The Durbin-Watson statistic in regression analysis

Statistics -- Watson-Durbin

Analysis of Why Social Networking Played

An Influence in the 2008 Presidential Election

The 2008 Presidential election forever changed the nature of campaigning, constituent management, fund raising, while also creating an entirely different platform for managing campaigns and getting new demographic segments to participate at a level never seen before. The 2008 U.S. Presidential election was the first to rely heavily on highly interactive, agile and quickly defined social networking strategies in an attempt to get those demographic segments who had voted the least in previous elections to participate. Barak Obama concentrated on the younger demographic segments through a concerted strategy of social networking that included sites on Facebook, MySpace, Twitter, YouTube and a multitude of microsites. John McCain stayed with more traditional strategies including lots of personal visits to states where electoral college votes would decide the election, Ohio being the focus of his and President Obama's attention the majority of the campaign. John McCain's social networking strategies relied more heavily on old school approaches to gaining votes and support, namely through face-time spent in pivotal states, e-mail blasts and the use of static websites. In the dichotomy of how each candidate managed their campaign is a representation of how new marketing practices are colliding with older approaches to selling. The new approaches of using social networking and social media to reach prospects and attain a level of trust with them is challenging the belief that face-time and much personal focus on actual customer interaction takes precedence over any electronic means of communication. The intent of this analysis is to evaluate how at least three dominant factors taken from the May 2008 Cloud computing, politics and adult social networking survey by Pew Research reflect how both presidential candidates varied in their approach to using social networking, and how social networking had a significant effect on President Obama winning the election.

II. Literature Review

During the height of the presidential campaign, both candidates continued to spend very heavily on all forms of media, often balancing expensive television time with a continual stream of updates and challenges to each other. Both had websites from the initial phases of the campaign, as did the majority of presidential candidates, all of which embraced the basics of social media rapidly, as this was shown in previous research to be correlated with the ability to reach multiple demographic segments quickly (Williams, Gulati, 2008).

Where the candidates differed however was in their approach for managing the more advanced aspects of social networking including Facebook and Twitter. One study specifically addresses the success rate of candidates using Facebook, MySpace and Twitter to better connect with voters, and is titled Social Networks in Political Campaigns: Facebook and Congressional Elections 2006, 2008 (Williams, Gulati, 2008). This study validates that websites are at a minimum required to be successful in managing campaigns, and that in the majority of cases (76%) those that win initial seats in the Senate or are successful in their re-election bids rely heavily on websites as a means to connect with voters and their constituents. There is a dividing line however between just using static and often push-based messaging that websites enable vs. using more interactive forms of communication. The rapid adoption of social networks has been driven by the design objectives of Web 2.0 (O'Reilly, 2006) as is shown in Figure 1: Design Objectives of Web 2.0 Technologies. The essence of these design objectives is to create applications that are much more focused on giving the user the opportunity interact how and through what means they choose to. Web 2.0 is the foundation of Facebook, Twitter and many other social networking sites that are all predicated on a very high level of interaction or as the mode says, the creation of an architecture of participation.

Figure 1: Design Objectives of Web 2.0 Technologies

Source: http://oreilly.com/web2/archive/what-is-web-20.html

The basis of this statistical analysis concentrates on whether the use of social networks that have Web 2.0 technologies as their catalyst had a statistically significant impact on one presidential candidate winning the election over another. The data set from the Pew Research Center from May, 2008, Cloud computing, politics and adult social networking, is used for the analysis.

III. Data and Method

The data set May, 2008, Cloud computing, politics and adult social networking from the Pew Internet and American Life Project is ideal for testing the hypothesis of whether Web 2.0- technologies and more specifically, social networking applications had a significant effect on the outcome of the 2008 U.S. Presidential elections. The data set has 47 different variables that span demographic, psychographic and attitudinal views of the Internet, social networks, and the use of social networks by candidates. The data set is based on a sample size of 2,251 adults that are aged 18 or over, and was conducted between the dates of April 4, 2008 and May 11, 2008. The sampling frame was the United States.

As there are 47 variables in this data set and the objective is to evaluate the effects of the three most relevant on the outcome of the U.S. presidential election. Using the parametric correlation functions of the Statistical Package for the Social Sciences (SPSS) Version 18 downloaded from the SPSS website, the following variables were seen as having the most significant impact on the outcome of the election. For a complete listing of all variables in the study and a longitudinal analysis of them, please see Appendix for a link to the zipped file of the questionnaire document, Spring Tracking Survey 2008. The link is provided out of convenience as the document is over 3,600 words long and 27 pages in length, and provides excellent longitudinal analysis of the questions asked in the survey.

Based on an analysis of all survey responses, the following three variables provided the most predictive accuracy as to which candidate would win the presidential election based on their use of social networking and the early adoption of Web 2.0 technologies for communicating and motivating voters to choose them. These three variables are as follows:

Q16. Have you ever created your own profile online that others can see, like on a social networking site like MySpace, Facebook or LinkedIn.com?

Q38. How many total profiles do you have online, counting all web sites?

Q39. Do you have a single profile on each site, or do you have multiple profiles on one web site?

Each of the three variables is now analyzed according to the specific statistical characteristics of each in the data set, followed by an analysis of the relationships within the data to the predictability of which presidential candidate would win the election. In the following section the hypotheses are also defined and tested through statistical analysis as well using SPSS Version 18.

From an analysis of the data, it is clear that when a presidential candidate chooses to use highly interactive media, they accidentally become some of the world's best evangelists for the technologies they are using. While President Obama had no intention of creating a groundswell of participation in the Internet in general and social networking sites specifically, this is exactly what he accomplished. The analysis later in terms of relationships shows how President Obama was able to create so much interest in the campaign that the majority of these on the survey saying they would vote for him for President had been on the Internet well under two years, with a significant number using the Internet for less than a year. Not only did this accidental evangelism lead to the recruitment of a new generation of Internet users, it also significantly increased the use of social networks as well. While McCain's supporters were evenly matched on the use of social media as consumers of Web 2.0-based and social networking-based content, Obama's supporters quickly emerged as creators of content, and this is seen in the three variables analyzed in this section. The first variable illustrates this dynamic of one candidate becoming the accidental evangelist for social networking site and Internet use:

Q16. Have you ever created your own profile online that others can see, like on a social networking site like MySpace, Facebook or LinkedIn.com?

The mean value of 1.74, median value of 2 support the finding that of those that did fill out profiles on social networks, they were exceptionally polarized in their views of the candidates. Only 29.6% of the respondents have a profile on a social networking site, 70.5% do not. Yet this 26.9% are the most vocal and focused in their views and discussion of the issues relative to the 70.5% that do not have profiles created. Keeping with the finding that President Obama excelled as an evangelist for social networks, the next question Q38. How many total profiles do you have online, counting all web sites? The following figure illustrates graphically the distribution of responses to this question. The mean value of this variable is 2.32, and median is 2.0. 80% of respondents have three or more profiles, and when this variable is controlled by number of years online by presidential candidate preference, it is clear that many profiles were created to see how Facebook and Twitter were being used during the campaign. Again, the presidential campaign acted as a catalyst for technology adoption, and led to first-time social networking users getting online. It also led to existing users creating additional profiles as well.

On the third most significant variable in determining the effects of social networks on which candidate would eventually win Q39. Do you have a single profile on each site, or do you have multiple profiles on one web site? Provides the most fascinating data from the three variables included in the analysis. Notice that the distribution of results is bifurcated from one spectrum to the other. This signals that through the use of the participative aspects of social networks, candidates were able to get the support of the top 5% of social networkers, as evidenced by having more than one profile on a given network. The mean value of this variable is 1.17 and it has a mode of 1, which again shows that the presidential campaign had the impact of creating higher adoption of social networks, with those early adopters being Obama voters.

When all three variables are taken together it is apparent that Obama's use of Web 2.0 technologies and social networks significantly increased the adoption of these technologies and websites. The data also shows that Obama, by using social networking as a core part of his strategy, adopted early adopters of other technologies as well. The use of cell phones for testing is an example

Analysis of relationships and their role in getting texts from the candidates shows that for variables 30B, 30C and 30D Obama dominates the cross-tabulation tables for the analysis. Literally, no one received text messages in any of the categories reflected by questions 30B, 30C and 30D from John McCain. This dramatically underscores how effective Web 2.0 technologies are as a means of connecting with and staying in contact with voters over a campaign.

Relationships in the Data

Appendix B provides a correlation analysis of the variables Q16, Q38, Q39 and Q41. The results show that at the .01 level of confidence that Q16 is the leading indicator of whether a respondent is pre-disposed to vote for Obama vs. McCain. Across the demographic factors and use of technologies for conventional tasks, there is no statistically significant difference between Obama and McCain voters at either the .01 or .05 levels. Only in the areas of social networking participation is the significant most prevalent and where there is a high level of statistical significance to the use of social networks for campaigning and the development of a voter base. The use of social networks as a catalyst for creating more participating online has been proven to be predictive of a candidates' ability to win an election based on this analysis.

Using regression analysis and Analysis of Variance (ANOVA), it becomes clear that the greater the level of familiarity or mastery of social networks, the greater the tendency to align with Obama's social media strategy. This is evident in the ANOVA completed that takes VOTE01 (Who the respondent would vote for) and completing regression and ANOVA analysis of variables Q16, Q38 and Q39. The following tables and scatterplots illustrate this point. What is also significant about this finding is that Obama, acting as an evangelist for social networks by accident actually galvanized the early adopters to create profiles for the first time, while also attracting the most experienced users of social networks who had multiple profiles on the same site.