Knowing someone who is transgender

In our last article a week ago, we stated that the estimate for the transgender population is roughly <½%, which is nearly 1/3 less than the estimate from a popular UCLA think tank that has it at nearly 0.7%.  Another popular survey topic, from Pew, Human Rights Campaign, GLAAD, and others, converses the question and asks how many people know someone who is transgender (and returns responses in a range from 15% upward to 35%).  So if a rare <1% of the population has a certain trait, what’s the likelihood someone in your friends network would fit that trait?  Larger than you might expect, yet a clear minority!  And from a policy perspective we know that being cognizant of someone you personally know who is transgender then breaks down the isolative barriers that may cause some to otherwise discriminate and oppress this minority.  That truly applies across many sections of the population (e.g., based on race, ethnicity, religion, other types of sexual orientation, etc.) and is particularly important now given the impulsive rise in hate crime following President Trump’s election.  This exposes hate equally in all of us (not always just in some other group of Americans!)  And we all need some soul-searching to find a more harmonious path in our actions and our thoughts.

And while the theoretical math results in the end do tie to our generous survey responders on this topic and continue to work less in the favor of the UCLA model, there are obviously gaps to any sort of survey endeavor such as this.  Namely people who were not invited to fill out the survey (selection bias), those who falsely provide societally acceptable responses (adverse selection), those who simply don’t know if they don’t know the sexual orientation of friends, etc.  Additionally the idea of “knowing” someone in the age of social media is difficult to delineate.  How much weight do you put on a personal friend, versus a coworker, versus a solely social media connection, versus a celebrity or a friend of a friend (but not necessarily a first-degree-of-separation friend).  Nonetheless on this website we’ve looked at the topic of social networks (here, here) and have a strong sense of how to consider this in the social media world (where at least it is easy to verify the most critical variable, which is how large one’s network is).

If it is true that most people have social media networks averaging low 200s, then we would expect that there is a ¾ or so chance that one knows at least one transgender person.  But again this should be highly dependent on how large one’s network is to begin with, with some loners in the community having a different probability than social magnets.  Some of the other surveys noted above look at generational differences as a critical explanatory factor but this is highly dubious and with less significant results.

In order to get a sense of how this would work for our own social network, we ran a survey last week across all the major social media accounts and returns >75 responses and a wealth of discussion engagement.  The most popular of these surveys can be seen here.  But the results are tabled below (and tabled at the bottom).  The survey question was “do you know someone in your social media network who is transgender?”

 

We’ll return to the survey responses in a moment, and note than a binomial distribution formulae could help identify the probability of knowing someone who is transgender based on the size of your social media network.  Clustering into large (>200 friends) and small (<200 friends) networks:

p(knowing|small network) = from all k>0,

_nC_k p^kq^(n-k)

~ 1- ₆₀C₀ (0.06%)⁰(99.94%)^(60-0)

~ 1-1*1*.96

~ 4%

p(knowing|large network) ~ from all k>0,

_nC_k p^kq^(n-k)

~ 1- ₂₀₀₀C₀ (0.06%)⁰(99.94%)^(2000-0)

~ 1-1*1*.30

~70%

So a large difference in knowing someone, based on whether you know many people or just a few.  If one has say 100 friends who are all fairly less communal, then we would expect that only 4 will reveal that they know a transgender.

Now returning to our survey, we should note that most who provided their age stated that they were less than 45 years old and the age-providing respondents were biased towards stating they knew someone who was transgender (more so than the general population).  This is all in line with the statistically significant breaking point of the non-profit surveys noted above.  The survey showed that roughly 37% of people stated they knew someone who was transgender, which is exactly at the average of the larger network and smaller network probabilities above (4% and 70%).  though if we reweight the theoretical probabilities closer to the respondent weights than the theoretical probability is in the low 30s%.

Additionally, for small networks (58% of our survey responders), the probability of knowing someone who is transgender was in the low 30s%.  While for the large network (42% of survey responders), the probability of knowing someone who is transgender was in the low 40s%.  This suggests that there is not as much homogenous random distribution of knowing one is transgender in our networks.  Also the probability of knowing someone being transgender in our survey could be biased based on the demographic bias of our network (political leanings, occupations, location clusters in bi-coastal cities, etc.)  And there could be a bias in only those wanting to reveal certain socially acceptable choices could easily tilt the results.  But in any event, the survey results are on the high side of the competitive 15%-35% range, and but not significantly and they excitingly match the mathematical theory discussed above at the top-line values!  It also shows -once again- that the UCLA survey of 0.7% transgender population is stretching well above what we can reconcile in any of these surveys.
 

Reference (table of picture above that can be pasted into Excel):

Social media survey	Don’t know a transgender	Know a transgender	Total
<200 friends	38%	20%	58%
>200 friends	25%	17%	42%
Total	63%	37%	100%