Fracking (Hydraulic Fracturing) 4 of 7

Last spring I did a pretty thorough look into fracking so here a copy. Necessarily due to the length it is broken into multiple parts. Here are the links for all parts: Fracking: 1, 2, 3, 4, 5, 6, 7, Citations.

Methods & Results I

The idea behind this experiment was to examine the literature in general in regards to fracking. The design was inspired by the research of Dr. David Sachsman (Sandman et al, 1972). The actual methods were greatly modified but the idea is similar in that the purpose is to examine and take a snapshot of the current state of the fracking in the literature and to identify any trends that might be apparent.

To collect data, the University of Tennessee at Chattanooga’s Lupton Library’s website was used to query 37 different literature databases (listed below) for more details regarding what the journals cover, please see their websites. Combined there are 10’s of millions if not 100’s of millions of individual items that were queried. Here is a url with more detail and links to information on the individual databases that will be left up for a while

 

ABI/INFORM Complete

Academic OneFile

ACM Digital Library

Annual Reviews

BioMed Central

Business and Industry

Business and Management Practices

Business Insights: Essentials

CQ Weekly

Early English Books Online (Digital only)

Education Abstracts

Emerald Group Publishing Limited

Expanded Academic ASAP

Gale Virtual Reference Library

General OneFile

General Reference Center Gold

General Science Abstracts

Humanities Abstracts

IEEE Publications Database

Latino Literature

M.E. Sharpe Journals

MEDLINE

MLA International Bibliography

NCJRS Abstracts Database

OAIster

Oxford Journals

Oxford Reference Online

Project Muse

ProQuest Biology Journals

PsycARTICLES

Readers’ Guide Abstracts

ScienceDirect

Snapshots

Social Sciences Abstracts

Taylor and Francis Journals

Wilson Business Abstracts

WorldCat.org

 

 

Each of these databases was queried simultaneously for each of 14 individual Boolean searches. Each of these searches were then broken into two groups: all results and peer-reviewed. This dataset provides a look at what is occurring in the world of fracking .The following are the key word(s) used in the searches:

 

Fracking

Fracking Benefit

Fracking Chemical

Fracking Court

Fracking Economy

Fracking Environment

Fracking Hazard

Fracking Leak

Fracking Policy

Fracking Pollution

Fracking Problem

Fracking Regulation

Fracking Risk

Fracking Rule

 

 

Each of these 28 data sets had articles sorted by year published (1-1-1994 to 12-31-2013) and the number of articles per year for each of the 20 years. 2014 was not included because it would not have contained a full year. The full data is located in appendix 2. A keyword with “PR” at the end indicates the peer-reviewed subgroup. The data sets “Fracking” and “FrackingPR” were then scatter plotted against the year (Graph 1.), and polynomial regression equations of best fit and  values were calculated and included:

Graph 1. Scatter plot of “Fracking” and “FrackingPR” data sets vs “Year” from Table 1.
Graph 1. Scatter plot of “Fracking” and “FrackingPR” data sets vs “Year” from Table 1.

Fracking

y = 0.0265×5 – 264.64×4 + 1E+06×3 – 2E+09×2 + 2E+12x – 8E+14

R² = 0.9838

 

 

FrackingPR

y = 3E-05×6 – 0.409×5 + 2038.2×4 – 5E+06×3 + 8E+09×2 – 6E+12x + 2E+15

R² = 0.9905

Graph 2.  Scatter plot of data set “Fracking” vs “Year” from Table 1. The numbers above the dots indicate the raw number of articles.
Graph 2. Scatter plot of data set “Fracking” vs “Year” from Table 1. The numbers above the dots indicate the raw number of articles.
Scatter plot of data set “FrackingPR” vs “Year” from Table 1. The numbers above the dots indicate the raw number of articles.
Graph 3. Scatter plot of data set “FrackingPR” vs “Year” from Table 1. The numbers above the dots indicate the raw number of articles.

Following this initial graph, two more scatterplots of the total articles and of the peer-reviewed articles were generated independently (Graph 2. and Graph 3.) and the article counts per year were marked on the scatter plots.

After these first three charts, the charts and data were examined for any apparent trends or variations. The first trend noticed is that the total articles per year were negligible prior to 2009, which had 67 (almost double the cumulative total of the previous 15 years). This was similar to what was also observed within the peer-reviewed literature data set, except the increase started in 2010 instead of 2009. From 2009 (2010 for peer reviewed) onward, the general trend was that there were more articles produced per year as the year approached 2013. The year 2013 had the largest quantity of literature produced within a one year period for both categories. Next, the data in the raw data chart was examined & summed for the total number of categories per year with results greater than zero. In 2008, there was one of the 26 remaining categories which contained results. 2009 had 8 categories with results, 2010 had 16 categories, and 2011 had results greater than 0 in all categories. This expansion of results into the subcategories is indicative that more research in more detailed topics -depth and breadth of information- was being examined. After this the statistical program R was utilized to generate 28 independent scatterplots of the individual subcategories for the years 2009-2013 against the year. These graphs are located in appendix 1.

These 28 scatterplots were then individually examined for interesting trends. The first trend observed was that generally speaking, the categories increased in number of articles each year. This was the expected observation. However, only 20 of the 28 five year plots followed this trend. The plots for seven of the eight remaining subsets (Chemical vs Year, Court vs Year, CourtPR vs Year, Hazard vs Year, LeakPR vs Year, PollutionPR vs Year, and RulePR vs Year) all trended up from 2009 to 2012, but in 2013 they showed a decrease in the number of articles. The last remaining scatterplot (ChemicalPR vs Year) increased in the number of articles from 2009 to 2010 and 2010 to 2011, but remained constant from 2011 to 2013. After examining the data three multivariate analysis of variance (MANOVA) were conducted. In brief, a MANOVA is a statistical test used to determine if there exists a difference within groups of data. The first MANOVA was done comparing all subsets within the year categories, checking to determine if a sub category had a significant change in the number of articles from one year to the next (p-values 1.). The second chart compared the 26 subsets changes to the changes in the “Fracking” set (p-values 2.). This effectively shows which growth curves vary significantly from the fracking curve. Finally, the third MANOVA combined the year and the “Fracking” data set curve against the 26 sub categories (p-values 3.) This shows which subsets had significant differences when both “fracking” article counts and the year were considered simultaneously.

p-values 1. Significant Change in Number of articles between years 2009, 2010, 2011, 2012, 2013 are highlighted in yellow.
p-values 1. Significant Change in Number of articles between years 2009, 2010, 2011, 2012, 2013 are highlighted in yellow.
p-values 2. Significant difference when compared to Fracking Articles in general is highlighted in yellow.
p-values 2. Significant difference when compared to Fracking Articles in general is highlighted in yellow.
p-values 3. When comparing individual subcategories to both fracking articles in general and the individual year of the articles following chart of p-values is generated significant values are highlighted in yellow.
p-values 3. When comparing individual subcategories to both fracking articles in general and the individual year of the articles following chart of p-values is generated significant values are highlighted in yellow.