Associative Network Method

Abstract

Background

The associative network method is a technique for studying social representations of a phenomenon. Although it is little known and used, it can be widely used in nursing research, particularly to understand the representations of a population on a disease or a professional practice.

Objectives

This article aims to describe the associative network method, proposed by De Rosa in 1995, through a concrete example.

Methods

The associative network is a technique that allows us to identify the content, structure, and polarity of social representations of a phenomenon. It was used with 41 participants to describe their representations of urinary incontinence. First, the data were collected following the four steps described by De Rosa. The analysis was then carried out manually and using Microsoft Excel. The different themes evoked by the 41 participants, the number of words in each theme, the rank of appearance of the themes, the indices of polarity and neutrality, and the rank of hierarchy were thus analyzed.

Results

We were able to describe in depth the representations of caregivers and people from the general population on urinary incontinence and, more specifically, their content and structure. The spontaneous answers of the participants allowed us to explore several dimensions of their representations. We were also able to obtain rich information, both qualitatively and quantitatively.

Discussion

The associative network—which is easy to understand and implement—is a method that can be adapted to various studies.

The associative network is a method used to study the social representations of a phenomenon. It is defined as a study technique suitable for identifying the structure, contents, and polarity of the semantic field associated with a certain representation (De Rosa, 2002). Its specific objective is to allow the investigation of different components of social representations. Although it is little used in comparison to other methods such as focus groups (Kidd & Parshall, 2000), interviews (Imbert, 2010), or questionnaires (Gubrium et al., 2012), associative network methods allow for an intermediary between qualitative and quantitative research (De Rosa, 2002). Indeed, it allows the research to obtain both textual information and numerical data that can be statistically analyzed. Although relatively unknown (Rateau et al., 2012; Vergès & Bastounis, 2001), this technique can be widely used in nursing research, particularly to understand the social representations of a phenomenon.

Jodelet (1984) defines social representations as a form of knowledge that, socially elaborated and shared, has a practical aim and contributes to constructing a reality common to a social group (Rubira-García et al., 2018). The elaboration of social representations comes from interactions between individuals; they are essential to communication and understanding in human relationships. Each individual has a personal representation of a phenomenon. Thus, everyone has an idea, a belief, or a way of seeing an object, and these thoughts often differ between persons, depending on the family environment, culture, or personal and professional experiences of each person. An individual’s representations are what a person thinks or knows about an object (Stewart & Lacassagne, 2005). Many disciplines study social representations because they are multiple and allow for a better understanding of phenomena (Churchill & Iacobucci, 2004; Roland-Levy et al., 2001). Indeed, although social representations were initially studied in the field of social psychology (Moscovici, 1984), they are now studied in many other fields. In the field of health, social representations were used very early to better understand the concepts of health and illness outside medical models (Herzlich & Pierret, 2010). In the very first study of social representations in health, the author developed a new approach to health and illness and also health behaviors (Galand & Salès-Wuillemin, 2009). The study of social representations thus makes it possible to provide answers very quickly on the reasons why certain health problems become more important in society or how they are managed. In an exploratory and descriptive study in Brazil, for example, researchers showed that multiple objects of representations are studied in the field of nursing. These include illness, professional practice, care services, health-disease process, and nurses’ professional identities (Marques et al., 2006). Several methods are recommended to study these representations, including the associative network method. Our study aimed to use this technique to describe the French social representations of urinary incontinence of elderly people in nursing homes.

METHODS

Our university’s institutional review board approved this research.

DESIGN

De Rosa (2002) described four main steps for realizing the associative network. The first step consists of asking the participant to write down all the words that spontaneously come to mind around a stimulus word proposed by the interviewer. In this first step, the participant must also write down a number next to each word to specify the order in which the words came to mind. The second step is for the participant to make connections between the different words and groups of words, if necessary. With the help of arrows, the participant creates connections between the terms. The third step consists of polarizing the participant’s words, during which the participant uses a symbol corresponding to the image of the term as “+” for positive, “−” for negative, and “0” for neutral. Finally, in the fourth step, the participant is asked to rank the importance of the words using numbers.

In the case of our example, the aim was to conduct a qualitative study using the associative network. This associative technique—projective (De Rosa, 2002)—allows the participant to spontaneously appropriate a simple stimulus (a word or an image) and give information about it without the participant providing any personal information.

SAMPLE

In our example, the method of the associative network was used before conducting focus groups with the same sample. The aim was to interview caregivers of older adults in nursing homes and people from the general population to collect information about their representations of urinary incontinence. After a call for participation in five geographical regions of France, a list of volunteer nursing homes was established. Seven were then randomly selected so that each French region would be represented. Then, with the help of the nurse coordinators of each nursing home, caregivers and people from the general population were recruited. The sampling was therefore nonrandom and accidental.

DATA COLLECTION

Stimulus Word

The first step was to determine the “stimulus word” or words to be proposed to a participant; it is around this word that the participant created an associative network. The choice of the “stimulus word” was made according to the objectives of the research; it can be a word, a short sentence, a short text, or even an image (Rateau et al., 2012). This is then written in the middle of a blank sheet of paper without any other information about it. In our study, the word “urinary incontinence” was chosen as the stimulus.

Occurrence

Occurrence is the number of times the participants mention a word or theme. It is used to assess the importance of an idea/representation (Lemaire & Denhière, 2006).

Rank of Appearance

During the first stage of the associative network, the participant is asked to write down spontaneously all the words that come to mind when the stimulus word is read. These words can be adjectives, verbs, and so forth. The participant is also asked to write the word down around the stimulus word and to number them as they come to mind. The purpose of this instruction is to study the content of the participant’s representation relating to the stimulus word. The numbers affixed to the word provide information on the structure of the representation. The order of appearance of a word gives information on the “prototypical accessibility” (Rateau et al., 2012) or socially shared order (De Rosa, 2002).

Links and Connections Between Words

During the second step of the associative network, the participant indicates, if necessary, whether certain words or groups of words are linked to each other. This makes it possible to define a framework, establish a situation around the words used, and allow certain groupings.

Polarization of the Representation

By asking the participant to indicate a +, −, or 0 next to each word, the participant provides an orientation to the words associated with the stimulus word—positive, negative, or neutral.

Ranking of Importance or Hierarchy

The participant ranks the words written around the stimulus word in order of importance. To do this, the participant was asked to write the order using Roman numerals to avoid confusing it with the numbers used before. If this instruction was too complex, participants could use the numbers in a new color. The ranking order can, of course, be different from the order of appearance. This essential step makes it possible to attribute unequivocal importance to the terms (Rateau et al., 2012). These four steps are generally carried out in about 10 minutes in order to obtain spontaneous responses from the participant(s).

DATA ANALYSIS

In our example, the data analysis was carried out both manually and using Microsoft Excel. It was done by following the different steps of the associative network. However, it is possible to use data analysis software such as Iramuteq (Moracchini et al., 2020) or EVOC 2000 (Castro et al., 2020). De Rosa suggests the following steps for data analysis (De Rosa, 2002).

Occurrence

The first step is to look at the occurrence of words proposed by the participants. The more the participants cited a word, the more it represents what is shared by a community. In our example, the words collected were grouped into themes.

Rank of Appearance

The next step is to look at the rank of appearance of the words. The closer a word is to zero, the earlier it appears in the association network. To calculate this rank, in our study, we calculated the average obtained for each theme and then classified the themes according to the average rank.

Links and Connections Between Words

The connections established between the different words and groups of words are then noted to clarify the participant’s representation.

Polarization of the Representation

The polarity and neutrality indexes indicate the orientation that the participants gave to their words (positive, negative, or neutral). These indices range from −1 to +1 and are calculated as follows.

Polarity Index

If the index is between −1 and −0.05, it means that the terms are mostly negative. If it is between −0.04 and +0.04, the results should be interpreted as follows: Positive and negative terms tend to be equal. Between +0.05 and +1, it means that the terms are mostly positive.

Polarity Index (P) = (Number of positive words − Number of negative words) / (Total number of associated words).

Neutrality Index

If the index is between −1 and −0.05, few words are neutral. If it is between −0.04 and +0.04, the neutral terms equal the sum of the positive and negative words. Finally, between +0.05 and +1, most of the terms are neutral.

Neutrality Index (N) = (Number of neutral words − [Number of positive words + Number of negative words]) / (Total number of associated words).

Importance Rank or Hierarchy

The closer the word’s rank is to zero, the more important it is to the person making the associative network. We calculated the average obtained for each theme and then classified the themes according to the average rank.

RESULTS

Sociodemographic Characteristics of Participants

A total of 41 people participated; 19 were caregivers of institutionalized older adults, and 22 were from the general population. Participants were 23–97 years of age (M = 54.73 years, Mdn = 57, SD = 18.6); 90.2% were women. Participants were executives in the intellectual professions of higher education, intermediate professions, employees, and retired. In addition, all the French regions were represented. Table 1 shows the themes, the number of words in each theme, the rank of the themes, the polarity and neutrality indices, and the rank of hierarchy.

TABLE 1.

Summary of Themes, Occurrences, Ranks, and Indices of Polarity and Neutrality

Themes	Number of words	Prioritization rank	Rank of appearance	Polarity index	Neutrality index
Imperiousness	10	1	2	−0.50	−0.33
Leaks	8	2	1	−0.81	−0.93
Consequences	7	3	5	−1	−1
Negative feelings	35	4	8	−0.87	−0.97
Medical/anatomy	22	5	7	0.07	−0.44
Organization	21	6	6	−0.58	−0.92
Age	6	7	3	−0.17	0.17
Types/causes	9	8	9	−0.08	0.50
Treatments	9	9	10	1	−0.85
Protections	11	10	4	−0.16	−0.79

Content of the Field of Representation of Urinary Incontinence and Order of Appearance of the Words

Around the stimulus word urinary incontinence, the 41 participants associated 146 words. Only 51 of these words were mentioned at least twice (i.e., one third of the words), whereas 26 were mentioned at least three times (less than one word in five). As a result, most of the words were mentioned only once, although there were apparent similarities between some of the words. The words were grouped into 10 themes: imperiousness (urgency), leakage, consequences, negative feelings, medical/anatomy, organization, age, causes, treatments, and protections. The theme with the most representation was negative feelings; it was mentioned 68 times by the participants.

The themes mentioned first by the participants were leakage and imperiousness, followed by age and consequences. In contrast, treatment and negative feelings came last.

Connections Between Themes

One of the particularities of the associative network is that it provides structural information because of the links made by the participants between their words. In this study, all themes had at least one link with another theme. As seen in Figure 1, the most essential link was made between the theme’s protections and negative feelings; it was seen six times between the words of the two themes. Other major links are negative feelings and medical/anatomy, negative feelings and organization, medical/anatomy and age, medical/anatomy and types/causes, and protections and organization. From this, it can be inferred that protections and organization suggest negative feelings. Similarly, there is a strong link between the medical/anatomy theme and the age and types/causes themes.

FIGURE 1.

Links between themes. The thickness of the lines between links is equal to the number of links with the associated theme.

Polarity and Neutrality Indices

After calculating the indices of polarity and neutrality, a ranking from the most positive to the most negative was made (see Table 2). The theme treatments were considered positive by the participants, with a polarity index of 1 and a neutrality index of −0.85. The medical/anatomy theme had a polarity index of 0.07—a positive connotation. However, the medical/anatomy theme also had a neutrality index of −0.44, indicating that participants rarely perceived this theme as neutral. Participants viewed the themes types/causes, age, and diapers negatively. Imperiousness had a negative polarity index of −0.50, with few associated words considered neutral. The themes leakage, consequences, negative feelings, and organization had a polarity index of <−0.50, meaning they were considered negative by participants; the neutrality indices ranged from −1 and −0.90, suggesting only a very small proportion of words were considered neutral.

TABLE 2.

Classification of Themes According to Their Polarity

	Treatments	Medical anatomy	Types/causes	Age	Protections	Imperiousness	Leaks	Consequences	Negative feelings	Organization
Polarity index	1	0.07	−0.08	−0.17	−0.16	−0.50	−0.81	−1	−0.87	−0.58
Neutrality index	−0.85	−0.44	0.50	0.17	−0.79	−0.33	−0.93	−1	−0.97	−0.92

Order of Importance Words

The ranks of appearance and prioritization of themes can be quite disparate, reflecting the fact that the first idea that came to participants during the association network was not necessarily the one they found most important in terms of the overall theme being discussed. The theme classified as the most important was imperiousness, even though it was represented by only 10 words and with an occurrence of only 12. The second and third most important themes were leakage and consequences, with occurrences of 27 and 14, respectively. The theme of negative feelings was ranked only fourth, with a very high occurrence of 68 citations. The theme classified as the least important according to the participants was protection, although it was cited 38 times.

Specific Analysis of the Negative Feelings Theme

A focus was made on this theme because 35 words were part of it and it was the most important in terms of occurrence (cited 68 times by the participants). The specific analysis of this theme was able to disentangle eight subthemes: isolation (represented by seven words), shame (five words), degradation (five words), judgment (four words), fear (four words), integrity (four words), depression (three words), and annoyance (three words).

DISCUSSION

Urinary incontinence raised negative feelings on the part of participants in this study. Feelings of shame and isolation were the most immediate feelings, but the feelings considered most important were those of fear and judgment—particularly from others. This finding is consistent with the results of other studies in which people with urinary incontinence experience negative feelings, including fear of others (Hägglund & Wadensten, 2007). Pads, too, were linked to negative feelings in our study. Although some participants considered them neutral—possibly because they provided some solution—most considered them negative, as they do not solve the problem and sometimes cause other problems, such as reorganizing one’s daily life. In our study, protections suggest primarily feelings of shame, isolation, and judgment, findings that are also found in other studies (Toye & Barker, 2020).

The associative network method has many advantages for studying social representations of a phenomenon. First, this technique allows participants a great deal of freedom of expression. Through simple, rapid instructions and the use of a single inducing word, the method allows participants to express themselves freely and spontaneously. For this reason, and when used with other techniques, the associative network method should be administered first to take advantage of its projective nature (De Rosa, 2002). Second, the method results in rich information that makes this tool an excellent technique for understanding social representations. Indeed, beyond a simple association of words, the associative network also allows researchers to obtain other essential information, such as the structure of the semantic field of the representation or its polarity. This is not possible with the other association techniques (Clemence et al., 1993).

It is also possible to carry out the associative network using another starting element besides the stimulus word. For example, the starting element can be textual, iconographic, filmographic, or even musical (Cohen & Moliner, 2017). Therefore, if we wished to study the social representations of urinary incontinence differently while using the associative network technique, it would be perfectly feasible to replace the stimulus word—urinary incontinence—with an image representing this urinary incontinence. The choice of starting element can thus be adapted according to the participant’s age or social condition. It is also possible to use several stimulus words—especially when the researcher is interested in the study of objects of representation—which are supposed to be mutually defined by the effect of a process of categorical differentiation (De Rosa, 2002). In one of her studies, De Rosa proposed describing the objects of social representations of young Europeans about the European community (De Rosa et al., 2005). To do this, she used several stimulus words, including European Community, European Union, and North–South.

Limitations

The study had some limitations. First, the participants were voluntary and were concerned about the subject. This concern may have increased the amplitude of the polarity indices. In addition, the sample was mostly women. Therefore, we cannot extend our results to men. The associative network method also has some limitations. Although its main advantage is related to the links that can be made between words, it is not possible to know the nature of these links. Also, this technique does not allow for easy interpretation of the results produced by the participants because of the lack of semantic context. Thus, it is preferable to use this technique within a multimethod approach to allow for exploration of all the dimensions of social representations and to bring more reliability to the results (De Rosa, 2002). In our case, we followed this analysis with focus groups.

Conclusion

The associative network method can be an asset for research. It is simple to administer and can be used in studies of many different types of participants—older adults, middle-aged adults, adolescents, or children. It can also be used in a variety of research settings as it is adaptable and quick to set up and carry out. The method could be particularly useful for studying the social representations of many specific nursing phenomena, including specific illnesses, professional issues, and care organizations.

ORCID iDs

Lea Peroni https://orcid.org/0000-0003-1862-2521

Didier Armaingaud https://orcid.org/0000-0002-3857-6975

Charlotte Henry-Ducos https://orcid.org/0000-0003-3821-7208

Monique Rothan-Tondeur https://orcid.org/0000-0002-8356-1713