*3.5. Data Analysis*

Once the information was obtained, analysis of the information and e ffective treatment of the data were carried out. In order to provide a more complete picture [55], a data-analysis triangulation, consisting of the combination of two or more data analysis methods, was carried out [56]. Specifically, for Questions 1 to 7, Iramuteq software was used to conduct a descending hierarchical classification (DHC) [57]. For Questions 8 to 10, MAXQDA software was used to conduct a constant comparison analysis [58]. Results obtained by each type of analysis were then combined at the interpretive level.

3.5.1. Instrument Used to Analyze Students' Primary Reflections on Their Voluntary Extracurricular Experience in Tangier (Questions 1–7)

In order to carry out a lexical analysis of the reflections expressed in the interviews, Iramuteq software was used to conduct the Reinert method. This analysis proposes that every discourse expresses a system of lexical words (or a group of words) that gives coherence and rationality to what the speaker expresses. The objective of the Iramuteq software algorithm is to perform textual data analysis, through the repetition of lexical footprints (word succession), in order to identify the most frequent lexical words shared by interviewees [59].

The software divides complete words such as verbs, nouns, adjectives, and adverbs from tool words such as articles, prepositions, pronouns, and conjunctions and includes only the former in the analysis. In the same way, it decomposes the corpus into elementary contextual units (ECU), which are one or two sentences long (30–50 words). In this way, the software performs an analysis of the complete words in each ECU. These words are then used to create a contingency table showing how the vocabulary is distributed by the ECU. From that table, the software creates a square matrix of distances and groups the ECUs according to the complete words they share [60].

Using the Reinert method, a DHC was performed to classify and group the ECUs into classes and consequently reveal the most characteristic vocabulary in each class. In this classification, each class was made up of di fferent groups of words that were included based on the frequency of words already lemmatized, and the association with the class was determined by a chi-square value equal to or greater than 3. This provided a margin of error of <0.05 (of a degree of freedom = 1) [59]. Using this classification method, several classes were obtained based on words and ECUs with statistically significant values.

Operations within this program are transparent and replicable until the moment of interpretation when researchers assign a tag to each class depending on the most significant complete words and the ECUs. Likewise, the Reinert method, which uses independence tests, calculates the relationship between each class and independent variable. If one class has a significantly higher proportion of ECUs belonging to a variable, that class is considered to be associated with that independent variable [61]. In this case, the group was used as a variable; assigning Group 1 to students who went to Tangier in January 2019 and Group 2 to students who went in June 2019.

3.5.2. Instrument Used to Carry Out an Analysis of Perceptions Expressed in Relation to Voluntary Extracurricular Experience in Tangier (Questions 8–10)

MAXQDA software was used to carry out a constant comparison analysis of the students' perceptions. This type of analysis is commonly used when analyzing qualitative data [62]. This software is used in qualitative research for its ability to synthesize, sort, and organize information collected with selected instruments and to present research results [63]. The program organizes and examines information by means of categories formed by a system of codes.

This software was chosen because of the possibility it o ffered of creating three separate categories for analysis. For the purposes of this research, it was important to separate the students' perceptions regarding the value of the experience in Tangier; their perceptions of the role of their peers; and their perceptions of the role of the instructor in their reflective development.

A code system was then utilized to examine the data. With regard to the second research objective, the code system emerged from the literature findings. However, as the data were being coded, new (sub)codes were incorporated into the system and existing ones modified. Table 2 shows the final code system used to carry out the analysis.


**Table 2.** Code system used for the analysis of Questions 8 to 10.
