Human physiology is amazingly diverse. The body has many systems and changing activity levels in one system often do not correlate well with changing activity levels in another. We will employ just one of these many measures but readers should bear in mind that additional measures will be necessary before more complete assessments are possible. The particular physiological measure we employ is based on EDA. EDA manifests itself as small increases in the secretion of sweat at various points on the body; most research, including our own, measures EDA at the fingertip (see Dawson et al. 2007). EDA is especially appropriate for our purposes because other physiological systems, such as the cardio-vascular, lie under control of both the sympathetic nervous system (SNS) and the parasympathetic nervous system. EDA, on the other hand, lies entirely under SNS control (Dawson et al. 2007). As a result, EDA increases are widely accepted in the psychophysiological community as reliable indicators of arousal (Kreibig 2010) and as linking closely “with the psychological concepts of emotion, arousal, and attention” (Dawson et al. 2007). The highly questionable use of EDA as an indicator of whether or not a single, designated individual is telling the truth should not be confused with its acknowledged value as an indicator of the situations under which a group of people, on average, is physiologically aroused.

EDA has been widely used in the field of psychophysiology to get at, for example, overall cognitive engagement with one’s environment (Nikula 1991), task vigilance (Dawson et al. 1989), and conditioned anxiety among those individuals suffering from post-traumatic stress disorder (Blechert et al. 2007). In terms of political behavior more specifically, EDA has been used as a measure of disgust response that predicts support for sexual morality policy (Smith et al. 2011a), as a gauge of threat response to predict support for socially protective policies (Oxley et al. 2008), and as a measure of anxiety in the face of negative political campaigning (Mutz and Reeves 2005).

It should be noted that, in and of itself, EDA is incapable of indicating valence (and this is one of the reasons it is problematic as a lie detector). Both an image of a snake and an image of a loved one typically increase EDA. Though this situation can be a problem for some research designs, the goal here, as stated above, is to acquire a broad measure of physiological response and not one that is keyed to, or strives to measure, responses to a discrete category of stimulus content. Thus, to obtain our measure of broad EDA response, we recorded mean changes in EDA of each participant occasioned by the presentation of a lengthy series of very diverse still images. We know the stimuli were diverse because an independent panel of 126 individuals rated their own emotional responses to these images in terms of valence, with one being “happy/positive feelings” and nine being “unhappy/negative feelings,” and also on intensity, with one being “no reaction” and nine being a “strong reaction.” The valence ratings ranged from a mean of 1.83 (SD = 1.178) for an image of a sunset to 8.42 (SD = .924) for an image of an anorexic woman. Strength of emotional reaction ranged from a mean of 2.79 (SD = 2.01) for a clean toilet to 8.11 (SD = 1.208) for the anorexic woman. The raters also were asked to select specific emotions that were elicited by each image, which resulted in an assortment of reported, evoked emotions, including happiness, disgust, satisfaction, amusement, anger, fear, sadness and anxiety. The ratings show that the images selected for presentation run the gamut from decidedly positive (sunset, bowl of fruit, smiling child, cute animals) to decidedly negative (wounds, vomit, physical fights and dangerous animals). Each image appeared for 12 s on a computer screen directly in front of the participant and was preceded for approximately 10 s by a fixation point on a blank screen.

Because individuals vary dramatically in their pre-stimulus EDA levels, and because our focus is on the degree of response, we constructed a ratio of change in EDA for each individual by dividing the average skin conductance level (SCL) during presentation of each stimulus by the SCL obtained during the previous inter-stimulus interval (or ISI). Thus, at stimulus onset, increases in SCL relative to the previous ISI provide a ratio above 1, while decreases will produce a ratio below 1. We then calculated overall tendency to display physiological arousal by computing the mean change in this ratio across all non-political images for each participant. These procedures are standard in EDA analyses and the mean magnitude of the EDA response we recorded for our sample is typical (see Dawson et al. 2007). What is perhaps less typical is our practice of averaging this response across such a wide range of images in order to get a broad measure of EDA responsiveness. As argued above, however, we believe that doing so will provide both useful information on its own and also serve as a valuable baseline against which physiological responses to particular categories of stimuli can be compared.

Our basic measure of political participation was obtained from a survey that participants completed long before the physiological exercise. Participants were presented with 11 items bearing on political participation, with a particular bent toward issue-driven participation: “how often do you discuss issues with other people,” “are you registered to vote,” “do you usually vote in elections,” “do you feel strongly about political issues,” “have you voted for particular candidates because of their position on a political issue,” “have you campaigned for particular candidates,” “have you contributed money to particular candidates,” “have you contacted elected officials to encourage them to support a position on a political issue,” “have you tried to persuade other citizens to support a position on a political issue,” “have you joined an organization that promotes a position on a political issue,” and “have you attended meetings that promote a position on a political issue.” The item gauging frequency of political discussion was a five-point scale but all of the others were dichotomous “yes–no” items. Each of the individual items was standardized and factor analyzed using principal-components factoring, with the overall measure of political participation created by multiplying respondents’ scores on each item by its factor loading and summing the products of the individual participation items together.2 Weighting the individual participation items by their factor scores makes it possible to create a participation index that includes only the items’ shared correlation with the underlying construct of political participation while discarding the portion of covariance not explained by that construct.3

Participants in this project were obtained in the following manner. We contracted with a professional survey organization to randomly contact individuals within convenient distance from Lincoln, Nebraska, to see if they would be willing to travel to our lab for a 90 min session in exchange for $50. In this fashion, 200 subjects were recruited and completed an extensive, computer-based survey of their political beliefs, personality traits and demographic characteristics. This is the survey that included the items on political participation mentioned above. It was intended that these participants would serve as a pool from which we could recruit smaller groups for physiological testing as money and lab time became available. Several months after the larger group completed the survey, 37 individuals with relatively strong political ideologies (either liberal or conservative) were invited back. A year later 51 additional participants from the pool were invited back, with this group consisting of some who were politically disinterested and uninvolved and others who were more engaged in the political process. We make no claim that these 88 individuals constitute a random sample. Their willingness to travel to participate in lab exercises on two separate occasions, for example, is an obvious source of bias. Still, for our purposes recruiting a purely random sample is less crucial than securing a group of people with a wide range of political participation levels.

On the second trip, research participants were brought to our physiology lab one at a time and sensors measuring EDA were attached to the distal phalanges (fingertips) of the index and middle fingers on participants’ non-dominant hand. After an acclimation period, the series of images was presented. The participant was not required to perform any behavioral task—only to pay attention to the images on the screen. Data from two participants were not usable. One was discovered to have a health problem and the readings for the other were corrupted perhaps because of a misplaced or malfunctioning sensor.