Technological Advancements and the Electoral Connection: How do citizens want their representatives to keep in touch with them?
Conor M. Dowling Yale University Institution for Social and Policy Studies 77 Prospect Street, PO Box 208209 New Haven, CT 06520-8209
[email protected] (203) 432.4811 (voice) (203) 432.3296 (fax)
October 15, 2011
Working Paper Analysis to be replicated using a nationally representative sample as part of the 2011 Cooperative Congressional Election Study.
Technological Advancements and the Electoral Connection: How do citizens want their representatives to keep in touch with them? Abstract Social media and other technological advancements have changed the way people interact with one another. Advances in communications technology can also change the relationship between representatives and their constituents, as they make it possible for representatives to keep in touch with their constituents in numerous ways on a regular basis. But do their constituents want them to? I present evidence from pilot survey data that people think representatives should communicate with their constituents in a variety of ways, but that web site updates and email are the most preferred methods of contact. I also present results from a survey experiment that shows people respond favorably to representatives who have embraced new technology as a means to inform their constituents of their activities. Taken together, these findings, which are currently being replicated with a nationally representative sample, speak to the broad issue of whether advances in communications technology are altering the “electoral connection.”
1
Social media (Facebook, Twitter, etc.) and other technological advancements (smartphones, tablets, etc.) have changed the way people interact with one another.1 New communications technology has the ability to change the way politics is conducted too (Evans and Oleszek 2002; Johnson 2004; Lipinski 2004). Some scholars, for example, have suggested that social media and more generally the Internet will result in greater political participation levels (e.g., Garcia-Castañon, Rank, and Barreto 2011; Tolbert and McNeal 2003). The impact of social media can also be seen in its utility for helping to coordinate political activity, as evidenced by the recent “Occupy Wall Street” protests (Karimi and Sterling 2011). Advances in communications technology, such as social media, can also change the relationship between representatives and their constituents. Recent technological advances have made it possible for representatives to “connect” with their constituents in numerous ways on a regular basis. But are representatives using new technology to do so? And do their constituents want them to? This paper takes a first step at addressing these questions. In so doing, it speaks to the broad issue of whether advances in communications technology are altering the “electoral connection” (Mayhew 1974; also see Fenno 1978; and Lassen and Brown 2010 for a recent review of this literature), as many academics, pundits, and politicians have suggested they have and will continue to do so (e.g., Croal 2009; Hansen 2009; Johnson 2004; Rhoads 2009; Smith 2010). In this paper, I present the findings from a pilot study that included unique survey items that focused on how people want their representatives to communicate with them. The study also included a survey experiment that was designed to see if people would reward a representative for using new technology to stay in touch with constituents. The findings provide three main 1
See, for example, the Pew Internet and American Life Project (http://www.pewinternet.org). 2
contributions to our understanding of the way in which advances in communications technology may influence how elected representatives connect with their constituents. First, I find that respondents think representatives should communicate with their constituents in a number of different ways. I also find evidence, however, that web site updates and email are the most preferred methods of contact, and methods that respondents think representatives should be using on a regular basis. For instance, I find that over 50% of respondents think a representative’s web site should be updated once a week or more. Second, I also find substantial variation within a communication method in terms of how often respondents think representatives should be using it to communicate with their constituents. This suggests there is not one preferred method of communication for constituents. Third, the survey experiment documents that respondents do respond more favorably to representatives who embrace the use of new technology to communicate with their constituents, suggesting representatives have an (electoral) incentive to adopt new technology. Recent work has already begun addressing several questions related to the use of new communications technology by politicians. For example, Druckman, Kifer, Parkin, and colleagues have investigated many aspects of the role candidate web sites play in election campaigns, from their adoption (Druckman, Kifer, and Parkin 2007) and what issues they engage (Druckman et al. 2010), to how incumbents and challengers differ in their use (Druckman, Kifer, and Parkin 2009), and whether candidates “go negative” more or less often on the web (Druckman, Kifer, and Parkin 2010). In a similar vein, other researchers have attempted to explain when and why representatives adopt the use of even newer social media technology, such as Twitter (Lassen and Brown 2010). I take a different approach in this paper. Rather than focusing solely on what representatives do, I also ask what their constituents want from them,
3
with the aim of elucidating on what form(s) constituents would like their connection to their representative to take. The remainder of this paper proceeds as follows. In the next section I describe my survey and experimental research design. I then present the results from the survey, followed by the survey experiment. I conclude with a brief discussion of some of the implications of the results, the limitations of the present study, and directions for future research.
1. Survey and Experimental Research Design I conducted a pilot survey on Amazon.com’s Mechanical Turk (MTurk) interface, an online platform for recruiting and paying subjects to perform tasks. (See Buhrmester, Kwang, and Gosling 2011 for a discussion of using MTurk to recruit participants for experiments.) The MTurk population is a convenience sample that appears more representative than student samples, but is not completely representative of the U.S. population. An MTurk sample is typically younger, less likely to own a home, and more likely to report no religious affiliation (Berkinsky, Huber, and Lenz 2010). A total of 1600 participants were recruited to take the survey.2
2
The survey was fielded from 7/19/2011 to 8/7/2011. Respondents were paid $0.50 to participate. The
text of the MTurk request read: “This survey will ask you a series of questions about you and your feelings about a variety of topics. The survey is here: [URL]. Once you finish the survey you will be provided with a code. To get paid, please enter the code below and click ‘Submit’. DO NOT CLOSE THIS WINDOW WHILE YOU ARE TAKING THE SURVEY. Payment is auto-approved in 5 days.” Only U.S. residents were permitted to take the survey.
4
After completing an informed consent page, all participants completed a few demographic questions and other items (listed in Panel A of Table 1). Toward the end of the survey, participants were randomly assigned to either (1) answer several questions about the ways in which representatives communicate with constituents (37.5% of the total sample), (2) take a survey experiment related to the ways in which representatives communicate with constituents (25% of the total sample), or (3) take a separate survey experiment not reported in this paper (37.5% of the total sample). Table 1 outlines the survey and experimental research design, including relevant question wording. The Appendix includes complete question wording for all survey items not reported in Table 1 or discussed in the text. [Table 1 about here] 1.1. Opinion Questions The participants who were assigned to answer the survey questions responded to three sets of items, described in the Panel B of Table 1. First, they were asked how often they thought members of Congress should communicate with their constituents in each of five ways: (1) “send a letter to constituents’ homes,” (2) “hold a town hall meeting,” (3) “send an email to their constituents,” (4) “provide an update on their web site,” and (5) “provide an update on the social networking web site, Facebook.” Participants chose from six response options that ranged from “do not need to do this” to “should do this more than once a week.” I collapse the top two response categories (“should do this once a week” and “should do this more than once a week”) for presentational simplicity and because there were not many responses in those two categories. Second, participants were asked how many times their district’s U.S. House member had communicated via the same five means as they were asked in the previous question (i.e., letter, town hall, email, web site, and Facebook) since the beginning of the year (January 1, 2011).
5
Again, the six response options ranged from “has not done this” to “more than once a week,” and I again collapsed the top two response categories. Third, I asked participants to rank the five communication methods from one to five, where one represented their “most preferred way to learn about [their] representative's activities” and five their “least preferred.” Note that this question differs from the first set of items by focusing on the participant’s own member of Congress, rather than the generic form in which the first set of items was asked. These three sets of items allow me to not only see what methods of communication members of Congress are using, but also compare that to what their constituents would like them to use. To the extent that there are any important differences between what representatives are doing and what their constituents would like, it suggests representatives may not be staying connected with their constituents in the best manner to fulfill their (reelection) goals. 1.2. Survey Experiment A separate set of participants was assigned to an embedded experiment, described in Panel C of Table 1. Participants were asked to read an experimental vignette designed to look like a newspaper story about the use of technology by Congress.3 This vignette was modeled after an actual story that appeared on June 17, 2011 on thehill.com, a web site devoted to covering Capitol Hill (Siegelbaum 2011). The short vignette that participants read, titled “Congress and New Technology,” began with the sentence “One big shift for Congress this year 3
Specifically, participants were told: “We would like to ask you to read part of a story that we found in a
newspaper. After you read it, we will ask you some questions. We have changed the name of the member of Congress in the story because we want you to answer these questions based on what you read, rather than on anything else you might know about the member of Congress.”
6
has been the ability to use iPads on the floor. But this change has been met with different responses by members of Congress.” The remainder of the vignette described one particular member of Congress’s (with an altered name) response to this change. The experimental manipulations followed a two-by-two design. Half of the participants were assigned to read about a female member of Congress (“Sarah Jones”), the other half a male (“Michael Jones”). The other manipulation was the member’s response to the use of iPads on the floor. Half of the participants were randomly assigned to the iPad treatment condition and read the following: Representative [Michael Jones / Sarah Jones] recently announced the release of an official mobile application for the iPhone, iPad and iPod Touch that allows [his / her] constituents to find [his / her] position on hot topics of the day, and easily access [his / her] contact information, news articles, e-newsletters, videos, and photos, the first application to be offered by a member of Congress.4 The other half of the participants were randomly assigned to the no iPad treatment condition and read the following: Representative [Michael Jones / Sarah Jones] does not use an iPad on the floor, and when asked recently stated, “I wouldn’t even know how to begin using an iPad.” These two treatments are meant to highlight the difference between a member of Congress who is attempting to connect with her or his constituents via the latest technology versus one who is not. Three items appeared on the page after the newspaper story asking the participant to evaluate the representative on scales from zero to ten. One asked the participant to rate the representative’s job performance (“How would you rate the job the representative is doing?” 4
The actual member of Congress who was the first to do this, according to the actual article, was Senator
Saxby Chambliss (R - GA).
7
[“poor” to “excellent”]). Another item asked the participant to rate the representative on a personal level (“How do you feel about the representative as a person?” [“negative” to “positive”]). The last item asked the participant to rate how effective the representative was at communicating with his or her constituents (“Do you think the representative does an effective job communicating with their constituents? [“very ineffective” to “very effective”]). Finally, participants in the experiment sample were asked standard job approval rating questions at the very end of the survey for President Obama, the U.S. Congress, and the U.S. Supreme Court. These three items appeared in a grid format with five response options— “strongly approve,” “somewhat approve,” “not sure,” “somewhat disapprove,” and “strongly disapprove.” These questions were asked to see if the treatment would spill over and affect broader evaluations of these institutions.
2. Results of the Opinion Questions In this section and the next, I present the findings from this pilot survey. Please note that any population estimates reported in this section should be treated with caution because of the non-representative nature of the MTurk sample. These items will be included on the 2011 Cooperative Congressional Election Study (CCES), a nationally representative opt-in Internet survey, to obtain more accurate population estimates. On the other hand, the results of the survey experiment (presented in the next section) are likely valid in other (survey) contexts as well, given the ability of researchers to replicate prior experiments using the Mturk population (Berinsky, Huber, and Lenz 2010). Appendix Table A1 presents summary statistics for both the opinion (column [1]) and experiment (column [2]) samples, as well as for each of the experimental treatment conditions.
8
2.1. What Methods Do Members of Congress Use to Communicate with their Constituents? I begin by presenting responses to the items asking participants to report what their member of the House had done to communicate with them since the beginning of the calendar year. Figure 1 presents these responses in graphical form (and Panel A of Appendix Table A2 reports the percentages). The bars are clustered by communication method, increasing in the amount of reported communication from left (lighter bars) to right (darker bars) within clusters. [Figure 1 about here] Two findings from Figure 1 are of particular note. First, most people reported either receiving very little communication from their House member since the beginning of the calendar year (a span of approximately 7 months) or that they did not know. The most common communication method respondents reported was a web site update (26.2% of respondents reported at least one web site update), but still 73.8% of respondents reported that their House member had not updated their web site (16.8%) or that they did not know if their House member had (57.0%). It is likely that during an active election year in which members of Congress engage in greater communication efforts and citizens pay more attention to politics more respondents would be aware of their representatives’ communication and their representative would engage in more communication with their constituents as well. In addition, the high percentage of respondents who report no Facebook updates from their House member (82.5%) is not because participants were not Facebook users. Even among those participants who report that they had been on Facebook on one or more days in the last week, 80.6% still report no Facebook updates from their House member (analysis available upon request). In future work I intend to investigate what percentage of Facebook users who have “liked” a member of Congress pay attention to that member’s updates. This will allow me to see
9
what percentage of the subset of people (i.e., active Facebook users) who might plausibly be able to learn about their representative’s activities on Facebook, actually utilize this format to obtain information. The second finding from Figure 1 that is of particular note is that the communication methods that involve newer technology (email, web site, and Facebook) were reported to have occurred more frequently than the other communication methods. Only 6.7% of respondents said their House member had sent a letter once a month or more often and 9.6% reported a town hall meeting had been held that often.5 Conversely, 14.6, 21.2, and 15.0% of respondents reported that their House member had sent an email, updated their web site, and updated Facebook, respectively, once a month or more often. Given that the lure of new technology is the ability to stay in contact with others on a more regular basis, this difference makes sense. 2.2. How Do People Want Their Members of Congress to Communicate with Them? But what do people think members of Congress should be doing to communicate with their constituents? Figure 2 presents responses to the items asking participants how they think members of Congress “should” communicate with their constituents in graphical form (and Panel B of Appendix Table A2 reports the percentages). The bars are clustered by communication method, increasing in the amount of desired communication from left (lighter bars) to right (darker bars) within clusters. [Figure 2 about here] Three aspects of Figure 2 stand out. First, 26.8% of respondents said sending a letter to their home was unnecessary. This is 10 percentage points more than the next highest category 5
This estimate is likely high, perhaps inflated by people reporting town hall meetings held by different
politicians, not those held by a single politician.
10
that respondents said was unnecessary. 15.8% of respondents said providing an update on Facebook was unnecessary (this figure only falls to 11.9% when restricted to those participants who report that they had been on Facebook on one or more days in the last week). While this does not suggest that sending letters to constituents’ homes should be considered a thing of the past (especially given the nature of the sample—active Internet users), it does indicate that many people say they prefer other methods, even town hall meetings (perhaps because they are found to be more informative) to receiving a letter in the mail. Still, the vast majority of people (59%) think that members of Congress should be communicating via all these methods, although they do think that some should be used more than others. Second, people think members of Congress should communicate via “new” technology (email, their web site, and Facebook) more frequently than “old” technology (letters sent to the home and town hall meetings). This is consistent with the pattern of responses (but not the percentages) for what their House member had done (see Figure 1). The modal response both for sending letters and holding town hall meetings was once or twice a year, while for email the modal response was once a month, and for updating both their web site and Facebook the modal response was once a week or more. In short, there is an expectation that members of Congress should keep in touch via new technology on a regular basis. The question that asked participants to rank “the various ways in which your member of Congress can communicate with you” results in a slightly different pattern of results. Figure 3 displays the mean rankings to this question for the five communication methods, with the whiskers indicating 95% confidence intervals. (Panel C of Appendix Table A2 reports the percentages for each communication method for each ranking, one through five.) The figure shows that web site updates are clearly the most preferred communication method, with email
11
second, town hall meetings third, followed by Facebook updates and sending letters. Taken together, the results presented in Figures 2 and 3 suggest that web site updates and email are currently the most preferred methods of contact, and methods that people think representatives should be using on a regular basis. [Figure 3 about here] The third aspect of Figure 2 that stands out is that in addition to variation across communication methods, there is also a great deal of variation within communication methods in terms of what people think members of Congress should do. For example, while most people (36.2%) think members of Congress should email their constituents once a month, an additional one-fifth think (20.6%) they should do so two or three times a month, and 18.3% think they should do so once a week. At the same time, more than 10% think that members of Congress either do not need to email constituents at all (10.8%) or only once or twice a year (14.1%). What accounts for these differences in beliefs about how members of Congress should communicate with their constituents? 2.3. What Explains How People Want Their Members of Congress to Communicate with Them? I conducted an exploratory analysis as a first attempt to address this question. I estimated separate OLS regression models for each outcome (communication method) using demographic (age, age-squared, gender, race, and education) and technology (hours spent on the Internet per day and days per week on Facebook) characteristics of the individual as predictors. These regressions are presented in the odd-numbered columns in Table 2. The even–numbered columns in Table 2 add measures of political interest (standardized, M=0, SD=1) and party identification (-3=strong Republican to 3=strong Democrat) to the original specifications.
12
[Table 2 about here] Three aspects of this exploratory analysis are worth highlighting. First, in terms of demographic characteristics, age has a U-shaped relationship with sending letters (columns [1] and [2]) and an inverse U-shaped relationship with updating Facebook (columns [9] and [10]). These results suggest that with increases in age, people initially prefer updates to Facebook over letters, but this relationship reverses as age continues to increase (i.e., the oldest people in the sample prefer letters to Facebook).6 Interestingly, holding all else constant, female respondents reported that members of Congress should engage in all five of the activities at lower rates than male respondents. This relationship is statistically significant for four of the five communication methods, with Facebook being the exception. Increases in education are also associated with decreases in the amount of communicating via letters (columns [1] and [2]), town hall meetings (columns [3] and [4]), and email (columns [5] and [6]) respondents think members of Congress should do. However, education is not statistically significantly associated with web site (columns [7] and [8]) and Facebook updates (columns [9] and [10]). Second, the two measures of technology characteristics of the individual are, for the most part, unrelated to beliefs in how members of Congress should communicate with their constituents. The number of hours a participant reports spending on the Internet per day is not statistically significantly associated with beliefs about any of the five communication methods. The days per week a participant reports using Facebook, as one might expect, is associated with the frequency with which a respondent thinks a member of Congress should update Facebook (columns [9] and [10]), but is not statistically significantly associated with beliefs about any of the other four communication methods. 6
It is important to keep in mind that approximately 90% of this sample is between the ages of 18 and 50. 13
Third, and perhaps most interestingly, political characteristics of the individual are associated with the frequency with which people believe their members of Congress should communicate with their constituents. This is the case even though I control for the other demographic and technology characteristics. As political interest increases, people think members of Congress should update web sites (column [8]) and Facebook (column [10]) more often (p<.01 for each, two-tailed: all p-values reported in the text are two-tailed). Increases in political interest are also associated with increases in the frequency with which respondents think members of Congress should hold town hall meetings (column [4], p=.060) and send email (column [6], p=.052). In addition, partisan affiliation is positively associated with how often respondents think members of Congress should engage in each of the communication methods. This association is particularly strong for sending letters (column [2], p<.01), holding town hall meetings (column [4], p<.05), and updating Facebook (column [10], p<.05), and less so for sending email (column [6], p=.097) and updating a web site (column [8], p=.424). This suggests that Democrats would like members of Congress to communicate with them more often than Republicans and Independents do. 2.4. Summary of Findings from Opinion Questions All of the preceding analyses will be replicated with a nationally representative sample. Nevertheless, the findings suggest that many individuals want their representatives to embrace new technology and use it to communicate with their constituents. Web site updates and email appear to be the most preferred method of contact (see Figure 3), and people expect their representative to use these two methods to communicate with their constituents on a fairly regular basis (see Figure 2). Additionally, the results presented in Table 2 suggest that certain types of individuals prefer certain communication methods more than others.
14
These findings raise the question of just how much people want their representatives to embrace and use new communication methods. Do they reward (i.e., think more highly of) representatives for doing so? Do some people reward more than others? In the next section I present the results of the survey experiment that are a first attempt to address this question.
3. Results of the Survey Experiment In this section I present the results of a survey experiment that was designed to see if manipulating a member of Congress’s use of new technology would result in that member of Congress being evaluated more favorably. I restrict analysis of the survey experiment to the 308 respondents who provided responses to all of the items used in the analysis presented below. I first tested for balance across treatment conditions using a multinomial logit model with a nominal experimental treatment condition variable as the outcome. The covariates were gender, age, race, education, income, income missing, political interest, party identification, hours spent on the Internet per day, and days per week on Facebook. (Columns [3]-[6] of Appendix Table A1 report summary statistics by treatment condition.) This analysis did not identify any imbalance for any of the covariates (the lowest p-value was for age: chi-square=6.62; p=.085). To predict each outcome measure (job performance, personal evaluation, and effective communication), I include indicators for three of the four conditions generated by the two-bytwo study design. “Male, iPad” is set as the omitted, reference category. The results of these OLS regressions are presented in the odd-numbered columns of Table 3. (I also present, in Appendix Table A3, the means and standard deviations for the outcome measures for each treatment condition.) The results presented in the even-numbered columns of Table 3 include controls for demographic and other pre-treatment measures of respondent characteristics to reduce the
15
standard errors. Specifically, I include the same set of covariates used to check for balance across treatment conditions (see above). As the results are consistent between the odd- and evennumbered columns, I focus discussion on the even-numbered columns. [Table 3 about here] Column (2) of Table 3 displays the results for Job Performance, which ranges from 0 (“poor”) to 10 (“excellent”). Compared to a male member of Congress described as saying that he “wouldn’t even know how to begin using an iPad” (i.e., the omitted category), both the female and male member of Congress who are described as embracing new technology (“Female, iPad” and “Male, iPad”) are evaluated more favorably on this measure (p<.01 for both). In addition, both the “Female, iPad” and “Male, iPad” treatment conditions are significantly different from the “Female, no iPad” treatment condition (p<.01 for both). There is no statistically significant difference between either a female described as embracing new technology and a male who is (“Female, iPad” v. “Male, iPad”, p=.364) or a female described as not embracing new technology and a male who is not (“Female, no iPad” v. “Male, no iPad”, p=.405). This suggests that neither female nor male members of Congress are rewarded (punished) more than the other for embracing (failing to embrace) new technology.7 Substantively, the coefficient of 1.291 for the “Female, iPad” treatment condition represents a shift of approximately 60% of a standard deviation of the job performance measure (see column [1] of Appendix Table A3 for means and standard deviations of the outcome measures in the full sample). In addition, the .951 effect of the “Male, iPad” treatment condition 7
This is also the case for the other two measures—Personal Evaluation and Effective Communication.
Tests of the equality of coefficients in columns (4) and (6) of Table 3 reveal no statistically significant differences between either “Female, iPad” and “Male, iPad” or “Female, no iPad” and “Male, no iPad.”
16
represents a shift of a little more than one half of a standard deviation. These are rather sizable shifts for a simple treatment in which a member of Congress is described as embracing a technological innovation in order to communicate with constituents. Analysis of personal evaluations of the representative, which ranges from 0 (“negative”) to 10 (“positive”), is presented in columns (3) and (4) of Table 3. The pattern of results for this personal evaluation is the same as it was for the job performance measure. A member of Congress described as embracing the technological innovation is thought better of “as a person” than a member of Congress who is described as technologically inept.8 And, once again, the effects are rather sizable. The 1.023 (.733) effect of “Female, iPad” (“Male, iPad”) is approximately half (one-third) of a standard deviation of the personal evaluation measure. Finally, the results presented in columns (5) and (6) indicate that participants viewed a member of Congress who embraced new technology as someone who does a more “effective job communicating with their constituents.” Moreover, these effects are even larger than for the other two measures. The effect of embracing new technology, whether a male (coefficient = 2.285) or female (coefficient = 2.511) member of Congress, is approximately one full standard deviation of the Effective Communication measure. This represents a sizable shift in evaluations of the member of Congress. The results of this survey experiment suggest that people respond to a representative’s use of new technology. They think better of their representative in terms of the job he or she is doing, as a person, and in terms of his or her effectiveness at communicating with constituents.
8
p<.01 for “Female, iPad” v. “Male, no iPad,” “Female, iPad” v. “Female, no iPad,” and “Male, iPad” v.
“Female, no iPad.” p<.05 for “Male, iPad” v. “Male, no iPad.”
17
These positive effects do not, however, carry over to having more positive evaluations of Congress as an institution or approval of other governmental actors (see Appendix Table A4). 3.1. Heterogeneous Treatment Effects Do some people evaluate members of Congress who embrace new technology more favorably than others? I investigate two likely possibilities for such heterogeneous treatment effects. The results presented in Table 2 suggest that political interest may be a likely source of heterogeneity of treatment effects because it is associated with thinking members of Congress should update web sites and Facebook more often. However, I did not find this to be the case. Compared to participants who said they were “somewhat” interested in politics, participants who said they were “very interested” in politics did not respond more favorably to the “iPad” treatment.9 Another possibility is that people who are more frequent social media users responded more favorably to the treatment. The primary measure in the data to capture social media use is the frequency with which participants report visiting Facebook. Approximately half of the participants reported visiting Facebook everyday. I created an indicator variable in which these individuals were scored one and all others were scored zero. I then interacted this variable with a treatment indicator for the iPad treatment condition, scored one for respondents who were assigned to this treatment condition. (Because the results presented in Table 3 suggest there was no interactive effect of the iPad treatment condition with the gender manipulation, I collapse the gender conditions.) Table 4 presents the results of OLS regression models in which each outcome measure was separately regressed on the iPad treatment indicator, the visit Facebook 9
Results available upon request. Only 22 participants (of 308) said they were “not at all” interested in
politics. They are excluded from this analysis.
18
everyday indicator, and the interaction between the two (in the even-numbered columns), along with the same set of control variables used in Table 3. [Table 4 about here] The odd-numbered columns in Table 4, which do not include the interaction variable, show the positive effect the iPad treatment condition has on each outcome measure. These columns also show that, compared to people who do not visit Facebook everyday, people who visit Facebook everyday do not evaluate the member of Congress described in the newspaper story significantly more favorably on any of the three outcome measures. Turning to the evennumbered columns, I find that the interaction between the iPad treatment condition and the indicator for participants who visit Facebook everyday is in the expected positive direction for all three outcome measures. More concretely, the coefficients for the iPad treatment condition indicator, which are positive and statistically significant for all three outcome measures, is the effect of receiving the iPad treatment condition among people who do not visit Facebook everyday. So, even for those who do not visit Facebook everyday, the treatment had a clear and sizable effect. The effect of the iPad treatment condition for those who visit Facebook everyday is the linear combination of the iPad treatment coefficient and the interaction coefficient. For example, in column (2), the effect of the iPad treatment condition among those who visit Facebook everyday on Job Performance is 1.609 (.909 + .700), which is not statistically significantly greater than the .909 effect of the treatment among those who do not visit Facebook everyday (p=.135). The effect of the iPad treatment condition among those who visit Facebook everyday on Personal Evaluation (column [4]) is also not statistically significantly different from the effect of the treatment among
19
those who do not visit Facebook everyday, as the coefficient of 1.211 (.711 + .500) is not significantly larger than .711 (p=.245). People who visit Facebook everyday do, however, think the member of Congress in the iPad treatment condition does a more effective job of communicating with constituents than people who do not visit Facebook everyday (see column [6]). The effect of the treatment among those who do not visit Facebook everyday is 1.926, while among those who visit Facebook everyday the treatment effect is 3.134 (1.926 + 1.209). This difference is statistically significant at p<.05. Thus, the results presented in Table 4 suggest that people who frequent Facebook everyday give the member of Congress who embraces new technology slightly, but not statistically significantly more favorable job and personal evaluation ratings, and significantly more favorable ratings in terms of engaging in effective communication with constituents. This finding gives greater credibility to the notion that some people not only prefer different communication methods (see Figure 2 and Table 2), but also view the communication they prefer as more effective.
4. Discussion, Limitations, and Future Directions The preceding analyses offer three main contributions to our understanding of the way in which current advances in communications technology, particularly social media, may influence the ability of elected representatives to connect with their constituents. First, it shows that most people think representatives should communicate with their constituents in a number of different ways. However, I also find that email, web sites, and social media should be used to communicate with constituents on a more frequent basis than other methods (letters and town
20
hall meetings). For example, over 50% of respondents stated that a representative’s web site should be updated once a week or more. Second, the analysis also documents that there is substantial variation even within a specific communication method in terms of how often people think representatives should be using it to communicate with constituents. In an exploratory analysis, political interest was a notable predictor of the frequency with which people thought representatives should update newer communication technologies (Facebook, web sites, and email). This may suggest that newer communication technologies will not necessarily expand the ability of representatives to connect with more constituents because only those who were already interested (and presumably paying attention) may want and pay attention to the messages sent by representatives, no matter what the communication method. Finally, the survey experiment documents that people respond more favorably to representatives who embrace the use of new technology to communicate with their constituents. This may suggest that simply remaining “with the times” sends a signal to voters (at least in this sample) that they respond to favorably. It could also be the case, however, that there is an expectation among citizens that the more constant contact that new social media provides, will permit individuals to be more informed about their representative’s behavior. In future work, I intend to more directly ask about and tease apart these two possibilities. To that end, there are several other important limitations to the preceding analysis. In addition to the non-representative nature of the MTurk sample, the survey data is also only a snapshot at one point in time. It is possible that people have different expectations for what they want from their representatives in terms of communication during election years. The survey did not include an exhaustive list of possible communication methods either. For instance, TV spots
21
were not included. While TV spots are typically only aired during election campaigns, they nevertheless are important sources of information for citizens. In future surveys, particularly those conducted during an election, including TV advertisements (and other communication methods) may prove fruitful. The survey also did not ask how “important” each communication method is to citizens. It could be the case that, although people think web sites and Facebook should be updated frequently, they gain more information that is useful to them for making decisions about, for example, who to vote for, from other sources. In future work, I intend to include a question to measure the relative importance people assign to the various communication methods. In addition to extending this work by collecting more representative and comprehensive survey data, another direction for future research is to compare actual communication from representatives to what people remember receiving. This may reveal an important disjoint between the effort representatives put in to communicating with their constituents and what is actually memorable for their constituents. In sum, this project begins to address the question of what role advancements in new communications technology, such as social media, will have for politics.
22
5. Appendix Age: What is the year of your birth? (in years) Gender: What is your gender? (Female=1; Male=0) Race: What racial or ethnic group or groups best describes you? (White=1; Black, Hispanic, Asian, Native American, Mixed, Other=0) Education: What is the highest level of education you have achieved? (1=no high school diploma; 2=high school graduate; 3= some college, no degree; 4=2-year college degree; 5=4year college degree; 6=post-graduate degree) Income: What was your total FAMILY income in 2010? (1=Less than $10,000; 2=$10,000$14,999; 3=$15,000-$19,999; 4=$20,000-$24,999; 5=$25,000-$29,999; 6=$30,000-$39,999; 7=$40,000-$49,999; 8=$50,000-$59,999; 9=$60,000-$69,999; 10=$70,000-$79,999; 11=$80,000-$99,999; 12=$100,000-$119,999; 13=$120,000-$149,999; 14=$150,000 or more; 15=prefer not to say) Hours Spent on the Internet per day: On a typical day, about how many hours do you spend on the Internet? (0=Less than 1 hour; 1 hour; 2 hours; 3 hours; 4 hours; 5 hours; 6 hours; 7 hours; 8 hours; 9 hours; 10 or more hours) Days per week on Facebook: During a typical week, how many days do you visit each of the following web sites? (0=0 days a week; 1=1-2 days a week; 2=3-4 days a week; 3=5-6 days a week; 4=Every day) Political Interest: How interested are you in politics and current events? (Very interested; Somewhat interested; Not at all interested; Standardized, mean=0, standard deviation=1) Party ID: Generally speaking, do you usually think of yourself as a Democrat, a Republican, an Independent, or what? …. (-3=Strong Republican; -2=Weak Republican; -1=Leaning Republican; 0=Independent; 1=Leaning Democrat; 2=Weak Democrat; 3=Strong Democrat)
23
6. References Berinsky, Adam J., Gregory A. Huber, and Gabriel S. Lenz. 2010. “Using Mechanical Turk as a Subject Recruitment Tool for Experimental Research.” Typescript, Yale University. Buhrmester, Michael D., Tracy Kwang, and Samuel D. Gosling. 2011. “Amazon’s Mechanical Turk: A New Source of Inexpensive, yet High-Quality, Data?” Perspectives on Psychological Science 6: 3-5. Croal, N’Gai. 2009. “All the President’s Tweets: How to Plug Americans into the White House.” Newsweek, February 3. Druckman, James N., Martin J. Kifer, and Michael Parkin. 2007. “The Technological Development of Congressional Candidate Web Sites: How and Why Candidates Use Web Innovations.” Social Science Computer Review 25: 425-442. Druckman, James N., Martin J. Kifer, and Michael Parkin. 2009. “Campaign Communications in U.S. Congressional Elections.” American Political Science Review 103: 343-366. Druckman, James N., Martin J. Kifer, and Michael Parkin. 2010. “Timeless Strategy Meets New Medium: Going Negative on Congressional Campaign Web Sites, 2002-2006.” Political Communication 27: 88-103. Druckman, James N., Cari Lynn Hennessy, Martin J. Kifer, and Michael Parkin. 2010. “Issue Engagement on Congressional Candidate Web Sites, 2002-2006.” Social Science Computer Review 28: 3-23. Evans, C. Lawrence, and Walter J. Oleszek. 2002. “The Internet and Institutional Change.” In James A. Thurber and Colton C. Campbell, eds., Congress and the Internet. New York: Prentice Hall. Fenno, Richard F. 1978. Home Style: House Members in their Districts. New York: Little, Brown. Garcia-Castañon, Marcela, Alison D. Rank, and Matt A. Barreto. 2011. “Plugged In or Tuned Out? Youth, Race, and Interent usage in the 2008 Election.” Journal of Political Marketing 10: 115-138. Hansen, Liane. 2009. “Republican Politicians Make a Social Media Push.” NPR, December 27. Johnson, Dennis. 2004. Congress Online: Bridging the Gap between Citizens and their Representatives. New York: Routledge. Karimi, Faith, and Joe Sterling. 2011. “Occupy protests spread around the world; 70 injured in Rome.” CNN, October 15.
24
Lassen, David S., and Adam R. Brown. 2010. “Twitter: The Electoral Connection?” Social Science Computer Review. DOI: 10.1177/0894439310382749. Lipinski, Daniel. 2004. Congressional Communication: Content and Consequences. Ann Arbor, MI: University of Michigan Press. Mayhew, David R. 1974. Congress: The Electoral Connection. New York: Yale University Press. Rhoads, Christopher. 2009. “Playing Catch-Up, the GOP is All Atwitter About the Internet: Republican Hopefuls Ponder a ‘Tech Gap’; Chuck DeVore’s ‘Tweets’ Raise Campaign Cash.” Wall Street Journal, January 30. Siegelbaum, Debbie. 2011. “Lawmakers say House Must Move More Quickly to Embrace New Technology.” TheHill.com, June 17. Smith, Craig Allen. 2010. Presidential Campaign Communication: The Quest for the White House. New York: Polity. Tolbert, Caroline J., and Ramona S. McNeal. 2003. “Unraveling the Effects of the Internet on Political Participation.” Political Research Quarterly 56: 175-185.
25
Figure 1. Since January 1, 2011, Members of Congress DID Each of the Following How Often?
90%
80%
70%
Percentage
60%
50%
40%
30%
20%
10%
0% Send letter
Hold town hall meeting
has not done this/DK Note: See Table 1 for question wording. N=481.
once or twice
Send email
once a month
Update web site
Update Facebook
two or three times a month
once a week or more
Figure 2. How Often SHOULD Members of Congress Do each of the Following?
80%
70%
60%
Percentage
50%
40%
30%
20%
10%
0% Send letter
Hold town hall meeting
do not need to do this Note: See Table 1 for question wording. N=481.
once or twice a year
Send email
once a month
Update web site
two or three times a month
Update Facebook
once a week or more
Figure 3. Average Rankings of the Five Communication Methods (1 = least preferred method; 5 = most preferred method)
Mean Ranking (1=least preferred; 5=most preferred)
4.0
3.5
3.0
2.5 Send letter
Hold town hall meeting
Send email Communication Methods
Note: See Table 1 for question wording. N=481.
Update web site
Update Facebook
Table 1. Survey and Experimental Design, Including Question Wording Panel A. Demographic Characteristics Demographics age, race, gender, education, income, state of residence, political ideology, and party affiliation Technology Demographics
hours per day spent on the Internet and days per week visiting various web sites (e.g., Facebook)
Panel B. Survey Items (N=481) Communication Methods Asked About Send letter; Hold town hall meeting; Send email; Update web site; Update Facebook What SHOULD members of Congress do?
Members of Congress are able to communicate with their constituents in a variety of ways. How often do you think members of Congress should communicate with their constituents in each of the following ways? (Response options : Do not need to do this; Should do this once or twice a year; Should do this once a month; Should do this two or three times a month; Should do this once a week; Should do this more than once a week)
What DID members of Congress do?
Since the beginning of this year (January 1, 2011), approximately how many times has your district’s representative to the U.S. House of Representatives done each of the following? ( Response options : Has not done this; Once or twice; Once a month; Two or three times a month; Once a week; More than once a week)
Ranking
Thinking about the various ways in which your member of Congress can communicate with you, please rank the following from 1 to 5, where 1 is your most preferred way to learn about your representative's activities and 5 is your least preferred. ( Response options : Ranking from 1 to 5)
Panel C. Survey Experiment, 2 x 2 (N=308) Experimental Congress and New Technology Vignette One big shift for Congress this year has been the ability to use iPads on the floor. But this change has been met with different responses by members of Congress. [iPad condition] For instance, Representative [Michael Jones / Sarah Jones] recently announced the release of an official mobile application for the iPhone, iPad and iPod Touch that allows [his / her] constituents to find [his / her] position on hot topics of the day, and easily access [his / her] contact information, news articles, e-newsletters, videos, and photos, the first application to be offered by a member of Congress. OR [no iPad condition] For instance, Representative [Michael Jones / Sarah Jones] does not use an iPad on the floor, and when asked recently stated, "I wouldn’t even know how to begin using an iPad."
Job Performance
Outcome Measures How would you rate the job the representative is doing? (Response scale [0-10]: poor - excellent)
Personal Evaluation
How do you feel about the representative as a person? (Response scale [0-10]: negative - positive)
Effective Communication
Do you think the representative does an effective job communicating with their constituents? (Response scale [0-10]: very ineffective - very effective)
Job Approval To what extent do you approve of the way each of the following is doing their job? (Response (Obama, options : Strongly Approve; Not Sure; Somewhat Approve; Somewhat Disapprove; Strongly Congress, Disapprove) Supreme Court) Note: Question wording and coding details not reported in the text or this table are included in the Appendix.
Table 2. What Predicts Beliefs in How Members of Congress Should Communicate with their Constituents? (1)
(2) (3) (4) (5) (6) (7) (8) (9) (10) Send Letters Have Town Hall Meeting Send Email Update Web Site Update Facebook (0=do not need to do this; 1=once or twice a month; 2=once a month; 3=two or three times a month; 4=once a week or more) Age (years) -0.058 -0.048 -0.029 -0.020 -0.026 -0.018 0.005 0.008 0.068 0.084 [0.023]* [0.022]* [0.025] [0.024] [0.031] [0.031] [0.023] [0.024] [0.040] [0.042]* Age-squared/100 (years) 0.069 0.058 0.028 0.016 0.008 -0.002 -0.011 -0.016 -0.099 -0.119 [0.028]* [0.028]* [0.031] [0.030] [0.039] [0.038] [0.028] [0.029] [0.052] [0.054]* Female (1=Yes) -0.354 -0.367 -0.214 -0.196 -0.379 -0.349 -0.223 -0.165 -0.131 -0.079 [0.100]** [0.102]** [0.090]* [0.089]* [0.123]** [0.124]** [0.088]* [0.087] [0.147] [0.145] Race (1=White) -0.291 -0.244 -0.172 -0.148 0.076 0.089 0.064 0.041 -0.056 -0.028 [0.118]* [0.117]* [0.108] [0.107] [0.149] [0.153] [0.102] [0.102] [0.154] [0.152] Education (1=No HS; 6=postgrad) -0.129 -0.140 -0.075 -0.086 -0.104 -0.115 0.009 0.002 -0.026 -0.046 [0.035]** [0.035]** [0.033]* [0.033]** [0.048]* [0.048]* [0.036] [0.037] [0.062] [0.062] Income (1=<$10k; 14=>150k; 15=prefer not to say) 0.014 0.018 -0.010 -0.004 0.014 0.021 -0.001 0.005 -0.015 -0.003 [0.014] [0.014] [0.012] [0.012] [0.017] [0.017] [0.013] [0.012] [0.021] [0.021] Missing: Income 0.052 0.071 -0.161 -0.136 -0.339 -0.312 -0.103 -0.080 0.097 0.146 [0.247] [0.249] [0.205] [0.201] [0.274] [0.277] [0.202] [0.195] [0.354] [0.351] Hours spent on the Internet (0=< than 1 hour; 10=10 or more hours) -0.017 -0.018 -0.001 -0.005 -0.025 -0.030 0.021 0.015 0.004 -0.004 [0.018] [0.019] [0.017] [0.017] [0.023] [0.023] [0.017] [0.017] [0.028] [0.027] Days per week on Facebook (0-4) 0.003 0.000 0.051 0.047 0.057 0.052 -0.009 -0.013 0.260 0.252 [0.030] [0.030] [0.028] [0.028] [0.040] [0.040] [0.031] [0.030] [0.053]** [0.052]** Political Interest (M=0, SD=1; low-high) -0.008 0.083 0.120 0.189 0.208 [0.051] [0.044] [0.061] [0.048]** [0.068]** Party ID (-3=S. Rep.; 3=S. Dem.) 0.064 0.057 0.052 0.019 0.098 [0.024]** [0.023]* [0.031] [0.024] [0.040]* Constant 3.099 2.864 2.795 2.605 3.411 3.246 3.408 3.374 1.582 1.266 [0.446]** [0.441]** [0.486]** [0.482]** [0.575]** [0.571]** [0.455]** [0.456]** [0.747]* [0.767] Observations 415 415 415 415 415 415 415 415 415 415 R-squared 0.107 0.121 0.066 0.089 0.082 0.098 0.022 0.069 0.090 0.126 Note: OLS regression coefficients with robust standard errors in brackets. * significant at 5%; ** significant at 1%; two-tailed.
Table 3. Evaluations of the Member of Congress Described in the Survey Experiment (1) (2) Job Performance (0=poor; 10=excellent) Female, iPad
1.399 [0.347]** -0.161 [0.299] 1.042 [0.340]**
(3) (4) Personal Evaluation (0=negative; 10=positive)
(5) (6) Effective Communication (0=very ineffective; 10=very effective)
1.291 1.097 1.023 2.625 2.511 [0.347]** [0.356]** [0.347]** [0.386]** [0.386]** Female, no iPad -0.243 -0.106 -0.144 -0.159 -0.195 [0.291] [0.321] [0.299] [0.357] [0.340] Male, iPad 0.951 0.765 0.733 2.319 2.285 [0.336]** [0.342]* [0.341]* [0.363]** [0.360]** Age (years) 0.062 0.067 0.066 [0.038] [0.039] [0.041] Age-squared/100 (years) -0.059 -0.059 -0.065 [0.044] [0.044] [0.048] Female (1=Yes) 0.223 0.103 0.457 [0.247] [0.253] [0.265] Race (1=White) -0.405 -0.299 -0.065 [0.299] [0.311] [0.317] Education (1=No HS; 6=postgrad) -0.020 -0.056 -0.063 [0.097] [0.099] [0.102] Income (1=<$10k; 14=>150k; 15=prefer not to say) -0.082 -0.100 -0.071 [0.036]* [0.038]** [0.037] Missing: Income 0.377 0.529 0.341 [0.466] [0.472] [0.587] Hours spent on the Internet (0=< than 1 hour; 10=10 or more hours) 0.005 -0.041 -0.014 [0.042] [0.039] [0.045] Days per week on Facebook (0-4) 0.000 0.040 -0.029 [0.078] [0.083] [0.083] Political Interest (M=0, SD=1; low-high) -0.152 -0.210 -0.265 [0.134] [0.137] [0.145] Party ID (-3=S. Rep.; 3=S. Dem.) 0.007 -0.038 -0.047 [0.066] [0.068] [0.068] Constant 5.093 4.678 5.173 4.988 4.867 4.250 [0.228]** [0.870]** [0.242]** [0.888]** [0.280]** [0.920]** Observations 308 308 308 308 308 308 R-squared 0.095 0.133 0.055 0.106 0.256 0.295 Note: OLS regression coefficients with robust standard errors in brackets. Omitted treatment condition is the "Male, no iPad" condition. * significant at 5%; ** significant at 1%; two-tailed.
Table 4. Frequent Facebook Users Evaluate the use of New Technology by a Member of Congress More Favorably (1) (2) Job Performance (0=poor; 10=excellent) iPad condition (1=Yes)
1.233 [0.237]** -0.033 [0.249]
(3) (4) Personal Evaluation (0=negative; 10=positive)
(5) (6) Effective Communication (0=very ineffective; 10=very effective)
0.909 0.943 0.711 2.486 1.926 [0.341]** [0.239]** [0.351]* [0.248]** [0.349]** Visit Facebook Everyday (1=Yes) -0.367 0.126 -0.112 -0.171 -0.748 [0.317] [0.255] [0.328] [0.265] [0.356]* iPad Condition X Visit Facebook Everyday 0.700 0.500 1.209 [0.466] [0.467] [0.486]* Age (years) 0.062 0.061 0.067 0.066 0.066 0.063 [0.038] [0.038] [0.039] [0.039] [0.040] [0.039] Age-squared/100 (years) -0.057 -0.056 -0.058 -0.057 -0.063 -0.062 [0.044] [0.043] [0.045] [0.044] [0.047] [0.046] Female (1=Yes) 0.234 0.262 0.107 0.127 0.477 0.525 [0.241] [0.244] [0.249] [0.250] [0.256] [0.257]* Race (1=White) -0.377 -0.350 -0.283 -0.265 -0.035 0.010 [0.304] [0.302] [0.319] [0.319] [0.318] [0.312] Education (1=No HS; 6=postgrad) -0.015 -0.010 -0.050 -0.046 -0.062 -0.053 [0.096] [0.095] [0.097] [0.096] [0.101] [0.100] Income (1=<$10k; 14=>150k; 15=prefer not to say) -0.079 -0.078 -0.099 -0.099 -0.067 -0.066 [0.037]* [0.037]* [0.038]* [0.039]* [0.038] [0.039] Missing: Income 0.393 0.366 0.557 0.538 0.332 0.285 [0.475] [0.474] [0.478] [0.481] [0.597] [0.593] Hours spent on the Internet (0=< than 1 hour; 10=10 or more hours) 0.006 0.007 -0.041 -0.040 -0.011 -0.009 [0.042] [0.042] [0.040] [0.040] [0.046] [0.045] Political Interest (M=0, SD=1; low-high) -0.152 -0.144 -0.212 -0.206 -0.262 -0.247 [0.135] [0.136] [0.138] [0.138] [0.145] [0.145] Party ID (-3=S. Rep.; 3=S. Dem.) 0.001 0.005 -0.045 -0.042 -0.049 -0.043 [0.065] [0.065] [0.067] [0.067] [0.067] [0.067] Constant 4.448 4.576 4.882 4.973 4.058 4.279 [0.841]** [0.834]** [0.848]** [0.852]** [0.878]** [0.864]** Observations 308 308 308 308 308 308 R-squared 0.128 0.135 0.103 0.107 0.294 0.308 Note: OLS regression coefficients with robust standard errors in brackets. Omitted treatment condition is the "Male, no iPad" condition. * significant at 5%; ** significant at 1%; two-tailed.
Table A1. Summary Statistics (1) (2) (3) Opinion Sample Variable Full Sample Full Sample Female, iPad Age (years) 31.877 33.029 35.762 [11.3209] [11.9577] [12.1506] Age-squared/100 (years) 11.440 12.335 14.242 [8.7677] [9.7046] [9.8034] Female (1=Yes) 0.580 0.669 0.698 [.4942] [.4714] [.4626] Race (1=White) 0.782 0.770 0.825 [.4133] [.4219] [.3827] Education (1=No HS; 6=postgrad) 3.925 4.162 4.318 [1.3188] [1.3038] [1.4122] Income (1=<$10k; 14=>150k; 15=prefer not to say) 7.545 7.760 7.857 [4.0589] [3.8321] [4.0555] Missing: Income 0.069 0.078 0.095 [.253] [.2685] [.2959] Political Interest (M=0, SD=1; low-high) 0.005 0.085 0.075 [1.0176] [.9718] [.9965] Party ID (-3=S. Rep.; 3=S. Dem.) 0.606 0.529 0.333 [1.8786] [1.9192] [1.9838] Ideology (-2=v. conservative; 0=moderate/not sure; 2=v. liberal) 0.329 0.316 0.097 [1.0399] [1.109] [1.0356] Hours spent on the Internet (0=< than 1 hour; 10=10 or more hours) 4.924 4.893 5.095 [2.6024] [2.7141] [2.7633] Days per week on Facebook (0-4) 2.743 2.477 2.222 [1.533] [1.6332] [1.6308] Observations 481 308 63 Note: Cell entries are means with standard deviations in brackets. See Appendix for question wording and coding details.
(4) Experiment Sample Female, no iPad 30.978 [10.9201] 10.775 [8.4047] 0.685 [.467] 0.719 [.452] 4.202 [1.2896] 7.596 [3.985] 0.079 [.2707] 0.019 [1.0001] 0.719 [1.809] 0.438 [1.0761] 4.742 [2.7074] 2.607 [1.628] 89
(5)
(6)
Male, iPad 30.494 [9.2967] 10.152 [6.6354] 0.654 [.4786] 0.741 [.441] 3.938 [1.2283] 7.247 [3.7667] 0.062 [.2422] 0.098 [.9533] 0.815 [1.8105] 0.482 [1.0967] 5.222 [2.824] 2.580 [1.5721] 81
Male, no iPad 35.907 [14.4017] 14.939 [12.6683] 0.640 [.4832] 0.813 [.3923] 4.227 [1.3005] 8.427 [3.4842] 0.080 [.2731] 0.158 [.9501] 0.160 [2.0602] 0.173 [1.1897] 4.547 [2.5537] 2.427 [1.7099] 75
Table A2. Percentages for Figures 1, 2, and 3 A. Since January 1, 2011, Members of Congress DID… has not done this/DK Send letter 78.2% Hold town hall meeting 75.3% Send email 78.4% Update web site 73.8% Update Facebook 82.5%
once or twice 15.2% 15.2% 7.1% 5.0% 2.5%
once a month 2.9% 5.8% 6.7% 5.0% 2.9%
two or three times a month 1.9% 1.7% 3.3% 6.7% 4.6%
once a week or more 1.9% 2.1% 4.6% 9.6% 7.5%
once or twice a year 45.9% 41.2% 14.1% 3.1% 2.3%
once a month 17.7% 40.3% 36.2% 11.2% 7.7%
two or three times a month 6.2% 8.5% 20.6% 16.8% 13.7%
once a week or more 3.3% 5.6% 18.3% 67.4% 60.5%
2 26.8% 23.9% 17.3% 15.0% 17.0%
3 13.3% 22.0% 26.4% 21.2% 17.0%
4 17.3% 13.3% 17.9% 31.0% 20.6%
5 (most preferred) 14.8% 17.3% 26.6% 27.2% 14.1%
B. Members of Congress SHOULD… Send letter Hold town hall meeting Send email Update web site Update Facebook
do not need to do this 26.8% 4.4% 10.8% 1.5% 15.8%
C. Members of Congress RANKING… 1 (least preferred) Send letter 27.9% Hold town hall meeting 23.5% Send email 11.9% Update web site 5.6% Update Facebook 31.2% Note: See Table 1 for question wording. N = 481.
Table A3. Evaluations of Representative Described in Survey Experiment and Job Approval Ratings, by Treatment Condition Variable Job Performance (0=poor; 10=excellent) Personal Evaluation (0=negative; 10=positive) Effective Communication (0=v. ineffective; 10=v. effective) Approval of Obama (-2=strongly disapprove; 2=strongly approve) Approval of Congress (-2=strongly disapprove; 2=strongly approve) Approval of Supreme Court (-2=strongly disapprove; 2=strongly approve) Observations Note: Cell entries are means with standard deviations in brackets.
(1) Full Sample 5.607 [2.1305] 5.568 [2.1321] 5.968 [2.5175] -0.071 [1.4778] -0.818 [1.1805] 0.104 [1.2329] 308
(2) Female, iPad 6.492 [2.0781] 6.270 [2.0729] 7.492 [2.1165] -0.286 [1.539] -0.746 [1.2822] 0.111 [1.3212] 63
(3) Female, no iPad 4.933 [1.8203] 5.067 [1.9874] 4.708 [2.0955] 0.000 [1.4538] -0.809 [1.1066] 0.011 [1.2293] 89
(4) Male, iPad 6.136 [2.2735] 5.938 [2.1814] 7.185 [2.0863] 0.136 [1.4209] -0.827 [1.202] 0.185 [1.1949] 81
(5) Male, no iPad 5.093 [1.974] 5.173 [2.0949] 4.867 [2.4236] -0.200 [1.5067] -0.880 [1.1736] 0.120 [1.2188] 75
Table A4. Job Approval Ratings as a Function of Experimental Treatment Conditions (1) (2) Approval of Obama (-2=strongly disapprove; 2=strongly approve) Female, iPad
-0.086 [0.260] 0.200 [0.232] 0.336 [0.235]
(3) (4) Approval of Congress (2=strongly disapprove; 2=strongly approve)
(5) (6) Approval of Supreme Court (2=strongly disapprove; 2=strongly approve)
-0.123 0.134 0.199 -0.009 0.044 [0.208] [0.211] [0.204] [0.218] [0.214] Female, no iPad -0.104 0.071 -0.020 -0.109 -0.163 [0.184] [0.179] [0.172] [0.192] [0.195] Male, iPad 0.039 0.053 0.008 0.065 0.077 [0.196] [0.190] [0.182] [0.193] [0.196] Age (years) -0.032 -0.023 -0.057 [0.024] [0.025] [0.024]* Age-squared/100 (years) 0.028 0.033 0.066 [0.028] [0.029] [0.027]* Female (1=Yes) 0.309 0.062 0.231 [0.151]* [0.137] [0.155] Race (1=White) -0.406 -0.272 -0.197 [0.170]* [0.161] [0.165] Education (1=No HS; 6=postgrad) 0.059 -0.055 0.080 [0.056] [0.055] [0.061] Income (1=<$10k; 14=>150k; 15=prefer not to say) 0.040 0.006 0.015 [0.022] [0.021] [0.023] Missing: Income -0.638 -0.382 -0.314 [0.282]* [0.263] [0.319] Hours spent on the Internet (0=< than 1 hour; 10=10 or more hours) -0.007 -0.041 -0.008 [0.026] [0.021] [0.025] Days per week on Facebook (0-4) 0.032 0.202 0.129 [0.048] [0.039]** [0.048]** Political Interest (M=0, SD=1; low-high) -0.050 -0.322 -0.151 [0.073] [0.063]** [0.074]* Party ID (-3=S. Rep.; 3=S. Dem.) 0.420 0.027 -0.017 [0.034]** [0.035] [0.038] Constant -0.200 0.018 -0.880 -0.403 0.120 0.498 [0.174] [0.503] [0.135]** [0.532] [0.141] [0.536] Observations 308 308 308 308 308 308 R-squared 0.012 0.396 0.001 0.178 0.003 0.085 Note: OLS regression coefficients with robust standard errors in brackets. Omitted treatment condition is the "Male, no iPad" condition. * significant at 5%; ** significant at 1%;
