The Elections Department introduced two new ways of voting in the 2023 Presidential Election – special polling stations at nursing homes and voting by postal ballot for overseas voters. Both of these innovations represented significant departures from the Government’s traditional stance that voters must present themselves in person at designated polling stations to be allowed to vote. I volunteered as a polling agent and counting agent for Mr Tan Kin Lian and observed the conduct of the polls at one of the nursing homes, and the counting of votes from nursing homes on polling day. I also observed the counting of overseas votes including postal votes eleven days later.
Voting for nursing home residents
Special polling stations were set up at 31 nursing homes to facilitate voting by about 4,000 nursing home residents, and I observed the voting process as a polling agent at one of these homes, Ren Ci (Ang Mo Kio). About 200 residents were eligible to vote at this nursing home. Based on the copy of the electoral register provided to me as a polling agent, these residents were from 28 out of the 31 constituencies in Singapore.
For residents who were able to walk or get around on a wheelchair, the process was very similar to that in an regular polling station, except that the residents voted in a small polling station that had been set up in the function room of the nursing home. Many of the residents did require some level of assistance from nursing home staff.
About one-fifth or 40 of the voters were bedridden and the very special aspect of a “special polling station” was that the polling booth went to them, and they could vote from their beds. According to the Elections Department(ELD), this was an evolution of the process that had been used during the Covid pandemic to allow voters under Stay Home Notice to vote in the 2020 General Election. A team of Presiding Officers (POs) and nursing home staff went bed-to-bed to deliver ballot papers to bedridden residents so that they could vote from their beds.
ELD must be applauded for its effort to make it possible for elderly Singaporeans with limited mobility to excercise their civic right and duty to vote. Unfortunately, I have grave concerns as to whether the voters were fully aware of what they were doing. Even among voters who were able to walk or get around in a wheelchair, I would estimate that 10-20% were not fully aware that they were voting in an election or of who they were voting for. Among the bed-ridden patients, my estimate was that at least three-quarters of the patients were not fully aware of what was happening around them.
ELD was aware of this issue, and when it first proposed polling in nursing homes, it said it would ”align the approach to handle nursing home voters who may lack the mental capacity to vote with the current approach adopted at regular polling stations. For example, the voter will not be issued a ballot paper if he does not respond to the election official’s repeated requests for his identification document and poll card.”
Unfortunately, this may not be sufficient in a nursing home setting. In a regular polling station, there is already some selection in that a voter has to be reasonably aware to even be able to present himself at a polling station. When the polling booth goes to a voter’s bed, however, that pre-selection does not take place. According to ELD, nursing homes assessed the mental capacity of patients before putting their names on the list of voters eligible to vote. Yet many of the voters who had been pre-selected by the nursing homes were not allowed to vote because they were unable to identify themselves when asked to do so by Presiding Officers(POs). And even among those who were given ballot papers, about 15% were unable to mark their ballot papers properly so their votes ended up being rejected during counting.
Just as in regular polling stations, POs are permitted to assist voters in marking their ballot papers. Unfortunately, given the poor mental and physical health of many of the patients, and the resulting difficulties in communication, I am not fully convinced that all the votes recorded fully represented the deliberate choices of the voters.
Counting of votes from nursing homes
Votes from all 31 of the nursing home polling stations were counted at a single counting centre, New Town Secondary School (NTSS), following the close of polls on polling day. The counting process was identical to that in a regular counting centre, except that ballots from 15 or 16 nursing homes across Singapore were mixed together at each counting table before counting. This is in contrast to the procedure at regular counting centres where typically only balllots from a single polling district are mixed together on one counting table. Overall results from all of the nursing homes are shown in Table 1.
| Ng Kok Song | Tharman | Tan Kin Lian | Total Valid Votes | |
| Votes received | 659 | 1,486 | 530 | 3,157 |
| Percentage of Valid Votes | 24.6% | 55.6% | 19.8% | 100% |
Mr Tharman’s vote share was significant lower than his national average of 70.4% and even lower than the 59.3% he received at one regular polling station in Hougang. It is possible that this genuinely reflects his support among nursing home residents, but I am unable to exclude the possibility that this result was due to a large proportion of the nursing home residents picking a choice at random.
High proportion of spoilt votes
| Valid Votes | Rejected votes | Total Votes Cast | Rejected votes/ Total Votes Cast | |
| Special Polling Stations (nursing homes) | 2,675 | 482 | 3,157 | 15.3% |
| Regular polling stations in Singapore | 2,478,085 | 49,670 | 2,527,755 | 2.0% |
This hypthesis is supported by the very high percentage of rejected votes or “spoilt votes” from the special polling stations – 15.3% compared to only 2% in regular polling stations (see Table 2). In my experience as a counting agent in past elections, the great majority of spoilt votes appeared to have been deliberately spoilt (e.g., X drawn in between the boxes, multiple X’s marked, or obscenities written on the ballot paper) so that they would not be counted. During the counting of votes from special polling stations in 2023, however, I observed that almost all the rejected votes appeared to have come from voters who were genuinely unable to mark the ballot papers correctly.
X-stamps were provided to the nursing home voters but the stamps were often placed randomly all over the ballot paper, and multiple stamps were often made on the same ballot paper. Pens were also provided for voters who had difficulty holding the stamps, but the pen marks were also poorly formed and not interpretable in many cases. My conclusion, unfortunately, is that a signficant proportion of the nursing home voters were not able to make a conscious choice as to who to vote for, and even if they did, they were unable to clearly indicate their choice on the ballot paper.
In special polling stations, just as in regular polling stations, POs were permited to assist voters in marking the ballot paper if asked to do so. Unfortunately, my observation on polling day was that many of the nursing home residents were unable to speak or to be understood by the POs due to their poor state of health. The POs certainly tried their best, in multiple languages and Chinese dialects, but despite the POs best efforts, I have grave reservations as whether all the marked ballots accurately represented the residents’ choices, and whether many of the residents were even aware of what was going on.
High incidence of nursing home residents with lack of mental capacity
Prior to polling day, the nursing homes put forward the names of residents who they assessed as being able to participate in polling. On polling day itself, POs made another assesment of the residents’ mental capacity by asking them to confirm their name and identity card number before issuing a ballot paper. Despite the pre-selection, turnout at special polling stations was only 77% compared to the national average of 94%. The most likely causes of the low turnout were that the resident’s health deteriorated before polling day, or the PO assessed on polling day itself that the resident did not have the mental capacity to vote.
| Valid votes | Votes cast | Eligible voters | Turnout (%) | Valid Votes/ Eligible voters(%) | |
| Special polling stations (nursing homes) | 2,675 | 3,157 | 4,087 | 77% | 66% |
| Regular polling stations | 2,478,085 | 2,527,755 | 2,698,671 | 94% | 92% |
After excluding rejected votes (spoilt votes), only 2,675 valid votes (or 66%) were received out of 4,087 eligible voters in special polling stations. This is significantly lower than the corresponding figure of 92% for regular polling stations becase of the high proportion of spoilt votes and the low turnout for nursing home residents.
Eighteen elections officials were present at the special polling station that I observed, assisted by about half as many nursing home staff. Given that there were 31 special polling stations altogether, I estimate that about 560 elections officials were required for all the nursing home polling stations. In addition, about another 30 elections officials were deployed for the counting of votes. All in, about 1 elections official was required for every 4.5 valid nursing home votes. In comparison, only one elections official was required for every 70 votes accepted in the nation as a whole.
| Valid Votes | Estimated number of election officials | Valid Votes/ Elections official | |
| Special polling station (nursing home) | 2,675 | 590 | 4.5 |
| Overall Total | 2,484,481 | 36,000 | 69.0 |
ELD must re-examine voting at nursing homes
In a reply to a Parliamentary Question by NCMP Hazel Poa, ELD acknowledged that “election officials did face challenges including managing voters who may lack the mental capacity to vote” during voting at nursing homes. I concur with that assessment, and I urge ELD to re-examine the pros and cons of having special polling stations at all. At the very least, ELD must reconsider the value of having mobile polling stations going to residents’ bedsides.
As described above, I had serious doubts as to the mental capacity of many of the nursing home residents who voted in the 2023 election. In particular, I estimate that at least 75% of the bedridden voters did not have the mental or physical capacity to vote. While ELD tried to overcome the physical barriers by allowing POs to mark ballot papers on behalf of residents who were unable to do so themselves, I am not completely convinced that the residents were able to communicate their intentions to the POs properly.
Risk of a disputed election
In the 2023 Presidential Election, Mr Tharman beat the next highest candidate, Mr Ng Kok Song, by over 1.3 million votes so the impact of votes from special polling stations was negligible. In 2011, however, only 7,832 votes separated Mr Tony Tan from Mr Tan Cheng Bock. In other words, it would have only taken 3,691 voters to have shifted their vote from Mr Tony Tan to Mr Tan Cheng Bock for the outcome of the election to have been different. If we ever had such close elections again, any doubts over four-thousand-odd votes from nursing homes would raise questons over the results of the election. There may even be legal challenges to invalidate the election reminiscent of the legal battle over “hanging chads” in the 2000 Presidential Election in the USA. To prevent anything similar happening in Singapore, I suggest that ELD,
- Reconsider the use of mobile polling stations. The manpower and other resources required to conduct mobile polling is extremely high when viewed against the number of valid votes that are eventually accepted from bedridden voters. It is also difficult for candidates’ counting agents to observe mobile polling even though mobile polling is the highest-risk area with regard to the mental capacity of voters and the conduct of polling.
- Tighten the criteria used to pre-select residents for inclusion in the electoral register for special polling stations. Even among the residents who were mobile enough to be brought to special polling stations on the nursing home premises, I estimated that at least 10% did not have the mental capacity to vote. About three-quarters of the bedridden voters also did not have the capacity to vote. While these voters would not receive ballot papers if POs determine on the spot that they lack the mental capacity to vote, it is a strain on the nursing home and on elections officials to have to make accomodations for these residents even though there is a high chance that they may not be able to vote at all, or that their ballot may eventually be rejected.
- Re-examine the costs and benefits of special polling stations. Fifteen times as many elections officials were required at special polling stations than regular polling stations for every vote that was eventually accepted. It may be more cost-effective for the Government to consider other means of giving nursing home residents the chance to vote. For example, the Government could consider paying for transport and staff to bring nursing home residents who are capable of voting to regular polling stations than it is for ELD to set up 31 polling stations and deploy almost 600 elections officials for less than 2,700 valid votes.
Counting of overseas votes and postal votes
Voting by post was introduced in the 2023 presidential election to make it more convenient for overseas Singaporeans to vote, compared to having to make their way to one of only a handful of overseas missions to cast their vote in person. Ten days were given for overseas ballot boxes and postal ballots to reach ELD, and counting of both overseas votes and postal votes was conducted on 12 September 2023.
The procedure for counting of ballots from overseas polling stations was similar to that for ballots from regular polling stations in Singapore, and all the overseas ballots were mixed together and counted on one counting table. As shown in Table 5, the 76.2% vote for Mr Tharman was above the national average of 70.4 % and the incidence of rejected ballots (spoilt votes) was almost the same as that for ballots cast in Singapore (see Table 2).
| Ng Kok Song (a) | Tharman (b) | Tan Kin Lian (c) | Valid Votes (a+b+c) = (d) | Rejected (spoilt votes) (e) | Votes cast (d + e) = (f) | |
| In-person votes at overseas missions | 409 | 1832 | 183 | 2424 | 30 | 2454 |
| Percentage of valid votes | 16.0% | 76.2% | 7.8% | 100% | ||
| Percentage of votes cast | 97.9% | 2.1% | 100% |
The results from the postal ballots were even more in Mr. Tharman’s favour, with him gaining 77.3% of the vote. However, the proportion of spoilt votes was higher, at 3.6% or 1.75 times the proportion of spoilt votes in the country as a whole.
| Ng Kok Song (a) | Tharman (b) | Tan Kin Lian (c) | Valid Votes (a+b+c) = (d) | Rejected (spoilt votes) (e) | Votes cast (d + e) = (f) | |
| Postal votes | 186 | 1002 | 109 | 1297 | 48 | 1345 |
| Percentage of valid votes | 14.3% | 77.3% | 8.4% | 100% | ||
| Percentage of votes cast | 96.4% | 3.6% | 100% |
One of the reasons for the higher number of spoilt votes is that section 33(1)(c) of the Presidential Elections Act,
33.—(1) The Returning Officer must reject as invalid the following ballot papers only ….
(c) any ballot paper on which anything is written or marked by which the voter can be identified except the printed number on the back;
Presidential Elections Act
requires that any ballots where the voter could be identified must be rejected. Some voters had signed their names on the face of the ballot paper, so those ballot papers had to be rejected. Some voters had printed their ballot papers double-sided with the return label on the other side, but since the return label included a QR code identifying the voter, these ballot papers also had to be rejected.
ELD’s instructions to postal voters do specifically mention that the ballot paper and return envelope should be printed on separate sheets of paper, but they do not emphasise that the vote would not be counted at all if they are printed double-sided on the same sheet of paper
The bigger concern with the postal ballots was that 41% were disallowed even though they had reached ELD on or before the September 11 deadline.
| Valid votes | Postal ballots accepted | Postal ballots disallowed | Postal ballots received by Sept 11 | Registered postal voters | |
| Postal votes | 1,297 | 1,345 | 918 | 2,263 | 3,432 |
| Percentage of postal ballots received | 57.3% | 59.4% | 40.6% | 100% | |
| Percentage of registered postal voters | 37.8% | 39.2% | 26.7% | 65.9% | 100% |
ELD has told the Straits Times that the postal ballots were disallowed because the “return envelopes were found torn, unsealed or opened, or had absent, faint, illegible or late postmarks” even though they had arrived on time. From what I observed as a counting agent, problems with the postmark were the biggest cause leading to rejection.
Postal voters could download postal ballots starting the day after nomination day and were required to post their ballots by the eve of polling day so any postal ballots which were received by ELD on or before August 31 would have been allowed. For postal ballots received between September 1 and September 11, ELD would examine the postmark to determine whether the return envelope was posted on time. The return envelope provided to voters to download was a business reply envelope but many countries do not routinely postmark business-reply envelopes, and even for those which had been postmarked, the date was often not clear so ELD could not accept them if they were received after polling day. Some voters went to the expense of returning their ballots by courier service or express mail. Unfortunately, because the official ELD return envelope was enclosed inside another envelope provided by the courier service, those return envelopes would also be disallowed because they were not postmarked.
The problem with postal ballots being rejected because of missing postmarks is not unique to Singapore and even if voters affix postage stamps to the return envelopes as advised by ELD, there is no guarantee that the stamps would be postmarked legibly or at all.
Apart from these reasons, some postal ballots were also disallowed because the signature on the outside of the return envelope did not match the signature that the voter had uploaded to the ELD website when he or she registered as a postal voter. Some ballots were also rejected because two ballot papers were found inside the same return envelope.
Turnout of overseas electors
| Overseas votes cast | Registered overseas electors | Turnout | |
| 2023 Presidential Election (postal) | 1345 | 3432 | 39.2% |
| 2023 Presidential Election (in-person) | 2454 | 3217 | 76.3% |
| 2023 Presidential Election (postal + in-person) | 3799 | 6649 | 57.1% |
| 2020 General Election | 4794 | 6570 | 73.0% |
| 2015 General Election | 3415 | 4868 | 70.2% |
| 2011 Presidential Election | 3375 | 5504 | 61.3% |
| 2011 General Election | 2683 | 3453 | 77.7% |
| 2006 General Election | 335 | 1017 | 32.9% |
When measured as the number of votes cast as a proportion of registered voters, the turnout for overseas polling in 2023 was dismal at 57% – the lowest since 2011. Only 3,799 overseas votes (including spoilt votes) were eventually accepted for counting, fewer than the 4,794 overseas votes accepted in 2020 when overseas polling was conducted in the midst of Covid restrictions around the world. This may have been because the election was a presidential election and some potential voters may have seen it was a lower stakes election than a general election. In 2011, the presidential election also had a lower turnout percentage than the general election which had been held just three months earlier.
The bigger factor, however, was the large number of postal ballots which either arrived late or were disallowed for postmark and other reasons. Less than 40% of registered postal voters eventually submitted postal ballots which were accepted and counted. If all of the 2,997 postal ballots which had been downloaded were eventually accepted and counted, the turnout for overseas voters including postal voters would have been 5,451 or a very credible 82% of registered overseas voters.
Shih Tung's Wordpress Blog 