category has four items: numbers 9, 4, 17, and 1.

Suitability level item

The suitability level item with the model (item fit) explained whether the item functions normally in taking measurements so there would be no misconceptions.

about the individual regarding the items. Suitability level items could be examined based on the data in Table 5 Item Fit Order, namely the Outfit Mean-square (MNSQ), Outfit Z-standard (ZSTD), and point measure correlation (PT MEASURE CORR) columns. The criteria for checking item fit or item misfit (outlier or misfit) according to Boone, Staver, & Yale (2014) are as follows: (1) The Outfit Mean-square value is greater than 0.5 and less than 1.5. The closer to 1, the better; (2) The Outfit Z-standard value is greater than − 2.0 and less than + 2.0, the closer to 0, the better; and (3) Point measure correlation values are more than 0.4 and less than 0.85. An item can be considered fit if it meets at least 2 of the three criteria [31]. Table 6 shows the details.

Table 6 Suitability level item

Table 6 shows the 1st criterion, which was that there are no misfit items; each item has a score greater than 0.5 and less than 1.5. According to the second criterion, three items are misfits; they are item numbers; 16, 6, and 13. Meanwhile, based on the third criterion, seven items are misfits; they are; 1, 4, 7, 9, 17, 22, and 25.

Rating scale diagnostic

This diagnosis determines whether the participants understand the differences in the answer choices. Respondents understand the difference in answers if the observed average and Andrich threshold values are increased by their level; the detail of the Andrich threshold values is presented in Table 7.

Table 7 Rating scale diagnostic career self awareness

Table 7 shows the suitability and the same increase in alternative levels 0, 1, 2, 3, and 4. The analysis results showed that the levels on the CSAI instrument correspond to the real behavioral conditions of victims of violence.

Item bias detection

Overall the logit position for each item based on gender is in Fig. 2.

Fig. 2

Logit Position of Each Item Based on Gender

Another measure of validity is that the instruments and items used do not contain bias because one individual with certain characteristics is more favorable than another. An item statement is said to contain bias if the probability value of the item is below 0.05 [36]. In this study’s context, bias only appears in terms of gender. The results of the bias analysis based on gender show that 25 items have a probability value > 0.05.

Discussions

Validity test

The validity test aimed to measure the instrument’s accuracy in research. The validity of an instrument that uses the Rasch model analysis can be comprehended by the test results for unidimensionality, rating scale, item bias detection, and item suitability level [28, 31].

The analysis of the unidimensional test shows that the construct of the instrument used can measure what needs to be measured, specifically career self-awareness. This result is shown in the results of raw variance explained by measures with a value of 36.8% and unexplained variance in 1st to 5th, each of which has a value of less than 15%. Dimensionality is a fundamental measure for determining the construct validity of an instrument [37]. Thus, the CSAI described each aspect of career self-awareness without deviating from what was determined to be: (1) Personal self-awareness, (2) World of work awareness, and (3) Self-attitude. Based on the criteria presented by Sumintono & Widhiarso, the CSAI instrument is included in the sufficient category. Hence, the CSAI instrument measures one variable, career self-awareness, without being influenced by other variables.

In addition, the analysis results show that the participants understood the differences in the response options. This result was based on the rating scale diagnostic results, which showed that the Andrich threshold increased with each level. Regarding the detection of item bias, it is known that the items of the CSAI instrument have a probability value of more than 0.05. This result indicates that the CSAI instrument neither benefits nor harms a particular gender [36]. Biased items could discriminate against individuals in other categories, leading to different results [30]. Therefore, the CSAI instrument could be used for all types of gender in victims of sexual violence. The use of instruments for two different groups is based on Ginzberg’s study, which represents that between the ages of 11–17, adolescents’ career development is in the tentative stage, which is a transition period from the fantasy stage of childhood to the realistic decision-making stage of young adulthood. Ginzberg argues that adolescents progress from the stage of evaluating their interests (11–12 years) to the stage of evaluating their capacities (13–14 years) and then to the stage of evaluating their values (15–16 years). Around the ages of 18 to 25, their thinking shifts from more subjective career choices to realistic career choices. During this time, individuals extensively explore available careers, then they focus on a particular career, and finally choose a specific job within that career.

In addition, the items in the suitability level were found to meet two of the three criteria mentioned by Boone, Staver, and Yale [36]. Briefly, the MNSQ (mean-square) is a chi-square calculation (measuring the degree of association) for the outfit and infit statistics [36]. Namely, the Mean Square Outfit value is used to test the consistency of responses with the difficulty of the statement items. The ZSTD (z-standardized) value provides a t-test statistic that measures the probability that the MNSQ calculation is random when the data fit the Rasch model [36]. It could be understood that the Z standard fit is used to describe how many of the measured column results are an outlier, unmeasurable, too easy, or too difficult. In addition, point measure correlation describes how good (SE) the items are by checking whether statement items are not understood, respond differently, or are confused with other items. From the results, several items did not meet the second criterion (numbers 16, 6, and 13) and the third criterion (numbers 1, 4, 7, 9, 17, 22, and 25). However, referring to the view of Sumintono and Widhiarso, 25 CSAI items are declared suitable because they usually work and can be properly understood by respondents and measure what needs to be measured, which in this case is career self-awareness [28].

Reliability test

Reliability is the constancy or consistency of a series of measuring instruments. A measuring instrument is reliable if measurements are made repeatedly, and the results remain consistent. The reliability test was carried out using the Rasch model according to the criteria presented by Sumintono and Widhiarso, which was understood from the results of the person measure test, the Cronbach Alpha value, the Person Reliability and Item Reliability values, and finally the result of the separation value [31].

The participants in this study were 50 individual victims of violence of different gender, ages, and ethnicity. The results of the person measure showed that the respondents generally have higher ability than the difficulty level of the items, as the mean value is above the logit value of 0.0. It is known that five items fall into the “very difficult” category: item numbers 2, 16, 15, 19, and 21. It is known that eight items are included in the difficult category: numbers 12, 18, 8, 20, 24, 6, 10, and 11. It is known that eight items are included in the easy category: numbers 13, 14, 23, 22, 5, 25, 3, and 7. In contrast, four items are in the “very easy” category: 9, 4, 17, and 1.

Moreover, the interaction between a person and an item is included in the category “excellent”. This interaction is shown by the results of Cronbach’s alpha value of 0.89, which falls into the “excellent” category. The reliability of the instrument based on the Cronbach Alpha value is divided into five categories, namely: (1) Bad if the value is less than 0.5; (2) Ugly if the value is between 0.5 and 0.6; (3) Enough if the value is between 0.6 and 0.7; (4) Nice if the value is between 0.7 and 0.8; and (5) Very good if the value is greater than 0.8 [28].

Moreover, person reliability as an indicator of the respondent’s answers consistency is included in the low category. This result is indicated by the person reliability value of 0.63. Meanwhile, item reliability as an indicator of the quality of the items in the instrument belongs to the very good category. This finding is shown by the item reliability value of 0.92. Instrument reliability on the value of person reliability and item reliability is divided into five categories as follows: (1) Low, if the value is less than 0.67; (2) Sufficient, if the value is between 0.67 and 0.80; (3) Good, if the value is between 0.81 and 0.90; (4) Very good, if the value is between 0.91 and 0.94; and (5) Excellent, if the value is greater than 0.94.

Reliable data can be used as a basis for obtaining data that can describe clearly so that the results of the description can be used as a reference to improve efforts to improve the quality of education, especially for victims of bullying.

subsequently, the person and the item have an average value of the infit mean-square and the outfit men-square, which are close to 1. therefore, the person and the item questions are close to the ideal criteria. the z-standard infit and z-standard outfit values for person and item are close to 0. thus, the quality of the person and item questions is in a good category.

The last instrument refers to the separation or grouping of persons and items. Individual separation indicates how well the set of items in the career self-awareness instrument is distributed across the logit ability range. The greater the individual separation, the better prepared the instrument is because the items in it can reach individuals with high to low levels [31]. Item separation shows how the measured sample is distributed along a linear interval scale. The higher the item separation, the better the measurement. This index is also helpful in defining the meaningfulness of the measured construct. Based on the results of calculating the separation of individuals and items in the CSAI instrument, it was found that respondents were divided into two kinds of abilities. In contrast, the items were divided into five groups, starting with the most difficult item and ending with the most accessible item.

The basis for considering the use of the CSAI instrument based on the context of Education, the instrument is very sustainable and unique to be followed up by counselors in schools to identify career readiness in victims of bullying, the data produced can be used as guidance and counseling services (curriculum guidance, individual planning and support system). While the CSAI instrument is studied based on the culture of implementing values ​​oriented towards individualism-collectivism and horizontal-vertical orientation, individuals from the support process, while vertical orientation is less able to internalize ways of thinking and acting properly.

Conclusions

The results of the unidimensional test showed a sufficient category. In addition, the results of the rating scale test showed that the respondents could understand the different answer choices and that there was no bias in the items. Furthermore, finally, the suitability level test showed that the items of the CSAI instrument were suitable. These results show that the CSAI instrument can be used to explore career self-awareness.

The person-measure test results showed that the respondents’ ability level was higher than the difficulty level item. Then the Cronbach Alpha results are very good in the category. Moreover, the reliability value for the person is low, but the reliability value for the item falls in the very good category. Finally, the grouping of persons and items has a high value. These results show that the CSAI instrument can be as a data collection tool.

Based on these results, a career self-awareness instrument can be used to measure the level of career self-awareness of victims of violence. This instrument could meet the measurement criteria to empower individuals to understand themselves and their work environment to achieve optimal career success.

The implication value of the instrument is that the data on the CSAI instrument is an analysis process using an interactive Rasch model that is carried out repeatedly until the researcher finds the composition of the instrument that is optimally developed because all criteria have been met. Researchers can follow up on the measurement of instrument data using classical theory and compare data processing using the Rasch model with the Winsteps application examine the level of CSAI through a qualitative approach with a phenomenological design. Data in the form of the same interval determines the quality of accurate analysis results as an effort to improve the quality of education.

Limitations and suggestions

The development of career self-awareness instruments is still limited to disseminating instruments online via a Google form, and few respondents are included. Further research is needed to use more respondents to make the data obtained more relevant. For future research, it is suggested to develop career self-awareness instruments at different educational levels.

The population size has a value of more samples in 50 sample which indicates the results of instrument calibration are very unstable and insensitive or biased. but the sample size in Rasch modelling can provide a solution that can be analysed through a linear scale, namely the logarithm odds unit. Theoretically, the stability of the item calibration corresponds to the model of the standard error or SE. If larger samples can be obtained, instrument testing can be done by grouping on smaller homogeneous groups such as by gender or age, and to determine the stability of item calibration in different measurement situations.