Could Candidates’ Visits to New Hampshire Indicate Who Will Win the Democratic Primary?

Introduction:

In the last Democratic Debate, Joe Biden noted in regards to his fourth place finish in the Iowa caucuses and his chances of winning the New Hampshire primary: “I took a hit in Iowa and I’ll probably take a hit here. Traditionally, Bernie won by about 20 points last time, and usually it’s the neighboring senators that do well.” In 2016, Hillary Clinton’s surrogates dismissed Bernie Sanders’ victory as “a matter of geography”. These statements are problematic in two levels. 

First, in 2020, Bernie Sanders is not the only candidate from New England in the race. While Clinton did not win the primary in 2016, she pulled off a surprise victory in 2008. Thus, she should have been able to compete against Sanders in 2016. Second, these views are an insult to New Hampshire voters’ intellect. Primary-goers proudly study the candidates’ positions very carefully, they expect candidates to spend a lot of time answering their questions, and they usually make-up their minds late in the game.

In this post, I will look at the candidates’ campaign events in the Granite State, especially those events that took place after the Iowa caucuses. To try to gauge the popularity of these events, I tap on data collected on each candidate’s Facebook page, which includes information on the number of Facebook users (i.e. guests) who demonstrated an interest in many of these events. 

The Data:

I scraped the data on campaign events from the NECN’s 2020 New Hampshire Candidate Tracker. I cross-referenced each event with the list of event indexed in the candidates’ Facebook page. I then collected the number Facebook users who said they would attend these events. 

It is important to add two caveats at this point. The Facebook “event guests” data does not represent the total number of people that attend an event. I suspect that the number of attendees is higher as the candidates use many tools to advertise their events. I use these data to help us make sense of each candidate’s support in Facebook. Second, the events listed in the Candidate Tracker are not included in some candidates’ Facebook pages. In these cases, I relied on the NECN’s reporting to construct my dataset.

My dataset also classifies each event into different types, including “town halls”, “rallies”, “speeches” and so forth. Although this analysis will not use these data points, the dataset includes specific information of the venues that hosted these events and the time of day. 

Who Will Likely Win the Primary?

Before we start our analysis, the RealClearPolitics average indicates that Bernie Sanders will more than likely win the primary. Pete Buttigieg should place second, while Massachusetts Senator Elizabeth Warren, Minnesota Senator Amy Klobuchar, and former Vice President Joe Biden compete for third place. The rest of the field is in the single digits. 

Given this forecast, we should expect that the leading candidates organized more events than the weaker ones. But this is not the case. Hawaii Congresswoman Tulsi Gabbard, who is expected to receive around 3% of the vote, held 126 events since she announced her candidacy last year. Colorado Senator Michael Bennet, who is likely to get less than 1% of the vote, held 82 events.

It is worth highlighting that the Biden campaign organized only 42 events in New Hampshire since April 2019. This is the lowest number of appearances for one of the major contenders in my dataset. 

Rather than looking at all of New Hampshire’s campaign events, I focus on those events held in January and February of this year. In addition, given their low polling average, this does not include events for the following candidates: Bennet, the philanthropist Tom Steyer, and former Massachusetts governor Deval Patrick.

Total Number of 2020 Campaign Events Per Candidates:

The next figure summarizes each campaign’s events for January and February 2020. In line with the figures above, Yang and Gabbard organized more events than their counterparts. For Gabbard, a poor showing in New Hampshire will more than likely end her presidential hopes and it may explains why she devoted some much time and resources in the state. 

Yang’s long-term prospects are better than Gabbard’s but it is not clear why he organized so many events. The polling average suggests that he will win around 4% of the vote. Does his internal polling suggest he receive a higher share of the vote?

Geographical Distribution of 2020 Campaign Events:

The next map shows that most of the events were held in the counties bordering Massachusetts, were the majority of the state’s residents live. More specifically, 76 events were held in Hillsborough County, where the city of Manchester is located, closely followed by 30 events in Merrimack County and 24 in Rockingham County. The candidates did not visit Coos County this year, though several candidates held events in this part of the state last year. 

In the last days of the primary, the candidates concentrated most of their time and resources in Manchester and Concord. This following bar graph shows that Yang visited many of the listed towns and cities. The same applies to Buttigieg, Sanders and Klobuchar. In contrast, the Biden campaign prioritized more densely populated communities over rural areas. It indicates that the former vice president may have difficult connecting with rural voters. 

Which Candidate Held the Most Events in February?

Most of the candidates participated in at least two events per day. The exception is February 7, 2020, as most of the candidates used this day to prepare for the debate. As expected, the majority of campaign events since the Iowa caucuses took place in the last days of the race. 

Yang appeared in 28 events, while Sanders had 22 and most of these took place in the last three days of the race.

Main types of campaign events and their popularity:

Town halls are the most popular type of campaign events in my dataset. The events labeled as “speech” include events or forums, where the candidate was been invited to speak. Not included in this graph are informal “meet and greets”, canvassing events and so forth. What can we learn from these data?

The candidates with the least recognition — Gabbard and Yang — held the most town halls in January and February. The Klobuchar campaign organized more rallies than town halls, which is surprising given her strong debate performance.

 In contrast, it is not accidental that the Biden campaign favored rallies to town halls as the former Vice President has in the past confronted attendees who have asked questions about his son’s connection to a Ukrainian gas company or challenged participants who questioned his electability. 

What type of activities did the campaigns selected in the closing days of the primaries? And which events attracted the most attention? To answer these questions let’s look at the collected Facebook data. 

Keeping in mind that the Facebook Guests data are not representative and that many more people attended these events, Sanders seems to have received the most total support. Nonetheless, Yang’s numbers are very impressive too. But if we average the total number of guests by the type of events, his popularity declines relative to the rest of the field. 

To gauge each candidate’s popularity, we can average the number of Facebook guests who expressed interests one of these types of events (i.e. see second column in the graph above). Although these data are noisy, these findings are in line with the primary’s forecasted outcome. 

Concluding Remarks:

In this short analysis, I have used different data points try to measure the campaign’s overall strengths and the candidates’ popularity. In reaction to Biden’s opinion that New Hampshire voters prefer New England candidates over nominees from others part of the country, this analysis shows that this is not the case. Sanders’ popularity is an outcome of his high name recognition as well as his campaign’s organizational capacities. 

It will be interesting whether the forecasts are correct. While Facebook data are not representative, they could be one way to measure and contrast candidates’ level of popularity. This is an area for further exploration and it will be interesting to see how this analysis applies to events data from Iowa or the upcoming nomination contests in Nevada and South Carolina, which require the candidates to spend some time in those states engaging voters.