Structural Equation Modeling and Management Role On The Control of Marek’s Disease Outbreaks in Amhara Region Ethiopia

Marek’s disease is a highly contagious and economically important oncogenic or paralytic viral disease of poultry and it is becoming a serious problem in the poultry industry of Ethiopia. A questionnaire was designed based on the framework to conrmed how the farm management is reduced the outback of the mark’s diseases. Each model construct was measured using a set of rating scale items (questions). A sample size of 200 farmers from different production systems were chosen for data collection. From the analysis, Cornbrash’s Alpha (coecient of reliability) based on the average inter-item correlations were evaluated for each parameter. This structural equation model examines the relationship between risk factors (ock size, litter management, number of staff) and the risk of each farm (numbers of sick, drops in egg production, and the number of death).


Introduction
Marek's disease (MD) may be an international threat to poultry production and is characterized by malignant neoplastic disease development in visceral organs and nerves, leading to paralysis and death in inclined chickens (Othman and Aklilu, 2019). So, so as to advance management strategy of Marek's diseases in poultry would force totally different methods: as an example higher application of the presently accessible tools, that is embrace acceptable measures of biosecurity, acceptable vaccination practices, and management of alternative immunological disorder viruses (Gimeno, 2004). For the higher and also the best choices to stop the Marek's diseases is that the vaccines have brought monumental advantages to poultry production (Bublot and Sharma, 2004).
After the isolation of Marek's sickness herpesvirus (MDV) within the late Nineteen Sixties the primary vaccines were developed in England (Schat, 2016) and conjointly at the tip of Nineteen Sixties, us Department of Agriculture (USDA) had established the fundamental purity and safety standards for poultry biologics (Espeseth and Lasher, 2013). Despite over forty years of usage, the precise mechanism(s) of MD vaccine-related immunity and anti-tumor effects don't seem to be proverbial (McPherson et al., 2016). All vaccines, with the exception of the cell-free lyophilized HVT, have to be administered parentally in the form of living infected cells. Under eld conditions, exposure to eld virus occurs early in life requiring immunity in the young bird (Zhou et al., 2019). For this reason vaccination takes place at one day of age (Şandru et al., 2008). Vaccinal protection against MD is complicated and in uenced by several factors. These embrace the host genotype which may in uence each host susceptibleness to MD and responsiveness to vaccination (Bavananthasivam et al., 2018).
Quanti cation of patterns of viral shedding and virus-induced host mortality are necessary for a rigorous understanding of the epidemiology of a disease, not least to identify increases in virulence (Atkins et al., 2011). The disease is an airborne poultry pathogen primarily affecting chickens and causing losses to the industry of $1-2 billion annually (Atkins et al., 2013).
The virus is then transferred by the bloodstream in alveolar macrophages to B and T lymphocytes. After the primary infection stage, between 7 and 14 days postinfection (dpi), MD may become latent in infected lymphoid cells, which proliferate in different parts of host, especially in the liver, spleen, kidney, proventriculus and ovaries. This leads to tumour formation after reactivation of the virus to the transformation stage. Subsequently, the virus is then transferred to the environment as ne particles of skin and feather debris (Woźniakowski and Samorek-Salamonowicz, 2014). The direct relationship between MDV strains of higher pathogenicity and greater immunosuppression (Li et al., 2011).
Indigenous chickens kept under village, scavenging production systems represent 97% of the chickens farmed in Ethiopia. In this system, micro-and macro-parasitic infectious diseases, combined with limited extension and veterinary services, are major constraints to village chicken production. Poor management and limited resources hinder the development of chicken production by reducing outputs and by predisposing to disease outbreaks (Luu et al., 2013). Increased productivity of the poultry subsector by using exotic breeds in Ethiopia failed to become a sustainable option mainly because this strategy recurrently faced the problem of birds not being adopted widely by the rural farmers due to several socioeconomic and environmental challenges. The management conditions under which the animals are produced vary along the existing production systems which were broadly classi ed into the village, smallscale commercial, and large-scale commercial systems based on ock size, production objectives, and level of specialization and/or technology use (Dana et al., 2010). Among the impediments of the poultry development in this country constitute diseases like Marek's disease with frequent outbreaks and presenting a serious threat and challenge to the young growing poultry industry in Ethiopia. Therefore, the objectives of this study were, to explore the relationships between farm size, litter management and number of staff as a risk factors for their contribution to develop risk, egg production loss and chicken death that intentions to implement MD control measures in the different chicken production systems of Ethiopia, and to identify potential management factors about the disease and its control measures that in uence the intentions to implement control measures using the SEM framework.

Theoretical Framework
The basic concept of SEM is a powerful, multivariate technique found increasingly in scienti c investigations to test and evaluate multivariate causal relationships. The key in the regression analysis is to determine how much of the change in the dependent variable is explained by the independent variable or variables. Although multiple regression analysis can only be applied to observed variables, the basic principles can be applied to structural equation modeling (Kline, 2011).
As a new statistical analysis technique, allows testing research hypotheses in a single process by modeling complex relationships among many observed and latent variables. However, in the method of

Questionnaire Design
A questionnaire was designed based on the framework described above. Each model construct was measured using a set of rating scale items (questions). The supposed ock size about the likelihood of getting MD disease, reduced egg production, or increased the death rate, was elicited by three rating questions. The rst question assessed the frequency of MD outbreaks as experienced in different farms' ock sizes. The second question referred to the experienced frequency in the small ock size (<50). Both questions were answered on a four point rating scale, which indicated the occurrence of having an outbreak every <50, 51-1000, 1000-10,000, or >10001 chicken ock. The third question measured the chicken ock size associate with MD occurrence by a three points rating scale (high, medium, low). The perceived number of staff or ability was measured using two rating questions on a three points scale (large, medium, and small): One the question was about the impact of MD relative to general chicken egg production problems and the other question about the impact of MD relative to the impact of other poultry diseases. The patterns of correlation among a set of litter management of the advised action for the control of MD was measured by two rating scale questions on a three points scale (high, medium, low) for the proposed control measures related to litter management, and in uence chicken performance and important for the welfare chicken. To con rms the correspondence of the data of the relations in the theoretical model for the control of MD. The objectives were weighed for three suggested speci c MD control measures, including 1) management of chicken at farm level, 2) MD vaccine address, 3) biosecurity of chicken farms.

Data Collection
Chicken Production systems Poultry production has a major role in the economy of developing countries, including an important role in poverty alleviation by means of income generation and household food security. More than half of Ethiopian households both in rural and urban areas keep chickens, although there is considerable variation in the distribution of chicken keeping, with most households in highland areas keeping chickens, and far fewer doing so in lowland pastoral areas (Sambo et al., 2015). Whilst village chickens are widely accessible and require few inputs, productivity is low and constrained by, among other things, disease, predation and scarcity of feed. Interventions to improve production include vaccination; bird distribution; management interventions; cross-breeding programs; and combined programs, but few interventions have been demonstrably sustainable in village chicken production systems (Bettridge et al., 2018).
Chicken production is an integral part of most rural families' livelihoods (an estimated two-thirds of Ethiopian villagers keep poultry) and birds are commonly kept at night on perches within the family dwelling, frequently in the kitchen (Brena et al., 2016). Poultry industry in Ethiopia is infant but fast growing sector. The industry faces various challenges such as shortage of feed in terms of quality and quality, poor husbandry practices, prevalence, and wide distribution of infectious and noninfectious diseases. Poor veterinary services and lack of appropriate breeding practices are assumed to be additional challenges. Moreover, the government has given less attention. Disease is resulting in as high as 20-50% estimated mortality rate ranges (Berhe et al., 2019).

Sampling
A sample size of 200 farmers from different production system, was chosen for data collection. This sample size was based on a pragmatic consideration of SEM feasibility and reasonable power of test for the intended statistical analyses, In addition, the minimum sample size for structural equation modeling is suggested as 15 (Bentler and Chou, 1987), and Some researchers suggest that the sample size for SEM should be 200-500, at least 200 (Çelik and Yılmaz, 2013). The farmers were sampled from 11 zones (North Shewa, North Gondar, East Gojjam, Central Gondar, Awi, West Gojjam, West Gondar, South Gondar, Bahir Dar, South Wollo, and North Wollo) in the both intensive and extensive farming system. Selection of zones and cities was based on the authors' subjective judgment of representativeness of the production systems and on convenience of accessibility. The difference in the number of zones sampled from the different production systems was a re ection of the proportion of zones in the different production systems. This selection procedure was continual in other zones until the required sample size (200) for each sub-city was reached.
The questionnaire was prepared in English and translated into Amharic language for all administration zones. It was directed to the selected farmers and farm owners by face to face interviews. The study proposal was ethically reviewed and approved by the Institutional Review Board of the University Gondar. Oral informed consent was obtained from each participating farmer after reading a written consent form.
The use of oral consent was approved by the Institutional Review Board considering the fact that most of the study participants could not read and write to give their consent in writing. The interviewers con rmed the participants' oral consent by signing on the respective consent form for each interview as per the Board's guideline.

Model Speci cation
In the present study, the six basic constructs of the SEM were used to assess farm owner perceptions of MD and its control in North West Ethiopia. In the evaluation of the effect of these factors on the motivation of farm owners to implement control measures against the disease by improving farm management, the intention to participate in hypothetical MD control measures was considered as a proxy of the actual behavior. This is due to absence of any o cial control in practice to measure the behavior directly. Although intention does always explain to farm management system was always related to disease outbreak. In the analysis, socio-demographics and husbandry variables were used as modifying factors of the perception.
Furthermore, supposed the preliminary study has hypothesized the measurement model described the relationships among these six variables in Figure 15. This has described a model that can be accounted for the observed relationships between ock size, litter management (LM), number of staff and between the number of sick, drop in egg production (DEP), number of death (both were indicators of their respective correlated factors). However, that there was a moderated relationship between all variables. Our model was accounted for the risk factor relationship. In essence, these has created a model that says the relationship between ock size, litter management (LM), and the number of staff and between the numbers of sick, drop in egg production (DEP), a number of death is equal to 1.0 to scale latent variables. Our model determined to the extent that the relationships hypothesized was captured the observed relationships on the parameter path (Kline, 2005).

Data Analysis
The valuations of this model were to endorse appropriate poultry health management to control the emergence of highly pathogenic diseases like MD and measures bio-security for better performance and quality of poultry production in a competitive world. Analyses of the management skills of poultry production to evaluate the proposed risk factor correlation to the occurrence and the reduction of poultry product by MD. By measures, the management risk factor to control the disease was performed in North West Ethiopian situation.
There are 13 correlation coe cients between the variables that used in the SEM, the number of staff, LM,

Variables Summery
The observed variable (manifest variable) is the measured variable in the data collection process; latent variables are variables that are measured by connecting to observed variables because they cannot be directly measured. Latent variables must be represented by more than one observed variable since they represent abstract concepts. Latent variables represent hypothetical constructs in a research model (Raykov, 1997). In the case of this study, the model has the following observed or endogenous variables:  (21), the number of distinct parameters to be estimated (19), and degrees of freedom (2).
The skewness and kurtosis values are examined to determine whether the variables in the data set are normally distributed. It is su cient to mark the "test for normality" and "outliers" options so that these test values can be obtained in a tabular form (Sarstedt and Mooi, 2014). This is because tests such as these are highly sensitive to sample size, with larger sample sizes being more likely to produce signi cant results. In light of this, it is recommended that the signi cance tests be used in conjunction with descriptive statistics, namely the kurtosis values for individual variables (Pituch and Stevens, 2015). Kurtosis values less than 3.00 in magnitude may indicate that a variable is normally distributed (Westfall and Henning, 2013). So, the result of this SEM was normally distributed. The reliability of the new and modi ed items was tested carefully before evaluating the research model. In order to have a valid construct in the model, each of the items comprising the model was checked to see if it was unidimensional, since this would have help to produce a consistent result.

Normality Test
To determine whether univariate normality exists, the model examines the distribution of each observed variable for skewness and kurtosis. Skewness is the degree to which a variable's distribution is asymmetrical, with positive skew describing a distribution where many scores are at the low end of a scale (the score distribution for a very di cult test). For the skewness index, absolute values greater than 3.0 are extreme (Chou & Bentler, 1995). Kurtosis is an index of the peak and tails of the distribution. Positive kurtosis re ects very peaked distributions (leptokurtic), with short, thick tails (also called heavytailed distributions) representing few outliers. Negative kurtosis exists when the distribution is quite at (mesokurtic), with long, thin tails, indicating many outliers. Absolute values higher than 10.0 for the kurtosis index suggest a problem, and values higher than 20.0 are extreme (Kline, 2005). Univariate normality is especially important to consider because distributions can vary from normality in at least four ways. In our model all variables were positively skewed (Table 1).  Correlation is the coe cient that indicates the power of linear relationship between variables. This coe cient must be statistically signi cant in order to be able to say that there is a relationship between variables. The correlation coe cient takes a value between -1 and +1 (Sipahi et al., 2010). All the variables that used in the SEM, the number of staff, LM, FS, number of sick, number of death, DEP and LM, all are positive relationship.in the positive direction (+1). Therefore, linearity, which is the most important assumption of regression analysis, also applies to structural equation modeling. The sample correlation coe cient result in the model showed liner relationship between dependent and in dependent variable (Table 3). Direct effect is the effect of one variable on another variable without any mediation. However, the indirect effect arises from the intervention of a variable which is playing mediator role between independent and dependent variables (Raykov and Marcoulides, 2006).  In this section, the ndings of the model are presented in (Table 4). Regarding the variables the between FS and LM was estimated to be (-17.159) and between LM and number of staff is estimated to be (-.019), they were indicted inverse consequence, the other were positive effect to risk and disease causing factors.

Matrices Factor Score Weights Estimates
The other issue is the direct effect of an independent variable (exogenous) on a dependent variable (endogenous) as reported on ( Table 5). The ndings of the results showed that the between number of death and LM, FS, and number of staff were not all statistically signi cant (p > 0.05). The other issue is the indirect effect of an independent variable (exogenous) on a dependent variable (endogenous) as reported on (Table 6)

Model Fit Summary
All of the risk factors, all the variables that used in the SEM, the number of staff, LM, FS, number of sick, number of death, DEP and LM cause the risk the total number of the respective items and the data ts the model reasonably well (Table 7). In the method of structural equation modeling, the measures that assess the compliance of the models with the data are called t indices or t statistics. There are many t indices in the literature. Below there are de nitions of the most commonly used of these t indices? The size of the sample were considered in the analyzes to be done by the structural equation modeling. Because many of the t indices are affected by sample size (Jayaram* et al., 2004).

Discussion
Marek's disease is a disease of chickens produced by a herpesvirus that produces a reduction in the immune response in acutely infected chicken followed by the production of tumors in many of the infected birds. Very virulent strains (vvMDV+) have been reported in a number of countries around the world and have affected broilers, breeders and commercial layers. This disease extensively limits the productivity of both egg producing and meat producing birds resulting in great economic impact in poultry industry. Vaccination, in conjunction with good farm cleaning and disinfection, proper reception practices, adequate downtime between ocks, all-in-all-out policy, accurate vaccination programs adapted to the type of bird and eld situation, good vaccine preparation and administration practices and strict biosecurity measures can greatly reduce the incidence of MD and thereby prevent the economic losses due to the disease. In problematic areas with monovalent vaccine, bivalent or polyvalent vaccine is recommended for the effective prevention of virulent MD. However, vaccine failures do occur as eld strains continue to evolve towards pathotype of greater virulence. The constant evolution of MD has pressed us for the development of new vaccines or vaccine strategies that control the more virulent emerging strains. However, the competition between the development of vaccines and evolution of MD is a major threat for poultry industry (Purchase and Biggs, 1967 Kurtosis values less than 3.00 in magnitude may indicate that a variable is normally distributed (Westfall and Henning, 2013). Hince, this SEM result was normally distributed.
Previously, guidelines for acceptable t included a non-signi cant CFI greater than 0.90 (Hu and Bentler, 1995), RMSEA less than 0.10 with a maximum upper bound of the 90% CI of 0.10 (dan Cudek, 1993).
Although many still follow these guidelines (as evidenced by the positive evaluation of models with t indices at or near these values), readers should be aware that debate exists among statisticians (and likely among reviewers) regarding acceptable t. Recent studies (Hu and Bentler, 1998)  The research uses questionnaires to assess different farms from different zones such as construction, infrastructure, feeding status, vaccine access and research and development model in Ethiopia. From the literature, three different input factors ( ock size, litter management, number of staff) were identi ed that affect the impact of numbers of sick, drops in egg production, and the number of death used to study the effects of different risk factors on the farming activities. Afterward, the relationship between each of the risk factors and the farm risks were studied. Therefore, it can be concluded that the model must give due attention to the risk factors such as ock size, litter management, number of staff which is different from the previous research result and this might be due to the fact that the effects of environmental factors is more on procurement and integration than the other farm risk factors.

Declarations
Data sharing statement Data will be made available upon request of the primary author.

Ethics statement
For the research team members to conduct the current study, after permitted ethical approval and statement given by the University of Gondar, Ethiopia. The current study was reviewed by the Institutional Ethical Review Board of the University of Gondar for its ethical soundness, and it is found to be ethically

Disclosure
The authors declare that they have no competing interests for this work. The constructs of SEM model in the performed factors analyses on the intention to implement MD control measures.