Questionnaire Development and Validity to Determine Intention Among Indian Youth to Help a Road Trac Accident Victim

Delay in receiving timely care after road trac accidents (RTAs) result in higher morbidity and mortality, suggesting need of educational interventions for improving provision of rst aid. Based on the theory of planned behaviour (TBP), the present study aims to develop and validate the scale to Measure of Intention to Help Road Accident Victim (MIHRAV) of young adults in a fast urbanizing Indian city. A cross sectional survey was conducted among the college students of Jodhpur, Rajasthan. The initial MIHRAV questionnaire developed for study comprised of 26 self-reported items encompassing four main constructs: intention to help; social norms; attitude and self-ecacy. The validation process included face validity, content validity, exploratory factor analysis (EFA), conrmatory factor analysis (CFA) and reliability analysis. Descriptive analyses were performed on all items, followed by exploratory factor analysis (EFA), and reliability analysis. Kaiser-Meyer-Olkin (KMO) test and Bartlett’s test of sphericity were done for aptness of data for EFA. EFA was performed using maximum likelihood with oblique rotation (Promax). Factor retention was based on the Eigenvalue more than one, and point of inexion on the Scree plot. Variables with factor coecients of 0.40 or more were retained. The reliability of the questionnaire was determined by internal consistency (Cronbach’s alpha) coecient and a Cronbach’s alpha coecient

systems that serve as the rst point of contact is essential to ensuring timely and equitable access to care for the injured.
As per Sasser et al. 4 the deaths of severely injured RTA patients occurring commonly due to airway compromise, respiratory failure or uncontrolled hemorrhage can be reduced by providing basic first aid measures. These rst aid measures can be administered by trained volunteers and non-health professionals.
Given the gaps in organized prehospital care in India and other low-or middle-income countries, it has been suggested by the WHO that bystanders and laypeople must be taught the basics of care to trauma victims. 5 Bystanders are de ned as people who attend at the crash scene without any knowledge and skill about relief, rescue, and rst aid to the RTAs victims. 6 Sometimes, the congestion of laypeople at the crash scene can affect the relief forces' function leading to new crash for those who are observing the scene. 6 However, in some cases, bystanders with adequate knowledge and understanding about aiding the injured can perform basic life support and initial measures for injury before arrival of the emergency organizations and in some cases, provide transportation. 7,8 Therefore, they can play a signi cant role in reducing the consequences of RTAs.
Prosocial behaviour also known as helping behavior was de ned by Stang et al. 9 as a voluntary behavior performed with the intention of bene ting another person or group of people. Situational characteristics in form of "diffusion of responsibility"; "ambiguity in interpreting situation" and "cost reward analysis of the helping act" along with individual personality traits such as "adherence to norm of social responsibility"; "empathy"; "pro-social values"; "level of moral development" are hypothesized as predictors determining helping behavior especially in emergency situation. [10][11][12][13] It is thought that providing training to laypeople in basic life support will improve the quality of care at the scene of RTAs, although little consideration has been given in these interventions to date on securing community engagement and maximizing participation. When teaching a new skill such as rst response, educators focus mostly on the what and how (cognitive and psychomotor skills) rather than the why, which falls under the affective domain of learning. 14 Previously several studies observed that information about helping or acquisition of skills does not always increase bystander responsiveness. 8,15,16 The current approach to provide bystander training in form of self-directed learning or instructor-led course focuses upon building individual skills acquisition and retention. However, no evaluation of a learner's intention of volunteering help in incident of RTA or theory-based attempts to strengthen their intentions to engage in this behavior are reported so far. Further, currently there is no clear understanding of predictors likely to in uence the strength of intention of young adults for providing care at the site of RTA.
According to well-validated theory based models (e.g., The Theory of Planned Behavior and Theory of Reasoned Action), an individual's behavioral intentions in uence their motivation for engaging in a particular behavior. 17 The theory of planned behavior (Ajzen, 1985(Ajzen, , 1988) is a social cognitive model used to predict people's behavior. It predicts that an individual's action largely re ect their attitudes, perceived norms and accepted behaviors and how well they believe they can perform the task. 18,19 Intention focused model to enhance performance of CPR by bystanders was conceptualized by Panchal et al. 20 Given the motivational component in giving rst aid at the site of road tra c accident, we have selected the theory of planned behavior (TPB) as a useful framework to investigate the young adult intention to help at the scene of a road tra c accident. To our knowledge, TPB has not previously been applied to the prediction and understanding of bystander intention to help at the site of RTA. Objective: The aims of the present study were threefold: First, we used the TPB to model the relationship between norms, attitudes and perceived behavioral control on intentions to help the road tra c accident victim.
Second, we evaluated whether attitude, subjective norm, perceived behavioral control could be regarded as good predictors of intention as speci ed in the Theory of Planned Behaviour. Third, we examined whether gender of an individual and their previous experience of helping an accident victim differ systematically with respect to the variables of the TPB.

Methods
This study was conducted as part of a funded project carried out to evaluate pre-hospital trauma care systems in Jodhpur, Rajasthan. This study consisted of two phases: Phase 1-questionnaire development and expert panel review and Phase 2-validation process.
Phase 1: Questionnaire development A literature search was conducted to identify instruments that contain item related to factors in uencing bystander's intention and helping behavior toward road tra c accident victims. The search was guided by using combination of different keywords including helping behavior, accidents, tra c accidents, bystanders, theory of planned behavior. The search engines included PubMed, Science Direct and Google Scholar. As research team was unable to identify any questionnaire related to measure of intention among young adults as bystanders, items for new instrument was framed through literature review as per the constructs of theory of planned behavior. 22 Items and scoring were formulated in accordance with the paper by Francis et al. 23 The provisional questionnaire contained questions on background information (age, gender, year of course, institute) and 27 self-reported items measuring the TPB constructs and perceived barriers of participants towards providing care to an accident victim. 24,25 . The initial instrument included a total of eight question domains. One item was removed after discussion with investigators because of unclear meaning leaving 26 items in the administered questionnaire(supplementary le 1). See Table 1 for initial instrument strati ed by domain and item identi er. Each item was measured on a 7-point Likert scale (1=disagree/ unlikely/ unpleasant, 4=moderately and, 7=agree/ likely/ pleasant), high scores indicated a positive evaluation of the survey questionnaire. Previous experience of helping an accident victim was assessed with a single item dichotomously coded as "yes" and "no".
Two forward translations into Hindi were done by a linguistic expert and a medical doctor. Subsequently, the two sets of translated Hindi questionnaires underwent a backward translation into English by another linguistic expert and a medical doctor. The aim of this procedure was to identify translated items that closely match the original English version and to produce the Hindi version of the questionnaire Phase 2-Validation process The validity of MIHRAV was assessed using content validity, face validity, exploratory factor analysis (EFA), reliability analysis and con rmatory factor analysis (CFA). For content validity the peer review was invited by experts in eld of emergency and trauma care and edits were made based on provided feedback. During face validity the investigator distributed the pre-nal version of the questionnaire to the target population for reviewing grammar and appropriateness of the questionnaire. During this phase, ten young adults were cognitively debriefed. Ten respondents are su cient for pilot testing using an interview type of pilot testing (Eremenco, 2005). 26 The respondents were interviewed and asked to identify words or sentences that they did not understand. The respondents were encouraged to suggest alternate suitable words with which they were familiar. The feedback of participants was discussed with other investigators in the research team, and the nal version of questionnaire was decided.
For subsequent factor analysis, from July 2019 till October 2019 the investigators administered the questionnaire among students studying in academic institutions of urban Jodhpur, Rajasthan. The data of educational institutions in urban Jodhpur was availed through annual report of year 2018 -2019 on Government of Rajasthan website for higher and technical education. 21 The colleges of medical, engineering, and other allied disciplines (pharmacy, nursing, and dental) were excluded, as students of these institutes couldn't be considered representative of general population due to their chosen study of discipline. Out of twenty-four professional colleges in the urban Jodhpur, six colleges were randomly chosen by lottery method (without replacement). These participants were likely to possess negligible or limited training to act at RTA site as learning opportunities are rare and are not part of academic curriculum for non-health students in this part of country. All the students present during data collection period in these institutes were approached for the study purpose. Students were invited to participate through the college administration and were repeatedly reminded to participate. A print version of the questionnaire was administered to the students in 30 minutes' class room setting on pre-speci ed days.
Two research assistants were present throughout to support students. Sample size: Sample size was estimated on the basis of planned procedure for factor analysis. Thus, as recommended to ensure a conceptually clear factor structure for analysis, a sample of minimum 5 participants per item was considered. Thus, the required minimum sample size was estimated to be 150 (26 items were in the questionnaire). 27 Ethics approval: This study was part of an Indian Council of Medical Research project approved by the AIIMS Jodhpur Ethics Committee. Ethical clearance for performing this study was obtained from the Institutional Ethics Committee (AIIMS / IEC/2018 /1188, dated 02.05.2018).
Informed Consent: All the eligible participants were informed about the purpose of the study, and were assured regarding the con dentiality of the information obtained. Written informed consent for participating in the study was obtained for all individual participants included in the study. No monetary reward was provided for participant willing to take part in the study. The items were not compulsory and participants had the option of abstaining from each question. The con dentiality of the dataset was maintained by removing the identi ers from the interview schedule.
Statistical analysis: Responses were coded, and entered in Microsoft excel. No items were reverse coded for data analyses. The distribution of the responses was inspected for each item to eliminate items with a low discriminative power. These were (a) items with 95% or more of the given answers in the same category, and (b) items with a standard deviation lower than .75.
Statistical analysis was performed using STATA version 13.1 (StataCorp LP, College Station, TX, USA). Descriptive analyses were performed on all items, followed by exploratory factor analysis (EFA), and reliability analysis. Kaiser-Meyer-Olkin (KMO) test and Bartlett's test of sphericity were done for aptness of data for EFA. EFA was performed using maximum likelihood with oblique rotation (Promax). Factor retention was based on the Eigenvalue more than one, and point of in exion on the Scree plot. Variables with factor coe cients of 0.40 or more were retained. The reliability of the questionnaire was determined by internal consistency (Cronbach's alpha) coe cient and a Cronbach's alpha coe cient > 0.70 and < 0.90 was considered an indicator of reliable scale. 28 Con rmatory factor analysis (CFA) was performed to evaluate relationships between structural paths and latent variables (factors) using AMOS 26.0. The hypothesized model was derived using the latent constructs of attitudes, norms, perceived con dence, and intentions as per TPB model. CFA was performed with the mean values of each domain identi ed and its respective factor identi ed in the EFA, however no cross-loadings were allowed.
Parameters were estimated for the CFA model based on the maximum likelihood procedure (sometimes called path analysis) involving tting the variances and covariance's among observed scores. It was essential to identify the four-factor model in order to estimate the model parameters. In conducting CFA, no warning messages were received from AMOS 26.0 regarding parameter estimates. Based upon this information, the four-factor model passed the "rules" for identi cation. We also examined the assessment of universal t pertaining to the quality of the model in SEM in order to support or reject its appropriateness for the population examined. The next step was to illustrate the observed (items) and unobserved (factors) in the hypothesized model ( Figure 1). The observed variables are represented as rectangles; ellipses represent the unobserved variables (latent variables) and the circles represent measurement error. The arrow between the unobserved variable and the observed variable represents a regression path and its number represent the standardized regression weight. The arrow between a small circle and the observed variable represents a measurement error term.
For CFA, the recommendations of Hu and Bentler (1995) were followed and the model t was evaluated by means of various goodness of t indices. 29 A statistically signi cant chi-square test suggests that model lack the t to data. In addition to chi-square, use was made of chi-square divided by the degrees of freedom, which is less sensitive to sample size. For this estimate, values below three are considered satisfactory (Bollen&Long, 1993). 30 Other indices were the root mean square error of approximation (RMSEA), the tucker Lewis Index (TLI) and the comparative t index (CFI). For the RMSEA, values below .05 are considered as a good t, values between .05 and .08 as acceptable, and values higher than .08 as reasonable errors of approximation in the population (Browne & Cudeck, 1992). 31 CFI compares the t of a null model (i.e., when unobserved variables are uncorrelated and independent) with the t of the researcher's model. 32 A CFI value of greater than 0.90 shows a psychometrically acceptable t to the data. 33 The GFI shows the degree of variance and covariance together explained by the model. The value of GFI ranges between 0 and 1. A value of 1 indicates a perfect t.

Results
A total of 695 participants responded to the questionnaire, of which 662 were included after removing the data with missed responses for one or more questions. Item analysis: Mean and standard deviation of instrument items by behavioral domain is illustrated in Table 1. All showed su cient variation across the response categories (i.e., less than 95% of responses on a single category and SD more than 0.75).
All participants with previous experience of helping an accident victim scored signi cantly higher (p<0.05) on all items as compared to participants with no experience. For majority of items, male participants scored signi cantly higher(p<0.05) as compared to female participants except for following; perceived ability to learn and perform in pre-hospital trauma care training course (item c and q); attitude toward administering care to road accident victim (item h, n, s) and in uence of friends and bystanders at an accident site towards helping a RTA victim (item b and bb). (Table 1) Exploratory factor analysis: The Kaiser-Meyer-Olkin (KMO) analysis was carried out to examine the criteria for identifying the factor structure. Since KMO index was 0.929 ('superb' according to Field, 2009) 34 All KMO values for individual items were well above the acceptable limit of .5 (Field,2009) 34 except for one item which was therefore removed ["item p"]. Bartlett test of sphericity was signi cant (p<0.001). Therefore, the data was t for identifying factors using exploratory factor analysis. EFA was conducted using maximum likelihood analysis with Promax rotation. For factor extraction, scree plot and Kaiser criterion with eigenvalue of more than 1 showed that the instrument contained four factors which represented 49.6% cumulative variance.
The item designation criteria (factor loading of more than 0.4, and cross loading of less than 0.35) was used for reduction of the instrument to simple factor structure. Table 2 shows the pattern matrix of factor loading of each item with factor loading of more than 0.4. Out of the original 26 items, a total of 20 items were retained in the EFA.
The rst factor included nine items; the second factor included ve items; the third factor and the four factor both included 3 items.
Reliability analysis: The summation of items within each factor yielded four subscales whose internal consistency ranged from 0.6 to 0.8 and thus indicated satisfactory internal reliability. 35 Scale descriptive statistics and reliability for the scales derived and con rmed by factor analysis techniques is shown in Table 3.
Con rmatory factor analysis: Based on visual display of factor loading, items related to "training attitude", "situational intention" and "situational con dence "were allowed to load on Intention; items related to "pro-social belief" and "willingness to provide care" were allowed to load on Attitudes; items related to "normative beliefs" and "training intention"were allowed to load on Social Norms; items related to "Execution con dence" and "universal situation likelihood" were loaded for Perceived con dence Focusing on Table 4, the signi cant chi-square value (p <0.0001) does not imply support for the fourmodel factors in total sample. However, it should be noted that empirical studies showed that the p-value becomes signi cant if sample size is large enough. 32,36,37 Based on Cohen's guidelines 38 , perceived con dence had moderate effect on intentions (standardized beta, β = 0.464, SE =.062, p<0.001), attitude also had moderate effect on intention (standardized beta, β = 0.558, SE =.057, p<0.001). Inter-relationship between perceived con dence and attitude (standardized covariance =0.597, SE=0.053, p<0.001), perceived con dence and social norm (standardized covariance =0.592, SE=0.055, p<0.001), and attitude and social norm (standardized covariance =0.841, SE=0.037, p<0.001) were signi cant. In reviewing values of GFI, CFI, TLI and RMSEA in Table 2, it is evident that the four-factor model represents a very good t to the total sample. The model accounted for 83.5% of variance in participant's behavioral intentions. (Figure 1).
Inter scale correlations: The Pearson correlations coe cients (r) between the four components were calculated using summated respondent scores on individual scale components. The proportion of linearly explained variance (r 2 ) between components was estimated by squaring r ( Table 3). As this table indicates, the determinants are signi cantly inter-correlated and signi cant correlations are also observed for behavioral intention. The association between intention, perceived con dence, attitude and social norm were found to be substantial (with proportion of explained variance > 0.20) indicating that young adults with intention to help at the site of RTA possess higher levels of execution con dence, pro-social beliefs and willingness to ful l social responsibility. (Table 5) Substantial association was also seen between normative in uence and attitude (r=0.54) thus, indicating that those who are willing to provide care and have pro-social beliefs are likely to have favorable training attitude and normative beliefs. Further, the results showed that attitude, perceived control and social norm are different constructs as inter-correlations was around 0.35 (Table 5) Group differences: Accepting the changes concerning item loadings on speci c factors that were revealed by the CFA, group differences were analyzed with respect to the changed composition of the scale. Intention, social norms, attitudes and perceived con dence differed signi cantly both for gender of participants and for individuals with and without previous experience of helping an accident victim.
Group differences in responses to the items of the four scales are depicted in Table 1. Differences on item level are consistent with the direction of differences on scale level.

Discussion
The purpose of the current study was to develop and validate a questionnaire determining intention of educated youth to engage in pro-social behavior at RTA site. The MIHRAV instrument was successfully developed and validated for use among Indian youths. Using factor analysis techniques, the investigational instrument was able to explain large proportion of the variance in intention to act in an incident of RTA. Bloom et al. (1956) postulated that behavioral learning is composed of three domains (knowledge, skills, and affective) and thus the ability for this instrument to explain more than half of variance exceeded a priori expectations. 39 The EFA yielded four factors aligning with the TPB constructs. Standardized internal consistency of each subscale was satisfactory (Cronbach's alpha >0.6). Thus, instrument developed based on constructs from the TPB was observed to have strong construct validity and internal consistency.
The convergent and discriminant validities of these formulated items were further evaluated by con rmatory factor analysis. The items designed to assess a given latent variable (e.g., attitude toward the behavior) loaded highly on a single factor and had no strong cross-loading, that is, items designed to measure one construct (e.g., perceived behavioral control) did not exhibit signi cant loadings on other latent variables (e.g., intention). 40 With factor analysis, the initial number of instrument items was reduced by 27% with nal scale constituting 19 items out of 26 items.
This study establishes that the TPB is a useful tool to evaluate behavioral intention of young adults for providing assistance and care at RTA site.
The Intention construct was based on a 9-item subscale from factors based on "situational intention"; "situational con dence" and "training attitude". Based on these ndings, intention-to-help at the site of RTA is a multidimensional construct dependent on the participant perceived con dence in communicating with bystanders, law authorities and health authorities for arranging help at accident site. The construct for Perceived con dence included four items from the domain of "execution con dence" and "universal situation likelihood" suggesting that the learner perception of own and others' possessing necessary skills required to help RTA victim drive this construct as discussed by Ajzen. 40 The construct for social norms included three items from domain of "normative beliefs" and "training intention". This suggest that the learners' belief about social norms and their motivation to comply with these beliefs constitute this construct similar to behavioral intention model (BIM) proposed by Fishbein and Ajzen. 41 The construct for attitude included three items mainly from domains of "pro-social belief and willingness to provide care". This was also in accordance to BIM where attitude toward the behavior is determined by the belief about the outcome of intended behavior and evaluation of these outcomes.
Male students and those with previous experience of helping an accident victim scored signi cantly higher for multiple items assessed. Ajzen has discussed personality traits, intelligence, demographic characteristics and other variables as background factors and hypothesized that TPB components mediate the effect of these background factors on intention and behavior. 41 Considering the lifesaving nature of pre-hospital trauma care education, it is recommended that countries with evolving trauma care systems should train community volunteers as rst responders to facilitate early and accurate care. 5 Attention to behavioral intention and measuring it within population and participants is essential to identify and ll gap in understanding of the learner and to engage them better in future rst aid related educational activities. 42 For the health educator, the current instrument along with knowledge and skill-based assessments will be helpful in providing a holistic understanding while training youth in responding to incident of road tra c accidents.
Few of the major limitations in this study is that the MIHRAV questionnaire was tested on groups of college students in Rajasthan, India and thus the performance of this questionnaire on youths who are out of formal education or not in college is not known. Also, further studies are needed to investigate this instrument validity across other social determinants as race, age, education, socio-economic background in various population subgroups. Future studies may be planned using this instrument to understand the changes in learner attitude and intention towards responding to an RTA incident in pre-post learning environment. It is also known that presence of an intent-to-act may not necessarily result in an act in an actual emergency because of mitigating factors. 41 Thus, utility of MIHRAV need to be further explored in form of longitudinal studies regarding helping behaviour at the site of road accident or the use of experimental deception-based studies to identify actual behaviours.
This study established the utility of well-established psychological theory (TPB) on a heterogeneous group of young adults. These students represent the section of educated Indian youth residing in fast urbanizing districts of India with limited or negligible experience of helping any RTA victim or acting as rst respondent and represents an important section of population prone to road tra c accidents. Use of TPB to develop sub-scales of instrument offers a validated mechanism to carry out future research in similar or other population subgroups. This is also essential to establish meanings within each subscale for better interpretability by curricula developers, instructors, and program evaluators working to reduce mortality due to road tra c accidents.
The construction and validation of the designed instrument in various settings can provide researchers and pre-hospital trauma care curricula designers an insight regarding the affective state of bystanders present at the site of RTA incident and also essential learning experiences and inputs to instructors to inform content, engagement, and outcome goals of pre-hospital trauma care courses for the lay public and to maximize a volunteer's propensity to act in case of road tra c accident. Thus, we recommend integrating the instrument with bystander's educational program for helping road tra c accident victim, in order to help learner's, re ect on their own beliefs as well as societal norms and expectations while responding to RTA incident.
Informed consent: "Written Informed consent was obtained from all individual participants included in the study." Ethical approval: "All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards." amendments. Approval was further obtained from the authorities of selected colleges. This study only involved participants who provided their written informed consent. The questionnaire was treated as con dential and anonymous; there was no personal information that could link the responses with any of the participants in the study. Each completed questionnaire was returned to the researcher on the same day of data collection. Author Contributions: NR and AJ conceived the manuscript, performed the analyses, and wrote the manuscript. ND, VRP, AS, PR and MSR contributed to the data collection and measurements. NR, AJ and DK involved mainly in data analysis and its data quality management. NR provided overall supervision, edited the manuscript and had the responsibility for submitting for publication. All authors read and approved the nal manuscript.   Note. EFA = exploratory factor analysis; CFA = con rmatory factor analysis Note. RMSEA = root mean error of approximation; SRMR = standardized root mean square residuals; GFI = goodness-of-t index; TLI = the Tucker Lewis Index; CFI = comparative t index.