Theoretical background: Narrative mindset and narrative knowledge

The narrative mindset is a tendency to interpret social information in the frame of stories. Stories are an essential tool for structuring and comprehending social behavior and a person’s experiences [22–28]. The building blocks of stories are motives, and the complications that induce, impede or block their fulfillment. In construing understanding of events and situations as stories, or people as stories’ protagonists, an individual uses general knowledge about human motives and specific contexts of evoking and pursuing them [29–33]. This narrative knowledge is developed during social interactions [34, 35] and is influenced by culture. At least in contemporary societies, it is specific for different social domains [36–38]. An example is the self-story: a stabilized knowledge structure that influences a person’s identification and decisions [39–43]. In most parts, it is shared with others (among members of one’s group or society).

When the narrative mindset is activated, the narrative knowledge is ready to understand and memorize incoming social information. As a result, a narrative mindset tends to structurize social information in story plots: the protagonists with intentions, obstacles, conditions of fulfillment, and timely ordered routes of striving for them [44–47]. There are several other influential and common tendencies in processing social realities. Some of these non-narrative processes are called paradigmatic [23–48]. The diverse examples for this kind of information processing include comparing objects, traits, and psychological states, trait attribution in person perception, social categorization, stereotyping, scientific explanations, creating and applying the law and regulations, and learning knowledge in school. However, it seems that the narrative mindset and knowledge greatly influence our everyday social interactions [26].

The empirical background: Impact of the narrative mindset on social cognition

Narrative knowledge and mindset result in understanding events as stories, but it is also a basis for other kinds of social cognitions [49]. The narrative mindset is likely to be a very human frame for making an impression of another, understanding their internal states, motivations, problems, and creating possible future simulations or finding solutions for social problems.

The basic building blocks of the narrative mindset and knowledge are intention categories like wishes, goals, plans, and emotions. Goal priming experiments provide evidence of the narrative mindset activating the intention category. If goals or motives are a major mental category of narratives, a narrative mindset should be a potent means of goal priming. Evidence shows that once primed, goals are accessible and have the potential to influence behavior for some time [e.g., 50, 51], but once a goal is fulfilled, goal-related information is deactivated, becoming less accessible and less likely to influence subsequent behavior [e.g., 52, 53]. Laham and Kashima [54] used this observation to distinguish the priming of a goal as a mental category from simple semantic priming by the goal’s term. Participants were primed with high achievement either after the narrative mindset activation or in a non-narrative context. They then completed either a behavioral task, followed by a measure of construct accessibility, or a behavioral task after a delay. Indicative of goal category priming, narrative-set participants showed more significant post-delay behavioral assimilation and less post-fulfillment accessibility than those exposed to the non-narrative context.

The intention category is a key factor in structuring social reality as meaningful stories and comprehensive events. The crucial role of the intention category in processing incoming information is evidenced by observations from cognitive psychology and psycholinguistics [55]. For example, being unable to discern the intention behind a segment of actions disturbs recognizing these actions as a single event [56, 57]. People spontaneously impose default motives when detecting complex facts and actions as a story or an understandable event [58]. Furthermore, different information regarding motive or emotion affects observers’ interpretations of the same chain of behaviors. Jury transcripts provide naturalistic observation. It appears that hypotheses for the defendant’s motive determine what evidence the jury is looking at and how it is interpreted, which subsequently influences the defendant’s conviction of guilty or innocent [59, 60].

The above data indicate that understanding in a story form means knowing the intention and applying this knowledge in processing information. The narrative mindset’s priming should increase the significance of the intention and emotion’s mental categories in understanding another person [61]. We assumed that it effects positively the quality of person impression. In one experiment, participants received a description of a young person presented with a professional label, e.g., "a policeman" [62]. They were asked to freely describe their impressions of the person and choose from the provided list a specified number of words that characterize the person best. Before the presentation, they participated in the narrative priming (writing a story based on provided pictures) or one of two neutral conditions (grouping social objects, no activity). After narrative priming, the impressions were scored by independent coders as more psychologically coherent, easier to imagine, and more complex compared to the neutral conditions. Furthermore, their authors were more confident in the validity of the impressions and the distinctness of the person. The narrative primed participants chose more words related to the motive domain as best descriptions of the person’s character and fewer words stereotyping the person e.g., specific to the professional category. In another study, participants obtained a person’s description that was divided into two sets, portraying two different personality profiles, and were asked to describe their impression of the person. Independent coders evaluated the descriptions after the narrative priming as more effectively integrating the inconsistent characteristics [63]. In conclusion, it seems that–compared to other everyday contexts of person perception–the narrative mindset facilitates impressions organized around a person’s motives and emotions. These are more individualized and less stereotyped, and easier to visualize than in other person perception contexts. These impressions are more coherent psychologically, even when based on inconsistent data.

The following studies observed the impact of the narrative mindset on the readiness to evoke motive-related categories in an impression of a person, measured by reaction time [62]. The attribution was in the form of fast yes vs. no decisions to state whether the person may possess a specified characteristic from a provided list. Before presenting the psychological portrait of the person, the participants were primed by a task inducing the narrative mindset, a neutral task, or were not primed. Provided characteristics belonged to general personality traits (quiet, straight, pessimistic) or broad motives (loves, strives for success, jealousy). It appeared that the narrative priming shortens the decision reaction time case of motive attributions, but not trait attributions. This effect was observed at the level of differences in fast reactions, which are not controlled consciously. It allows a conclusion that the impact of the narrative mindset and activated narrative knowledge on social cognition is independent of the level of conscious and effortful control, which is typical for schema-driven mental processes. When activated, e.g., by priming, they regulate conscious thinking [63].

We further expected that, because the narrative mindset accentuates the motives and emotions and provides a framework to integrate them into a meaningful vision of a person, this mindset may also induce a better understanding of a person in their specific situation, for example, in difficulty. It may take the form of empathizing with a person in need and may lead to helping. It should be easier to understand and feel a person’s problem or tragic situation, and subsequently help them if we know this person’s story, even if it is only indirect. Mass media provide a lot of examples. TV news about a critical drought and hunger in Ethiopia in 1991 illustrated by one victim’s life story, a small girl from an Ethiopian village, lead to higher financial donations than presenting some tragic statistics [30].

In one experiment, participants were informed they would be participating in an informal meeting organized by the medical service in which they would be asked to consider helping individuals with leukemia [64]. Each subject was randomly given one of two texts: a description of the life of an ill man with leukemia or information about leukemia and the role of bone marrow transplantation in therapy. Next, they were asked to decide whether they wanted to participate in a bone marrow donation program. Finally, the participants filled out a test measuring their general ability to detect the narrative structure in provided material [65]. The data confirmed that the presentation of a story of an ill person related to a higher percentage of declarations of help (which was verified after two days). Still, this effect occurred almost exclusively among those who had a higher narrative competency. It may suggest that the observed effect’s main factor was the narrative mindset activated more easily in those who could extract and understand the narrative message of the text in a more elaborate way. We observed from the literature that reading a story about a person, or category of persons in need, increases the probability of helping. Narrative priming before the presentation of a person in need strengthens the willingness to spend time on the phone soliciting money for an ill person and readiness to take a course preparing to be a member of the Big Sisters–Big Brothers program [62].

We may conclude that being in the narrative mindset evokes a more straightforward empathic understanding of a person in need. Activated narrative knowledge provides a framework that enables identifying and integrating a person’s motives, emotions, and situational context in a coherent vision. It attracts attention, arouses feelings and moral convictions, and increases the willingness to help.

We supposed that there is still another kind of narrative mindset that influences understanding of other people’s circumstances. The narrative framework may facilitate imagining oneself in the other person’s situation [66]. We called it the personalization of other people’s problems. This facilitation results from structural similarities between understanding other people’s problems within a story frame and knowledge emerging from a person’s life experiences–their experiential knowledge. Experiential knowledge is organized in episodic memories and scripts based on repetitive social events [30]. This knowledge is transformed creatively and is enriched by memories shared with close ones. [67, 68]. It is hypothesized to have a narrative-level organization proven by a better recall of events that form coherent narratives than ones that do not [69]. Linguistic analyses of protocols with spontaneous descriptions of memories collected in different life situations indicate that experiential knowledge is essentially about trajectories of motivated actions and related emotions modified, blocked, or helped by incoming situations [e.g., 70–72]. When the narrative mindset facilitates story framing of other people’s difficulties, thinking about someone else’s difficulty activates their experiential knowledge more effortless because of that similarity. We posit that personalization of someone else’s problem is especially useful in situations with no ready-to-use advice, such as common-sense advice learned and socially shared as behavioral scripts or proverbs [73]. This "(…) forces the problem-solver to abstract elements from his own subjective experience and to utilize these elements as the foundation for highly subjective solutions" [5, p. 232].

Our experiential knowledge plays a significant role in reasoning, such as social problem-solving and imagining and planning the future [74]. Neurological studies show that damage to brain areas supporting episodic memory coincides with a lowered ability to imagine future and possible events [75, 76]. Episodic memory processes, mediated by the medial temporal lobes, contribute to open-ended problem-solving [18]. Furthermore, after a brief training session for recollecting details of past experiences, young and older adults perform better on memory and imagination tasks, as well as means-end problem-solving tasks [15, 77]. On the other hand, lowered autobiographical memory after posttraumatic stress disorder relates to ineffective social problem solving [78]. Parasuicide patients tend to retrieve overgeneral autobiographical memories (i.e., with lesser specific emotional content), which relates to problem-solving deficits [79, see also: 80]. High alexithymia (inability to verbalize emotions) correlates with deficits in both episodic memory [81] and mentalizing skills, such as perspective-taking mental imagery, fantasizing, and narrative transportation [82–84].

In conclusion, experiential knowledge is a helpful resource in thinking, especially in the social domains. Structural similarities between the narrative understanding of a person coping with a problem and our personal experiences will spontaneously prompt us to wonder how we would feel and cope in the same situation, making it easier to find a solution. We expected that activating personal experiential knowledge would result in a more elaborate understanding of the problem and, therefore, more effective thinking about solving it.

The hypotheses

Based on the presented data and its interpretations, we expected that the narrative mindset facilitates an empathic understanding of someone experiencing trouble and seeking a solution. The activated narrative knowledge enables one to better understand and efficiently think about another person’s intentions, emotions, and situational context. The narrative interpretation of the problem also opens a way to personalize it: to use personal experiences to understand the dilemma and offer good advice.

In two experiments, we verified the connected hypotheses that evoked narrative mindset facilitates (1) empathic understanding of another person with a problem, (2) personalization of the problem, and subsequently (3) increases the solution’s effectiveness in the advice form. Effective advice is advice that (probably) resolves the problem when applied. In our research, the measure of advice’s effectiveness was based on the coders’ evaluation of the extent to which the advised action may solve a given problem in reality. To broaden the verification data, we observed the impact of the activated narrative mindset in two cases of social problems often analyzed in classical research: interpersonal (connected with other people) and emotional (struggle with emotions) [2, 5].

Experimental activation of the narrative mindset

To observe the effects of the narrative mindset, we used an experimental procedure of mindset activation. Diversified data illustrate how easily simple word stimuli used as primes can activate a mindset that regulates thinking and behavior in a subsequent situation. Such word stimuli addressed a complex knowledge structure (e.g., the word "mother"). It was shown that using even simple words as primes (like mother vs. colleague or friend vs. co-worker) unconsciously activate different knowledge mindsets that provide criteria of person evaluation, affect the decision to help, stipulate specific goal-oriented behavior or behavioral stereotypes, or affect the likelihood of aggressive or polite behavior [52, 85–88].

The narrative mindset is frequent and influential among people. It may be induced when listening, reading, or watching pictures or movies and making a story [89]. In our study, the narrative mindset was activated by a task to create a story illustrated in the provided pictures. Providing a task to perform specified action (hammer a nail) and then observing transitory readiness to use the same action to cope with a different task (repairing a watch) is classical experimentation on a mindset effects on problem-solving [90, 91].

There are three benefits of using such a procedure. First, the task used to evoke the mindset, like "create a story illustrated in provided pictures", is a simple instruction, understood equally by all participants. Similarly, the stimuli–pictures in this case–are easily understood. Finally, the control condition might be closer to fulfilling "the smallest difference" requirement as participants are working on the same set of pictures as the narrative set participants, performing a known ordinary task of grouping objects according to their similarity.

As in many studies on social problem solving, a participant was asked to provide a solution to a problem presented in a 3^rd person form, meaning it was a solution for the other person, not their own problem [92]. This may eliminate the influence of different uncontrolled personal factors. We took an open approach in observing the impact: the participants were presented with a problem and had freedom to invent the best advice. This allows better inquiring about the expected effects of the narrative mindset on the advice’s content characteristics.

Ethics statement

The studies involving human participants were reviewed and approved by Research Ethics Committee at the SWPS University, Department of Psychology. The patients/participants provided their written informed consent to participate in this study.

Plan of the two experiments

Two experiments had the same structures and similar tasks for the participants. They differed in the kinds of problems to be solved. There were interpersonal problems in experiment 1 and emotional problems in experiment 2. The experiments consisted of two parts. In the first part, all participants were assigned randomly to one of two conditions: inducement of the narrative mindset or the control condition.

In the narrative mindset condition, each participant was presented with six pictures depicting expressive scenes with people interacting. The same woman and man appeared in each picture, together or alone, and with other people in one of the pictures. A participant was asked to look at the pictures and imagine a story that might be illustrated on them and then describe the story: "Describe the story that you thought about when looking at the pictures. Describe how the story started, what is happening in it, and how it ends. Describe what the characters are all about and what is bothering them, and why all of this is important for them. Do not restrain yourself in any way when searching for a meaningful story".

In the control condition, each participant obtained the same array of pictures as used in the narrative mindset condition and was asked to categorize them into all subcategories that come to mind that have a common feature and then to describe each feature: "Your task will be to find such sets of pictures that have a common unique feature, a feature that the rest of pictures don’t have. You can take various things into consideration, even unobvious ones: the type of place, the accessories, the way the people appear or behave, their emotions. It could be the way they are dressed. You can take into account the wider environment in which the pictured places might be. Do not restrain yourself in any way when searching for the common features".

Participants could take as much time as they needed to complete the task. In the pilot studies, no significant differences were observed between the two experimental conditions on scales of instruction understandability or task difficulty, or task completion.

At the end of this part, the participants were told this was the end of study one and were invited to participate in study two (which was the second part of the same experiment). The second part of the experiment was a social problem-solving task. In experiment 1, participants were asked to provide advice on how to solve four interpersonal problems, and in experiment 2, on how to solve four emotional problems. In each case, the problem’s description included brief information specifying a particular person’s needs, like some characteristics of the person and the social environment. Participants’ task was to invent the best solution in the form of advice. Therefore, it was a divergent version of the problem-solving procedure: a participant was free to create the solution. In other studies, this procedure is usually more convergent: a participant is provided with the need and positive ending and is asked to find steps leading to such an end [4]. We assumed that the divergent version would facilitate a more elaborate search for the expected mindset-related differences in the advice content.

In both experiments, the sample sizes were determined following previous studies on social problem solving with experimental manipulation of episodic memory specificity induction [14, 15] and on the influence of narrative mindset on various social cognition tasks [62, 63].

An objective analysis of the attentional focus of advice-givers

To provide a further assessment of problem personalization, independent from coders’ evaluations, we performed an analysis using the Linguistic Inquiry and Word Count (LIWC) 2001 program [93]. It analyzes texts by scanning them in a word-by-word manner and calculating words within specified categories. Verb tense and personal pronouns are indicators of a person’s focus. We believe that the number of 1^st person singular pronouns (like I, my, me, myself) and past tense verbs (like walked, were, had) would be good metrics for the extent to which advice-givers utilized personal experiences [94, 95]. The use of 1^st person singular pronouns is characteristic for situations when a person is focused on oneself and takes personal ownership of the reported experience [95, 96]. Personal pronouns and past tense verbs are characteristic of informal communication with colleagues or close ones, based on sharing personal experiences. Simultaneously, they are infrequent in complex and analytical thinking used in collaborative work on shared tasks. Next, past tense verbs are not common in communication with strangers that lack disclosure [94]. Texts describing personal stories can be differentiated from the other types of social media posts based on a higher frequency of past tense verbs and first-person pronouns [97]. The higher number of them, the closer psychologically the person is to the described events [98]. We expected that significantly more 1^st person singular pronouns and past tense verbs would be present in the advice written by narrative mindset condition participants. However, based on previous studies using the LIWC [95], we supposed that correlations would be small to moderate rather than strong.

Participants

The experiment finally included 109 mTurk users (70 men and 39 women) who correctly followed the instructions. Before analysis, 12 participants (7 from the experimental condition) were excluded by independent coders for one or two of these reasons: (a) possible cheating (the same text in more than one participant), (b) provided text was irrelevant to the task and was probably copied from other sources. Participants were recruited from the English language pool of the mTurk workers and paid $3.5 for their input. The experiment was conducted online and was done individually. Table 1 presents detailed demographics.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics of the participants in the interpersonal problem-solving study.

https://doi.org/10.1371/journal.pone.0253729.t001

Procedure and materials

Participants were asked demographic questions (gender, age) and then randomly assigned to the narrative mindset activation (N = 53) or a control condition (N = 56). After fulfilling the first part (a task with the pictures), they were invited to the second one. Next, they received the following instruction: "You will be presented with a description of four persons, each of them having specific problems to overcome. Your task is to invent a solution to each problem. Imagine that you give advice to each person. What would you advise them to do? Consider each problem carefully and then provide the best solution. Try to be detailed in the description of this solution. Imagine that each person has the same gender as you".

Then four problems from four different people were presented. Descriptions of problems were based on the MEPS task [4]. Descriptions were gender-neutral.

Four interpersonal problems were as follows:

1. "A realizes that a friend is avoiding her/him. A wanted very much to make up with him/her.
2. B had just moved in that day and didn’t know anyone. B wanted to have friends in the neighborhood.
3. C must prepare materials for a presentation in 2 weeks together with a shy classmate with whom C has rarely communicated.
4. D was listening to people speaking at a meeting about how to make things better in her/his neighborhood. D wanted to say something important and to have a chance to be a leader too."

The problem’s description ended with the task: "What is your advice for [letter pointing on a person from task]?". Participants were given an unlimited amount of time to describe their advice.

Advice evaluation–coders’ assessment

Three pairs of coders, all of whom were postgraduate clinical psychology students experienced with elementary psychological training and advising in everyday interpersonal and emotional problems, worked independently on the provided texts with advice. They assessed the advice on three scales (each pair working on one scale):

The scale of effectivity–evaluates how well given advice would solve the person’s problem if applied. Coders were asked to assess the extent to which an application of given advice would solve the problem on a 5-points scale ranging from "1"–not solves to "5"–solves completely. They were instructed to consider the specificity of the person’s problem.

The scale of empathy–evaluates how deeply the provided advice is based on an empathetic understanding of the person in need. Coders were instructed to assess the extent to which participants referred to the person’s feelings, emotions, motives, and specific way of understanding and thinking (current or possible at the course of getting to a solution). The 5-point scale ranged from "1"–absent to "5"–very highly.

The scale of personalization–evaluates the degree to which participants introduced their own experience when formulating the advice. Coders evaluated how often a participant uses personal references such as recalling personal experiences and stating opinions and beliefs that stem from life experiences, both their own and those of friends or family. They might be explicitly used as an argument for the provided advice or read between the lines. The 5-point scale ranges from "1"–absent to "5"–very highly.

Before the scoring, coders were trained on material from the pilot studies. They were given detailed instructions and example pieces of advice written by participants of the pilot studies. Their first assessments were a topic of shared discussion to ensure their full understanding of the scales’ character and to calibrate the scales’ evaluation in a coordinated way. After that, coders worked independently.

Two coefficients were used to assess the inter-rater reliability, in line with previous studies that applied the qualitative scale assessment of problem-solving results [e.g., 9, 92, 99]. Pearson’s correlation coefficient was used to estimate the degree of agreement on the coders’ overall evaluations. These were the means of the ratings for the four pieces of advice which were calculated separately for each of the three dimensions. Coders’ evaluations correlated as follows: efficiency scale r = .89 (p < .001); empathy scale, r = .92 (p < .001), and personalization scale r = .87 (p < .001). The range of the correlations is similar to those reported in previous studies [e.g., 92].

The weighted kappa coefficient was used to assess the level of agreement on each piece of advice separately. This allowed for a more detailed evaluation of the inter-rater reliability. We chose the weighted kappa because it computes the adjacent ratings as partial agreements; thus it provides more accurate estimates for ordinal variables than the standard Cohen’s kappa [100]. We chose the linear weighting because it gives more reliable estimates when there is no reason to assume that the adjacent categories’ differences are of different magnitude [101]. The results can be found in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Interrater reliability of the coders’ evaluations of the pieces of advice—linearly weighted kappa (the interpersonal problem-solving study).

https://doi.org/10.1371/journal.pone.0253729.t002

In line with the Cohen’s guidelines, the level of agreement between the coders for the majority of evaluations can be judged as substantial (.61 - .80), in three cases as moderate (.41 - .60) and in the case of the third advice on the personalization of problem scale, as fair (.21 - .40). The final measures used in the statistical analysis were computed by calculating the average of the four ratings for each of the two coders in a given pair separately and then taking the mean of the two resulting values.

Examples of pieces of advice that illustrate high and low results on each scale are presented in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Examples from participants’ pieces of advice from both conducted experiments.

https://doi.org/10.1371/journal.pone.0253729.t003

Data analysis

Analyses were performed using IBM SPSS 25 software and RStudio v1.2. The data used in the analyses can be found here: http://cabanski.pl/files/NM_Emotional_and_interpersonal_problem-solving_databases.zip. The assumption of the normal distribution of residuals was examined using Q-Q plots. Whenever there was at least a moderate deviation from linearity, we used a non-parametric test. Otherwise, a parametric test was used. Comparisons between the narrative mindset and control conditions were analyzed using either t-test for independent samples or Mann-Whitney U. The effect size for t-test comparisons was computed using Cohen’s d. Correlations were computed using either Pearson’s or Spearman’s coefficients.

The mediation models were assessed using the PROCESS macro for SPSS created by Preacher and Hayes. Specifically, models four and six were used for simple and dual mediation analyses, respectively. The indirect effects were estimated with the bootstrap method of 5000 samples with a 95% confidence interval. The indirect effect is deemed significant if the 95% confidence interval does not cross zero. The estimated regression coefficients were standardized for the continuous variables and partially standardized for the dichotomous variables. The indirect effects were presented in the unstandardized form.

Correlations and group comparisons

Across all three scales the narrative mindset group was rated a higher score. The narrative mindset group was judged to have more effective advice (M = 3.49, SD = .49) than the control group (M = 3.19, SD = .87), t(87.57) = 2.19, p = .031, d = .42; to be more empathic (M = 2.23; SD = .79) compared to the control group (M = 1.81, SD = .64), t(107) = 3.10, p = .002, d = .58; and to be more reliant on personalization of problem (Mdn = 1) than the control group (Mdn = 1), Z = 2.66; p = .008. There were no statistically significant differences in 1^st person singular pronouns (Z = 0.13; p = .893) and for past tense verbs (Z = 1.74; p = .082). The results are presented in Table 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Comparison of control vs narrative mindset conditions on the effectivity of advice, empathy, personalization of problem and the LIWC measures (the interpersonal problem-solving study).

https://doi.org/10.1371/journal.pone.0253729.t004

The effectivity of advice, empathy, and personalization of problems all correlated positively. Empathy and effectivity of advice were strongly associated (r = .66; p < .001), while personalization of problem was moderately associated with both effectivity of advice (rho = .37; p < .001) and empathy (rho = .46; p < .001).

When tested separately for each of the two conditions, all correlations remained significant and positive, but there was a change in strength in the association between personalization of problem and effectivity of advice. In the narrative mindset condition, the correlation was moderate (rho = .45; p = .001), while in the control condition it was weak (rho = .28; p = .036). Empathy correlated strongly with effectivity of advice in both the narrative mindset (r = .70; p < .001) and control (r = .68; p < .001) conditions, and the association between personalization of problem and empathy was moderate in both the narrative mindset (rho = .47; p < .001) and control (rho = .38; p = .004) conditions.

Spearman’s correlations analyses revealed a positive and weak association between the number of 1^st person singular pronouns and the problem’s personalization. The correlation occurred for each of the coders’ computed ratings separately as well as the combined rating of the two. No more correlations were found to be significant. The results can be found in Table 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Correlations between LIWC measures and the coders’ assessments on each scale in the interpersonal problem-solving study–Spearman’s coefficient (N = 109).

https://doi.org/10.1371/journal.pone.0253729.t005

Simple mediation analyses

When analyzed using separate simple mediation models, both personalization of problem, BCa CI95% [.04, .19], and empathy, BCa CI95% [.10, .45], mediated significantly the effect of narrative mindset on the effectivity of advice. Each of the models yielded a complete mediation effect, as the direct effects of the narrative mindset on the effectivity of advice were not significant for the model with personalization of problem (β = .26, p = .171) and empathy (β = .03, p = .832) respectively. The mediation models are presented in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Mediating effect of personalization of problem and empathy on the association between narrative mindset and effectivity of advice–two separate simple mediation models (the interpersonal problem- solving study).

Note: Presented regression coefficients were partially standardized for the narrative mindset predictor. The remaining were fully standardized. Coding for the independent variable: 0 –control, 1 –experimental. * p < .05. ** p < .01.*** p < .001.

https://doi.org/10.1371/journal.pone.0253729.g001

Dual-mediation model

Inducing a narrative mindset led to a stronger personalization of the problem (β = .54, p = .004). Both narrative mindset (β = .35, p = .048) and personalization of problem (β = .41, p < .001) positively predicted empathy, but there was no indirect effect of narrative mindset on empathy through personalization of problem, BCa CI95% [-.05, .04]. When combined together, empathy positively predicted effectivity of advice (β = .65, p < .001) while narrative mindset (β = .03, p = .835) and personalization of problem (β = .00, p = .995) were not significant. Therefore, there was a complete l mediation effect of the narrative mindset on the effectivity of advice, which was partly mediated through empathy only BCa CI95% [.01, .33] and partly through the dual mediation of personalization of problem and empathy BCa CI95% [.03, .21]. The mediation model and the indirect effects are presented in Fig 2 and Table 6 respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mediating effect of personalization of problem and empathy on the association between narrative mindset and effectivity of advice–dual-mediation model (the interpersonal problem-solving study).

Note: Presented regression coefficients were partially standardized for the narrative mindset predictor. The remaining were fully standardized. Coding for the independent variable: 0 –control, 1 –experimental. * p < .05. ** p < .01.*** p < .001.

https://doi.org/10.1371/journal.pone.0253729.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. The indirect effects for the dual-mediation model (the interpersonal problem-solving study).

https://doi.org/10.1371/journal.pone.0253729.t006

Participants

This experiment had the same structure, rules of participant recruitment, and incentive as experiment 1. In total, 120 mTurk workers (53 men and 67 women) participated in the study. Independent coders decided that data from 13 participants (8 from the narrative condition) cannot be used because of one or two reasons: (a) possible cheating (b) providing text that was irrelevant to the task, possibly copied from the Internet. Finally, there were 56 participants in the Narrative mindset and 64 in the Control condition. Detailed demographics can be found in Table 7.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Demographics of the participants in the emotional problem-solving study.

https://doi.org/10.1371/journal.pone.0253729.t007

Procedure and materials

After the mind-setting part, participants were invited to the second part of the experiment: the emotional problem task. They were asked to provide advice for four people dealing with particular states with a high emotional component. Problem descriptions were inspired by the works of Appel and Kaestner [2] and Siegel and others [5]. They were:

"A dreams about a career in acting, but worries, because during public speeches, A is always very anxious about others and feels embarrassed.

The husband/wife of B died five years ago in a car accident. Although a lot of time has passed, B is depressed. Almost nothing can put him/her in a good mood, s/he does not see much sense in anything s/he does. B would like to feel like s/he used to.

C is very gullible, which is often exploited by others. Every time s/he was taken advantage of, s/he decided that in the future s/he will be guided by her own interest. Unfortunately, such situations still happen to her/him.

D often explodes with anger and becomes aggressive towards others, even towards loved ones. S/he has the impression that people are starting to avoid him/her. D feels that s/he cannot control his/her bad emotions."

Data analysis and advice evaluation

Coders’ assessment and LIWC analyses were performed in the same manner as in experiment 1. Each pair of coders worked on the same scale in both experiments. The same statistical analyses, as in experiment 1, were used.

Evaluations provided by coders correlated significantly. Effectivity scale r = .94 (p < .001); empathy scale, r = .89 (p < .001) and personalization scale r = .93 (p < .001).

Like in experiment 1, the range of the correlations is similar to correlation in other studies using this measure. Example pieces of advice are presented in Table 3.

The interrater reliability was found to be satisfactory–the level of agreement between the coders was from moderate to substantial. Table 8 presents the computed weighted kappa coefficients.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Interrater reliability of the coders’ evaluations of the pieces of advice—linearly weighted kappa (the emotional problem-solving study).

https://doi.org/10.1371/journal.pone.0253729.t008

Correlations and group comparisons

Similarly to experiment 1, the narrative mindset group was rated higher than the control group across all three scales. The narrative mindset group was judged to be more effective (M = 3.51; SD = .81) than the control group (M = 3.20; SD = .64), t(118) = 2.40; p = .018; d = .42; more empathic (Mdn = 1.88) compared to the control group (Mdn = 1.50), Z = 4.21; p < .001; and more reliant on personalization of problem (Mdn = 1.63) than the control group (Mdn = 1.13), Z = 3.12; p = .002. Additionally, the narrative mindset group (Mdn = 0.90) used significantly more 1^st person singular pronouns in their advices than the control group (Mdn = 0.00), Z = 2.49; p = .013. There were no significant differences in the use of past tense verbs, Z = 1.48; p = .139. The results are presented in Table 9.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Comparison of control vs. narrative mindset conditions on the effectivity of advice, empathy, personalization of problem, and the LIWC measures (the emotional problem-solving study).

https://doi.org/10.1371/journal.pone.0253729.t009

The effectivity of advice, empathy, and personalization of problems, like in experiment 1, all correlated positively. This time, a strong association was between empathy and personalization of problem (rho = .64; p = < .001), while effectivity of advice was moderately associated with empathy (rho = .43; p < .001) and personalization of problem (rho = .35; p < .001).

However, there was a change in correlations when they were tested independently for each of the two conditions. In the control condition, effectivity of advice ceased to correlate with empathy (rho = .16; p = .219) and personalization of problem (rho = .23; p = .067), which is in contrast to the narrative mindset condition, where it correlated strongly with empathy (rho = .53; p < .001) and moderately with personalization of problem (rho = .37; p = .005). The association between empathy and personalization of problem remained strong in both the narrative mindset (rho = .63; p < .001) and control condition (rho = .55; p < .001).

The number of past tense verbs was positively correlated with the personalization of problem and empathy. The associations were mostly moderate and occurred for each of the coders’ ratings separately and for the combined evaluations. The number of 1^st person singular pronouns was positively and weakly associated with personalization of problem and effectivity of advice (for each of the three evaluations) and the 1^st coder’s evaluation of empathy. The results can be found in Table 10.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. Correlations between LIWC measures and the coders’ assessments on each scale in the emotional problem-solving study–Spearman’s coefficient (N = 120).

https://doi.org/10.1371/journal.pone.0253729.t010

Simple mediation analyses

When analyzed using separate simple mediation models, both personalization of problem, BCa CI95% [.02, .25], and empathy, BCa CI95% [.14, .41], mediated significantly the effect of narrative mindset on the effectivity of advice. Each of the models yielded a complete mediation effect, as the direct effects of the narrative mindset on the effectivity of advice were not significant for the model with personalization of problem (β = .27, p = .145) and empathy (β = .07, p = .682) respectively. The mediation models are presented in Fig 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Mediating effect of personalization of problem and empathy on the association between narrative mindset and effectivity of advice–two separate simple mediation models (the emotional problem-solving study).

Note: Presented regression coefficients were partially standardized for the narrative mindset predictor. The remaining were fully standardized. Coding for the independent variable: 0 –control, 1 –experimental. * p < .05. ** p < .01.*** p < .001.

https://doi.org/10.1371/journal.pone.0253729.g003

Dual-mediation model

Inducing a narrative mindset led to a higher personalization of problem (β = .57, p = .002). Both narrative mindset (β = .39, p = .004) and personalization of problem (β = .64, p < .001) positively predicted empathy and there was no indirect effect of narrative mindset on empathy through personalization of problem, BCa CI95% [-.13, .10]. When combined together, empathy positively predicted effectivity of advice (β = .49, p < .001) while narrative mindset (β = .07, p = .677) and personalization of problem (β = -.03, p = .788) were not significant. This means that the effect of the narrative mindset on the effectivity of advice was fully mediated—partly by the indirect effect through empathy only BCa CI95% [.05, .27] and partly by the indirect effect through personalization of problem and then empathy BCa CI95% [.04, .26]. The mediation model and the indirect effects are presented in Fig 4 and Table 11 respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Mediating effect of personalization of problem and empathy on the association between narrative mindset and effectivity of advice–dual-mediation model (the emotional problem-solving study).

Note: Presented regression coefficients were partially standardized for the narrative mindset predictor. The remaining were fully standardized. Coding for the independent variable: 0 –control, 1 –experimental. * p < .05. ** p < .01.*** p < .001.

https://doi.org/10.1371/journal.pone.0253729.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 11. The indirect effects for the dual-mediation model (the emotional problem-solving study).

https://doi.org/10.1371/journal.pone.0253729.t011