Mid-terms Part I: Will healthcare deliver a blue wave?


Trump’s divisive rhetoric and a comparatively disciplined issues-driven campaign from the Democrats have presented voters with a stark choice in tomorrow’s mid-terms: turn out to vote because you fear a tidal wave of immigration and criminality, or go to the polls because you fear losing your access to affordable health care. Analysis conducted by Signify indicates that healthcare is more important to voters than any other issue and may swing results all over the country.

Signify is an ethical data science company. We apply machine learning technology to public data sets in order to understand what really matters to people and help our clients build resonant, helpful campaigns. To apply an issues lens to the mid-terms we analysed coverage and engagement in three bellwether races – congressional races in North Carolina (9th) and Pennsylvania (7th) and the Senate race in Arizona – in order to determine what it was about each candidate that was covered by the media, and in turn what aspects of that coverage really excited or upset voters.

We used a simple metric to compare volume of coverage for a topic, with the level of engagement. Namely how often each article, blog or video was ‘shared’ by voters. In terms of online behaviour, a ‘share’ is a more important identifier than a ‘view’, a ‘like’ or a ‘comment’ because it indicates the sharer really cares about a topic. In crude terms, a ‘share’ also equates to an exponential increase in eyeballs for that coverage.

Our work in these mid-terms has no predictive element. However, we will publish a second blog after the results to see if the patterns we have identified here bear any relation to results. If so, it could be good news for Democrats.

In North Carolina our analysis of media coverage found very little of note. Only the GOP candidate Mark Harris has significant cut through, with his right-wing views on women’s rights and reproductive rights. Polling is wildly variable for this district, but Harris is definitely polarising. In contrast, Democrat Dan McCready is vanilla. Three of the four top issues he gets coverage for are Gender, LGBTQ and Social Justice – where he is being entirely defined in relation to Harris. On just one issue – tax cuts for the rich – he put his head above the parapet and seen much greater engagement, with voters sitting up and taking notice of his critique of this policy 1 .

In Pennsylvania, our findings on healthcare were supported but it seems unlikely that this issue will prove to be the key deciding factor in the race. Marthy Nothstein (R) has faded dramatically thanks to allegations of assault when he was an athlete. The only other issue that he tracks for at all is Healthcare – and that is hurting him as well. Susan Wild (D) meanwhile is a popular candidate whose most livewire issues are Women’s Rights and Abortion. However, hernext most significant issue is Healthcare – coverage on this topic is 14 times more likely to be shared than issues like social mobility, climate change or human rights 2 .

In the Arizona race we found a richer, more newsworthy story. DEM candidate Sinema has been relentlessly attacked as unpatriotic and soft on terrorists and immigration by her opponent and backers 3 . This has been highly effective and Sinema has lost significant ground; she is basically defined in terms of online profile by the things people have said about her, and the things that have been said back in her defence. It’s not a pretty picture and McSally has gained on her steadily in the polls until last week.

Real+Clear+Politics+-+Arizona+2018+Mid+Terms.png

In the past week the direction of travel has flipped and Sinema has regained her lead and opened up a two-point gap in aggregated polling. The race is still way too close to call but the downturn in McSally’s fortunes exactly coincides with the moment when her inconsistency on pre-existing conditions was exposed, and then highlighted in the national media.

McSally is a decorated veteran and she is perceived as tough on security and immigration, with a highly effective campaign focussing on her service and heroism. In a border state which went solidly for Trump in 2016 4 this toughness should appeal – and her opponent Sinema has been portrayed in a series of well-back attack ads as soft on terrorism and immigration. This has proved resonant with voters and contributed to a big surge in support for McSally between July and September.

However, starting in September questions began to be asked about whether McSally could be trusted to protect the interests of voters who rely on Medicaid. In Arizona that equates to 1 in 6 adults and 3 out of 8 children 5 . McSally has said she is passionate about protecting people with pre-existing conditions (which applies to many of the two million retirees in her state) but she is now attracting national attention for her voting track record which exactly contradicts this.

The graph below shows that national security – an issue on which McSally has rallied and attacked her opponent – is the leading topic in conversation about her, but only a little ahead of healthcare. On this issue, and particularly in regard of her determination to roll back affordable care, McSally’s own voting record and statements have done more to harm her than any position taken by Sinema.

McSally+coverage+chart+by+Signify.png

Of course there are thousands of factors at play in each mid-term race, but we are interested to see whether healthcare is as important to voters as the data indicates. With this in mind, we will keep an eye on key races and the NBC ‘reasons for voting’ exit polls. If it turns out that a focus on healthcare helps the Democrats achieve some big results on Tuesday night, it will change the politics of repeal and could improve the lives and health of millions of Americans.

We will link to pt I and analyse results later this week.

If you’re interested in ethical data science and the application of public data to better discourse, please get in touch or follow us @signify

1. NC data set: Signify research based on 25 relevant articles and 36,000 shares
2. PA data set: Signify research based on 20 relevant articles and 21,670 shares
3. AZ data set: Signify research based on 300 relevant articles and 881,000 shares AZ
4.
Trump wins Arizona
5.
Arizona Medicaid stats

Previous
Previous

Signify featured in Guardian mid-term coverage

Next
Next

‘Woke’ advertising: does it work?