July 2015 - Intercepts for Social Change; situating marriage equality data in Ireland through random polling.


Aengus Carroll & George Robotham[1]



A significant deficit in social policy and social change research and advocacy is the lack of reliable data, and this is particularly true regarding sexual orientation and gender identity issues. Funding, resourcing and policy development rely on credible and reliable data and documentation. RIWI’s random domain intercept technology (RDIT), which has been used or identified by various civil society organizations as an ICT method to engage new citizen voices,[2] addresses several core challenges to data collection amongst LGBTI populations, especially those in hostile States, in terms of cost, reach, security and reliability.

This is of particular resonance in the same sex marriage context in the US and Ireland, where the remarkable speed of attitudinal change amongst the general public is notable.[3] In both countries, the campaigns have been going on for eleven years (the first marriages were in Massachusetts in 2004, the same year the seminal marriage equality court case was lodged in Ireland). RIWI’s technology demonstrates an important accuracy in its polling of the previously unengaged, and it is clearly capable of capturing attitudinal shifts as they occur with a minimum of resources. As such, and importantly to social change agendas, advocates and activists can shape their campaigns supported by this data, by drawing data from traditionally hard to reach populations.[4]


Irish data

Close examination of RIWI’s 2014 figures against the final outcome of the 2015 referendum in Ireland, indicate that the RIWI random intercept technology was an incisive tool to measuring emergent public opinion on the issue of marriage equality in mid-2014. A number of other factors, as we will discuss, came into play later in the campaign process that increased the percentage actually realized in favor of a ‘Yes’ vote, but these could not have been foreseen in 2014. In fact, RIWI’s randomized data suggested, and subsequently largely reflected, the shape that the actual campaign took.

In the 12 months before the vote, poll headline figures indicating an over 70% ‘Yes’ vote were not considered to be realistic by analysts on either side of the campaign.[5] Generally, these exceedingly high projections were, firstly, dependent on the exclusion of the undecided (the ‘Don’t Knows’), and, secondly, these ‘Yes’ responses were entirely surface-level to question of agreement/rejection of marriage equality and did not account for other reservations on which the ‘Yes’ campaign was subsequently challenged (children, surrogacy, etc).

Traditionally, in Irish referendum campaigns the ‘Don’t Know’ vote in opinion polls largely went against the proposals in the actual vote. As such on the Marriage Equality referendum their exclusion in opinion poll headline figures skewed the projected ‘Yes’ vote positively to such an extent that the figures were clearly unrealistic. This is a problem that was addressed by serious commentators throughout the referendum process.[6] A reason for this state of affairs is that on constitutional issues, people tend to give an instinctive reaction, an answer they feel they should give (so-called ‘social desirability bias’), or an answer that would place them in the presumed majority (so-called ‘herd bias’). [It should be noted that, unusually in constitutional law provisions around the world, the Irish constitutional text can only be changed by a referendum of the people (i.e. not by courts).][7]

A January 2015 Red Sea poll that drilled into the figures somewhat, concluded that the same sex marriage referendum could, in fact, be lost.[8]  It showed 59% strongly agreeing with same sex marriage but a significant percentage of these still had fundamental concerns around parenting, adoption and social construction (the ‘best option’ argument). They stated that if one considers all those in the sample who had concrete reservations on specific question the core ‘Yes’ vote could be as low as 44% while reading for unqualified support for same sex marriage itself the vote stood at around 48%. An Irish Times poll of May 2015 presented that 58% intended voting ‘Yes’ (while actually presenting a 70% headline figure which discarded ‘Don’t Knows”). It also reported 25% No and 17% ‘Don’t Know’ in coming to its 58% figure.[9]

An earlier poll by IPSOS/MRBI from November 2012 for the Irish Times (IT)[10] produced figures amazingly close to RIWI’s random intercept survey of May 2014:

RIWI   May 2014                    52%,    29%,    19%.

IT        November 2012          53%     30%     17%

These two polls were both clearly closer to the realities on the ground. Instantly of course, the question emerges of the many implications of the research instruments for respondents and research practice. The second important question that emerges is about the affordable reach, randomness and representativeness of both these, and other, polling methods.

It was clearly identified by the ‘Yes’ campaign in the summer of 2014 that two key goals had to be achieved: high turnout and the youth vote. The first of these was that the ‘Yes’ vote required at the very least a 50% turnout to ensure a chance of success against the core No vote (around 25%) which was long identified and certain to come out to vote. Experience at the Lisbon and Nice referendums showed that low turnout ensured oppositional wins (thereby triggering reruns).[11] The second goal related to the demographic issue regarding young people – young people were clearly the most in favor of marriage, but traditionally the least likely to come out to vote.[12] Hence, a concerted effort would be required to change this long established pattern for this cohort of young voters.

We now need to examine variables that did come into play on these two key goals that affected the final outcome in terms of percentiles. These confirm that RIWI’s polling in 2014 were incisive, and the variables we now describe were unprecedented in Irish political campaigns. The first of these was a push that resulted in 106,000 new voters,[13] and the second variable that came into play was the enhanced voter turnout through the #hometovote campaign that motivated a wider demographic to come out to vote.

The first area to look is that of the youth vote, identified as a traditional dilemma on matters of Irish social concerns; here we had a two-pronged approach that came into operation.  The first part was a voter registration push in October and November 2014 strongly focused on students around the country and supported by the Union of Students in Ireland and the LGBT youth work organization BeLonG To.[14] The second part was a push to get late voters onto the supplementary register between February and May 2015. The first student registration drive resulted in over 40,000 new voters, and the second phase produced just under 66,000 new voters - again mainly young voters.[15] As this latter cohort were on a separate register, they were identified on polling day as having voted in much higher numbers than the general population - up to 90% at many polling stations reaching 99% in one.[16] It is fair to presume that the earlier cohort of 40,000 young voters having made the effort to register voted in equally high numbers.

In RIWI’s 2014 randomized intercept poll, 79% of the 18-24 year olds presented as voting ‘Yes’. This high figure was supported by conventional opinion polls – many indicating a ‘Yes’ support percentage well into the 80s for this age group. In relation to the 106,000 newly registered voters it was conceded by most commentators (including the main spokesperson for the ‘No’ campaign) that they were predominantly ‘Yes’ voters.[17] Assuming (conservatively - in light of reports from polling stations and opinions of informed commentators) an 80% turnout and an 80% ‘Yes’ vote from these voters, the yield is a total of 67,840 new ‘Yes’ votes. This figure is almost 6% of the total ‘Yes’ vote of 1,201,607 recorded on polling day. The high overall turnout of 60.5% on the day indicates that the turnout of the entire 18-24 years old age group confounded the expected pattern.

The second area to look at was the question of voter turnout. As mentioned, it was understood that for ‘Yes’ to win, the turnout would need to be over 50%. This priority was ratcheted up in the last weeks of the campaign. A strong contributory factor in the final high turnout is considered to be the #hometovote campaign.[18] This was a very strong social media campaign with a worldwide reach targeted at recent emigrants.[19] It had an unprecedented impact in the days immediately preceding polling day with uploaded to social media shots of emigrants coming back to Ireland on trains, planes and boats to vote. Images of many thousands caring deeply enough to travel home to vote featured heavily in traditional media also, with interviews featuring on main news bulletins. This intensified the ‘Yes’ messages of the need for equality and human rights. Combined with canvassing teams telling their own and other’s personal testimonies on the doorstep and media reporting many of these testimonies, there was a tonal softening that had a mobilizing effect on the less motivated ‘Yes’ voter in particular. In Dublin working class areas - where inequality and discrimination are systemically entrenched - the ‘Yes’ vote was particularly high at over 80%. The overall ‘Yes’ vote for the Dublin region at 71% compared to an average for the rest of the country of 58.79% and this higher vote increased the national ‘Yes’ vote by over 3% to 62.1%.[20]

An important question that emerges from the RIWI exercise on attitudes to same sex marriage in Ireland, is ‘can this random intercept technology be relied upon to provide accurate data that can feed into the shaping of advocacy strategies and practices, and into the design and roll-out of policy for marginalized or minority populations?’. From the case study of Ireland regarding attitudes to same sex marriage, the answer would have to be a ‘yes’. The data gathered from a small sample in May 2014 turned out to be surprisingly astute after the two factors described above are taken into account. These two factors delivered over 9 per centile points extra to the total ‘Yes’ vote on polling day. Importantly, the data that RIWI’s intercept produced would have suggested much the same strategic priorities as those that did in fact shape the actual campaign.

General application

On other thematic research areas that stem from a focus on how sexual orientation and gender identity is observed in UN Member States, the RIWI intercept approach appears to address a number of traditional challenges to data collection. Firstly, it is important to note that documentation of human rights situations of sexual and gender minorities, across ethnicities, ages, races and social classes, has been repeatedly identified as crucial to understanding the challenges these populations’ experience.[21] The Trans Murder Monitoring (TMM) Project Map demonstrates this very well; on first glance at this map one might think that trans murders mostly happen in South America, but what the TMM analysis is actually showing us is a map of where trans murders are being documented as such.[22] Clearly, within public policy spheres, core data are essential to development and inclusion of minorities, especially when that inclusion is contested within the framework of a State’s traditional values, as internalized by agents and agencies of the State.

In every UN Member State and territory, discriminatory law or policy operates (directly, or through omission in spheres such as health, education, employment, partnership, documentation, and so on), thereby comprising the work of frontline advocate organizations.[23] In the more LGBT-hostile territories social networks have been hugely useful tools in information dissemination and activist organizing. For surveys and polls, however, organizations have used various techniques to reach LGBT respondents, such as ‘snowball’ methods to reach gay and lesbian constituencies.[24] A number of reasons feed into this methodology, primarily financial as well as skills-based, but one of its drawbacks is that its cumbersome and often overly biased. However, if funders already working with social movements in these territories were aware of a robust and credible data collection method, such as RIWI, where the data remain anonymized, yet verifiable, there is great potential for partnerships at national, transnational and regional levels. RIWI can collect population-centric data in nations where risk organizations would otherwise need to deploy intelligence personnel in-country to obtain reliable ‘situational intelligence’ data for population-centric opinion using expensive surveys. Furthermore, a major issue in using bodies of documentation gleaned under separate projects for comparative purposes is the uneven nature of terminology embedded in questions, and the capacities to thereafter discover and utilize the data. With the dangers attached to exposure of individuals, locating representative samples has proved challenging.

There are many applications to which this technology could be put, particularly in hostile States, but also in garnering information on public attitudes in all States where discrimination based on sexual orientation and gender identity is identified by advocates. In terms of cost, reach, security and reliability, the RIWI random intercept technology has the potential to be a ‘game-changer’ for policy development and service delivery related to sexual orientation and gender identity. In partnership with the lived experience of advocate organizations in territories, this technology has transformative potential for social change agendas.


How RIWI’s Global Survey Technology Works

RIWI is a global survey technology and risk measurement company. When users navigate the Web by typing an Internet domain name, whether it is a gTLD (e.g. www.anyURLtyped.org), a TLD of any kind (e.g. www.anyURLtyped.xyz), a ccTLD (e.g. www.anyURLtyped.co), or an internationalized domain name (“IDN”), into the URL (“address”) bar, they may type in a domain or sub-domain (e.g. www.anyURLtyped/example.com) that takes them to an unintended IP destination. That is, the intended IP destination either does not exist or is inaccurate — such that the user randomly encounters a RIWI survey on that page, which RIWI controls at that given time. RDIT accesses the dynamic and highly scalable flow of online users around the world every day. When mistakes on non-trademarked URLs/IP addresses occur on any device in any country, such as input errors on URLs (technically, a ‘masked domain’ that is resolved to an IP address) or other errors during manual URL input (i.e., people go to registered domains other than the ones they intend to visit), users commonly are shown or ‘land on’ sites/IP addresses that deliver ads, vacant domains, parked domains, expired offers for coupons or non-existent full-page sites or web-browser page displays (e.g. ‘this page does not exist error’), interstitial displays or domains that may be in escrow prior to the transfer of the domain from the domain owner or registry to the future owner. The vast scale of random RDIT sample to which RIWI has access as a result of its intellectual property and sample consists of individuals interacting with devices coming in from a Web browser. Thus the response back to the user going to any location or clicking on any link or button can go anywhere and be processed in any way before presenting a RIWI survey back to the user. Further information can be found at www.riwi.com.



[1] Aengus Carroll, LL.M is an Irish author and editor with particular interest in the socio-legal aspects of international human rights advocacy, especially relating to sexual orientation and gender identity and expression. He is co-author of State Sponsored Homophobia 2015: A world survey of laws: criminalisation, protection and recognition of same-sex love (Geneva; ILGA, May 2015). George Robotham is a long-term LGBT activist in Ireland with a background in community and voluntary sector development initiatives for the Irish civil service, a founding director of Dublin’s LGBTI community resource center, and has a particular interest in the capacity of LGBTI communities to reach the most vulnerable in the provision of frontline services.

[2] Examples: “Emerging Opportunities: Monitoring and Evaluation in a Tech-Enabled World” by Linda Raftree and Michael Bamberger (https://www.rockefellerfoundation.org/report/emerging-opportunities-monitoring/); http://openinggovernment.com; and, http://blogs.worldbank.org//opendata/it-takes-village-taking-open-data-offline-community-indonesia;

[3] This short article looks at the Irish marriage referendum only. The recently released results of Pew's Same-Sex Marriage study in the US (http://www.pewforum.org/2015/06/08/graphics-slideshow-changing-attitudes-on-gay-marriage/) are almost identical to RIWI’s US findings as part of its 51 country, 50,000 respondent Same-Sex Marriage study released in May 2015 (http://ilga.org/immediate-release-equality-marriage-survey-support-sex-marriage-countries-already/). Among those expressing an opinion in 2014 in the US, the Pew survey found 56.5% support and the RIWI study found 57.7% support.

[4] At the United Nations’ launch of the International Lesbian, Gay, Bisexual, Transgender and Intersex Association’s (ILGA World) world survey of laws pertaining to sexual orientation in May 2015, RIWI data on a survey of attitudes to same sex marriage in 51 countries around the world was presented, see video (5) at http://us10.campaign-archive1.com/?u=fbf5a2a6a3ba3d30af7a9ea01&id=a4bc66a95a&e=635a56a5db.

[5] News, “Support For Marriage Equality Increasing” The Irish Times, Monday 7 April, 2014.


[6] Noel Whelan, “Same Sex Marriage: Why the Proposal May Not Be Accepted by Voters”, The Irish Times, 12 December, 2014. http://www.irishtimes.com/opinion/same-sex-marriage-why-the-proposal-may-not-be-accepted-by-voters-1.2034752.

[7] See Article 47 of the Constitution of Ireland. www.irishstatutebook.ie/en/constitution/

[8] Red C Poll.“Same Sex Referendum Could Still Be Lost” 26 January 2015. http://www.redcresearch.ie/news/same-sex-referendum-could-still-be-lost#sthash.p458xQdS.dpbs.

[9] Stephen Collins “Seismic Shift Needed For No Side to Carry Referendum” The Irish Times, 16 May 2015. http://www.irishtimes.com/news/politics/seismic-shift-needed-for-no-side....

[10] Ibid IT, fn 1.

[11] Summary of Referendums 1937-2012. http://electionsireland.org/results/referendum/.

[12] National Youth Council, “Youth Vote Will Be Key to Referendums” 21 May, 2015.


[13] RTE news, “Surge in People Registering to Vote” 18 May, 2015.


[14] Register to Vote “Campaign Launched for Civil Marriage Equality Referendum”. http://www.iccl.ie/news/2014/11/03/register-to-vote-campaign-launched-for-civil-marriage-equality-referendum.html.

[15] For this part of the campaign, the work of LGBTI youth organization BeLonGTo was critical.

[16] Fiach Kelly “Young Voters Engaged, But For How Long?” The Irish Times 25 May, 2015. www.irishtimes.com/news/politics/the-young-voter-engaged-but-will-they-stay-with-politics-1.2224498.

[17] News “Nearly 66,000 New Voters Register Before Same-Sex Vote” The Irish Times 18 May, 2015 http://www.irishtimes.com/news/politics/nearly-66-000-new-voters-register-before-same-sex-vote-1.2217021.

[18] Clara Kenny, “The Journey #HomeToVote: Selection of the best Tweets” The Irish Times 23 May, 2015.  http://www.irishtimes.com/life-and-style/generation-emigration/the-journey-hometovote-selection-of-the-best-tweets-1.2222250.

[19] In Irish law, unlike many other States in the world, emigrants are only entitled to vote for a period up to 18 moths after emigrating, and must physically return to Ireland to do so.

[20] Home News, “Marriage Equality Referendum”, The Irish Times, 25 May, 2015. Print edition.

[21] European Parliamentary Research Service “The rights of LGBTI people in the European Union” Briefing, May 2015. http://www.europarl.europa.eu/EPRS/EPRS-Briefing-557011-Rights-LGBTI-people-EU-FINAL.pdf; UN Human Rights Council, Discrimination and violence against individuals based on their sexual orientation and gender identity, 4 May 2015, A/HRC/29/23 at paragraph 27: “Data are patchy but, wherever available, suggest…” http://www.refworld.org/docid/5571577c4.html.

[22] Transgender Europe (TGEU) “All Reported Murders of Trans People Since 2008”. Interactive TMM Map. (Updated). http://www.transrespect-transphobia.org/en_US/tvt-project/tmm-results/all-tmm-reports-since-2008.htm.

[23] ILGA-Europe, “Rainbow Europe 2015”. Map and other Resources, (Brussels), http://www.ilga-europe.org/resources/rainbow-europe/2015; ILGA World, State Sponsored Homophobia May 2015, (Geneva). http://us10.campaign-archive1.com/?u=fbf5a2a6a3ba3d30af7a9ea01&id=a4bc66a95a&e=635a56a5db.

[24] United States Agency for International Development, Testing the Waters: LGBT People in the Europe and Eurasia Region, (Washington, DC: USAID, 2013) at 37. http://www.usaid.gov/documents/2496/testing-waters-lgbt-people-europe-and-eurasia-region.



Subscribe to our newsletters

Get updates on LGBTI rights issues, and on our work to support LGBTI civil society worldwide