This research will run a total of six models to contend with the mediums and two variants of political participation, using path analyses. Associations are tested for using path analyses.
Sampling and Data Collection
This research makes use of a dataset collected by Pew Internet & American life project (November 2010 Post-Election Tracking Survey, 2010), conducted by the Princeton Survey Associates International, and collected between November 3, 2010 and November 24, 2010 via a telephone and mobile phones nationwide survey. The survey was based on a sample of 2, 257 of ages 18 and older, which includes 755 mobile phones interviews, with the total margin of error within plus or minus 2.4% percentage points.
The sample was drawn via a combination of telephone and cellular phone random digital dialling (RDD) to represent the American population (Smith, 2011b). Samples were obtained from Survey Sampling International, LLC (SSI) according to PSRAI specifications and phone numbers were selected calculating for the probabilities of their share of active blocks. The cellular sample was drawn through a systematic sampling of dedicated and shared wireless blocks (Smith, 2011b). All samples were made available daily and kept active for 5 days. Each number was called at least 7 times during different times of the day and week. Half the time interviewers asked to speak to the youngest adult male, if not available then to the youngest adult female. For the other half of the time the opposite strategy was followed. In the case of the cellular phone sample, the person who answered was the respondent. Landline numbers had a 13.7% response rate and cellular phone numbers has a 15% response rate (Smith, 2011b).
Due to the intent of this research to specifically focus on online mediums, the above sample is filtered using the question from the dataset that asks “Do you use the internet, at
•‧
國
立立 政 治 大
㈻㊫學
•‧
N a tio na
l C h engchi U ni ve rs it y
least occasionally?” in order to base all findings on those who actively use the internet, in order to avoid a non-use bias. This filtered the total number of respondents to 1, 628 with a sampling error within plus or minus 2.8%.
Measures
All measures included here are those directly relevant to the model, while demographics and traditional media are controlled for.
Table 1 about here.
For reasons identified by previous authors above, demographic variables as summarised in table 1 above are considered important variables in any type of research involving political participation (McLeod et al., 1999; Putnam, 1995; Steinberger, 1984).
This research thus controls for demographics. The sample consists of 1, 628 respondents, age 18 and older (M = 44.66, SD = 16.86) with gender 44.8% male and women slightly over
represented at 55.2% female, 79% white, 8.2% Hispanics, and 17.9% people of other ethnicities. Education (M = 4.98, SD = 1.48, Range = 6) is a continuous variable on a 7-point scale that ranges from 0 = none or grade 1 - 8, to 7 = postgraduate training or
professional school after college and the mean equal to some college. Income (M = 5.36, SD
= 2.09) investigates the total family income in the last year, before tax reductions and ranges from 1 = less than $10,000 to 9 = $150,000 or more, with the mean being equal to $40,000 to
$50,000.
Previous research has also identified the strength of political ideology as an important predictor of participation (Cho et al., 2009; McLeod & Perse, 1994; McLeod et al., 1996;
Shah, Cho, et al., 2007; Shah, McLeod, et al., 2007). This research thus includes strength of political ideology (M = .66, SD = 2) operationalised as a 3 point continuous variable recoded to compute the respondents’ distance from neutral (0 = neutral, 2 = very liberal and
•‧
國
立立 政 治 大
㈻㊫學
•‧
N a tio na
l C h engchi U ni ve rs it y
conservative). The original question asks the respondent how they would describe their political views: In general, would you describe your political views as...
Due to the long history of traditional media influencing political participation, even an online world where they are considered to have faded to the background (Caparini, 2004), traditional media is considered a necessary control variable for predicting any association of online mediums to political participation. Traditional media is operationalised as a
combination of two variables combining television, radio, magazines, and newspaper news sources (M = 2.04, SD = .92) into a single variable. Respondents were asked: Overall, how have you been getting most of your news about this year’s campaigns and elections- from television, from newspapers, from radio, from magazines, or from the Internet? [Accept two answers].
Online political information use (M = 2.37, SD = 1.72, KR20 = .75) is operationalised
as a summative scale of 6 dichotomous items (0 = no, 1 = yes), asking respondents if they did any of the following in the months leading up to the election: Look for information online about candidates' voting records or positions on the issues? Watch video online about the candidates or the election Use the internet to research or 'fact check' claims made during the campaign Get news online? Look online for news or information about politics or the 2010 campaigns?
Facebook political information use (M = .46, SD = .94, KR20 = .73) is
operationalised as a summative scale based on 4 dichotomous items asking the respondents to think about what they did during the November 2010 elections (0 = no, 1 = yes): Get any campaign or candidate information on social networking sites? Sign up on a social networking site as a 'friend' of a candidate, or a group involved in the campaign such as a political party or interest group? Post content related to politics or the campaign on a social
•‧
networking site? Join a political group, or group supporting a cause on a social networking site?
Mobile phone political information use (M = .62, SD = 1.13, KR20 = .69) has been
operationalised as a summative scale constructed from 7 dichotomous items that ask the respondents (0 = no, 1 = yes) firstly whether they a mobile phones in the months leading up the elections to: Keep up with news related to the election or politics? Share photos or videos related to the election campaigns? Download or use any software applications or 'apps' that provide updates from a candidate or a group involved in the campaign such as a political party or interest group? And whether they used it on the day they you voted to: Inform others that you voted? Let others know about conditions at your voting location, such as delays, long lines, low turnout, or other problems? Did you happen to use your mobile phones to monitor the results of the election as they occurred, or did you not do this?
As already argued above, two potential mediators are investigated through exposure to the communication variables. Wider range of views exposure has been recoded into a dichotomous variable with 1 = yes 64.9% and 0 = about the same 35.1%. It asks the
respondent: Do you think that the internet exposes people to a wider range of political views than they can get in the traditional news media, or is most of the political information you can find online the same as what you can get elsewhere?
The second potential mediator is that of credibility. Credibility has been
operationalised into a dichotomous variable with 1 = easy 37.3% and 0 = difficult 62.7%, that asks the respondent: Thinking about the political information you find online, would you say it's usually easy or difficult for you to tell what is true from what is not true?
As pointed out earlier, this research has two dependent variables. Online political participation (M = .49, SD = 1, KR20 = .68) has been operationalised as a summative scale based on 6 dichotomous items (0 = no, 1 = yes). It asks the respondents if they did any of the
•‧
國
立立 政 治 大
㈻㊫學
•‧
N a tio na
l C h engchi U ni ve rs it y
following in month leading up to the election: Sign up online to receive updates about the campaign or the elections? Send email related to the campaign or the elections to friends, family members or others? Contribute money online to a candidate running for public office?
Use the internet to participate in volunteer activities related to the campaign – like getting lists of voters to call, or getting people to the polls? Take part in an online discussion, listserv or other online group forum like a blog related to political issues or the campaign?
Share photos, videos, or audio files online that relate to the campaign or the elections?
Offline political participation has been operationalised as a single dichotomous
variable that asks the respondent whether he or she voted (1 = 67.6%), or did not vote (0 = 32.4%) during the November 2010 elections and asks: A lot of people have been telling us they didn't get a chance to vote in the elections this year on November 2. How about you...
did things come up that kept you from voting, or did you happen to vote?
•‧
To test the hypothesis for this research, path analyses were applied. Independent variables included online political information use, Facebook political information use, and mobile phones political information use, which were all tested for in separate models due to their conceptual overlaps. These variables were used to test for direct and indirect
association, through the two expected mediators namely wider views exposure and credibility, and results were collected in 3 figures and two tables, which constitutes a total of 6 models.
Each model represents a different communications variable. Each model contained 4 blocks, with block 1 reflecting demographics, block 2 ideology, block 3 communication variables, and block 4 the two secondary orientations.
This section begins by mentioning some additional results that are noteworthy, but which are not tested for by the research questions. Table 1 shows that age is a consistent indicator for the role of online political information use (β = .075, p < .001), Facebook political information use (β = .140, p < .001), and mobile phone political information use (β
= .117, p < .001) in online political participation, which suggests that those who participation in online political participation are mostly older. Another finding for online political
participation hints towards a civic divide with regards to mobile phone political information use, with race (β = .072, p < .05), and education (β = .064, p < .05) also weakly significant.
This suggests that white educated and slightly older respondents are more likely to use mobile phones in online political participation.
The same is also seen in Table 2 for offline political participation, being very strong indicators in the model for online political information use (β = .927, p < .001), Facebook political use (β = .926, p < .001), and mobile phone political use (β = 1.003, p < .001), again indicating than an older respondent is more likely to participate in offline political
participation. Also relevant in offline political participation, is the consistent indicator of