Abstract
The current study examines some Boundary Questionnaire 18 (BQ18) psychometric properties, a shorter version of the Boundary Questionnaire (BQ). The current study examines the BQ18 through factor analysis to understand more about the structure of the inventory. At this point, the BQ18 has not been examined in depth. Further, the authors suggest a shorter version of the BQ18 termed "BQ13" with stronger reliability. The current study also examines the utilization of the "BQ13" as a unidimensional scale. The authors discuss the positive outcomes of having a new scale to measure mental boundaries that are more up-to-date and can be utilized not only in research settings but clinical settings as well.
Introduction
The idea that boundaries may be firm, thick, or impermeable and more fluid, thin, or permeable is a well-known concept (Federn, 1953; Hartmann, 1991; Landis, 1970). Hartmann (1991) described the distinction between thick and thin boundaries as the difference between two processes or functions in the mind, with thickness and thinness existing on a continuum. At the thin end of the continuum, there is an increased connection of functions in the mind. At the thick end, there is more separation between functions in the mind.
An individual with very thin boundaries tends to be anything but rigid; these individuals may be involved in many tasks, struggling to focus on one at a time. This person has trouble differentiating between thoughts and feelings. This individual will probably have a high fantasy life, sometimes unable to separate fantasy from reality. In general, the thin-boundaried have a less solid sense of personal space, a tendency to merge with a partner or lose oneself in a relationship, and fewer psychological defenses (Hartmann, 1991, 1997).
An individual with very thick boundaries can typically easily focus on one thing at a time and keep thoughts and feelings separate. These individuals’ mental states are usually straightforward, for example, regarding the distinction between fantasy and reality; additionally, they commonly possess a clear sense of the separation between past, present, and future. An individual with thick boundaries typically also has a strong physical boundary (a definite sense of space around himself) and a clear group identity that they can bond with and have connections with others. Such individuals also usually have a clear autonomous sense of self, never losing themselves in a relationship; they tend to see the world in terms of black vs. white, with no in-between (Hartmann, 1991, 1997; Hartmann et al., 1998).
Thinner and thicker boundaries have been linked to different mental illnesses and negative psychological well-being. For example, Hartmann (1991) found that individuals diagnosed with schizotypal personality disorder or borderline personality disorder have thinner mental boundaries than the overall population, while persons diagnosed with obsessive-compulsive disorder score thicker than average.
There has also been evidence of correlations between mental boundaries and other personality traits. For example, Mccrae (1994) found a very high correlation (r = .73) between boundaries and openness to experience, wherein those with thinner boundaries were more likely to be more open.
The current study is part of a more extensive study on mental boundaries and the relationship between social support and self-esteem. The researchers were able to use the collected data from the larger study to perform a factor analysis on the BQ18 to identify the psychometric structure of the scale. By doing so, the researchers were able to identify a potential new scale, the "BQ13," which had better reliability and is shorter. The current study explores ideas and implications for developing a new boundary scale with fewer questions and is more reliable for measuring mental boundaries. The findings of this current study will push forth the development of a new boundary scale or revised boundary scale for researchers and clinicians. The researchers will discuss their findings in depth.
The Boundary Questionnaire
The Boundary Questionnaire (BQ) is a self-reported inventory designed to measure thick and thin mental boundaries. At the time of its creation, it was the first questionnaire that did not measure mental boundaries through projective measures (Fisher & Cleveland, 1968; Landis, 1970). The BQ is a 138-item inventory addressing 12 different content areas as follows: (1) sleep dreams wakefulness; (2) unusual experiences; (3) thoughts, feelings, and moods; (4) impressions of one's childhood, adolescence, and adulthood; (5) interpersonal distance and openness and closeness; (6) sensitivity; (7) preference for neatness and precision; (8) preference for clear edges, lines, and clothing; (9) opinions about differences between children and adults; (10) opinions about organizational lines of authority; (11) opinions about boundaries among groups, peoples, and nations; and (12) opinions about the identities among beauty, truth, and other abstract concepts (Harrison et al., 2006; Hartmann, 1991).
Participants are instructed to answer each question on a 5-point Likert scale ranging from 0 ("not at all true of me") to 4 ("definitely true of me"). However, the measure has limitations by its vague scoring instructions and lack of clear cut-off points for clinical use. Harrison et al. (2006) found the BQ has high reliability (α = .925); good construct validity, showing meaningful differences between theoretically thick and thin boundary groups of participants; and discriminant validity, showing significant correlations with some MMPI scales of psychopathology but not all. After review and feedback from participants taking the survey, expressed some questions to be unclear and resulted in the removal of seven questions leaving the BQ with 138 questions and increasing the alpha reliability to .930 (Harrison et al., 2006).
The Boundary Questionnaire 18
The Boundary Questionnaire 18 (BQ18) is a short form of the BQ with 18 items selected based on face validity, high correlation with total scores on the BQ, and distribution across the inventories’ categories (Kunzendorf et al., 1997). The BQ18 is scored by inverting the rating scales on questions 5–7 and 16 and then adding the ratings for all 18 items (Kunzendorf et al., 1997). The BQ was very highly positively correlated with the totals on the BQ18 (r = .87). Each item of the 18 selected for the BQ18 had a mean correlation of (r = .36, SD = .09). The BQ18's scoring for clinical purposes is more transparent than the full BQ (Aumann et al., 2012). There are no fixed cut-off points for the BQ18. However, Kunzendorf et al. (1997) suggested that thick boundaries score less than or equal to 29, medium boundaries (a combination of thick and thin boundary characteristics) between 30 and 42. Thin boundaries score greater than 42 (on a total scale of 0 to 72). However, as is the case with the BQ, the BQ18 does not specify whether the concept of boundaries is unidimensional or if boundaries might comprise two dimensions, namely thick and thin. The extant research on the measurement of boundaries has not, to date, addressed this important issue. Thus, to facilitate a better understanding of the construct, this study conducted an exploratory factor analysis of the BQ18. This closer and more critical examination of the measurement and thus conceptualizing the construct will advance the literature on boundaries and help future researchers more definitively understand what is being measured.
Methods
Participants
The sample consisted of 295 (98 males, 194 females, 3 "other") students enrolled in undergraduate courses at a medium-sized university in the United States. The average age of the sample was 20.34 (SD = 3.18). The sample was predominate of White, non-Hispanic origin (n = 251, 85.1%). The remainder of the participants identified as Asian/Pacific Islander (n = 18, 6.1%), Black/African American (n = 14, 4.7%), Hispanic/Latino (n = 9, 3.1%), and Another Race/Ethnicity (n = 3, 1%).
Procedure
Participants were recruited via email, with a link to Qualtrics to fill out a survey on Mental Boundaries. After providing informed consent, participants completed an anonymous online Qualtrics survey at their leisure. There was no time limit for questionnaire completion and no exclusionary criteria imposed. Data from all participants was used. SPSS 26 for Windows was used for all statistical analyses.
Statistical Approach
Coefficient alpha was used to examine the internal consistency of scales. A maximum likelihood factor analysis was used, given its usefulness in identifying latent factor structures (Costello & Osborne, 2005). A Scree plot was used to confirm the number of factors to be extracted in addition to using Eigenvalues. We used Varimax rotation when extracting factors in an attempt to increase the interpretability of the extracted factors.
Results
Preliminary Analyses
Coefficient alpha reliabilities were run for both the BQ18 and "BQ13". For the current study, the BQ18 yielded a coefficient alpha of .69, and the "BQ13" yielded a coefficient alpha level of .77. Scales in this study roughly met the conventional cut-offs for acceptable reliability. However, the reliability of the BQ18 was at the low end of this range (see Tabachnick et al., 2019). A correlation as run between the BQ18 and “BQ13” resulting in a high positive correlation (r = .954)
BQ18 Factor Analysis
An exploratory factor analysis was conducted to explore the underlying factor structure of the BQ18. To date, there are no known studies that have done so; however, such an exploration is vital toward establishing the validity of the scale. The BQ18 was factor analyzed using maximum likelihood factor analysis with varimax rotation, using SPSS 26. The Kaiser–Meyer–Olkin measure of sampling adequacy (.79) and Bartlett's test of sphericity (p < .05) analyses indicated that the items were appropriate for the factor analysis. Using the traditional criterion for several factors of retaining eigenvalues greater than 1 (see Tabachnick et al., 2019), the initial analysis yielded five factors explaining a total of 49.86% of the variance for the entire BQ18. However, the scree plot was ambiguous and showed an inflection point that would instead justify retaining two factors based on the clear "elbow" in the plot after the second factor, with both factors still having eigenvalues above 1 (Tabachnick et al., 2019). The factor analysis was then run again, forcing two factors only. When only two factors were retained, the factors had a total variance of 31.65%. Because the total variance is under 60%, this is a limitation to the study that would suggest more research on the psychometric properties of the BQ18. The limitation will be further discussed in the discussion section. Factor 1 was labeled Thin Boundaries based on the content of the 13 items with loadings above (.3), including the following items, which taken together explained 20.36% of the variance:
My feelings blend into one another. I am very close to my childhood feelings. I am easily hurt. I spend a lot of time daydreaming, fantasizing or in reverie. Sometimes it's scary when one gets too involved with another person. A good parent has to be a bit of a child too. I can easily imagine myself as an animal or what it might be like to be an animal. When I work on a project I don't like to tie myself down to a definite outline. I rather like to let my mind wander. In my dreams, people sometimes merge into each other or become other people; I believe I am influence by forces that no one can understand. There are no sharp dividing lines between normal people, people with problems, and people who are considered psychotic or crazy. I think I would enjoy being some kind of creative artist. I have had the experiences of someone calling me or speaking my name and not being sure whether it was really happening or whether I was imagining it. I like stories that have a definite beginning, middle, and end; A good organization is one which all the lines of responsibility are precise and clearly established; There is a place for everything, and everything should be in its place.
The second factor derived was labeled Thick Boundaries based on the content of the 3 items with loadings above (.3), including the following items, which together explained 11.30% of the variance:
The "BQ13" Factor Analysis
A second factor analysis was computed based on only the thin-worded items to optimize mental boundaries further. For the sake of this study, the reduced item scale composed of these 13 items will be called the "BQ13." The researchers then conducted a Reliability analysis on the BQ18 for item analysis for the new scale. The researchers then deleted any items with item-total scale correlations of <.3 for reliability purposes. The researchers found that five items fell below .3 and were further deleted from the scale. The five items that have been eliminated from this scale include those that are thick-oriented (i.e., the three comprising the second factor of the initial factor analysis) and those with low loadings (<.3) on either factor (which may have resulted from poor item construction). At first glance, the BQ18 is mostly thin-worded items. The removal of all thick-worded items may help justify using the BQ18 to be more of an inventory of thinness. The researchers took into account that the thick-worded items could be used as a subscale. However, a reliability analysis on the three thick items yielded a questionable Cronbach's alpha of .63. The researchers decided to use only thin worded items based on the number of thin worded items. The nature of the BQ18 may be more suitable for measuring the thinness of boundaries and not thickness. The three items that were eliminated due to thick wording include: "I like stories that have a definite beginning, middle, and end;" "A good organization is one which all the lines of responsibility are precise and clearly established;" and "There is a place for everything, and everything should be in its place." The two items eliminated due to low factor loadings include: "When something happens to a friend of mine or to a lover, it is almost as it has happened to me;" and "I am a down to earth no nonsense kind of person."
These "BQ13" items were factor analyzed using maximum likelihood factor analysis with varimax rotation. The analysis yielded two factors explaining a total of 36.00% of the variance. Further, The Kaiser–Meyer–Olkin measure of sampling adequacy (.83) and Bartlett's test of sphericity (p < .05) analysis indicated that the items were appropriate for the factor analysis. Eigenvalues were above 1 for the two factors; however, the scree plot indicated a clear "elbow" between the first and second factors, suggesting a single-factor structure (Tabachnick et al., 2019). This single factor for the "BQ13" had an eigenvalue of 3.48. It explained 26.80% of the total variance, and all factor loadings were above .3.
Eliminating the aforementioned five items of the BQ18 increased the scale's internal consistency and resulted in acceptable factor loadings for each of the remaining items and resulted in a more conceptually clear and interpretable unidimensional scale that coherently operationalizes the construct of mental boundaries, with high scores representing high "thinness."
Discussion
In previous literature, there have been attempts to understand the factor structure of the original BQ (Hartmann, 1991; Rawlings, 2001). However, there has been no known attempt to examine the underlying factors that make up the BQ18 (Rawlings, 2001). Therefore, as a preliminary step to understanding the psychometric properties of the BQ18, this study explored the factor structure of the BQ18 to gain a more in-depth understanding of what the BQ18 measures. The current study found evidence that there are at least two distinct underlying factors of the BQ18 and possibly several items that are not psychometrically sound. The first run of the exploratory factor analysis produced five factors. However, three factors had low eigenvalues (just above one) that fell after the "elbow" in the scree plot and showed some of the items double-loaded across the factors. Thus, the researchers forced a two-factor model that produced factors labeled "thick" and "thin," based on the content and wording of the items and informed by Hartmann's previous research on mental boundaries (e.g., Hartmann, 1991). The Cronbach's alpha for the full BQ18 in the current study was α = .69, roughly at the conventionally acceptable cut-off (Tabachnick et al., 2019). The researchers then eliminated all "thick" worded questions based on the notion that the BQ18 may be more inclined to measure thin boundaries only on a unidimensional scale. Two other items were eliminated from the BQ18; it was unclear what they represented, given that the factor loadings were low (<.3) on both thin and thick factors. It was thus determined that these items might be unfit or problematic for the inventory as a whole. Taken together, the elimination of these five items on the BQ18 resulted in an alternate scale deemed the "BQ13." With the alternative 13 question inventory, Cronbach's alpha increased to α = .77, which is more clearly in the acceptable range for internal consistency suggesting an improvement in measurement.
Eliminating five questions from the BQ18 reduced the number of items and may have changed the meaning of the inventory compared to the BQ18, which included a mix of items focusing on thin and thick boundaries; all items on the "BQ13" are thin-worded questions. Kunzendorf et al. (1997) created the BQ18 as a shorter mental boundary inventory. However, in further examining the structure of the BQ18, it appears the creators intended to create a unidimensional inventory with thinness on one end of the continuum and thickness on the other. Some items only focus on thick boundaries, while the majority of questions are more thin boundary related. The authors of the BQ18 may not have anticipated how individuals may respond in different patterns to the thick and thin boundaried items. By measuring mental boundaries with a more explicit focus on positive responses to all items reflecting thinness—meaning negative responses would reflect thickness—the inventory would function more cleanly as a unidimensional scale. Further suggestions for future research addressing the dimensionality of mental boundary measurement will be addressed later in this section.
Recommendations for Future Research
Firstly, there are still some uncertainties about the overall design of the BQ18. Because the BQ18 is the only known mental boundary inventory (aside from the cumbersome original BQ measure) developed decades ago, there is room for updating and improvement, such as including more modern language. The BQ18 has many questions that can be somewhat challenging to interpret. Using straightforward language and short questions or statements would make a new inventory more precise, easy to follow, and easy to understand.
Additionally, validation testing with non-clinical samples (such as college students) would be helpful toward assessing more diverse populations and using the scale more widely. The original BQ was first used for nightmare sufferers (Hartmann, 1991; Kunzendorf et al., 1997) created the BQ18 for the same purpose but with fewer questions. These scale development efforts may have been restricted to the intended scope of assessment or item wording. Though the BQ and BQ18 were created to measure mental boundaries based on nightmares and dreams, the present work shows the potential for the concept of mental boundaries to be utilized in broader clinical settings such as college counseling centers.
As noted earlier, one potentially more valid approach would be to develop a new inventory that assesses thinness only on a unidimensional scale. With this approach, a scale would be developed only using thin-worded items. Being high in thick boundaries would be represented by low scores. Alternately, a new inventory could be developed that has two distinct factors that are thin and thick. Ideally, the items of this scale would be constructed to create a two-factor structure, representing two dimensions of mental boundaries, thick and thin. Ideally, to make a new scale clean and clear, there would be the same number of thin boundary questions as thick boundary questions.
One example of a two-dimensional inventory that may be similar is the Positive and Negative Affect Schedule (PANAS). The PANAS scale uses 10 items for two subscales, one measuring positive affect and the other negative affect (Watson et al., 1988). The development of the PANAS may provide insights into the development of a two-dimensional mental boundaries scale that measures both thick and thin boundaries. Given the potentially problematic nature of the wording of at least some of the BQ18 items uncovered in the present study, a new scale development should involve generating an entirely new set of mental boundary items. These new questions may resemble the structure of the PANAS by having 10 clear statements about thick mental boundaries and 10 clear statements about thin mental boundaries.
Conclusion
There has never been an attempted to examine the BQ18 in-depth to the researcher's best knowledge. Rawlings (2001) examined the full BQ and was able to develop an empirical short version. Rawlings (2001) findings produced a short version of the scale with 46 items. We are proposing that there can be a shorter version for the measurement of mental boundaries that is psychometrically appropriate and has good validity.
Further, with proposing a shorter version of the BQ18, some aspects need to be considered before its construction. It is suggested to conduct a cross-validation study with a new sample to compare findings. It is also suggested that there would have to be norms for clinical context to evaluate the boundary thinness of one person regarding the general population.
Modifying or developing new scales to measure mental boundaries also leads to understanding more about mental boundaries. With the ambiguines of what scores mean on the BQ18, we have started examining what we call "Boundary Equilibrium." "Boundary Equilibrium" is the balance between thick and thin boundaries. Creating a scale that can help identify that area can help future researchers and clinicians work with individuals to help balance out their Mental Boundaries.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.
