What Interferes with the Similarity Between Free Lists Applied at Different Times?

Abstract Background: The free list is an ethnographic method that characterizes the local knowledge of a population about a given cultural domain. However, there is still much to be elucidated about the variables that can inuence the number of plants cited by the participants with the use of this technique. This study analyzed whether age and external stimuli inuence the similarity of free lists applied at different times. Methods: Data was collected from 103 farmers from the rural community Alto dos Canutos, in the municipality of Picos, PI. For that purpose, two free lists were applied at two different times, with an interval of three months between them. Subsequently, the similarity between the rst and second free lists of each participant was calculated using the Jaccard Similarity Index. The Generalized Linear Model (GLM) with binomial errors and stepwise approach was also used to analyze how age and external stimuli affect the collection of information when comparing the free lists applied at different times. Results: The age of the participants inuenced the information collected in the free lists, demonstrating that the older the participants, the lower the similarity among the free lists. Among the external stimuli analyzed, only the presence of third parties inuenced the content of the free lists at the time of the interview. However, contrary to expectations, it positively inuenced the similarity of the lists. Conclusion: The results show that the studied variables can inuence the capture of knowledge if the objective of the research is to inuence the individual knowledge of the participants.


Background
The free list is an ethnographic method used to list elements of a given cultural domain [1,2], whereas cultural domain is understood as the set of cultural information belonging to the same category [3]. It is widely used in ethnobiological research to list its elements, in the order in which it comes to the mind of the interviewee. Items are identi ed when the interviewee is asked everything they know about the "X", the "X" being the name of the domain to be questioned [4]. An appropriate question for using the free list to identify information in a category may be, for example: "which medicinal plants do you know"?
The elements captured from this question reveal information both about the items mentioned, as well as about the people who listed them, showing remarkable data in a community/culture [5], such as: the importance based on the consensus of information (frequency of convergent reports) [6]; the preference for components used in a shorter period of time [7]; the determination of local specialists, since large lists characterize the most knowledgeable people [6], and the listing order of the items, in which those mentioned last are considered less important in relation to the rst ones [8]. This order, together with the number of times the item is cited, serves to calculate the Salience Index [9,10] that determines the cultural relevance of a term or element. However, some studies that use the free list do not take into account the use of the Salience Index in their analyzes and these are now based only on the number of times that the same item was mentioned [11,12,13,14].
In many cases, the data listed may not be remembered during the interview [1,6,15], which brings limitations in the responses to the free lists, as this is a collection instrument that depends on the interviewee's memory. Thus, it is possible that individual knowledge is being captured in an incomplete way and this may have consequences on the results of the research, since the hypotheses may be con rmed or disproved based on responses that are biased by different factors.
In order to understand what in uences the responses of participants during the collection of ethnobiological data, it is essential to consider the age factor, since the literature has offered evidence of its in uence on inventories. This is evidenced, for example, when capturing information about plant species, since memory lapses are more frequent in the elderly [16]. Conversely, there are works that report elderly people as specialists and protectors of local knowledge [17], which shows they know a greater diversity of ethnospecies [18,19] and their applications [18,20]. However, other studies have not found great differences in the knowledge of plants among the oldest and youngest [16,21]. Thus, it is often not possible to know if this lack of differences is concrete, as it could be related to a memory lapse effect that affects more the elderly during the interviews.
Certain variables are believed to affect collection of information. However, it is noted that the approaches taken so far in the analysis of age interference, ascertain the level of knowledge only by comparing information obtained from the various members of a researched group. It is necessary to assess the effects of extrinsic variables on the technique by considering an analysis from the same individual, performed at different times.
As well as age, other variables are commonly mentioned in ethnobiological research as possible causes of interference during the capture of knowledge, such as the location and presence of third parties [22]. Miranda et al. [23] provides evidence that different locations where interviews are conducted may provide lists with different sizes and repertoires of plants, however, their interviews were conducted in different places (fairs, supermarkets, plant stores and a control location). However, it is still necessary to discover whether interviews conducted in closed and related places, such as the environment that makes up a community, may or may not lead to free lists with different characteristics.
The presence of third parties, during an interview, may be subject to biasing the data in some way, such as: changes in the respondents' responses [22] or when offering suggestions to them (interference from third parties) [6]. These changes in responses were con rmed in works developed by Boeji [24] and Aquilino [25], when researching the marriage subject. However, the study does not correspond to an approach directed to ethnobiological research. Thus, it is necessary to assess how much this presence, simple or with interference, actually in uences the content of free lists.

Material And Methods
Description of the study area The study was carried out in the state of Piauí, in the municipality of Picos (7°04'37"S and 41°28'01"W). This municipality has approximately 78,222 inhabitants [27] and is located on the right bank of the The rural community Altos dos Canutos (7º09'51''S and 41º33'5''W), which is located 14 km from the center of Picos, was chosen for this study because it is a representative rural community of the region, with easy access to the residence of all inhabitants and for having a local plant-dependent medical system (Fig. 1).
The community has 214 inhabitants distributed among 62 households, according to data provided by the agent of the Family Health Post that exists in the neighboring community (Vigia Community). Its economic activity is based on family farming, with the cultivation of beans (Phaseolus Vulgaris L.) and corn (Zea mays L.). The families consume their production while the surplus is traded. Only the main street of the community has pavement, recently built. All houses are made of masonry and have running water, electricity and regular garbage collection, carried out weekly. The community has a Health Post, where the doctor's visit takes place once a week; the nursery school and elementary school I has been deactivated, and the students have been transferred to the village of Torrões, located 9 km away from the community. They now commute to school by bus.

Community Recognition and Ethical Aspects
Before establishing contact with the community, an authorization was obtained from the Picos Rural Workers Union. Subsequently, the initial communication with the community was facilitated due to the assistance given by the representatives of the Community Residents' Association who promoted a meeting with the aim to present the research and the researcher.
The research was submitted to the Research Ethics Council (CEP) of the Federal University of Piauí and approved under opinion number 89553018.5.0000.8057. In compliance with the precepts of the National Health Council (Resolution n°466, December 12, 2012), which has the ethical aspects of research involving human beings, the objectives of the research were explained to members of the community. Those who agreed to participate had their permission requested by signing the Free and Informed Consent Form (TCLE) as well as their individual authorization to enter their residence.

Interview Data
The community has 62 households with 214 residents. As the research was restricted to the providers only, a total of 124 residents were interviewed. However, some residents are widowed, separated or live alone, so the initial sample would comprise 114 residents.
The three-month period of this survey was a decision made respecting the logistics of the work to make sure that all residents were interviewed in the same period in both stages. In addition, this interval was proposed so that it would not become tiring for the participants, since the lists were applied to the same interviewer using the same question.
The following protocols were used to conduct the interviews: 1. All interviews took place with the same interviewer; 2. There was a difference of three months from one interview to another for all participants; 3. The participants were informed that the interviews should be carried out individually, avoiding other people nearby. However, this was not always possible, as there were times when other people appeared during the interview. The fact that a third person remained in the same room as the interviewer during the interview was considered as a presence of third parties and interference by third parties if this person at any time intervened (trying to assist the informant); 4. The participant was informed that the interviewer would record their response in writing.
During the application of the free lists, the observation approach [31] was used, which allowed the interviewer to be aware of the possible variations that could occur and to record information in a discreet way. For this, the interview forms had elds to record the presence of third parties, their degree of kinship with the interviewee and if this caused any intervention, the place where the interview was conducted and the other possible factors that could interfere.
Regarding the pro le of the interviewees, the age group of the interviewees ranged from 21 to 85 years with 33% being young people (aged from 21 to 39 years); 36.9% adults (aged from 40 to 59 years) and 30.1% elderly (aged from 60 to 83 years). The number of species cited in the rst free list reached 116 species, 25 of which were exclusive to this list and 100 in the second free list, 11 being exclusive. Altogether 127 different species of medicinal plants were cited (Table 2-Additional le 1).

Botanical Material Collection
After the end of each free list, the plants cited were collected, giving preference to fertile specimens following the procedures used in the studies of plant taxonomy [32]. However, some plants were not found in the native vegetation of the community, as they were bought in the market or in a medicinal plant fair, in the municipality of Picos-PI.
Herbalization techniques were applied to the collected plants for exsiccates and duplicates, according to the usual procedure [33].
Botanical identi cations were made with reference in the specialized literature and by consultation with a specialist. The classi cation system adopted was the Angiosperm Phylogeny Group IV [34]. The Latin names were revised and the abbreviations of the authors' names acquired through the Missouri Botanical Garden database [35].

Preparation and Analysis of Ethnobotanical Data
A binary matrix in excel (presence and absence) was performed with the registration of all species cited in the two free lists for the preparation of the data.
The two free lists were compared and for each participant it was considered: (a) the plants that were cited in both free lists and (b) the plants that were cited in only one of the free lists. From these data, the similarity of the plants cited were calculated using the Jaccard Similarity Index (JI) [36] for each participant, through the following formula: JI= a/a+b. With the result of similarity data for each participant, the mean and standard deviation were calculated. From the results of similarity data obtained from each participant, the mean and standard deviation were calculated. When compared, the free lists showed low similarity with a mean of 0.26 and a standard deviation of 0.16.
The Generalized Linear Model (GLM) with binomial errors was also used. At rst, the explanatory variables would be age, place of interview, presence of third parties, interference by third parties during the interview and presence of factors that could affect the interview. However, we performed a multicollinearity test, using the function 'omcdiag' of the package 'mctest' of R and observed a positive result. Thus, we removed the third party interference variable from the model. The response variable was a double column with (1) the number of double presence of the species mentioned in the two free lists, and (2) the number of plants exclusive to a single list (associated columns using the "cbind" function of R). The stepwise approach was also used to leave the variables that gave the lowest AIC values in the model (Akaike Information Criterion). The Generalized Linear Model (GLM) with binomial errors was performed using Software R (version 3.5.0) [37].
Except for age, the other explanatory variables were coded for use in GLM as explained below:

Results
Does the age of the interviewee in uence on the recall of medicinal plants known to them? It was con rmed that age was one of the variables that made up the model, so that older people showed less similarity between their lists, reducing the double presence of the species mentioned in the free lists in relation to other age groups (Table 3). The location where the interview was conducted, the interference of third parties and the factors that may in uence the interview did not remain in the model, as they did not interfere with the content of the free lists during the collection of data. Still, contrary to what was expected, the presence of third parties positively in uenced the similarity in the lists (Table 3), making the free lists more similar.

Discussion
Does the age of the interviewee in uence the recall of medicinal plants known to him?
The data obtained for this research show that older people had less similar free lists, thus reducing the double presence of species between the lists. This indicates that even in the face of a relatively short interval (three months), as proposed in this research, elderly people are not always able to recall e ciently the names of the plants known to them. It is important to note that this study compared the similarity of cited species and as there was a time interval from one free list to another, it is necessary to consider that the difference between the lists may be due to something temporary in these individuals. Even with this observation, the result of this research indicates the need for greater caution regarding the collection and analysis of information from this individual knowledge, especially when related to this age group, in order to avoid bias in the results.
A possible justi cation for presenting different free lists in this study is due to a characteristic inherent to older people: impaired memory. Researches claim that elderly people are more forgetful than young people [18, 42] and this forgetfulness has been discussed in ethnobiology, noting that the recorded knowledge begins to decline after the group aged between 59 to 68 years [18]. Thus, the interviewees of this group do not remember all the medicinal plants they know, needing more effort to remember, which brings limitations in the answers.
In these cases, it cannot be a rmed precisely that a greater number of plants mentioned by one participant represents greater knowledge in relation to another participant who mentioned fewer plants.
Thus, in communities with a predominance of the elderly, it is necessary to avoid evaluations that consider individual knowledge, based only on the number of plants mentioned in a single free list.
It is suggested the use of more than one free list, with the same participants, in communities with high rates of elderly people. This will help to collect more complete data about plants known to this age group and will help to validate individual knowledge, thereby improving the quality of the data collected. Another suggestion would be to opt for the interview-checklist [43], a technique that provides more stimulus to the interviewee and provides less memory effort in relation to the free list. In addition, it allows the interviewee to be more comfortable and makes it easier to remember the question. Wiryono et al. (42) recorded the correlation between botanical knowledge and people's age. The authors observed that when using photographs to identify known plants within a community, the elderly recognize more plants and better report their respective uses in relation to younger people. However, there is still a few studies that link age and the use of visual stimuli, requiring further testing for this statement.
In view of the context presented above on the in uence of age on the collection of ethnobiological data, the researcher who is aware of this approach will be able to identify the technique that best suits their research, and it is up to them to select the best data collection instrument to capture the desired information.
Do external stimuli in uence the content of free lists?
Regarding the different places where the interviews were conducted, our results show that it did not interfere with the plants cited. In contrast, the study by Miranda et al. [23] reported that the place where the interview is conducted has a direct in uence on the responses mentioned by the interviewees, in which the participants listed the plants according to the local context in which the interviews were conducted. Thus, ethnobiological studies suggest that the presence of environmental stimuli during the execution of free lists presents a limitation to the method, as the participants can be in uenced by them at the time of the interview.
As in the present study, where the interviews were conducted with non-specialists and with native vegetation occurring close to the residences, the place of application of interviews was also the object of a study carried out in southeastern Madagascar. This study found that the proximity of the participants' house to the forest had no in uence on the medicinal plants mentioned [12]. In the rural community studied here, the interviews were conducted in closed spaces and in open spaces, where in the latter the participant had visual contact with nature and even it still did not interfere with the content of the free lists. However, studies need to be repeated in other communities to understand if these results represent a general pattern.
In ethnobiological research, interviews can take place in different environments, such as, at the participant's home, close to home, at the workplace, among others and in living spaces. Among those, the living space is the least suitable in opposition to the other places, where the participants have a greater control of the spaces [2] and feel more comfortable. In addition, a quiet place, away from the areas of greatest activities should be prioritized [44]. In the studied community, the interviews were conducted at the participant's house, the neighbor's house and at the workplace (within the community), a very familiar environment and perhaps that is the reason the place has not in uenced on the collection of information, as the selected space prepares a relaxed and enjoyable environment for the interview.
Regarding the presence and interference of third parties, there are related studies, which focused on the theme of marriage. Thiese have found that the presence of the spouse affected the responses of the participants, either because the presence of the partner brought positive evaluations about the marriage or because they had to agree with the partner, making it di cult the disclosure of negative or con ictiong positions [45,46]. However, although it is recommended that interviews take place individually, mainly in studies aimed at assessing individual knowledge, there are still no studies that have addressed the in uence of the presence and interference of a third person in ethnobiological studies.
Other studies suggest that the presence of third parties may interfere in the results. This may happen at the moment when responses are suggested to the interviewee, impairing the data collection, since the items must be said by the interviewee in the order that comes to their mind [6]. Thus, it is suggested that the presence of a third person should be avoided [47], although sometimes it is not possible to prevent it. In this study, it was found that the interference of third parties did not in uence on the collection of information.
The For example, people may feel intimidated in reporting about some plants, especially if they are related to something more intimate, such as diseases related to the genitourinary system (prostate, urinary in ammation, in ammation of the uterus, among others) and this can lead the participant to omit certain plants. However, it is worth mentioning that the question was very general: which medicinal plants do you "know"? Thus, they were able to cite the plants without directly relating to those that are used in their daily lives. This possibly did not interfere with the fact that a person was present during the interview.
In this context, the researcher will be aware of his actions when using this technique in the presence of third parties. The researcher may, depending on the circumstances, decide whether to continue the interview, return at another time and/or simply leave. Thus, the presence of a third party can be bene cial if the objective of the research is to capture the general knowledge of a community.
However, the need for more detailed research is emphasized, in order to verify in which situation the presence of a third party is positive or negative during the interviews with the use of the free list. For example, checking whether the presence of a person of the opposite sex intimidates the participant from reporting certain types of plants associated with the treatment of any gender-related illness.
Even though it is not the interviewer's desire, the interviews suffer external in uences during the research, which is beyond the control of the researcher [47]. During this work, factors were also present, although these did not interfere with the plant species mentioned in the free lists.
However, the low similarity found in the free lists at different times, indicates that something made the participants lose their attention at the time of the interview. Even though it is a data collection technique that is considered simple and allows information to be obtained quickly [6], only requiring short answers, which would decrease the chance of the participant not concentrating on answering them [48]. It is believed that these advantages reported on the technique did not make the participants feel uncomfortable, even performing activities at the time of the interview, a circumstance that also did not in uence the data provided. Despite this, it is understood that the presence of factors did not help to disperse the interviewee's attention and did not hinder the disclosure of information, not affecting the answers to the questions.
Therefore, it is believed that these factors have not interfered precisely because they make the environment as natural as possible for the interviewee. If the interviewees had been placed in an arti cial environment, that is, in an isolated location, it could have been di cult to report the information. This nding is consistent with the literature [49], which states that for data collection to occur effectively it is necessary that it takes place in the most familiar way to the interviewee.

Conclusion And Future Perspectives
To understand the variables present in the use of the free list, some were highlighted in this research. To cover all of them, a generalist approach was applied. The results analyzed can serve as a basis for future studies, even though we know that there are other variables in this scenario that also deserve to be analyzed, such as the socioeconomic pro le and education, to understand how they affect the collection of ethnobiological data. In addition to these, it is also necessary to investigate the environmental context in which the informant is inserted in relation to variations in the environment and the availability of plant resources. This could be achieved by raising the following question: will the plants present in the surroundings of a community be present in people's memory even when they are in a closed environment, then being remembered?
The age factor in uenced the data collection process during the execution of free lists. The interviewer, knowing the consequence that the free list has, can choose whether or not to use this data collection technique to capture the desired dimension. If this technique is selected, it is suggested that it is performed twice with the same sample to access more data, check the difference between the two moments and thus be able to validate the knowledge and increase the reliability of the data collected by people within this age range.
Regarding the presence of third parties, this study showed a positive in uence that caused the interviews to present more similar lists. With this, it is believed that the interviewer does not always need to worry about the presence of third parties, being able to decide whether or not they remain during the application of a free list. Therefore, if the researcher's objective is to recover the maximum knowledge of plants disseminated within the group of people being studied, the researcher need not have concerns about the presence of third parties. However, this presence needs to be rethought if the aim is to understand the participant's individual knowledge. Further investigations are still needed to verify in which situation the presence of a third party is positive or negative during the interviews with the use of the free list, for example, to verify if the presence of a person of the opposite sex intimidates the participant to report certain types of plants associated with some gender-related diseases.

Consent for publication
Not applicable.