Secondary menu

Explaining the Turnout Decline in Canadian Federal Elections: A New Survey of Non-voters


10. The Internet

The possibility of using modern Internet technology to enhance opportunities for voting has intrigued numerous observers. Access to the Internet has spread widely; for example, only about one quarter (26.5 percent) of the current sample reports no access to the Internet either at home, at work, or both (33 percent had access at "both"). Questioning people about their potential Internet use for voting requires adding the proviso that "technology allow enough safety and secrecy", as we did in our questions. It also falls into the realm of the hypothetical, and therefore the results should not be regarded as a totally accurate guide when people are asked about potential usage. Nevertheless, the potential for the Internet to provide additional access to the election process is sufficient to warrant exploration.

We will include a sample calculation about the effect of Internet voting on the overall turnout rate later in this section. Here, we can give the results when people were asked whether they would "likely use the Internet" to:



Table 46 Likelihood of Using the Internet to Register, by Vote in 2000 (percentages)


  Voted in 2000? Total
Yes No
Likelihood of using the Internet to check or modify your personal information or register on the list of electors Very likely
32.5
31.7
32.1
Somewhat likely
25.7
26.2
26.0
Not very likely
15.0
15.9
15.5
Not at all likely
26.8
26.2
26.5
Total
100.0
100.0
100.0
V = .015   p < .926



Table 47 Likelihood of Using the Internet to Vote, by Vote in 2000 (percentages)


  Voted in 2000? Total
Yes No
Likelihood of using the Internet to vote on-line rather than to cast a ballot at the polling station Very likely
33.1
43.2
38.5
Somewhat likely
14.2
17.0
15.7
Not very likely
14.8
12.0
13.3
Not at all likely
37.9
27.8
32.5
Total
100.0
100.0
100.0
V = .131   p < .000



As can be seen in tables 46 and 47, a majority of the public, as represented by our sample, declares that it is either "very" or "somewhat" likely that they would use the Internet to register or modify their information on the list of electors and also that they would use the Internet to vote on-line. With regard to the first topic, Internet registration, there is very little difference between those who voted or did not vote in the 2000 general election in their professed likelihood of doing this. Internet voting, on the other hand, is more likely to appeal to current non-voters. We can see from Table 47 that 43.2 percent of non-voters in 2000 claim they would be very likely to vote on-line, and a further 17 percent say they would be somewhat likely to do so. If we are to take these answers at face value, the introduction of Internet registration and Internet voting could have a beneficial effect on raising the turnout rate. Of course, the trick is to estimate how realistic such professed intentions really are.



Table 48 Likelihood of Using the Internet to Vote, by Voting Frequency in Last Three Federal Elections (percentages)


  Voted in none 1 2 Voted in all 3 Total
Likelihood of using the Internet to vote on-line rather than to go cast a ballot at the polling station Very likely
41.4
57.8
36.8
28.9
36.7
Somewhat likely
20.5
 9.4
18.2
13.3
15.5
Not very likely
11.3
 9.4
13.0
16.2
16.6
Not at all likely
26.9
23.3
32.0
41.6
34.1
Total
100.0
100.0
100.0
100.0
100.0
V = .133      p < .000



Table 48 looks, not just at the behaviour of the sample in 2000, but at the voting frequency of our respondents over the last three federal elections (this is the same variable used in Table 20). We can see from this cross-tabulation that those with lower voting frequency in those elections, those who voted either once or not at all, are more likely to say they would vote via the Internet. Even those who voted in two of three were more likely to express the intention of using the Internet than those who voted in every election. It is likely that those Canadians who make a point of voting in every election enjoy doing their civic duty by actually going to the polls, and would be less likely to make use of an Internet alternative. Interestingly, it is the group of people who voted in one of the three opportunities that is most likely to use the Internet to vote. Thus, the Internet might turn out to be a way in which "intermittent" or "transient" voters (see Harold D. Clarke, Jane Jenson, Lawrence LeDuc and Jon H. Pammett, Political Choice in Canada and Absent Mandate, various editions, for a discussion of "transient voters") could improve their participation rates. We know that these are not people who are "permanent non-voters", but rather electors who turn out when they can, or when their interest is piqued. Internet voting might make the difference for some of them.



Table 49 Predictors of Internet Use for Registration or Voting (multiple regression) (Betas only)


Predictors
Likelihood of:
Registration Voting
Age
-.175*
-.167*
Education
.146*
.153*
Income
.121*
.121*
Gender
.048
-.002
Urban-rural residence
.069*
.089*
Born in Canada
.017
.016
Mobility
.055
.004
Name on 2000 list
.032
.008
Active participation†
.047
-.007
Passive participation†
.145*
.099*
Voting frequency 1993–2000
-.019
.129*
N =
1 294
1 299
R 2 = .150
R 2 = .144
† = factor scores
* = statistically significant p < .01



Table 49 presents a summary of the predictors of Internet use in a multiple regression format, for conciseness using only the Beta, or standardized regression coefficients. The overall amount of variance explained is modest, about 15 percent in both cases. And the predictor variables associated with "likely" Internet use for registration or voting are quite similar. It is clear that younger, urban respondents, who are more educated and have higher incomes would be more interested in the Internet option. It is well known that these variables also correlate with use of the Internet in general. When it comes to possible Internet registration, the only other statistically significant predictor is having engaged in higher levels of "passive participation" in the past, that is, writing a letter to a newspaper, calling a radio talk show, or joining an Internet chat group. With Internet voting, there is one additional predictor, our "voting frequency" variable from Table 48. Here, the direction of the relationship is such that lower previous voting frequency is associated with higher levels of proposed use of the Internet for voting. This reinforces the relationship observed in the previous table, and shows that lower voting frequency has an effect independent of other variables. Once again, this suggests that some benefit might be gained from the introduction of Internet voting.

To explore the potential impact of the Internet in more detail, we have prepared two calculations regarding the potential contribution of the Internet to reducing two problems of a personal or administrative nature, that of not being on the list of electors and being "busy at work" on polling day. We have already seen, in tables 46–49, responses to the question "If technology allowed enough safety and secrecy" how likely would you be to use the Internet to a) Check or modify your personal information or register on the list of electors and b) Vote on-line rather than to go cast a ballot at a polling station. To the extent that absence from the list might be a deterrent to voting, Internet registration or information modification might alleviate that problem. And to the extent that busy people might find it difficult to get to the poll, Internet voting could allow them to participate.

Internet registration: To calculate the potential impact of Internet registration, we first cross-tabulated, for non-voters, the importance of the factor of not being on the list of electors by the likelihood of using the Internet to register. We considered those who said it was "very or fairly important" that they were not on the list, and those who said it was "very or fairly likely" that they would use the Internet to register. This calculation produces 130 people (out of an effective weighted non-voter sample of 1 065) or 12.2 percent of non-voters to whom this procedure might be applicable. To put this into terms of the whole electorate, we must multiply our non-voting group by .39 (since this was the percentage who did not vote). The result is 4.76 percent, which is the potential rise in the voting rate with Internet registration.

Before placing too much stock in this figure, we should qualify it in two ways. First, it seems reasonable that some degree of interest in the 2000 election be a necessary condition for someone to go to the trouble of engaging in an Internet registration process. We therefore went back to our 130 potential registrants mentioned above, and retained only those who said they were "very or fairly interested" in the election, removing those who were "not very or not at all interested." This leaves 53 people, which multiplied by .39 equals a potential rise in the voting rate of 1.94 percent with Internet registration.

Finally, we feel that it is also reasonable that only those with Internet access would likely engage in this Internet process. Using the measure in the dataset that reports whether respondents had access to the Internet at home, at work, or both, we found that we were left with 39 people from the above group who had access somewhere. Once again, multiplying by .39, we conclude that the rise in the overall voting rate that might reasonably result from Internet registration would be 1.43 percent. Thus, taking at face value people's expressed opinions that a) not being on the list was an important factor in their decision not to vote, b) that they would have registered by Internet if they could, and qualifying it by whether c) they were interested in the election, and d) they had access to the Internet, we conclude that there would be a small increase in the overall voting rate (we estimate less than one and a half percent) if Internet registration were permitted.

Internet voting: For those non-voters of 2000 who reported that being "busy at work" was an important reason not to cast a ballot (32.9 percent of non-voters, or 12.56 percent of the electorate) we investigated whether they said they would vote via the Internet, if this were possible. Taking those who said this was "very or fairly likely" we identified 238 people, or 21.7 percent of non-voters. Calculating this as a percentage of the total electorate (21.7 X .39) we get an initial total of 8.46 percent, representing the potential rise in voter turnout with Internet voting.

However, as with the previous analysis, it seems reasonable to make the calculation more realistic with a couple of qualifiers. When we sort those 238 people according to their expressed interest in the 2000 election, we find that only 84 of them were "very or fairly interested". Redoing the calculation reduces the potential impact of Internet voting on the overall voting rate to 3.07 percent. Finally, the number reduces further (to 77 people) if we apply the condition that the person have Internet access at a home or work location. When this restriction is made, the potential rise in the voting rate is 2.82 percent. Thus, taking at face value people's expressed opinion that a) being busy at work was an important factor in their decision not to vote, and b) that they would vote on the Internet if they could, as well as c) that they were interested in the election and d) they had access to the Internet, we conclude that there would be a rise in the overall voting rate of slightly less than 3 percent if Internet voting were implemented. This percentage could be somewhat greater if people from other non-voting categories (i.e. "out of town") not captured by the "busy at work" category are added in.