Wednesday, August 24, 2016

US News and World Report Hospital Rankings

Every year US News and World Report ranks the top colleges and hospitals in the US.  Those that are ranked high use these results to market their institutions to potential students and/or patients.  I've been asked by a friend to comment on the method that US News and World Report uses to create the annual rankings of hospitals in the US.  

According to their FAQ page for these rankings the rankings are based on a summary score from 16 hospital specialties.  Twelve of these specialties (cancer, cardiology & heart surgery, diabetes & endocrinology, ear, nose & throat, gastroenterology & GI surgery, geriatrics, gynecology, nephrology, neurology & neurosurgery, orthopedics, pulmonology and urology) have objective health outcomes and are used to rank hospitals within specialties and are factored into the rankings by health outcome data such as survival rates.  Another 4 specialties (ophthalmology, psychiatry, rehabilitation and rheumatology) have outcomes that are more subjective so surveys of physician specialists are used to rank hospitals within specialties and factored into overall rankings.  

I have two comments on these rankings.  The first comment is anytime a ranking is based on a summary score of different measures invariably some information is lost.  A hospital can have high ratings in geriatrics, ophthalmology and neurology while another can have similar high rankings in gynecology, cancer and psychiatry.  If the rankings for both hospitals on the other measures are the same, the two hospitals can have roughly the same overall rankings.  The overall rankings say little about the hospitals relative strengths and weaknesses which are different.  

Even within specialties outcomes can vary widely for different diseases.  There are many types of diseases that each specialty treats.  Overall rankings for specialties can vary widely among different diseases.  If one has a specific disease it is better to look at how that hospital treats that disease.

The second comment is that the four subjective specialties that are ranked based on a survey of physicians perceptions of which hospitals are the best at these specialties.  Presumably the physicians can rank their hospital first and can choose ones simply based on reputation (eg. Massachusetts General or the Cleveland Clinic).  Also the respondents may have no actual firsthand knowledge of the hospital's performance in these specialties.  A survey of the patients of the specialties of the respective hospitals may be more accurate but such a survey would not be practical or ethical to conduct.

**Related Posts**

Health Care and the 2014 Election in Pennsylvania

Pennsylvania Voters Want a Better Health Care Law

POLL: Dislike of healthcare law crosses party lines, 1 in 4 Dems want repeal - (But Doesn't Ask Why)

Podcamp Pittsburgh XI Video Now Posted

The video of my presentation at PodCamp can be seen on YouTube and here.  The slides are presented below.


**Related Posts** 

My Raging Chicken Press Interview 

Podcamp Session Feedback Part 2-The Video of My Session Has Been Posted


Tuesday, August 16, 2016

The Facebook and Twitter General Election

RCP % Avg
Facebook Followers
FB Engaged
Engagement Rate
Twitter Followers AUG
Hillary Clinton (D)
Donald Trump (GOP)
Gary Johnson (L)
Jill Stein (G)
Hal Boyd (I)

With the dust finally settled from the conventions, the poll numbers in the general election should be fairly stable now (depending on the next Trump gaffe).  I thought I would take a look at the state of the race in relation to the social media and the third party candidates Gary Johnson and Jill Stein.  Former CIA Operative Hal Boyd has announced an independent bid for the White House but at of this writing I was unable to find any polls, Facebook Pages, or Twitter profiles for him. 

As I did in the primaries, I looked at the Real Clear Politics Poll Averages (RCP) for the four candidates and looked at how they correlate with their respective following and engagement on Facebook and Twitter.  Facebook provides numbers of their following and engagement (those who click on, like and or share their posts) while Twitter provides numbers on their following.  

At the top of the post is the raw data for the analysis.  Below are the correlation coefficients for each social media variable and the candidates respective RCP polling average.  Of the four variables, the number of twitter followers was most strongly correlated with poll support at 0.94 (correlations are on a scale from -1 to +1). The correlation for the FB engagement rate is negative because Johnson and Stein have much higher numbers of followers engaged with their posts relative to their total followers than Clinton and Trump.  

RCP % Avg
Facebook Followers
FB Engaged
FB Engagement Rate
Twitter Followers AUG

The graph below shows the relationship between the Twitter followings of the candidates and their poll averages.  The r-square statistic of 0.89 says that 89% of the variability in the poll numbers is accounted for by the number of followers.   According to the website, 62% of Clinton's supporters are real, 59% of Trump's are, 62% of Johnson's, and 92% of Stein's are real.  

Of course those who follow a candidate may not necessarily support a candidate.  For the Facebook following of Donald Trump, of the 19 friends of mine who follow his page, I counted at least 6 who I'm pretty sure are not voting for him.  For Hillary Clinton's page, of 58 friends of mine who follow her page, I did not see any who would not be voting for her (Facebook does not provide the names of all the friends).  For Gary Johnson, 22 of my friends like his page, but I'm not sure which are actually voting for him.  67 of my Facebook friends like Jill Stein's page and I'm not sure how many are voting for her.

I don't have data on how Johnson and Stein's social media profiles have grown since February.  I hope to have that information in a few months.  However, I do have that information for Trump and Clinton.  Clinton had a 75.4% increase in her Facebook following while Trump had a 48.1% increase.  Trump had a 44.1% increase while Clinton had a 40.4% increase.  The rest of their numbers are presented below along with the RCP averages for a two way race for Clinton and Trump. 

RCP % Avg (2 way)
Facebook Followers
FB Followers April
FB Gain
% chg FB
Twitter Followers AUG
Twitter Apr
Twitter gain
% chg Twitter
Hillary Clinton (D)
Donald Trump (GOP)

As I've stated before, it is a chicken-egg problem whether or not the popularity comes before the large social media following or vice versa?  Regardless of which came first having a large social media following gives a candidate a cheap way of communicating with their core supporters.  Trump may have an advantage in social media but the other candidates have higher levels of engagement among their followers.

A post from the website argues that Facebook posts do not change friends opinions on the candidatesThe numbers that they present in support of their arguments are very small.  Of course those who follow a page are interested in what they have to say.  When they share posts with their friends they encourage discussion.  It takes a lot more than posts or commercials to get people to change their mbut it can influence those who are undecided.

  **Related Posts**

What Effect Do 3rd Parties Have on Trump and Clinton? 

Facebook Primary 2016, August Update, Does it Predict Support?

Facebook and Twitter Primary: The Final Five