English (pdf)
Article in xml format
Article references
Automatic translation
Send this article by e-mail
Cited by SciELO 
Cited by Google 
Similars in
SciELO 
Similars in Google 
Permalink
Romantic jealousy is defined as a set of thoughts, feelings and actions that follow a threat to the existence or quality of a relationship, generated by the perception of a potential romantic attraction between the partner and a real or imaginary rival (Salovey, 1991; White, 1981). Ben-Ze'ev (2013) describes them as a cluster of emotions whose adaptive value is to counteract the breakdown of existing social ties (Panksepp, 2013). Constitutes a sociocultural phenomenon present in varying degrees in all societies (Schmitt, 1988). However, when control is lost, intense and constant jealousy can become pathological (Mathes, 1992). The Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013) classifies jealousy in two diagnostic categories: (a) other obsessive-compulsive disorders and specified related disorders (Obsessive Jealousy) and (b) delusional disorder (Jealous Type).
Romantic jealousy at the pathological level generates high levels of violence (López-Ossorio, González Álvarez, Buquerín-Pascual, García-Rodríguez, & Buela-Casal, 2017; Rodríguez, DiBello, Overup, & Neighbors, 2015; Ureña, Romera, Casas, Viejo, & Ortega-Ruiz, 2015) which may be present since as early into the relationship as the courtship stage (Pazos, Oliva, & Gómez, 2014; Penado-Abilleira & Rodicio-García, 2017). According to the National Institute of Legal Medicine and Forensic Sciences, there were 47,248 cases of intimate partner violence in Colombia during the year 2015. This trend has been stable since 2005, and it indicated that 34.89% of the aforesaid cases were triggered by "jealousy, distrust and infidelity" (Forensis, 2016). In other countries, jealousy has also been the cause of intimate partner violence (Belus et al, 2014; Harris, 2003). Jealousy is also associated with various psychopathologies, including possible comorbidity with suicidal behavior, substance abuse, psychosis and emotional disorders (Costa, Sophia, Sanches, Tavares, & Zilberman, 2015).
The greatest interest in research on romantic jealousy dates back to the mid-1990s (Hart & Legerstee, 2013), despite the fact that specific symposia have been held since 1977 by a panel presentation at the Convention of the American Psychological Association (Salovey, 1991). Interest in this study has been growing. The most studies have focused on corroborating the evolutionary hypothesis of sex differences in emotional vs. sexual jealousy (Buss, Larsen, Westen, & Semmelroth, 1992; Fernández, Vera-Villarroel, Sierra, & Zubeidat, 2007; Sagarin et al., 2012). The characteristics of the rival that incites this emotion have also been studied (Buunk & Dijkstra, 2015), type of attachment (Miller, Denes, Díaz, & Buck, 2014), satisfaction and commitment to the relationship (Dandurand & Lafontaine, 2014), self-esteem (DiBello, Rodríguez, Hadden, & Neighbors, 2015), its association with alcohol abuse problems (Rodríguez, DiBello, & Neighbors, 2015) and the use of social networks in the increase of this emotion (Utz, Muscanell, & Khalid, 2015).
There has also been interest in Latin America in the cross-cultural study of romantic jealousy in the Mexican population (Hupka, Zaleski, Otto, Reidl, & Tarabrina, 1997), in the Chilean population (Fernández, Sierra, Zubeidar, & Vera-Villaroel, 2006) and in the Argentinian population (Bu-unk, Castro, Zurriaga, & Gonzáles, 2011). Sexual differences have been examined in Brazil (Carvalho, & Ambiel, 2016). In Colombia, sexual differences were addressed in relation to infidelity (Portilla, Henao, & Isaza, 2010) and a preliminary intervention study was also carried out (Martínez-León et al., 2016).
In terms of measurement of romantic jealousy, about 40 self-report instruments have been reported since the creation of the Self-Report and Projective Jealousy Scales (Bringle, Roach, Andler, & Evenbeck, 1979). Many instruments were used only once. The most used scales (Martínez-León, Peña, Salazar, García, & Sierra, 2017) are the Infidelity Dilemmas Questionnaire (IDQ; Buss et al., 1992) created to assess gender differences; the Multidimensional Jealousy Scale (Pfeiffer & Wong, 1989) developed to evaluate the three dimensions of jealousy (cognitive, emotional and behavioral dimensions); the Facebook Jealousy Scale (Muise, Christofides, & Desmarais, 2009) aimed at evaluating the probability of an event that causes feelings of jealousy participants related to their activity on Facebook; the Jealousy Scale (Buunk, 1997), used to evaluate three types of jealousy (reactive, anxious and preventive) and the Interpersonal Jealousy Scale (IJS; Mathes, & Severa 1981), which evaluates the negative emotion resulting from actual or threatened loss of a loved person to a rival. All of the instruments listed above have shown adequate levels of validity and reliability.
In the Latin American context, there is an adaptation in Chile of the Inventory of Behaviors of the Couple That Cause Jealousy (Fernández, Pavez, & Dufey, 2014). In Mexico, a Multidimensional Measure of Jealousy (Díaz-Loving, Rivera, & Flores, 1989) has been elaborated and a preliminary study was carried out with the Multidimensional Inventory of Romantic Jealousy (Mota, González-Arratia, Valdez, González, & Hernández, 2016). In Brazil, the Romantic Jealousy Inventory was revised (Bueno & Carvalho, 2012) As for Colombia, there are no psychometric studies of a specific measure on romantic jealousy. In addition, previous studies did not perform Factorial Confirmatory Analysis (CFA) so as to corroborate the factor structure found, and low reliability levels were found in their factors (Bueno & Carvalho, 2012) or small samples were used (Mota et al., 2016).
The IJS developed for Mathes and Severa (1981) is one of the most used scales for evaluating romantic jealousy, they found six factors that evaluate the susceptibility to threats related to: (a) Partner's infidelity and dating others (items 1, 3, 12, 14, 15, 16, 20, 22, 23, and 25); (b) Partner's popularity (items 2, 3, 4, 7, 10, 13, 17, 20, and 23); (c) An untrustworthy partner (items 1, 4, 6, 19, and 21); (d) Partner's old dates (items 11, 24, 26, 27, and 28); (e) Partner's indifference (items 8, 9 and 24); and (f) Partner's remaining friends with old dates and sex differences (items 5 and 18).
There are four studies that have examined in terms of their reliability and validity. Mathes and Severa (1981) present two studies, and the first study performed the construction of the scale with 39 items, which was reduced to 28 items after statistical analysis. Subsequently, in a second study, IJS was applied alongside other scales that evaluated romantic love-links, insecurity, self-esteem and an ad hoc questionnaire of separate identities, confirming, according to theory, that the higher level of jealousy, the higher the level of Romantic love and the lesser the separate identity of the couple. In this study, we report a coefficient of reliability of .92 for the total scale. In a third study, Mathes, Phillips, Skowran and Dick (1982), in order to extend the indications of validity, used the IJS and two behavioral measures: (a) threat when confronted by a rival and (b) expressions of possessiveness towards the partner, through a telephone call that they made to the person who participated in the study. The expected correlations were obtained, which again supported the validity of the scale. Finally, in a fourth study, Mathes, Roter and Joerger (1982) corroborated the convergent validity of IJS with other scales that evaluated romantic jealousy.
In this research, and in order to have external evidences of validity, we used the Self-Rating Jealousy Scale which had a correlation of .82 with the IJS (Hawkins, 1987). In addition, we included questions about the frequency and severity of jealousy displeasures and the person's control thereof, based on the association between jealousy and aggression (Kar & O'Leary, 2013; Llor-Esteban, García-Jiménez, Ruiz-Hernández, & Godoy-Fernández, 2016).
The evaluation of romantic jealousy is of great relevance both at scientific level and at the level of public health; for this reason, valid and reliable instruments are required which should be adapted to the population under study, in this case Colombia. This study sought to adapt to the Spanish language and examine the psychometric properties of the IJS. For this, the factor structure was tested by Factorial Confirmatory Analysis (CFA) and reliability indices were obtained. In addition, the analysis was enriched by the Item Response Theory (IRT), the differential function analysis (DIF) of the items by sex and, obtaining external evidence of validity.
Method
Participants
A sample of 603 people from three cities in Colombia was used (31% professionals, 56% postgraduate and 69% undergraduate students), aged 18 to 45, and obtained by quota sampling in nine private universities, a public university and health centers; 40,8% were men and 59,3% women, with an average age of 26 (SD = 7.51).
Instruments
Interpersonal Jealousy Scale (IJS; Mathes & Severa, 1981). Consists of 28 items answered on a 1-9 Likert scale, wherein 1 = strongly disagree and 9 = strongly agree.
Self-Rating Jealousy Scale (Hawkins, 1987). Self-rating jealousy scale, composed by a single-item: "How jealous do you consider yourself to be?" (5 = Very jealous and 1 = Not jealous).
Ad Hoc Jealousy Items. This questionnaire includes three Likert-type (from 1 to 5/7) questions about romantic jealousy: "How often do you have troubles because of your jealousy?" (7 = Daily and 1 = Never); "What is the severity of these troubles?" (5 = Too severe and 1 = Not severe at all); and "To what extent can you control your jealousy?" (5 = All times and 1 = Not once).
Procedure
Translation into Spanish and localization were done using the technique back translation (Muñiz, Elosua & Hambleton, 2013), with the collaboration of three bilingual professionals. Both the English version and the Spanish version were administered to bilingual students - with an interval of 15 days - and significant correlations were found. Calls were made in the schools of Psychology at different universities in Bogotá, as well as graduate programs in the area of Health Management and Health Centers, in order to recruit participants. Subsequently the main ethical aspects were laid down in terms of ensuring voluntary and informed participation, data confidentiality and the information to be given to the participants in the study. Administration for students was done in groups; professionals were contacted at health centers and several universities.
Statistical Analysis
Three factorial models were tested through CFA: (a) the original six-factor model proposed by Mathes and Severa; (b) the one-dimensional model, taking into account that the original authors had reported that the first factor explained 62.2% of the variance; and (c) a modified, one-dimensional model, based on descriptive statistics, discrimination indexes, Cronbach's alpha if the item was removed, adjustment parameters of each item (under the item response theory -IRT), according to the recommendations for the selection of Items (Lloret-Segura, Ferreres-Traver, Hernández-Baeza, & Tomás-Marco, 2014). We used the X 2 for purposes of evaluating the fit of the models. However, because this is a conservative statistics measurement and it is unlikely to achieve a good fit through this measure (Byrne, 2012), we used a combination of incremental adjustment rates as criteria for the evaluation of all the models proposed. We took into account the X 2 /df ≤ 3 index; GFI (Goodness of Fit Index); CFI (Comparative Fit Index), and the TLI (Tucker-Lewis Index). In general, the values for CFI, GFI and TLI ≤ .90 indicate a proper fit (Bentler, 1992). The Root Mean Square Error of Approximation (RMSEA between .05 and .08) was evaluated with a criterion of .05; a value less than this criterion indicates a good model fit (Byrne & Campbell, 1999) and is highly recommended for its sensitivity in identifying poor model specification (Hu & Bentler, 1999).
In addition, low-level analysis was performed with IRT, which is an excellent complement since both technologies coexist perfectly in the construction and analysis of the tests (Muñiz, 2010). Adjustment parameters were analyzed for each item of the instrument, considering that the item is calibrated when its Infit and Oufit adjustment parameters are between 0.40 and 1.40 logits (Bond & Fox, 2007).
In order to achieve the descriptive analysis, the correlation coefficients, and the obtaining of the reliability index, the SPSS program (V.22) was used. In addition, Mplus (v. 7.31) were employed to evaluate the construct validity through a confirmatory factor analysis. Moreover, Rasch model was applied using the program Winsteps (v. 3.69.1.13).
Results
CFA was performed in order to establish construct validity. First, the six-factor model proposed by Mathes and Severa (1981) was tested, but some indicators were not adequate. We then tested a unidimensional model including all the items, which also did not show a good fit. Therefore, the IRT analysis for the 28 items was performed, evaluating the Infit and Outfit parameters, which indicated that the items 5, 6, 9, 11, 15 y 16 do not conform to the model, since their Outfit values (1.56, 1.62, 1.45, 1.55, 2.36 y 1.87) were outside the accepted range (Bond & Fox, 2007). Subsequently, corrected item-test correlations were performed. The previous six items, together with items 12, 18, 19 and 21, obtained the lowest correlations with values under .20. Taking into account the analyses above, these ten items were debugged. Table 1 presents the indices of goodness of fit of each model. The one-dimensional model with 18 items showed an adequate fit (see Appendix).
Figure 1 shows the one-dimensional structure of the IJS. Standardized loadings ranged from .40 (item 28) to .70 (item 7).
Figure 1 One-dimensional confirmatory factor analysis (CFA) chart of the Spanish version of the IJS.
Item Analysis and Reliability
Table 2 shows the descriptive statistical elements (mean, standard deviation, skewness and kurtosis), as well as the corrected item correlation and Cronbach's alpha of the 18 items analyzed. All of them were within the accepted range in asymmetry and kurtosis. The corrected item-total correlations exceeded the value of .30 in all cases. Likewise, the elimination of some items did not increase the reliability index. The reliability for the full scale was .90.
Limitation
The sample of participants was a convenience sample of adults contacted in the streets of the main town of the country, and this sample was of moderate size. The present study was not epidemiological in character; that is, it was not intended to estimate precisely the proportion of people expressing each of the positions that were found. Such studies would require very large, representative sets of participants. The present study was instead a psychological study; its aim was to delineate the way in which participants utilized the information provided in realistic scenarios. For such studies, community samples of participants are sufficient because the different possible positions are limited in number and rough estimates of the percentage of participants who endorse each position are usually sufficient. The main interest of the study was to demonstrate that, in addition to total agreement or total rejection of forgiveness, alternative, more complex and more circumstances-determined positions can exist among Colombians.
Psychometric Analysis of the Scale Under IRT and DIF by Sex
The Infit / Outfit adjustment data showed that all items in the Spanish version of the IJS were calibrated (Infit ranged from 0.79 to 1.25 and Outfit from 0.78 to 1.44). Analysis of the differential functioning (DIF) of the items by sex indicated significant differences in the items 2 (x 2 = 1.26, p <.026), 4 fe2 = 1.02, p <.031), 13 (x 2 = 1.39, p <.023), 23 (x 2 = 4. 27, p <.038), 25 (x 2 = 1.64, p <.019) y 27 (x 2 = 1.17, p <.0278) Therefore, it can be affirmed that these items measure the latent trait differently in men and women.
Evidence of Validity
As expected, the correlations between total and factor scores with the Ad Hoc Jealousy Items and Self-Ratig Jealousy Scale were statistically significant (Table 3).
Discussion
This study provides the first Spanish version and psychometric analysis of the Interpersonal Jealousy Scale, using EFA and CFA. The results obtained provide a one-dimensional instrument that evaluates romantic jealousy as a negative emotion resulting from an actual or threatened loss of the person loved to a rival. This version made up of 18 of the original 28 items showed an adequate construct validity, good reliability (α = .90) and evidence of convergent validity. Six of its items measure latent trait differently in men and women.
The original version by Mathes and Severa (1981), made up of six dimensions, did not show a good fit, possibly because three of its factors (3, 5 and 6) consisted only of two and three items (Lloret-Segura et al., 2014). And in addition, seven of its factors (5, 6, 11, 16, 18, 19 and 21) had a negative syntax in their wording, which could increase their difficulty (Moreno, Martínez & Muñiz, 2004). The one-dimensional model - proposed because the first dimension of the original model already explained 62.2% of the variance - and the factors whose composition exceeded the three items (two and four) explained 12.9% and 7.2% of the variance respectively, was adjusted when the items with negative syntax were eliminated. However, this did not have a good fit in the IRT or in the corrected item-total correlations, which were also part of the factors with a very low number of items.
It is also noteworthy that some of the items deleted, such as item 5 "When___likes some of my friends, I am pleased" and item 11 "I want____to continue to maintain the friendship with the people with whom he used to go out" alluded to the subject of friendship, which is probably not generating romantic jealousy, unlike questions that indicate communication with ex-partners (Knox, Zusman, Mabon, & Shriver, 1999). Items such as 12 and 15, which had no negative syntax but which included very general statements such as "I would like____to be faithful to me", to which most people responded positively, failed to discriminate the construct being measured. Thus, the IJS, composed of its 18 items, evaluates jealousy in a single dimension understood as sensitivity to the threats of the rival a concept according to the essence of the construct (Salovey, 1991).
The analysis of differential functioning (DIF) of the items by sex indicated that six of their items measure romantic jealousy differently according to men and women, all of them refer to the threat to the relationship in a more Emotional type, which is in line with the evolutionary theory of jealousy (Bendixen, Kennair, & Buss, 2015; Buss et al., 1992). However, it is suggested to use the same two types of scenarios of possible infidelity (sexual and emotional), given the multiple studies that confirm the importance of the two scenarios.
As to external evidence of validity, positive and significant relationships were found - as expected - which support the relationships already demonstrated by Mathes (1992). The correlation of the results of the total score of the IJS with the Self-Rating Jealousy Scale was equally significant, as in the case of the evidence obtained by Hawkins (1987). The relationship between the score of the IJS and the number and severity of arguments due to jealousy - and the control the person has over their jealousy - are consistent with the indications given by Dutton, Ginkel and Landolt (1996). The latter demonstrated a significant relationship between the IJS and self -reported scores of anger. All the evidence above supports the findings of DeSteno, Valdesolo & Barlett (2006), who reported a relationship between jealousy and aggressive behavior. However, it is advisable to make comparisons with a similar instrument so as to ensure concurrent validity and determine overall construct validity (Messick, 1994). These studies will allot to reject or confirm the common variation of the items that make up the test with stronger supporting foundations.
Some limitations of the study should be noted. It is necessary to take into account the fact that most of the subjects were in early and middle adulthood, were highly educated, which does not make it possible to generalize the results to the entire adult population or to other sexual orientations; This required, as Hawkins (1987) points out, revise the wording of several of the items.
The results of this study provide opportunities for future lines of research, including the generalization of the IJS' one-dimensional structure to other samples and cultures, which has not been done as yet. It would be interesting to examine the cultural equivalence of the scale, both in collectivist and individualist cultures. Only one adaptation of the IJS in Hungary is currently reported (Orosz, Zoltán, Kiss, Farkas & Roland-Lévy 2015). Moreover, further research should include several questions related to the use of social networks such as Facebook or Snapchat, given the current importance of these communication media and the evocation of romantic jealousy (Dainton & Stokes, 2015; Utz et al, 2015) or to use parallel scales that evaluate them. In addition, it is necessary to confirm the relationships found in non-Colombian population with personal, interpersonal and sociocultural variables (Martínez-León et al., 2017) thereby continuing with the study of this important and complex human emotion
