Multiple readers have reached out to Oceanside News with regard to a recent online poll we held. The poll allowed visitors to the Oceanside News home page and Qualicum Beach Town Council byelection page to indicate which candidate they plan on supporting in the byelection. We would like to provide clarification about irregularities our staff and readers discovered.
Over the past few months, we’ve held several online polls on a broad range of issues of interest to our readership. Initially the byelection poll received the typical volume of votes. Then our staff and readers noticed a pickup in how quickly the numbers on the poll were changing.
We were initially pleased to see the high level of reader engagement and then disappointed to discover what appeared to be potential manipulation of this online poll.
Our polls are set up to allow anonymous voting. To mitigate against repeated voting, the software makes use of cookies. While this is not a bullet-proof method, it does provide a level of protection against repeated votes while also providing a level of anonymity for participants (as no account login or other verification is required).
Polls like this are a common feature among news outlets. They are not meant to be a scientific gauge of public opinion, but an opportunity to engage readers, encourage participation and get general sense of reader sentiment.
When we discovered that this poll showed indications of potentially deliberate manipulation, we decided to remove the poll a little ahead of schedule. We then conducted a technical analysis of the data.
Of the 5,670 votes cast, 197 appear to be from unique users. Of the 5,474 repeated votes, 4,630 were for one candidate, likely originating from 2-3 users.
If we count only the votes from unique users, the results were:
Brian Denbigh | 1 |
Sarah Duncan | 122 |
Peter Kent | 3 |
Anne Skipsey | 65 |
Jean Young | 5 |
These are the results without duplicate votes filtered out:
Brian Denbigh | 1 |
Sarah Duncan | 1,024 |
Peter Kent | 5 |
Anne Skipsey | 4,631 |
Jean Young | 9 |