How well are outgroup attitudes and behaviours toward bisexual individuals measured? A systematic review of the psychometric properties of binegativity measures

ABSTRACT

Bisexual individuals not only experience instances of invisibility and erasure within various social systems and structures but also encounter prejudice and discrimination that is fuelled by stereotypes. While gay men and lesbian women tend to only experience homonegativity from heterosexual individuals, bisexual persons are subject to binegativity from both heterosexual and gay/lesbian groups. Only recently has there been a concerted effort to increase the amount of research conducted on binegativity among heterosexual and gay/lesbian groups. However, since this body of research is only recently growing, it seems prudent to identify the binegativity measures currently available and evaluate how well each adheres to best practices regarding their psychometric properties. In the current study, 82 scales were identified and their adherence to best practices in psychometric assessment was examined. Most measures lacked sufficient details attesting to item development/refinement, factor structure, scale score reliability, and construct validity. No measure was found to follow best practices for all psychometric properties; those that were close are identified and recommendations are made for improving future binegativity scales.

KEYWORDS:

• Bisexual
• bisexuality
• binegativity
• biphobia
• psychometrics

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Although we are providing what we perceive as best practices in the context of psychometric property assessment based on a review of the literature, it is important to note that ‘best practices’ are at some level subjective and not all researchers may agree.

2. CFA requires an understanding of structural equation modelling and is more complex than EFA. Since such statistical techniques are beyond the scope of this paper, additional details regarding CFA will not be provided. For further coverage on CFA, see Harrington (2009).

3. Test-retest reliability is the only form capable of assessing an instrument’s temporal stability (i.e. similar scores are produced over time; Charter, 2003).

4. (Bisexual OR Bisexuality) AND (Biphobia OR Biphobic OR Binegativity OR Binegative OR Biprejudice OR Anti-bisexual OR Stigma* OR Discriminat* OR Prejudic* OR Bias* OR Opinion* OR Perception* OR Attitud* OR Belief* OR Negativ* OR Violence OR Harassment OR Aggress*) AND (Measur* OR Instrument* OR Scale OR Index OR Inventor* OR Questionnaire* OR Test*).

5. A score of zero was issued in such cases to highlight that specific psychometric data cannot be generated for a specific measure. While in other cases a score of zero indicates either an omission or calculation that does not follow best practices, these situations do not reflect any sort of oversight. Instead, a score of zero for any psychometric property that is not applicable serves to indicate the limitations of single or two-item measures. For example, receiving points for factor structure and scale-score reliability, despite not being calculable, would lead to inflated total scores that would misrepresent the utility of the measures.

6. This additional designation addresses concerns that are present in previous research that issued scores of zero when best practices were not followed exactly as described (e.g. Morrison et al., 2018b). ‘Mostly best practices’ describe situations where authors have adhered to the requirements discussed above but, for whatever reason, missed an element that would not result in a full point being issued. However, it would also not be acceptable to flag it as completely overlooking or erroneously carrying out the calculation or analysis. For example, it is uncommon for confidence intervals to accompany estimates of reliability despite them being recommended. Since many researchers are not taught that this information is valuable to readers, they may not be cognisant of their omission. Furthermore, if an estimate of reliability is provided without a confidence interval, should this be deemed as an oversight similar to those examples where no reliability estimates are provided at all? We feel there is a meaningful distinction here, hence the designation ‘mostly best practices’.

Additional information

Notes on contributors

CJ Bishop

CJ Bishop obtained his PhD in Applied Social Psychology from the University of Saskatchewan. CJ’s research interests include prejudice and discrimination towards sexual and gender minority (SGM) groups, psychometrics, and street-involved and homeless LGBTQ2S+ youth.

Emily Pynoo

Emily Pynoo is a Masters student in School and Clinical Child Psychology at the University of Alberta. Emily is currently a research assistant with the Community Health Empowerment and Wellness (CHEW) Project which provides support and services to LGBTQ2S+ youth facing barriers. Her research interests include community intervention with vulnerable youth populations.