Summary Investigation Report on Robocalls
3. The Investigation
3.1 Information Provided by the Complainants
56. As indicated above, investigators examined each of the more than 40,000 communications received by the Commissioner to distinguish actual complaints from the more general expressions of elector concern. In total, 1,726 complainants made 2,448 complaints (some reporting both incorrect poll location calls and nuisance calls). Investigators attempted to contact each and every complainant, at times repeatedly, in order to ensure the collection of all possible relevant details: further information on the complainant's identity; his or her telephone service provider; the content, time of day and date of the call; and any other relevant information that might be available.
57. It should be emphasized that, at all times, investigators assumed the good faith of each elector in submitting a complaint. In most instances, however, complainants could provide only vague and incomplete information. Only a limited amount of the information gathered could be used as evidence. Each complainant had typically engaged briefly with the caller in question months before. Few could provide an incoming calling number, and some could not identify their own telephone service provider. Only two were able to provide a recording of the call, while another provided a transcript of the call.
58. Many electors said their complaints were prompted by media accounts, raising the significant difficulty of ascertaining whether or not their subsequent recollection was influenced by the media stories. The investigators could not and did not rely on unsubstantiated information reported in the media that could not be independently verified, nor did they focus on any particular political party.
59. It is worth noting that, of the three calls for which complainants provided a recording or transcript, investigators found the content of two to be innocuous after reviewing the evidence provided. The remaining call was an instance of incorrect poll information sent to a party supporter. On further investigation, it was found to have likely come from the same party's local campaign. It would be unreasonable to infer from this specific evidence that the caller intended to prevent the elector from voting.
60. Finally, a review was conducted of those electoral districts having 15 or more complainants, which represented 667 (39%) of the 1,726 complainants. In those districts, 6 (0.8%) of the 667 complainants told investigators that they did not vote as a result of the calls in question.
61. In examining investigators' contact with the complainants, it may be useful to highlight the following numbers. Of 1,726 complainants:
- 163 (9%) could not be reached despite repeated attempts by investigators to do so;
- 1,556 (90%) could not provide the incoming calling number of the impugned call;
- 170 (10%) could provide what they believed to be the calling number;
- 273 (16%) could not identify their own telephone service provider.
62. Given that, in most cases, recollection of a call received months earlier was the only available evidence, it was essential as a matter of good investigative practice to carefully examine the information provided. To more objectively assess recollections, particularly in the context of extensive media reports, investigators consulted a subject matter expert in applied memory research. The expert advised caution on the basis that, among other things, a) delay from the time an event is experienced and then recollected may influence recall by allowing details to be misremembered or filled in, and b) details of media accounts could come to be incorporated in a recollection.
3.2 Follow-up on Information Provided by Complainants
3.2.1 Source of Incoming Calls
63. As noted earlier, 170 complainants provided investigators with a calling number for the alleged inappropriate calls they were reporting. Once duplicate numbers were accounted for, investigators were left with 96 unique calling numbers.
64. Some of these numbers were confirmed to be those of known political telemarketers, while others were linked to commercial telemarketers who were not engaged in political calling. Using link analysis software, investigators determined that the largest number of calls to complainants from one reported number was 13 calls. That incoming number was linked not to a political entity or someone acting on its behalf, but to a credit card phishing scam.Footnote 13
3.2.2 Complainants' Telephone Service Providers
65. In an attempt to identify the source of the impugned calls, investigators sought complainant telephone records from their telephone service providers by way of judicially authorized production orders or, in a few cases, through voluntary co-operation from service providers on the consent of complainants.
66. There is no legal requirement for a telephone service provider to retain call records. Service providers retain call records for varying amounts of time if the record relates to calls for which billing information is required. Otherwise, many do not retain records at all.
67. Only a few service providers, particularly those using digital data capture such as Internet logs, retained some data and were able to retrieve it for investigators on the basis of a production order. A majority of complainants subscribed to service providers, such as Bell Canada and Telus, that did not retain call records except for those having a financial component. This had no application in the case of complainant subscribers. In addition, a rather large number of complainants (273, as noted above) were unable to identify their own service provider. Finally, even where data was available, such as with Rogers, Shaw and Videotron, the service providers could not guarantee the data was complete, or that its integrity had not been degraded through the archiving process.
68. Figure 3 shows that call data was available for 342 complainants (20%) and unavailable for the remaining 1,384 (80%).
69. In most cases where call records were retained, recipient consent for the release of records for their own number was not sufficient authorization for most service providers. They would only respond to a judicial order.
3.2.3 Production Orders
70. Production orders are judicial orders obtained under the Criminal Code requiring third parties – that is, a person or entity that is not under investigation, such as telephone service providers – to produce specific documents or data. To obtain an order, an investigator must first swear an affidavit, known as an Information to Obtain (ITO), before a justice. The ITO must demonstrate an investigator's reasonable grounds to believe that:
- an offence has been or is suspected to have been committed;
- the data will provide evidence respecting the commission of the offence; and
- the subject of the order has possession or control of the documents or data being sought.
71. Investigators sought production orders for the complainants' telephone records in the relevant period as the sole means of identifying with certainty the telephone numbers of incoming calls. A production order could only be sought, however, for telephone records of complainants whose report was detailed enough to provide reasonable grounds to believe an offence is suspected to have occurred, that the data sought would afford evidence respecting the commission of an offence, and where there were reasonable grounds to believe that the service provider still retained a record.
72. Of the 342 complainants for whom call records existed, 129 had provided investigators with enough detail to establish reasonable grounds to seek a production order by the time in the investigatory process that production orders were being sought. Among the remaining 213 complainants for whom call records existed, a handful had alleged conduct for which call records were unnecessary – for example, several alleged misconduct relating to a provincial election. Most of the remaining complainants, however, had submitted complaints after the initial step of the production order process was underway (September 2012). Investigators intended to seek the records for these latter complaints in subsequent production orders.
73. Production orders were obtained for records from Rogers, Shaw and Videotron. Together, they provided records of 6,051 incoming calls received by the 129 complainants named in the production orders. Investigators determined that these calls originated from 1,597 different numbers. Each number was matched to a subscriber where possible. Some service providers gave subscriber information, but others refused to confirm subscribers without a production order. Some numbers originated with US service providers, all of whom similarly refused to co-operate. In the end, subscriber information for incoming calls was obtained for 949 numbers and could not be obtained for 648 numbers. Each number was also checked against political telemarketers' call log data, numbers known to have been used by political entities, and the CRTC and Canadian Anti-Fraud Centre databases of suspect numbers.
74. Of the 129 complainants whose call records were obtained by production order, political calls could be confirmed in some instances, but, in the majority of cases, not the call content. Of these calls, not all were determined to be problematic.Footnote 14 In some instances the call content retrieved from voice recordings was inconsistent with the allegations made by complainants.
75. Overall, no discernible pattern of misdirection, such as a single predominant calling number or constellation of predominant calling numbers, was noted that would suggest a potential breach of the Act. Investigators were able to establish that several political calls were directed to an elector living outside the electoral district targeted by the call, and they uncovered instances of multiple calls to the same elector. As well, investigators were able to confirm the content of one call, found in the complainant's call records, that provided incorrect poll location information. In this last case, however, investigators found no evidence that the call was intended to interfere with the elector's right to vote.
76. The analysis of the production order results led investigators to conclude that reasonable grounds to believe that an offence had been committed no longer existed in respect of the 129 complainants. A review of the information from the 213 complainants whose service providers retained records determined that the nature of their complaints was very similar to that of the initial 129 complainants, and did not provide enough new information to provide reasonable grounds to seek additional production orders.
77. This meant that it was no longer possible to seek further production orders or search warrants; that avenue had come to an end and could not be pursued further. The inability to access the call records of the 213 additional complainants for whom call records existed, but whose complaints came too late to be included in the initial ITOs, meant that some complainant reports could not be checked against concrete telephone records. This made reliance on the co-operation of political parties and telemarketers even more important.
3.3 Follow-up with Political Entities, Their Telemarketers and Other Relevant Actors
78. A significant amount of investigative effort and time was used in attempting to obtain the co-operation of individuals, political parties, telemarketers and telephone service providers. Those who agreed to co-operate sometimes took considerable time to come to that decision or to schedule a response to investigators' requests for meetings; in some cases, they declined to co-operate fully. Some telemarketers and telephone service providers refused outright to co-operate.
79. Investigators conducted over 65 extended interviews with various actors: individuals (some more than once) from each of the major parties, participants in local campaigns, telephone service providers, telemarketing companies and Elections Canada staff. As indicated above, investigators analyzed available data using link analysis software and consulted with subject matter experts in telephony, statistics and memory. Extensive and sophisticated spreadsheets were used to identify, organize and regularly compare each piece of data.
3.4 Calls Made on Behalf of Political Entities
80. Each of the major parties used telemarketing companies to make live or automated calls to electors. In some cases, calls were also made directly by campaign volunteers. Call processes varied by party.
81. At the national level, the Conservative Party, using data from its Constituent Information Management System (CIMS) database, called through two primary telemarketing companies: Responsive Marketing Group (RMG) for live calls, and RackNine for automated calls. Individual candidate campaigns used a variety of telemarketers. For its part, the Liberal Party used its Liberalist database and called electors through two telemarketing companies, Prime Contact and First Contact, while individual candidate campaigns used several telemarketers. The New Democratic Party used its NDP Vote database and called through one telemarketer, Strategic Communications.
82. Investigators also sought and generally obtained the co-operation of the three major parties and their telemarketers to provide access to party officials and to check party database records against the phone numbers of complainants, with the assistance of an independent expert third party. It must be noted, however, that in some instances co-operation was slow in coming. There were a few instances where it took several months or longer for investigators to receive the information they had requested, or for interviews they had asked for to be arranged. In one instance, a person who investigators believed could have provided very relevant information declined to be interviewed.
83. Through this process, investigators were able, in some cases, to track alleged inappropriate calls from the caller to the complainant through call centres and databases. In most of those cases, however, investigators were only able to confirm whether or not a complainant had in fact received one or more calls from a political telemarketer. They were unable to establish the content of that call or determine that the call received was the one about which the elector submitted a complaint. As a result, in many cases, it was simply not possible to gather any information confirming the allegation made by a given complainant that he or she had received an inappropriate call.
84. The next sections describe the results of inquiries with parties and co-operating telemarketers. The parties are discussed in alphabetical order.Footnote 15 Investigators determined that each of the three major parties called some complainant numbers and that, in some cases, electors were given incorrect poll location information. In a few cases, it was admitted that calls made could possibly be perceived as nuisance calls, but were not intended to be so.
3.4.1 Conservative Party of Canada
85. A number of the telemarketers providing the bulk of call services for the Conservative Party or Conservative Party candidates were contacted by investigators. Most agreed to co-operate. Investigators were told that no telemarketer had independent access to the Conservative Party's CIMS database, and that each relied on selected CIMS elector data provided to them by the party or by local campaigns for calling purposes. Two telemarketers, RackNine and RMG, retained recordings of their calls. Only one telemarketer, RMG, advised electors of their poll locations.
86. Campaign Research provided call services to 39 local Conservative Party campaigns and called 89 complainant numbers during the election. The company does not retain recordings of calls made. Campaign Research provided investigators with a list of call display numbers used. One matched a call display number reported by a complainant, but as no call recording was available, investigators were unable to determine the call content. Investigators were advised that Campaign Research callers worked from a script, a sample of which was provided. The text of the script raised no concerns relevant to the investigation. Investigators were told that callers did not have access to, or provide electors with, poll location information.
Front Porch Strategies
87. Front Porch Strategies is an American telemarketer with a Canadian operation. Investigators learned that Front Porch Strategies provided services to 10 Conservative Party campaigns for telephone town hall meetings, and did not provide poll location information to electors. The company provided their call display numbers, none of which could be linked to the impugned calls.
88. RackNine sent recorded messages on behalf of the Conservative Party and a number of its local campaigns during the 41st general election. In all, 405 complainant numbers were matched to RackNine-generated calls under the supervision of an independent expert third party. The role of the expert third party with RackNine, and with other telemarketers, was established by investigators to ensure that database matching and data extraction, where necessary, were done in a way that ensured the integrity of the process and the resulting data collection, and protected privacy interests. The RackNine calls to complainants were linked to 92 separate recorded messages. Upon listening to each of these messages, investigators discovered that 87 messages related to the 41st general election, while the remainder related to other political events or entities. The messages were solicitations of support or announcements of upcoming events; none were problematic and none provided poll location information.Footnote 16
Responsive Marketing Group
Responsive Marketing Group Calls and Communication of Poll Locations
89. RMG provided live calling services to the Conservative Party national campaign and 80 local campaigns, using CIMS calling data provided by the party or by local candidates. RMG was the only telemarketing service in the 2011 general election that made and retained recordings of live calls. The approved GOTV script from the Conservative Party, intended for use in the national campaign and provided to RMG for use by its callers, read as follows:
"Hi, I'm calling for (name from list)
This is (first name) with Prime Minister Stephen Harper and the Conservatives. Your candidate (Candidate name) asked me to call you.
Will you be voting on Monday?
(if yes) Great. Elections Canada has changed some voting locations at the last moment. To be sure could you tell me the address of where you're voting? (check against the address you have)
That's the right address. It's open until (time). What time will you be voting?
Do you need a ride to vote? (if yes, mark it, and have the campaign call them)
I'll have someone call you to arrange your ride.
Elections Canada reported that more than 2 million Canadians voted in the advance poll–up 36% from last election–more than any other advance poll in Canadian history. That means many of our opponents have already voted.
We need you to vote on Monday, because our opponents have already voted.
Will you please vote on Monday? (Wait for answer)
Can we offer you a ride to the polls on Monday?
Thank you for your time."
90. RMG provided investigators with a list of the call display numbers it used, against which complainant information could be matched. In February 2013 and again in September 2013, supervised by an independent expert third party, complainant telephone numbers were matched against RMG call logs and associated files. RMG allowed investigators to select, review and listen to all the calls made to any of the complainants' telephone numbers, as well as to a random sample of 1,000 completed calls to non-complainant telephone numbers. From this, investigators could determine if RMG callers stayed on script when speaking with electors and whether or not the calls made included incorrect poll location information or contained anything that could constitute a nuisance.
91. Information collected during the investigation showed that RMG attempted to call 289 complainant numbers. Not all call attempts were completed. For example, some calls were not answered, went to voice mail or resulted in a hang up. Investigators reviewed all dial attempts to complainant numbers. Investigators discovered that among those calls which did connect and were recorded, only some resulted in an agent talking to an elector about his or her poll location. There were many reasons for this. In some cases, for example, the intended recipient was not available. In other cases, recipients sometimes told the caller they had already voted, hung up, or ended the conversation for some other reason before a poll location could be discussed.
92. In total, investigators reviewed all 126 completed calls to complainants and 1,000 completed calls to other electors. Of the recorded calls in which RMG callers reached the intended elector, the callers identified themselves as calling on behalf of the Conservative Party in almost every case. None said they were calling from Elections Canada, although, as noted in the script, the words "Elections Canada" were used twice and there was a reference to a change in some poll locations.
93. Investigators found that a number of RMG callers told electors at which poll location they should vote, rather than asking electors to verify the poll location indicated on their VIC as outlined in the script. As well, a number of the RMG calls identifying a specific poll location provided incorrect information, usually directing electors to a location farther away than their assigned poll.Footnote 17 Other RMG calls directed electors to the same location as that on the elector's VIC. Most calls did not provide a poll location.
94. Investigators grouped call content into two different categories: calls that did not provide a poll location, and calls that did provide a poll location. Where a poll location was mentioned, investigators grouped call results into three subcategories: “RMG provided correct information,” “RMG provided incorrect information” and “impossible to determine if information RMG provided was correct or not.” This latter subcategory refers to a conversation in which electors did not confirm or refute, during the call, the information regarding their poll location. This subcategory applies only to 1,000 sample calls. For complainants, investigators were able to determine the assigned poll location and whether or not the information provided by RMG was correct. Figure 4 shows the results of the review.
|Category of Discussion||Sample Calls||Calls to Complainants|
|Poll location not provided||720||72.0%||65||51.6%|
|Poll location provided|
|RMG provided correct information||235||23.5%||27||21.4%|
|RMG provided incorrect information||10||1.0%||34||27.0%|
|Impossible to determine if information RMG provided was correct or not||35||3.5%||0||0.0%|
95. It is important to highlight some of the information contained in Figure 4, which was compiled by investigators when they listened to the recordings of calls made by RMG callers.
96. Investigators have determined that, with respect to the complainants, in 51.6% of the cases poll location was not provided and in 21.4% of the cases the poll location information provided by the RMG caller was correct. Put another way, in 27.0% of the cases involving complainants, RMG provided wrong poll location information.
97. Importantly, while some RMG callers who provided electors with incorrect poll locations told the elector where to vote, the majority of calls providing an incorrect location were more nuanced. In the latter cases, the RMG caller stated a different poll location than the electors saw on their VIC, but the caller acknowledged that the elector should check with Elections Canada to be sure.
98. Finally, as Figure 4 shows, in the sample of 1,000 randomly selected conversations that were listened to, investigators determined that in 72.0% of the cases poll location was not provided. In 23.5% of the cases the information provided was correct, and in 3.5% of the cases investigators were not able to determine if the information given was correct or not. On the other hand, investigators were able to confirm that wrong information was provided in 1.0% of the sample cases.
Responsive Marketing Group Calls and Conservative Party Officials
99. Investigators were told by the Conservative Party national campaign chair that Elections Canada had no authority to limit a party's use of the poll location data, as outlined in the Elections Canada caveats and restrictions (see paragraph 40 above) that accompanied the information. He said that the calling strategy was to tell supporters where to vote rather than refer them to Elections Canada, for fear the elector would not bother to call and consequently might not vote.
100. Through information obtained from the Conservative Party, investigators learned that during the final days of the election some electors complained to one local Conservative Party campaign of Conservative Party callers providing incorrect poll location information to party supporters. The local campaign challenged the accuracy of CIMS poll location data, and one campaign worker suggested to the national campaign that the callers stop giving electors "polling advice." In one e-mail response, the party's Coordinator of Direct Voter Contact noted that the incorrect information arose from CIMS placing electors in the wrong polling division. He concluded that "[t]his is a very small group of people, but there will be a handful in every riding." This statement is consistent with the number and distribution of complaints of calls providing incorrect poll locations outside the electoral district of Guelph.
101. The coordinator also told investigators that he recalled receiving some "notes," originally from Elections Canada, about calls providing electors with incorrect poll location information. These notes were provided to him by party officials. He advised investigators that his practice in such a case was to ask RMG if they had called the elector, and if so, to provide him with the call recording. In every case he could recall regarding complaints of an incorrect poll location, the issue had been a one-off error in the CIMS database and not a system error affecting many electors at once.Footnote 18 He said that having found that the problems related to data rather than a systemic error, he allowed RMG calling to continue; had he discovered a systemic error, he would have halted calling. He noted that all databases have errors despite best efforts at data cleaning.
102. The reference to some "notes" of calls from Elections Canada may refer to events of April 29 to May 1, 2011, when returning officers in 11 electoral districts reported elector complaints of incorrect poll location information coming from Conservative Party callers. When contacted by returning officers, local Conservative Party campaigns advised that the calls were from the national campaign of the party, and that the campaigns could not stop them. Through established channels, Elections Canada raised the issue twice with the Conservative Party. The party twice assured Elections Canada that they understood that some poll locations had been changed by returning officers and that, in consequence, candidates were confirming the proper poll location in calls to a number of supporters.
3.4.2 Green Party of Canada
103. No issue arose about inappropriate calls involving the Green Party. One candidate self-reported that the autodial company she employed had looped calls to a small number of electors so that the same electors were called repeatedly for a period of time. The campaign apologized to each elector and explained the problem. None complained to the Commissioner.
3.4.3 Liberal Party of Canada
104. A number of electors complained of receiving inappropriate calls from someone claiming to be from the Liberal Party or from one of its candidates' campaigns. The Liberal Party and its campaigns used telemarketers Prime Contact and First Contact as their primary call centres. First Contact calling numbers were matched to three complainant reports; however, without a recording of the call or a detailed report of the call content from a complainant, the substance of the calls is unknown. No Prime Contact calling numbers were available. Prime Contact and First Contact do not retain their calling data, but rather upload it into Liberalist (see below). Investigators were told that the telemarketers used by the Liberal Party at the national level did not advise electors of their poll location.
105. Investigators also learned that some local Liberal Party campaign callers did provide poll location information. Evidence was found indicating that the party's campaign callers in two electoral districts provided or appeared to provide incorrect poll location information to two electors. However, no evidence that would establish a malevolent intent was discovered (indeed, in one case investigators learned that the problem was discovered on election day and corrected by the campaign).
106. Liberalist is the Liberal Party database used to track information pertaining to electors, including their phone numbers and party support. Liberalist has the capacity to record calls, but this feature was not activated during the election and so it did not retain call recordings, except for one autodial call. Liberalist recorded the date calls were made, but not the time of day for most calls. Investigators were unable to match a specific Liberal Party call to a complaint in the absence, in most cases, of complainant telephone records and of any record in Liberalist of the time of day calls were made.
107. Complainant numbers were matched with Liberalist data under supervision of the independent expert third party. The analysis showed that 480 complainant numbers were called. Calling scripts were retained in only a few cases, and these were provided to investigators by both the Liberal Party and various Liberal Party campaigns. The content of the scripts raised no concerns on review by investigators. However, scripts only indicate what callers are supposed to say to electors, not what was actually said. The data matching process allowed investigators to determine whether a Liberal Party telemarketer called an elector, but not the content of the call.
108. As well, investigators looked into reports that persons believed to be posing as Liberal Party supporters called electors during what was for those electors a religious holiday period. For the purpose of this investigation, such a complaint would only be relevant if there was evidence to demonstrate that the caller knew or had reason to believe the call recipient was of a particular religious faith, and that the call was intended to prevent the elector from voting or, by some pretence or contrivance, to induce the elector to vote or not vote for a particular candidate. No such evidence was located. One relevant local Liberal Party campaign confirmed, during the election campaign, that they had called electors in the electoral district during the religious holiday period, and might have called a complainant as they could not differentiate electors by religious faith.
3.4.4 New Democratic Party
109. Four complaints related to calls that allegedly came from the New Democratic Party or some of its candidates: two complaints of calls providing incorrect poll location information, and two of nuisance calls. No independent evidence could be provided by the complainants, and their service providers did not retain call records.
110. The New Democratic Party uses a database called NDP Vote. Calling is primarily carried out by Strategic Communications, with limited volunteer calling by the New Democratic Party itself or its local campaigns. New Democratic Party callers did provide electors with poll location information, obtained from NDP Vote, but no call recordings were made and no elector call logs were retained. Strategic Communications provided the call display numbers used, none of which matched complainant reports.
Return to source of Footnote 13 A phishing scam over the telephone is typically one in which the caller purports to be calling on behalf of a legitimate company, but which is really intended to trick recipients into providing personal, financial or password data.
Return to source of Footnote 14 Political calls were confirmed, and in some cases their content determined, using database records of several organizations. This included voice recordings from the telemarketer Responsive Marketing Group, about which more will be said later in this report.
Return to source of Footnote 15 The Bloc Québécois is not referred to in this report as it was not named in any complaint.
Return to source of Footnote 16 RackNine was the autodial service that broadcast the voice message at the centre of the Guelph investigation. The independent expert third party examination of the RackNine database, for the purpose of the present investigation, excluded the Guelph message from consideration.
Return to source of Footnote 17 In some cases the difference was a few kilometres; in several cases the difference was more than 100 kilometres. The furthest location, provided to an elector called twice by RMG, was 740 kilometres from the correct location.
Return to source of Footnote 18 He provided the example of an Ontario elector whose name was listed twice in CIMS, at an identical street address but in different towns.