Structured personal conflict as a main determinant of intragroup conflict: a study in Mexico

Managers can spot and target the common driving forces of conflicting employees to ensure their efficiency and productivity. The aim of this research is to provide evidence for the existence of a personal profile, named in this work Structured Personal Conflict, which is present in intragroup conflicts, traditionally studied based on the factoring of conflicts of processes, results and relationships. More than 400 questionnaires were distributed by mail among workers from different organizations and jobs in Mexico, the methodology used the information of 201 usable questionnaires. Various goodness-of-fit tests obtained were applied and rival models were compared using Confirmatory Factor Analysis (AFC). As a result, it was found that structured personal conflict can be estimated from the intragroup conflicts of processes, results and relationships, and that it has a greater explanatory power than either of them individually. The importance of structured personal conflict in the personnel selection and in the performance of teams should be emphasized.


INTRODUCTION
According to some authors such as Greer  , intragroup conflict is a process that begins when an individual or group perceives differences and opposition between himself and others about incompatibilities of interests, resources, beliefs, values or practices. Pondy (1992) and Jehn (1994Jehn ( , 1997 identified the categories for the intragroup conflict: processes, tasks and relationships; De Witt et al. (2013) proposed considering individual perceptions in future research, since this variable could affect the collective phenomenon. Ayub et al. (2017) establish that the Big Five theory is one of the successful theories in predicting behavior at work. The authors identify that personality traits, during the conflict management process, determine performance.
This study contributes to the discussion about individual perceptions by demonstrating the existence of a conflict named Structured personal conflict, which can be distinguished from exogenous conflicts to the person. The study seeks to provide more evidence of how personal conflict stands out over any other type of conflict at work and requires consideration to define how to handle it and thus avoid group failure.

LITERATURE REVIEW
Some studies indicate that members of a group distinguish their intragroup conflicts within three categories: i) with internal processes (De Witt et al., 2012;Vodosek, 2007), however, different opinions may even be presented among the members regarding the severity of each of them (Amason, 1996 (2013) defined intragroup conflict as a process that begins when an individual or group perceives differences and opposition between himself and others about interests, resources, beliefs, values or practices. As an example, Pondy (1992) expresses it as a dynamic process in which at least one of the members of the group perceives, feels or behaves against another and defines the conflict as potentially functional, and even neutral, as well as possibly negative. In the same line, the classification by Pondy (1992) and later Jehn (1994Jehn ( , 1997 identifies three categories for the intragroup conflict: processes, tasks and relationships. And it has been used by many researchers ( De Witt et al. (2013) propose that conflict makes group members more likely to be rigid and to process information partially, using their own information during decision-making, rather than sharing the information with the group. Following this line, Smith (2014) proposes that the presence and persistence of antecedents of conflict lead to a higher level of intragroup conflict, and this level of perceived conflict increases negative behaviors and reduces positive behaviors. 1 Other Some studies indicate that group members perceive conflict in a similar way, but that they generally neglect potentially important differences between each of its members (Amason, 1996 When the conflict is already a threat, the members of a group seek to confirm their information, as well as the context for their decision-making (Fischer et al., 2011). Mikkelsen and Clegg (2017) add that conflict is an element of dissonance in the order, which can have positive and negative effects. McCarter et al. (2018), in contrast to classifying the conflict as good or bad, focus on identifying five models of intragroup conflict: diversity, behavioral negotiation, social dilemma, social exchange and economy of transaction costs. In particular, they highlight that the models of conflict of diversity and behavioral negotiation (lab) are normally used to analyze conflicts within groups made up of people and have been studied mainly at the individual level within the group, the most widely used methodology being surveys and data from archive 1 . Jehn and Chatman (2000) suggest that the asymmetry of the personality traits of the members of a group is more important than the conflict itself. This asymmetry of personality traits determines at least the experience of the conflict, this means perceiving the conflict differently, that is, a problem can be perceived as more or less conflictive by the members of a work group. Jehn et al. (2010) include in their studies the individual perception, as well as the attitudes of each of the individuals with the level of conflict, and identified that conflictive people tend to generate more complicated work environments, with less trust, respect, and commitment. In their research, they have found that better attitudes towards the group tend to improve individual satisfaction with it. According to the Dual concert theory, the CMS is a function that involves a person in himself and others. From this idea four styles of CMS can be understood: forcing, which implies a great personal concern and little for the other person, so it could be cheating, persuasive in excess, great concern for himself and another person, in this case he would be a problem solver looking for satisfactory solutions for both parties, taking great care of others and little of himself would be yielding, little concern for himself and others, he would tend to avoid conflict, for example, by reducing the importance of the subject, and the fifth is the compromise, intermediate concern for himself and another person.
Big Five Theory (Norman, 1963) identifies five personality traits that influence team behavior, namely, openness to experience, responsibility, extroversion, kindness, and neuroticism.
According to Barrick    members less open to ideas other members might have about task-related ideas. Relationship conflict develops in terms of misattribution, harsh and emotional language in arguments, or hurtful and aggressive tactics by some group members to convince others of their views (Jehn, 1997).
The aim of this study is to demonstrate the existence of a different element named Structured personal conflict. It is an individual and personal profile, endogenous, latent to the person, and can be distinguished from traditionally studied conflicts of process, task and relationships, which allows a better explanation of intragroup conflict. It is significant to underline that inter-construct correlations higher than 0.7 indicate the existence of some degree of multicollinearity that might suggest the elimination of some of the typical constructs for any multivariate analysis (Gujarati, 1993). This study is aimed at exploring the presence of a structured personal conflict within intragroup conflicts, based on the following hypotheses:

Research hypotheses
H1: The component of structured personal conflict explains a higher percentage of total variance than the component conflict of net processes, net results and net relationships individually.
H2: The components of structured personal conflict, net processes, net results and net relationships build a robust model to explain the phenomenon of intra-group conflict. Although personality significantly helps determine interpersonal interactions (Barrick et al., 2005), personality traits often fail to find a meaningful place in empirical studies (Hogan & Kaiser, 2005), so their research, unlike others on the sub-ject, focus on the exogenous perception of the person around the conflict of processes, results and relationships, or are justified from the findings that serve to determine how their members work together (Amason & Schweiger, 1994;Jehn, 1995).
The research seeks to contribute to the literature on an endogenous personal profile of conflict that affects the team.

METHOD
The method consists of five stages: 1) Constructing a research instrument in Spanish. 2) Sample.

Research instrument
Following the recommendation of Bostwick and Kyte (2005), the instrument begins with a series of neutral questions to locate respondents within the topic. A first open question was to find out the main conflict that the team has at that time. In the second instance, the subjects were asked: a) From which of the following three elements (processes, results or relationships) do you think this conflict began? b) If you were responsible for solving that conflict, from which of the following three elements (processes, results or relationships) would you start its correction?
To measure the constructs, a series of 18 statements was used (six items per construct). A 5-point Likert scale was used (where 1 means never and 5 always).
To measure the conflict construct with the processes, the items were adapted from Jehn and Mannix (2001)

Internal reliability test results
The three constructs tested in the sample resulted in adequate internal reliability, with Pearson item-total correlations between 0.638 and 0.840, and a Cronbach's alpha of 0.8417 to 0.9001.
The Sampling Adequacy Test gives a KMO value of 0.944 and Bartlett's Sphericity Test a Chi-square of 2037,117 (P-val < 0.001), which indicates that the matrix of 18 items may be subject to factorization.
Exploratory factor analysis confirms that with only three factors, an extracted variance of 62.8% can be obtained (Table 1) Applying a VARIMAX rotation, it is achieved that all the items turn out to be significant with individual loads greater than 0.4 within their factor, recommended by Hair et al. (1998). The exception is Item 12, which can be left aside, as the load is slightly below the 0.40 limit. However, it was decided to maintain it, since Cronbach's alpha does not improve substantially with its omission, and the structure of six items per construct would be broken. Note: The rotation is Varimax of the factor loads.

External reliability
The discriminant validity of Fornell and Larcker (1981) was adequate as the square roots of the average variance extracted of each factor were superior to the pairwise correlations against the other factors.

Confirmatory Factor Analysis (CFA)
When applying a CFA to the sample information, to check the grouping of the 18 items within their corresponding theoretical factors, and using the LISREL 10.1 (64 Bit) system with Maximum Likelihood estimators, it was found that the indicators of goodness-of-fit tests were marginally acceptable, according to the following results (Hair et al., 1998, p. 687): 1) All the coefficients of the items individually are relevant and significant to their theoretical construct (R 2 > 0.25 and a P-val < 0.001).
2) The mean square error of approximation RMSEA = 0.0825 with a 90% CI of (0.0707; 0.0943) is marginally acceptable (what is applicable is an index between 0.05 and 0.08).
4) The standardized fit index NFI = 0.883 is marginally acceptable (what is applicable is NFI > 0.9).

Sample data descriptive
The composition of the sample (201 generated) was as follows: 122 men (60.7%), 77 women (38.3%), and two did not declare; the average age was 34.8 years (SD of 11.0); 39 (

Correlations matrix
Direct correlations between outcome conflict and process conflict is r = 0.680 (PV < 0.05) and relationship conflict and process conflict is r = 0.682 3 According to Nieves and Dominguez (2010), a strong positive and significant correlation is r > 0.7 and P-val < 0.01. 4 It is used to form a smaller number of uncorrelated variables from a large data set. The objective of principal component analysis is to explain the maximum amount of variation with the fewest components. This analysis is commonly used in the social sciences when using large data sets. It is commonly used as a step in a series of analyses to reduce the number of variables and avoid multicollinearity (MINITAB Note).
(PV < 0.05); the correlation between outcome conflict and relationship conflict is r = 0.705 (P-val > 0.05) 3 . This presupposes a problem of multicollinearity. Table 2 presents the respective inter-construct correlations.

Test hypothesis 1
In particular, to address this situation, Gujarati (1993) suggests applying the latent variable -factorization process. For this reason, this investigation proposes to explore the existence of the fourth construct that can be created from the information of the three traditional theoretical constructs.
To test hypothesis 1, the Principal Component Analysis (PCA) of the LISREL 10.1 System (64 Bit) methodology is used 4 . Table 3 presents the first six latent components constructed based on the information from the 201 useful questionnaires.
To illustrate the theoretical composition of the first three main components, Figure 2 shows the average loads of items per construct with reference to their theoretical origins.
The first component (PC1) is constructed from the items of the process conflict (with an average of loads of 0.23), results (with 0.22) and relationships (with 0.25). This practically uniform composition verifies the existence of a component (endogenous) that explains the high correlation observed in the correlation matrix. This PC1 component, for the present investigation, will be called Structured Personal Conflict (SPC), as an extension of the works that explain that personality differences can be a major reason for conflict, as well as the perception of conflict and preference for managing that conflict. Note: The diagonal of each factor presents the square root of the AVE. ** -significant correlations at the 0.01 level.
The second component is mainly loaded with items related to the process conflict (with 0.27), which is attenuated by the items related to the relationship conflict (with 0.27 but with a negative sign). The third component is mainly loaded by the items related to the outcome conflict (with 0.32), which is attenuated by the items related to the other two types of conflict. Table 3 shows that the first component (PC1) reaches an Eigen value of 9.73 (DS = 0.94), which represents by itself 54.08% of the total variance. This value is significantly removed from the Eigen values of the other constructs. In the absence of an intersection between the confidence interval of the Eigen value of component 1 and the sum of the Eigen values of the following three, hypothesis 1 is approved.

Test hypothesis 2
To test hypothesis 2, it is proposed to use Confirmatory Factor Analysis (CFA) to evaluate four rival models composed of different quantification options of the Structured Personal Conflict (SPC) construct.
• Model 2: SPC = weighted average of 18 items based on the ACP coefficients for the first component PC1.
• Model 3: SPC = value of the item with the lowest score.
• Model 4: SPC = value of the first quartile of 18 items.
To quantify the Net conflict process, Net conflict task and Net conflict relationship constructs, the average of the respective items will be used, subtracting from each of them the value of the structured personal conflict (SPC), depending on the model.
In accordance with the recommendations of Hair et al. (1998), to locate the best model, the various goodness-of-fit tests obtained from the Confirmatory Factor Analysis (CFA) are com-pared through the LISREL 10.1 system (64 Bit) with estimators of Maximum Likelihood. The results of the goodness-of-fit tests of the four models are presented in Table 4.
Model 3 is the one that obtains the best scores in the goodness-of-fit tests, compared to rival models. It only requires the R 2 coefficient issue to be resolved.
As  construct. The CFA for Model 3, now limited to 15 items, presents the following improvement of goodness-of-fit indicators: 1) All the R 2 of the coefficients are greater than 0.25, therefore they are relevant.
2) All the P-values of the coefficients are less than 0.001, therefore they are significant.
3) The Conditional number 4,547 indicates that there is no multicollinearity.
5) The goodness of fit index GFI = 0.902 is adequate.
This model shows robust results, which is proven by Hypothesis 2. Figure 3 presents the relationship diagram (Path Diagrams) of the proposed model.

DESCRIPTIVE VALUES OF THE NEW CONSTRUCTS
Given model 3, Table 5 presents the descriptive statistics of the sample based on the 15-item instrument. The inter-construct correlations are presented in Table 6. Now, the correlations obtained have less interdependence than those identified under the traditional model (see Table 2). It is worth noting the negative signs of the correlations between the constructs of conflict in processes, task and relationships, against the structural personal conflict, which suggests that, the greater the personal structural conflict, the less margin to identify any other type of specific conflict, and vice versa. Another significant finding is the degree of independence found between personal structural conflict and conflict with relationships with others.

DISCUSSION
Personality traits remain a subject of study and have found significant support in empirical studies (Hogan & Sherman, 2020 A first finding, based on the sample data, was the important correlation by pairs, positive (0.65 > r > 0.70) and significant (P-val < 0.01), between the constructs of results and relationship processes. This result is similar to the meta-analysis on 116 articles prepared by De Witt et al. (2012), except for the correlation between the conflict of results and the conflict of relationship, which in this study with the Mexican sample was substantially higher than that of the meta-analysis (0.705 > 0.52, P-val > 0.05). This indicates an important component of multicollinearity. These correlations greater than 0.7 between pairs of regresses are a symptom of multicollinearity. Gujarati (1993) suggests, as a remedial measure, to omit some constructs that would go against the theories reviewed in the literature. A second remedial measure is to apply a latent variable factorization process.
From a Confirmatory Factor Analysis, four rival models were analyzed to quantify the structured personal conflict construct (SPC). The model that presented the best goodness-of-fit statistics was the one to quantify the structured personal conflict construct (SPC), based on the item with the lowest score, out of the 15 used. This model has the advantage of offering a series of constructs to measure intragroup conflicts with lower inter-construct correlations than is presented in the meta-analysis of 116 articles prepared by De Witt et al. (2012), recommended situation for multivariable analysis models.