How to make health surveys more inclusive
We work closely with occupational health care providers and international companies with thousands of employees in order to prevent illness from affecting productivity in the form of absences, reduced performance, undesirable retention rates, higher insurance costs and to boost a positive employer image through happy, healthy employees. We work with both traditional browser-based health surveys, paper surveys as well as an artificial intelligence empowered app that are all used to analyse health and well-being in order to find potential risks before symptoms occur. Our analytics provide scientifically valid results and effective feedback for individuals.
As you might have noticed, most health-related questionnaires and analyses ask about the respondent’s sex and only offer ‘male’ and ‘female’ as options. There are a multitude of reasons behind this, including difficulty of obtaining data concerning non-binary persons and lack of understanding why inclusion is important. We were no different and hadn’t taken the necessary measures to fix it – until now. We had, of course, thought about and it was on the to-do list, but just recently we proudly released an updated version of AlvinOne where we have placed our best efforts in making it more inclusive and less binary when it comes to sex and gender.
The shift into a more inclusive model turned out be a bit more of challenge than I had initially anticipated. Now that we have done it, I wanted to share how we, a small company with limited resources did it so that maybe someone else who is struggling with similar issues could have an easier time doing it. And maybe if it feels a little bit easier, more companies would take that step toward a less gender-normative and more inclusive approach in their products.
According to several studies, discrimination toward non-binary genders in health care is usually caused by lack of knowledge and training. In order to contribute to closing this gap in information or at least in an effort to not contribute to maintaining it, I’ll start by defining some basic concepts related to the matter.
Gender as a term means the cultural norm you identify with, while sex means the sex assigned to a person at birth. Not all languages, such as Finnish, even have separate words for sex and gender which might cause some confusion. Transgender by definition means that a person’s gender identity doesn’t match the sex they were assigned at birth. Cisgender means that a person’s gender identity matches the sex they were assigned at birth. Cis is a Latin prefix and means ‘on this side of’ while ‘trans’ means ‘on the other side of’. If a person identifies as transgender, it doesn’t necessarily mean that the person identifies as “the other gender” either. We have been taught to only recognize two genders and this binary view is what has led to sometimes unintentional discrimination against transgender persons through gender-normative practices and institutional structures. In order to break down these harmful structures, largely held up by dichotomous gender assumption, we need to start looking at gender as a multidimensional spectrum. This is a pretty wide area to cover and I’m by no means a qualified expert to cover it as such, but I want to share my experience on how we are attempting to make health surveys and analyses more inclusive at our company.
This is how we started
We started exploring how the changes in AlvinOne could be done by finding the places in the product where sex or gender is mentioned and how that information is used. The following matters came up:
- The section that analyses alcohol consumption uses information regarding the respondent’s sex directly as it is embedded in the question itself. There is one question specifically that addresses gender: “How often do you drink more than seven restaurant servings of alcohol (men) or over five restaurant servings (women) during one day?”
- The questions in this section are based on a globally used and scientifically validated model, Audit C, and we can’t change any questions without compromising the scientific validity our product.
- In the other parts of the product the respondent’s sex doesn’t have a significant impact on the questions asked or the analytics and no require major changes would be required.
- Lack of data. We have an incredible, unique database with over 10 million health factors from over 190 000 subjects. Beyond its size, the strength of the database stems from structured data collected from a fairly heterogeneous group of working people between the ages of 18 and 64. However, the respondents are categorized as either male or female so any trends or correlations found cannot be directly applied to non-binary persons.
- Adding more options in the registration phase where users enter their sex doesn’t solve the whole issue; we need to consider what kind of results and feedback we give the end-user regarding their alcohol use if we can’t base the feedback on our own data or the binary system that Audit C relies on.
When I drew up my first suggestion in flowchart-form and presented it to my colleagues and developers, there were so many unanswered questions that we had to take a step back and reassess the situation. If we just changed the question concerning the user’s sex in the registration phase, the alcohol question would still include binary sexes in the question itself. If we change the question itself, it might not be scientifically valid.
I had to return to doing more research, more trial-and-error, and what you do when you want the unbiased and honest opinion of others – ask people what they think on social media. We eventually had to accept that there is no perfect solution, but our efforts led to definitive and vast improvement, and sometimes the process can be more valuable than the initially desired result. If we couldn’t resolve the issue by completely fixing it – meaning that we’d be able to provide an equal amount and quality of information to all genders in the product at hand– we’d have to do the next best thing.
We’d also have to consider how to communicate openly why we ask about gender in the first place, give the end-user the all the information necessary to support their decision-making in how to proceed and the option to skip the questions that require any gender or sex information. And of course, ideally, we’d have to do this without making the use of the product more difficult and without overwhelming users with information if it is not relevant for them personally.
We ended up doing the following:
- We added ‘other’ and ‘rather not say’ as options when we ask about the user’s sex in the beginning when the user registers to use the app.
- If the user selects one of the new options described above, a text box with a second question pops up where we explain why we ask about the user’s sex and explain how sex assigned at birth might affect how the body processes alcohol. It’s not just size that differentiates how typical male and female bodies process alcohol. Several factors such as hormones, body-fat balance and medication used in a transition process may affect how our bodies respond to alcohol. We then ask if the user would prefer to have their answers analysed according to a male-typical or female-typical body or if they’d like to skip the questions regarding alcohol consumption.
- We couldn’t really change the question about alcohol itself or add a ton of information right next to the question (nor did we want for the user to have to have restate their sex every time they complete a health analysis in the app) so instead we made sure the app knows what the user’s preferences are for all future analyses and asks the right question about alcohol consumption accordingly.
- In the settings of the app, we added the possibility to later change the sex you selected or the preference of how your answers should be analysed when sex information is used.
As I have lately come to understand, instant perfection is near-impossible with these things. Regardless, we have a responsibility to constantly evaluate and challenge ourselves (and our businesses) on whether we are being as inclusive as we could be. If the answer is no, it should be openly discussed even when there are no ready solutions. Transparency and dialogue are crucial for improvement.
P.S. If you have any tips on the subject and how to further develop our solution or the issue in general, feel free to contact me at firstname.lastname@example.org.
Useful links about the subject:
GLAAD, Tips for allies of transgender people: https://www.glaad.org/transgender/allies
GurveyGizmo, how to write gender questions for a survey: https://www.surveygizmo.com/resources/blog/how-to-write-survey-gender-questions/
Author: Johanna Varje, COO & co-founder at AlvinOne