Research design, context and participants
The present exploratory psychometric study was conducted in Beiragh during 2018 after receiving approval from the Ethics Review board at Tarbiat Modares University. Beiragh village is located in the northern slope of Sahand Mountain in the south of Tabriz metropolis. About 5000 persons live in Beirgh, the employment and economic growth of most of whom depend on their livestock. In addition, livestock husbandry is considered as the main job of Beiraghians so that most of dairy products in the country are produced in villages such as Beiragh. The livestock of Beiraghians mainly included sheep and goats and rarely cows, which are vaccinated by two veterinarians and two livestock vaccinators in each year with paying no charge.
Table 1 summarizes the details related to the participants in each phase, which indicates the recruitment of different participants depending on the study phase.
Selecting specialists with valid articles or work experience in the intended field is considered as important. Therefore, the experts were invited in the study from different groups with the above-mentioned qualifications to help generate items, finalize the first draft of the questionnaire, and evaluate the face and content validity of the items [20, 21]. Further, livestock breeders were requested to participate in formulating items and examining their face and construct validity. Furthermore, all of the livestock breeders were male and the details of their selection in each phase are described as follows.
Developing the first draft of the questionnaire
The initial items of BPQ were achieved through a thorough literature review and interview with all stakeholders. Additionally, the databases including MEDLINE, PubMed, EMBASE, ERIC, and Cochrane Library, as well as the Cumulative Index of Nursing and Allied Health Literature (CINAHL) were searched to find the studies published about brucellosis prevention or animal vaccination. In this regard, a combination of the keywords of "prevent", "Brucella" and "vaccine" were utilized to explore in English and Persian. Thus, 816 articles issued between 2008- 2019 were obtained, their abstracts were read, and duplicate ones were removed, of which 110 more relevant ones were read in full text. Then, nine Persian and three English questionnaires were obtained by reviewing the articles and contacting corresponding authors [15–17].
Interviews were conducted to identify the factors influencing preventive behavior of livestock breeders. The conceptual framework for conducting the interviews was the concepts from the first four phases of PRECEDE model . Purposive sampling was employed to recruit participants  and directed content analysis was used to analyze the content of the interviews .
The volunteer livestock breeders, health educationists, veterinarians, and experts from a vaccine and serum production institute in the region participated in 30-45-min face-to-face interviews in their desired time and place. They were told that their information would be kept confidential and used anonymously. The items obtained from literature review were combined with the results of interviews. In addition, health educationists, veterinarians, and experts from a vaccine and serum production institute were asked to participate in three focus group discussions. Each session lasted 90 minutes, one member of the research team acted as the coordinator of the sessions, and another took note. Then, a directed analysis was performed on the content of the interviews, identical and duplicate questions were eliminated, and some questions were edited. Finally, the first draft of the study questionnaire was confirmed, and the anchor response of the items were discussed and finalized by the research team members.
Assessment of face and content validity of the questionnaire
The face validity of the intended questionnaire was examined qualitatively and quantitatively by co-operating livestock breeders and health educationists. In qualitative evaluation, any ambiguity in the meaning, wording, and scaling of the items, as well as grammatical errors and those in item allocation were identified and resolved based on the feedback from livestock breeders and health educationists. However, the impact score (IS) of each item was calculated for quantitative assessment.
The livestock breeders participating in the evaluation differed from those involved in examining the construct validity of BPQ and those co-operating in the cross-sectional part of the study.
In order to assess the face validity of the items, the appropriateness of each item was rated by an expert by using a five-point Likert scale, and the IS of each item was calculated by using the formula of
IS = frequency (%) × importance [22, 23]. In the fourmula, the frequency represents the number of the patients rating the appropriateness of the item as 4 or 5, while importance refers to the mean score of the item on a 1-5scale.
Additionally, the content validity of BPQ was evaluated qualitatively and quantitatively. In this regard, content validity index (CVI) and ratio (CVR) were calculated for quantitative assessment. To this end, BPQ was emailed to 20 veterinarians and health educationists for evaluating the validity. One expert failed to complete the questionnaire and two questionnaires were set aside by considering the precision of the data. (Response rate= 0.85%).
The CVI and CVR were determined based on the three and four-point Likert scales, respectively. The formula of (Ne – N/2)/ (N/2) was used to calculate CVR , in which N indicates the total number of panelists and Ne illustrates the number of those rating the item as “essential”. Further, items with CVR below 0.46 were removed based on the Lawshe table .
In order to compute the CVI of the items, the relevance of each item was rated in a four-point Likert scale by using the formula of CVI (the number of the specialists who assigned scores 3 and 4 to the items/N). Furthermore, the items with the CVI less than 0.79 were eliminated [25, 26].
Assessment of construct validity of the questionnaire
The construct validity of BPQ was examined through exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). It is recommended to perform each of EFA or CFA by participating at least 200 ones from the target group . Sampling framework in the phase of the study included 2122 livestock breeders living in Beiragh. Since livestock breeders deliver their dairy products to the local dairy production mini-factories in the region, the complete list of livestock breeders was prepared from forty cheese production mini-factories in the village. Due to 10-15% drop rate in the previous relevant studies, there was a need for 50 more participants [27, 28]. In the phase, 450 livestock breeders were randomly selected from Beiragh and invited for participation by using (www.randomizer.org software (.
Exploratory Factor analysis (EFA)
EFA was implemented on 42 binary items and 17 Likert-scale ones, which were intended to explain the preventive behavior of livestock breeders. In addition, the number of optimal factors was determined through principal component analysis and oblimin rotation method. Loadings with the significance lower than 0.5 were excluded from the analysis . If an item was loaded into different factors, it was related to the factor in which the item had the largest factor loading. After completing the analysis, the items were categorized and each category formed a construct or factor. Then, the extracted factors were named by team members based on the nature of their items, as well as the characteristics proposed by them to measure.
Confirmatory Factor analysis (CFA)
For the items with binary response anchors, the generalized confirmatory factor analysis  and WLSMV statistical estimation method were applied for the items with binary response anchors . Additionally, the intended conceptual model was tested by using M-Plus 7.4 software. A conceptual model was proposed to test by considering the factors recognized by the EFA and hypothetical relationships between the factors. Further, 59 items were grouped into three factors of awareness, attitude, and practice. Due to the improper fitness of the initial three-factor conceptual model, the awareness was divided into direct, indirect, and vaccine awareness. Furthermore, the suggested new conceptual model with 59 questions and 5 latent factors (direct, indirect, and vaccine awareness, attitude, and practice) was evaluated through confirmatory factor analysis. After removing the items having low factor loadings, the final model with five factors (behavioral constructs) and 53 items was confirmed.
The fitness of the proposed model was assessed by using the fit indices including the ratio of chi-square to the degrees of freedom (X2/DF) and root mean square error of approximation (RMSEA), as well as comparative fit (CFI) and Tucker-Lewis indices (TLI). The CFI and TLI at least 0.90 and RMSEA below 0.08 represent a good fitness [31, 32]. The final conceptual model was introduced after excluding non-significant items.
Assessment of reliability of the questionnaire
The internal consistency of BPQ was tested by calculating Cronbach’s alpha coefficient and composite reliability (CR). In addition, the stability of the results was evaluated by determining interclass correlation coefficient (ICC) [23,24]. Due to the need for about 30-40 participants to assess reliability , 42 volunteer livestock breeders completed the questionnaire twice in a two-week interval, and consequently the stability of the results was measured . Further, the standard error of measurement (SEM) was calculated to analyze the absolute reliability of the results. Furthermore, IBM SPSS statistics version 24 was utilized to perform data cleaning and compute reliability indices. P values less than 0.05 were considered as significant.