The sample consisted of a total of 20 adolescents, of which 10% were female (n = 2), and 90% male (n = 18), with an age measurement of 14.45 ± 0.61 (min = 14; max = 16).

First, regarding the results of cannabis use, the data obtained place the prevalence of cannabis use at 10% (n = 2). Of these, one individual’s last cannabis use was 40 days ago, while the other shows a more habitual use, being the last use 3 days ago. The age of onset of use was 13 years old. One of the two consumers used cannabis mixed with tobacco. When asked that, if cannabis consumption were legal would they consume it, 20% of the students said that they would consume it, while 15% of the sample had already tried it (including consumers). Finally, it is worth noting that one student consumed cannabis alone quite often, while the other indicated that he had never consumed cannabis alone. Regarding the cannabis use intention, the total values are 1.35 ± 1.13; however, we find a high maximum, that result in the equivalence table indicates that it is above the 90th percentile. This means that 90% of young people of the same age have a lower intention to consume than this one.

Secondly, the descriptive results obtained for the variable stress and are set out in Table 2. Stress obtained total values of 37.65 ± 18.45.

Thirdly, Table 3 shows the values of the structure of the consumption and friendship networks, the actors who may be in more central positions in the networks and with greater degree of influence. The following parameters are described: indegree, outdegree, degree of proximity (out/in closeness), betweenness, and influence through the eigenvector. It is found that networks are similar in terms of values and density, which may suggest that adolescents select their friends as the peers with whom they would go out to consume.

Figure 1 illustrates the distribution of normalized betweenness in the friendship network, showing that a large number of individuals have a very low intermediation capacity. In fact, 30% of the individuals have no intermediation capacities at all.

A correlational analysis between the different variables, stress, cannabis use intention and friendship and consumption networks is shown in Table 4.

No significant correlations were found for cannabis use with these data, given the small sample size of the consumers. Behavioral manifestations of stress involve behaviors, such as acting defensively, neglecting friendships or showing negative attitudes in different interpersonal relationships. In this regard, it is noteworthy that a statistically significant correlation was found between the behavioral manifestations of stress dimension and the metrics of out closeness (r = 0.541), outdegree (r = 0.530) and the degree of betweenness (r = 0.496). In addition, statistically significant correlations were found in this network for in closeness (r = −0.511), emotional manifestations and physical manifestations and outdegree (r = 0.446). The cannabis use intention appears statistically significantly correlated with different network metrics: for the friendship network, outdegree (r = 0.596), out closeness (r = 0.531) and betweenness (r = 0.598), while, in the consumption network, outdegree (r = 0.252), out closeness (r = 0.500) and betweenness (r = 0.549). On the other hand, it shows a statistically significant correlation with total stress (r = 0.622) and its different manifestations: emotional (r = 0.510), physiological (r = 0.687) and behavioral (r = 0. 0479).

Figure 2 represents the classroom friendship network, where males are represented with a blue color and females with a pink color. Cannabis users are identified as square-shaped nodes and nonusers as circles. The size of the nodes varies according to their total stress. The intensity of the relationships is measured through the strength of the ties (0–4), where a greater strength indicates that these students are united by a bond of friendship by spending a large part of their free time together. The size of the nodes varies as a function of the total stress scores. Thus, we find how individuals who are consumers are close and could be considered friends. These nodes, although they have relationships with those actors who are more central, establish ties of greater intensity with actors who are on the periphery. However, it stands out how the girl who uses cannabis establishes ties of friendship with the central actors, being, in addition, one of the individuals with higher levels of stress. In general, we found a dense network, where relationships of varying intensity were high. We found an isolated node and a node on the periphery that only maintains a relationship with two individuals from the others in the class, being also one of the individuals with the highest level of general stress.

Figure 3 shows the classroom consumption network from which students were asked to select, out of all their classmates, those individuals with whom they would consume. As in the previous case, the size of the nodes varies as a function of the total stress scores or the behavioral manifestations of stress. The structure is similar to the friendship network, since the central actors are maintained, in this case, the central actor being a cannabis user. Consequently, it may be an indication that the friendship network influences cannabis consumption, being that this consumption is accepted by those nodes with whom they share more time.

Figure 4 shows the two networks, but, in this case, the size of the node varies depending on the cannabis use intention. Although most of the subjects obtained scores according to the mean, we found nodes of greater size, whose scores were indicative of an intention higher than normal. These actors are found both in the periphery and in the center of the network and are individuals in whom action must be taken through different intervention strategies when they are in a situation of risk at the beginning of cannabis use.