A relational network also known by the term social network is an immaterial reality which, according to Granovetter (1973) , refers to a network that enables the individual to create and mobilize relationships within and outside the company in order to gain access to the resources of other social actors. According to Hirsch et al. (2008) , it is a set of relationships that links people who share a strong common objective and denominator: a system of values, interests, tastes, a past, qualities or a common project. In the same vein, Barabel et al. (2010) consider it as a community of individuals linked to each other by their origins, their centers of interest, their needs, and their close or similar points of view within the framework of interactions which can give rise to exchanges of information, logics of mutual aid and experience sharing, a source of learning. Senayah and Akakpo-Numado (2017) and Soulez (2002) simply consider it as the set of relationships that an individual maintains. We understand the relational network as all the people in relation to the ego and whose business he regularly discusses. Social capital, developed by Bourdieu (1980) and Coleman (1988) , refers to the social resources that individuals can mobilize through their relationships and social networks. In this research, social ties, defined as the set of relationships that the manager maintains with people in his or her social network, can be considered a form of social capital. In reality, these ties can provide the manager with resources such as access to information. As for ethnicity, it is often associated with the notion of ethnic diversity, which refers to the presence of different ethnic groups within a society or organization, such as the manager’s social network.

According to Géraudel (2011) , not all relationship portfolios are created equal. There are as many forms of relational networks as there are leaders, and not all of them lead to the same returns in terms of resources. The authors have developed a fairly rich literature on the properties required for a relational network to provide as many resources as possible. We limit ourselves within the framework of this work, to two dimensions relating to the content of interindividual links ( Granovetter, 1973 ) and to the characteristics of the actors present in the relational portfolio ( Lin, 1999 ).

Granovetter (1973) distinguishes between strong and weak ties. Weak ties are characterized by a small amount of interaction over time, low emotional intensity, little trust and mutual services, but they do allow them to get in touch with a larger number of individuals. On the other hand, strong ties, which come from embedding family or ethnicity, are compartmentalized within the same clique. According to Géraudel (2011) , ties are described as strong when the individual is close to his alter, spends time with them and sees them frequently, while ties are weak when the individual has no privileged relationships with the alter and does not invest time to maintain the relationship.

It is the characteristics or attributes of the alter that are important in the search for relational resources relevant to the individual ( Chollet, 2006 ; Lin, 1999 ; Géraudel & Gundolf, 2016 ). When we talk about the proximity of the players in the relational network, we are referring to the fact that the alter or not have characteristics identical to the ego ( Géraudel, 2011 ). These characteristics are assessed on a functional, sectoral, geographic, hierarchical, cognitive and organizational level, etc.

According to the theory of social capital developed by Lin (1995) , individuals through the relational structures in which they are embedded manage to access several types of resources. Many authors maintain in this respect that the manager’s relational network facilitates access to information ( Barlatier et al., 2013 ; Mayegle & Ngo Omam, 2015 ). We believe that it makes sense, given all of this, to make the following assumption:

H1: The manager’s relational network determines access to information.

Granovetter (1973) through the theory of the strength of links establishes the influence of the nature of social links on information. He argues that the stronger the relationship between two individuals, the greater the possibility that they know the same people and, obviously, access the same information. On the other hand, weak links increase the chances of the ego accessing people with interests, multiple and rich information. According to Géraudel et al. (2006) , the nature of the links determines access to information, and strong links have a positive influence in exploitation contexts while weak links seem beneficial in exploration contexts. Brüderl and Preisendörfer (1998) and Adler and Kwon (2002) along the same lines add that social relations and social contacts are important channels for accessing information. We find the justification to put forward the following hypothesis:

H1.1: The nature of the manager’s social ties positively determines access to information.

According to Burt (1992) , having links with very different people guarantees the manager access to information that does not overlap with that which he already has. For Chollet and Geraudel (2010) , a good way for managers to have a variety of information is to have links in different areas. They show that the more people a manager has in their network whose fields are very different, the more they access information. Lin (1999) explains that it is more the fact of having heterogeneous alter that is interesting in terms of access to social resources. Working with geographically, functionally, hierarchically or cognitively close individuals constrains the flow of information. All this leads us to formulate the following hypothesis:

H1.2: Proximity of actors negatively determines access to information.

The ethnicity represents a human population that is considered to have in common an ancestry, a history, a culture, a language or a dialect, a way of life; quite often several of these elements at the same time. Akono Evang (2014) describes it as a more or less homogeneous group whose members have a community of ways of being, thinking, feeling, and doing related to language, culture and other elements that present some stability, as opposed to other groups of the same nature. Many studies have been conducted on the influence of socio-cultural factors, including ethnicity in the conduct of business. They relate inter alia to commercial behavior ( Sogbossi Bocco, 2009 ; Darley et al., 2013 ), growth ( Boukar & Julien, 2009 ), decision-making and managerial practices ( Kamdem & Ongodo, 2007 ). According to Sogbossi Bocco (2009) , the dynamism and success of SMEs depends on their ability to reconcile the social and cultural values of Africa with the necessary economic efficiency. Ethnicity, as a determining factor in socialization, constitutes an essential element that could be at the origin of specific managerial behaviors. Hernandez (1997) concludes that the survival of SMEs is linked to their ease in adapting economic logic to socio-cultural logic. We believe that the relationship between the leader’s social network and information can be moderated by ethnicity since it represents a reservoir of customs, lifestyles and shared values between the alter. All this brings us to the following hypothesis:

H2: Ethnicity moderates the relationship between the manager’s relational network and access to information.

In the absence of a real basis, since it is a strand not yet explored, we have, from the main hypothesis, formulated two sub-hypotheses according to the characteristics of the network.

H2.1: Ethnicity moderates the relationship between the nature of social ties and access to information.

H2.2: Ethnicity moderates the relationship between the proximity of actors and access to information.

Our research model is presented in Figure 1 .

Cameroon constitutes our field of research. Our choice relates to SMEs in the agrifood sector. The choice of the agri-food sector is essentially linked to its importance in African economies in general and Cameroon in particular. The agri-food sector is one of the main economic sectors in Cameroon, contributing to GDP and employment. This sector is crucial for economic and social development in Cameroon, particularly with regard to food security, poverty reduction and employment promotion. Based on the classification of Broutin and François (2003) , we selected four branches of activity in order to respect the diversity of this sector: Bakeries-Pastry shops, Restaurants, Agricultural product processing companies and fishery products. These SMEs are continually in a process of value creation, which requires renewal and strengthening of existing information.

Unlike other sectors of activity, agri-food businesses in Cameroon are characterized by strong social networks, particularly in rural areas where communities are closely connected. These social networks play an important role in access to information and knowledge. The influence of social networks on access to information may differ in other sectors of activity in Cameroon, depending on the size, structure and culture of businesses, sectoral issues and the regulatory context that may influence the importance of social networks in accessing information. The leaders of the cities of Douala, Yaoundé and Ngaoundéré represent the discussion unit.

This research has two main phases. The first is carried out using an interview guide through semi-structured interviews lasting approximately one hour, with 27 managers of agrifood SMEs chosen appropriately. The quantitative phase representing the second phase is carried out using a questionnaire administered to 273 managers of agrifood SMEs obtained using the convenience method. This final administration is preceded by two pre-tests with 10 and 105 individuals. The first pre-test responding to the debriefing method led to eliminating misunderstood and reformulation of certain questions. The second made it possible to ensure the internal consistency of the constructions.

A five-point Likert scale ranging from “Strongly disagree” to “Strongly agree” was used to assess the measurement items of the constructs.

The scale for measuring the nature of social ties is adapted from the work of Marsden and Campbell (1984) and relates to emotional proximity, the length of the relationship and the frequency of contact. To this scale, we add two dimensions from our qualitative study (trust and reciprocity of exchanges). The items used to measure the reciprocity of exchanges are generated from the exploration, while confidence is measured by the scale of Gurviez and Korchia (2002) .

The scale for measuring the proximity of actors: Adapted from Géraudel (2008) , it includes four dimensions relating to geographic, sectoral, functional and hierarchical proximity. We have integrated the cognitive proximity that emanates from semi-structured interviews. To measure it, we used Bergadaà and Del Bucchia (2009) .

The information measurement scale: the Géraudel (2008) measurement scale is suitable because it highlights the different elements of the Cameroonian context. The ethnic measurement scale: the measurement scale used in this research is adapted from the exploratory study and includes three items.

We performed a content analysis to process the data from the interviews. To this end, we cut the speeches of the interviewees by theme and frequency of appearance. The analysis of the significant elements of these discourses allowed us to form groups of sentences and responses according to the themes. Sphinx IQ software is used.

To test the structure model, we have examined, with regard to direct links, the adjustment indices and have resorted to standardized coefficients of determination to student t, which indicates their significance and the percentage of variance of the endogenous variable. Multigroup analysis is used to validate moderating assumptions. For this, we used the Chi-square difference test using the structural equation method (EQS). The CMIN, CMIN/DF, RMSEA, GFI, TLI, CFI and NFI indices were used to assess the quality of the fit of the model. EQS modeling was chosen in this research for several reasons. The relationships between variables are complex and involve interactions between several factors. EQS allows these relationships to be modeled in detail and specific hypotheses to be tested. This method allows for testing causal relationships between variables, which is crucial for understanding how the manager’s social network and access to information influence results. EQS offers great flexibility in specifying relationships between variables, which allows for testing complex hypotheses, such as moderating hypotheses. We could have used other statistical methods, such as multiple regressions and analyses of variance, but these do not allow for modeling complex relationships and testing causal relationships in the same way as EQS. SPSS and Amos software are used to process quantitative data.

Table A1 summarizes the results of the exploratory and confirmatory factor analyses. This table shows that all the data can be factorized. All KMO indices are above the theoretical threshold (0.5). Barlett’s test tests indicate broadly acceptable significance. Relative to reliability and convergent validity, all the indicators respect the standards commonly accepted in the literature because all the values of Cronbach’s Alpha and Jöreskog Rhô are > 0.7. Similarly, the Rhô values of convergent validity (Fornell and Larker’s PVC) are all largely above the recommended threshold (0.5). All constructs have excellent internal consistency and good convergent validity.

The results relate to direct links and moderating relationships.

Table 1 summarizes the results relating to the links between the social network and information.

Source: study results, 2015.

The values of the regression coefficients, the probabilities, as well as the student t’s are all significant; similarly, all Fit indices are well adjusted. We can therefore conclude with the validation of hypotheses H1.1 and H1.2.

We divided the sample into two groups, namely, the absence and presence of ethnicity, and carried out a multigroup analysis. Only individuals with an average of 2.5 on this scale have ethnic relationships. Table 2 summarizes the results.

Source: study results, 2015.

We note an increase or a decrease in the intensity of the two relations. The value of the difference test in the nature of social ties -information relationship (0.249) is greater than the standard (0.05) and is less than the threshold recommended in the proximity of actors—information relationship (0.045). As a result, ethnicity has a significant and positive moderating effect on the nature of social ties—information relationship. We conclude with the validation of hypothesis H2.1. On the other hand, there is, a priori, no effect moderator of the ethnicity on the explanatory relation of the access to information by the proximity of the actors. Hypothesis H2.2 is rejected. The ethnic group is the subject of consideration and interpretation by the alter, members of the network. The alter may be more or less inclined to transfer to the leader the information he holds according to ethnic ties. In this hypothesis, two leaders with identical social networks cannot benefit from the same benefits if their ethnic relations are different. Research has not studied the moderating role of ethnicity on the relationship between social networks and information; however, work establishing the influence of socio-cultural factors on business management abounds ( Sogbossi Bocco, 2009 ; Darley et al., 2013 ). The results of this work are consistent with this work and help to further strengthen the contingency theory, which establishes the influence of the environment on the conduct of the company. In Cameroon, ethnicity is one of the variables that intensifies or diminishes the relationship between the social network and information.