Data and Findings

The information below consists of county geocode specifics, a Tableau visualization depicting the number of cases during the Tweet collection period, and a findings menu containing statistical analysis from SPSS, choropleth (heat) maps from R, and individual county sentiment analysis.

# County Geocode Area Radius

Coefficients Correlations Model Summary (Linear Regression) Normal Distribution (Positive Skew)

After SPSS bivariate correlation and linear regression modeling, several key insights were revealed (To view any of the SPSS results in greater detail, click on any of the images above to expand.). Through bivariate correlation, both positive and negative tweet sentiments maintained a significance (p) value greater than the established 0.05 critical value, indicating weak evidence against the null hypothesis, thus failing to reject the null hypothesis. In other words, these values also indicate greater coincidence over correlative genuineness. Similarly, the correlation (r) values of both positive and negative tweet sentiment counts indicate a more scattered and less linear-like relationship. Although, its positive direction indicates that the dependent ‘case’ variable increases as the independent sentiment variables increase. Approaching the r-square value, the lower value indicates that tweet sentiments predict only 11.4% of the variance of the dependent variable (number of cases), indicating a weaker (more scattered) fit of the regression line. Given the positive skew of the distribution plot, it appears that the method’s weakness - the number of tweets collected in each county - resulted in lower bounds affecting the overall skew of the distribution. This can be accredited to both potential limitations in the Twitter search methodology used, in addition to a lack of COVID-19, or “coronavirus” related tweets in a given county.

New Jersey Positive Tweet Sentiments, by County New Jersey Negative Tweet Sentiments, by County

The above choropleth maps (better known as heat maps) depict the positive and negative tweet sentiment count densities by county. Together, these maps reveal two interesting occurrences. In approaching the positive tweet sentiments choropleth map, the greatest count density is primarily evident in the southern region of the state. Examining the Tableau county case visualization above, this positive sentiment is consistent with a smaller number of confirmed cases in the same subregion. Peculiarly, the negative sentiment counts are also largely concentrated in a similar subregion, apart from a few counties in northern regions of the state. This finding could be consistent with some of the near-equal sentiment ratios uncovered in the individual county analysis below. In such instances, certain counties exhibited a near identical positive and negative tweet sentiment count from the tweet sample in that area. This could indicate neutral attitudes or disagreements among the populations in these counties, thereby explaining the coexistence of the opposing densities depicted in the choropleth maps. Another possibility is that this finding is representative of the fairly uneven population distribution in the state. Much of New Jersey’s population is concentrated in the north, thus explaining the increased outbreak severity in the northern counties. The southern region of the state is far less populated in comparison, but contains the greatest density of COVID-19 tweet sentiments. To better test this hypothesis, additional tweet collection over a longer period in the COVID-19 pandemic would be required.

Atlantic County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 99/104

Negative Sentiment to County Case Ratio: 99/382

Positive Sentiment to County Case Ratio: 52/191

Takeaways: Atlantic County's positive and negative sentiments closely matched each other. The level of activity within this county resulted in tweet sentiment counts that exceeded state averages.

Bergen County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 44/59

Negative Sentiment to County Case Ratio: 44/12163

Positive Sentiment to County Case Ratio: 59/12163

Takeaways: Bergen County's positive tweet sentiments exceeded its negative tweet sentiments. However, the county remains among New Jersey's most infected counties, despite the prevalence of positive tweet sentiments.

Burlington County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 79/126

Negative Sentiment to County Case Ratio: 79/1456

Positive Sentiment to County Case Ratio: 9/104

Takeaways: Burlington County is among New Jersey's largest counties in area, and is also among the states most positive with regard to tweet sentiments. Such an occurrence may be reflected in its lower-case count.

Camden County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 82/113

Negative Sentiment to County Case Ratio: 41/959

Positive Sentiment to County Case Ratio: 113/1918

Takeaways: Camden County is among New Jersey's most economically adverse counties, but has positive tweet sentiments exceeding negative tweet sentiments. This is especially reassuring, given the smaller case count in comparison to other economically stronger counties.

Cape May County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 11/7

Negative Sentiment to County Case Ratio: 14/199

Positive Sentiment to County Case Ratio: 22/199

Takeaways: Cape May County adheres to the intended predictive model, with its larger positive tweet sentiments count fostering case counts among New Jersey's lowest.

Cumberland County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 11/17

Negative Sentiment to County Case Ratio: 11/68

Positive Sentiment to County Case Ratio: 1/4

Takeaways: Cumberland County mirrors trends in many of the other Southern New Jersey counties, with positive tweet sentiments overtaking negative tweet sentiments, and fewer cases in comparison to other counties.

Essex County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 36/53

Negative Sentiment to County Case Ratio: 36/9901

Positive Sentiment to County Case Ratio: 53/9901

Takeaways: Essex County, peculiarly, remains among New Jersey's most infected counties. Despite the greater count of positive tweet sentiments, this county ranks among the worst.

Gloucester County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 103/109

Negative Sentiment to County Case Ratio: 103/683

Positive Sentiment to County Case Ratio: 109/683

Takeaways: Gloucester County's positive and negative tweet sentiments were very close, but exceeded the state county sentiment averages. At a near-equal ratio, the sentiments are clearly divided among the county.

Hudson County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 5/8

Negative Sentiment to County Case Ratio: 85/9956

Positive Sentiment to County Case Ratio: 34/2489

Takeaways: Hudson County, similar to other counties in the northern part of the state, exhibited a contradictory behavior. Despite the larger amount of positive tweet sentiments, the county is among the state's most infected.

Hunterdon County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 58/33

Negative Sentiment to County Case Ratio: 58/391

Positive Sentiment to County Case Ratio: 33/391

Takeaways: Hunterdon County exhibits negative tweet sentiments at an almost 2:1 ratio. However, the county is also among the state’s least infected.

Mercer County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 39/70

Negative Sentiment to County Case Ratio: 39/2215

Positive Sentiment to County Case Ratio: 14/443

Takeaways: Mercer County, located in the central portion of New Jersey, coincidentally ranks in the middle with regard to case severity. Its positive tweet sentiments greatly outweigh its negative ones.

Middlesex County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 86/65

Negative Sentiment to County Case Ratio: 43/3812

Positive Sentiment to County Case Ratio: 65/7624

Takeaways: Middlesex County, ranking among New jersey's most infected counties, adheres to the alternate hypothetical model, with negative tweet sentiment counts exceeding positive counts.

Monmouth County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 139/68

Negative Sentiment to County Case Ratio: 139/4528

Positive Sentiment to County Case Ratio: 17/1132

Takeaways: Monmouth County possesses a large negative to positive tweet sentiment ratio. It has a case amount in the thousands and a ranking that nears the sates infection severity midpoint.

Morris County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 10/33

Negative Sentiment to County Case Ratio: 5/996

Positive Sentiment to County Case Ratio: 11/664

Takeaways: Morris County, with around a 3:1 positive to negative tweet sentiment ratio, ranks near the state's midpoint in infection severity. Given the county's location in the COVID-19 hotspot region, the ratio indicates a likelihood of sentiment effects on case counts.

Ocean County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 22/39

Negative Sentiment to County Case Ratio: 11/758

Positive Sentiment to County Case Ratio: 39/1516

Takeaways: Ocean County, New Jersey's second largest county in area, exhibits a sentiment ratio moderately consistent with the alternative hypothetical model. Its positive tweet sentiments are noticeably larger in count than negative tweet sentiments.

Passaic County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 43/57

Negative Sentiment to County Case Ratio: 43/3968

Positive Sentiment to County Case Ratio: 57/3968

Takeaways: Passaic County possesses a near 3:2 positive to negative tweet sentiment ratio. Despite this occurrence, the county is located in a COVID-19 hotspot cluster in the state's northern region. Even with economic disadvantages, it avoids the top rank in outbreak severity.

Salem County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 11/7

Negative Sentiment to County Case Ratio: 11/105

Positive Sentiment to County Case Ratio: 1/15

Takeaways: Salem County is among the state's lowest in outbreak count, but possesses an 11:7 negative to positive tweet sentiment ratio. This may largely be due to the lack of COVID-19 related tweets in the area, given that it is the least populous county in the state.

Somerset County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 67/75

Negative Sentiment to County Case Ratio: 67/2283

Positive Sentiment to County Case Ratio: 25/761

Takeaways: Somerset County, with a near 8:7 positive to negative tweet sentiment ratio, adheres to the alternative hypothetical model. Its case outbreak count ranks on the lower end of the state's outbreak severity midpoint.

Sussex County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 41/62

Negative Sentiment to County Case Ratio: 41/626

Positive Sentiment to County Case Ratio: 31/313

Takeaways: Sussex County, with an approximate 3:2 positive to negative tweet sentiment ratio, adheres to the alternative hypothetical model. It ranks among the state's lowest in infection severity.

Union County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 14/37

Negative Sentiment to County Case Ratio: 56/8959

Positive Sentiment to County Case Ratio: 148/8959

Takeaways: Union County, despite its significant 37:14 positive to negative tweet sentiment ratio, does not remain consistent with the alternative hypothetical model. It ranks among the state's worst in county outbreak severity, despite its significant positive tweet sentiment count.

Warren County Twitter Sentiments

Negative Sentiment to Positive Sentiment Ratio: 37/41

Negative Sentiment to County Case Ratio: 37/543

Positive Sentiment to County Case Ratio: 41/543

Takeaways: Warren County, with a near 1:1 positive to negative tweet sentiment ratio, exhibits sentiment similarities similar to other large counties. Despite this, it ranks among the state's lowest in county outbreak severity.