To make estimates about what a certain group of people are thinking, pollsters need to speak to enough people from that group. Generally speaking, though, the smaller the group, the harder it is to get respondents. That’s why, when they’re gauging opinion among different racial groups, pollsters rarely manage to get further down the list beyond white (approximately 69% of all voters), black (12%) or Hispanic (12%).
But that last 7% is a significant share of the electorate that simply gets lumped together under the category “other”. According to the LA Times polling data (which tends to skew a little more Republican) it is this “other” racial group that is the most closely split about who they intend to vote for.
Just who is an “other”, though? The US Census Bureau’s 2015 Annual Estimates of the Resident Population uses more than three racial groups – it also includes Asian (a category that represents 5.7% of all US adults), American Indian and Alaska Native (1.1%), Native Hawaiian and Other Pacific Islander (0.2%) as well as two or more races (1.8%).
Sound like a complete list to you? Not to me it doesn’t. I’m Arab, an ethnicity that frequently doesn’t appear on forms like this (although the Census Bureau is considering adding it to their list of options in the 2020 census). Even if the list were comprehensive, lumping together those groups with potentially differing political interests in 2016 polling as one big “other” seems crude at best.
There have been some efforts to add nuance. Last week a group of US universities conducted a joint survey of 1,694 Asian Americans as well as using existing polling. The responses skewed heavily Democratic – when asked who they would choose in a race between Democrat Hillary Clinton and Republican Donald Trump, just 10% said Trump.
Because the survey was focused on only Asian Americans it meant they could understand differing opinions within this group (which is still a pretty crude category). Responses were broken down by state, foreign born/native born, age, gender as well as origin – Indian, Chinese, Japanese, Korean, Vietnamese, Korean, etc.
But such surveys are expensive and time consuming (remember what I said at the start – it’s hard enough to get people to respond to surveys and once you get picky about who you want to respond, things get even harder). That means there are around seven out of every 100 voters about whom we know very little. As if there weren’t already enough uncertainty around this election.
