This study aimed to test the underlying factors, provide evidence about its internal consistency, and analyze the fit and construct validity of the BAS in a sample of Spanish caregivers of people diagnosed with BPD.
Structural Validity Of The Model Obtained For The Bas
Pioneering work by Horwitz and Reinhard (1992) and Reinhard et al. (1994) obtained five-factor models for the BAS. As indicated above, these studies failed to meet some of the basic requirements for exploratory factor studies, such as not accepting factors with less than three items or not including the same item in two or more factors. In our opinion, the lack of methodological rigor in these studies suggested the need for a more rigorous analysis of the structure of this scale. Subsequent studies obtained two-factor (Guada et al., 2011), three-factor (Ivarsson et al., 2004; Kwak et al., 2013), four-factor (Hunger et al., 2016; Murdoch et al., 2014), and five-factor (Aydemir et al., 2011) models for the BAS.
We obtained a three-factor model for the BAS (Disrupted Activities; Worry, Guilt, and Being Overwhelmed; and Personal and Social Dysfunction) using exploratory procedures, as in Ivarsson et al. (2004) (Activity Limitation; Worry and Guilt; and Social Strain) and Kwak et al. (2013) (Activity Limitation; Social Strain; and Feelings of Worry and Guilt), with a similar distribution of items per factor, but small differences: in contrast to the studies by Ivarsson et al. (2011) and Kwak et al. (2013), in our study, item 1 was not included in the disrupted or limited activities factor, item 19 was not included in the social factor, and item 18 was not included in the worry-and-guilt factor.
It is noteworthy that: (1) the mean score on item 1 (M = 1.92) was low, although it was not the lowest (the mean for Item 10 was 1.51, and this item was included in Factor 2); (2) the highest mean score was on item 18 (M = 3.70), close to the maximum of the scale; and (3) the mean for item 19 (M = 2.70) was comparable to that of other BAS items that were included in any factor (e.g. items 16 and 17, among others). With regard to item 1 (Financial problems), its non-inclusion could be due to the fact that financial problems, even if they exist, are not significantly related to caring for a relative with an SMD (e.g. expenses for psychiatric medication, psychotherapy costs, help from external caregivers, among others). Regarding items 18 (Worry about future) and 19 (Stigma upsetting), it is possible that in the past 6 months the family caregivers have not felt a significant amount of worry about the future or upset due to the stigma of having a relative diagnosed with an SMD (it should be noted that the items on the BAS are responded to in relation to the statement: “Please, would you tell me to what extent you have had any of the following experiences in the past 6 months?”). It would be interesting to investigate how family caregivers interpret worrying about the future: is it concern about possible economic hardship, the evolution of their relative diagnosed with BPD, or the future understood in a vague and diffuse way? Likewise, it would be interesting to find out whether family caregivers are aware of the meaning of the term “stigma” and, therefore, respond appropriately to that item. It must be noted that, in the present study, the sample was exclusively composed of family caregivers of people diagnosed with BPD, unlike previous studies that used samples composed of people with different diagnoses of SMD. The different composition of the samples used in these studies may have led to the differences in the results of the BAS structural analyses.
Previous studies that analyzed the structure of the BAS used Principal Component Analysis (PCA) with Varimax rotation, with the exception of Hunger et al. (2016), who used CFA. Both PCA and Varimax assume uncorrelated factors (e.g. Jolliffle & Cadima, 2016). We used an EFA, specifically the Weighted Least Squares extraction method with Oblimin rotation method (Cfr. Gaskin & Happell, 2014), because we assumed that the factors underlying the BAS items were correlated (like Hunger et al., 2016). The assumption that factors are not correlated seems unlikely in the case of psychological variables, such as those measured by the BAS. Objective and subjective burden are correlated aspects or facets of the same burden experience. Therefore, it seems more appropriate to assume that the BAS factors are correlated and use an oblique rotation method (such as Oblimin) rather than an orthogonal method (such as Varimax) in the EFA (e.g. Gaskin & Happell, 2014).
We tested the model obtained for the BAS using confirmatory procedures in Subsample 2. Results showed the goodness of this model. Only Hunger et al. (2016) tested the BAS structure using confirmatory procedures, although in reality these authors analyzed the models proposed by Reinhard et al. (1994), but introducing a new parameter, i.e. correlations between the factors.
In conclusion, the present study offers a cross-analysis, both exploratory and confirmatory, of the BAS, and it proposes a reduced three-factor model with 16 items that shows good structural properties.
Internal Consistency Of The Model Obtained For The Bas
The model for the BAS obtained in this study showed good internal consistency, with estimations between .85 and .91 for the factors and .92 for the whole scale, which are similar to those obtained in previous studies that found Cronbach’s alphas between .89 (Aydemir et al., 2011; Reinhard et al., 1994) and .92 (Hunger et al., 2016) for the whole BAS. One exception is the study by Hunger et al. (2016), who found a Cronbach’s alpha of .64 for the Guilt subscale (which contained the items, 12, 13, and 17; these items were included in the Worry, Guilt, and Being Overwhelmed factor obtained in our study), an alpha of .74 for the Time Perspective subscale, and an alpha of .78 for the Personal Distress subscale.
Construct Validity Of The Model Obtained For The Bas
As expected, the factors in the model for the BAS obtained in this study correlated negatively with quality of life and positively with anxiety, depression, and stress. These results are comparable to those obtained in the study by Kwak et al. (2013), and support the construct validity of that scale.
Clinical Implications
The BAS can be a useful instrument for clinicians, who can assess burden in family caregivers of people diagnosed with BPD (or another SMD) and improve the efficiency of programs designed to provide resources and develop skills to manage the symptoms of the burden of caring, such as Family Connections (e.g. Fernández-Felipe et al., 2020, 2021; Hoffman et al., 2005), and positively influence their mental health and personal well-being (e.g. Akbari et al., 2018; Durmaz & Okanli, 2014). Scores on the BAS before and after the program can be a valid and reliable indicator of change in the family caregivers of people with a diagnosis of SMD.
Limitations Of This Study And Suggestions For Future Studies
Some limitations of the present study should be mentioned. Regarding the sample, it would be desirable to confirm the structure obtained for the BAS in a larger sample than the one we used. It would also be interesting to analyze the invariance of this scale with regard to sex and other variables of clinical interest. For example, in our study, most of family caregivers of people diagnosed with BPD were biological mothers, and as previous studies have found, this population could be more vulnerable to feelings of burden than other caregivers and, therefore, need more support (Jørgensen et al., 2021). It would be interesting to test the BAS invariance between groups of caregivers.
It would be interesting to have repeated measurements in a longitudinal design, in order to analyze the test-retest reliability of the model obtained for the BAS, and to confirm the construct validity of this scale using scales other than the ones used in our study.
It would be useful to take into account the psychological health of family caregivers of people diagnosed with SMD (e.g. Sansone & Sansone, 2009), and to know whether they are receiving pharmacological treatment and/or psychotherapy.
Future studies should consider the socio-economic status of the family caregivers, in order to identify factors that might facilitate or hinder the care of a relative diagnosed with an SMD, and assess the possibility of drafting statements that are more in line with the content of each item on the BAS.
It would be interesting to analyze the psychometric properties of the BAS model obtained in our study in family caregivers of people diagnosed with an SMD other than BPD, chronic disabilities or degenerative diseases, among others.