ESOMAR Guideline on Social Media Research

CONSULTATION DRAFT

DRAFT VERSION 5

1.  INTRODUCTION

The evolution of social media in recent years has fundamentally transformed the lives of billions of people around the world. The concept of consumers generating their own content on the internet has become ubiquitous. This has created new opportunities for researchers to observe, interact and gather information. Already many techniques have been developed to leverage social media such as community panels, crowd-sourcing, co-creation, netnography, blog mining and web scraping. Moreover it is likely that many more will evolve over the coming years as the internet continues to change.

The ICC/ESOMAR International Code on Market and Social Research requires that the same fundamental ethical and professional principles which govern face to face, mail and telephone research, are also applied to all types of online research (see Appendix 1). This document aims to provide guidance for the use of social media in market, social and opinion research.

It builds upon previously issued guidelines concerning Online Research and Passive Data Collection and supports ESOMAR’s mission to ensure effective self-regulation of our profession and industry. It aims to provide social media researchers with an awareness of the issues and guidance on how they can best apply the Code’s fundamental principles of respect for consumers, trust, transparency and professionalism to them. These guidelines also promote our consistent position to maintain a clear distinction between market, social and opinion research and marketing and PR activities.

Recognising that online research generally, and social media specifically, is continually evolving, these guidelines primarily endeavour to convey principles rather than be prescriptive over precise methodologies.

While regulations in this area are still evolving, the guidelines are based on the principles underlying relevant laws and regulations presently in operation, especially with respect to data privacy and intellectual property.

1.1. Scope

These guidelines cover the collection of social media data for research purposes.

Data collected by services involving the methods of social media research (e.g. brand sentiment monitoring) could be used by clients for non-research purposes. Researchers that are engaged in collecting social media data that will be used for purposes other than research must clearly differentiate this activity from their research activities and not misrepresent it as research. See section 2.1 for more on this issue.  

This guideline shall be read in conjunction with the ICC/ESOMAR International Code on Market and Social Research and other ESOMAR guidelines available at http://www.esomar.org/

1.2. Definitions

Social media is defined as internet based platforms and technologies that permit users’ interaction and/or facilitate the creation and exchange of user generated content. Whilst the scope is evolving, currently the most frequently used examples include:

·         Online forums/discussions, communities, blogs, social networks (e.g. Facebook)

·         Video/photo sharing (e.g. YouTube)

·         Multi-person/group communication and/or collaboration platforms (e.g. Twitter).

Social media data refers to the information (photos, comments, etc.) that users generate or share while engaged in or with social media.

Social Media Research (SMR) covers all research where social media data is utilised either by itself or in conjunction with information from other sources. Examples of current social media research include:

·       Monitoring or crawling social media platforms (from automated monitoring of brand sentiment through to ad-hoc desk research)

·       Ethnographic research (from observing online behaviour to collecting primary data in various forms, including ‘friending’ users)

·       Co-creational techniques

·       Online communities that generate or deliver consumer opinions, reactions, feedback on a regular, formal or systematic basis.

Social Media Data Services (SMDS) refers to the provision of data from the social media world provided for a non-research purpose (see section 2.1 for a more detailed definition and discussion).

Throughout these guidelines a number of specific terms will be used whose meanings are as follows:

Cloaking – is a masking technique whereby the original social media data such as comments, photos or videos is altered to a point that it cannot be traced back or attributed to the original user (using a search engine for example). See Appendix 2.

Client - the definition in the ICC/ESOMAR Code is any individual or organisation that requests, commissions or subscribes to all or any part of a market research project.

MROC (Market Research Online Community) - is one of the more frequently used terms used to describe an online community created specifically for the purposes of market, social and opinion research. Others include DORC (Dedicated Online Research Community).

Researcher – the definition in the ICC/ESOMAR Code is any individual or organisation carrying out, or acting as a consultant on a market research project, including those working in client organisations.

Respondent - the definition in the ICC/ESOMAR Code is any individual or organisation from which information is collected for the purposes of a market research project, whether they are aware of it or not, or is approached for interview.

Scraping – the process of extracting data from social media data for analysis. This can be automated or done manually.

ToU – the Terms of Use policy that a website or online service requires its users to accept.

Walled garden – an online service which requires users to register or apply for membership before being permitted to participate. A walled garden can only be accessed after the user has obtained a login and/or password, even if entry is automatic.

2.  KEY PRINCIPLES

All the core fundamental principles of the ICC/ESOMAR Code (see Appendix 1) apply to social media research. The following section explains the implications of them for this context.

2.1. Distinguishing market, social and opinion research as the purpose

Researchers must ensure that the data they provide is not used for any purpose other than market, social and opinion research. This can cause particular issues with social media.

The ICC/ESOMAR Code requires researchers to be transparent in their dealings and not to misrepresent as market research any project which has another purpose (see definition of market research in the ICC/ESOMAR Code). To aid clarity and avoid potential harm to the reputation of the researcher and of market research, it is recommended to present research services and the organisation or company carrying them out in such a way that they are clearly differentiated from any non-research activities (see ESOMAR guideline on Distinguishing Market Research from Other Data Collection Activities). Furthermore, when provision of social media data could include its use for other purposes such as sales and promotion, the project should be called a ‘social media data service’ and not ‘market, social or opinion research’.

In some cases those taking part in research-based communities could be exposed to sales and PR messages as part of the research process. This is permissible under the ICC/ESOMAR Code provided the purpose is for research (see section 4.4 for more guidance on how best to inform participants about the purpose).

2.2.  Protecting identifiable data

Social media platforms offer many opportunities to view personally identifiable data. Some people post information that overtly discloses their identity and they are aware of this. Others are not aware that the services they are using are open for others to collect data from or think that they have disguised their identity by using a pseudonym or username. However, online services are now available that make it easy to identify a “poster” from their username or comments and can link that to many other aspects of personally identifiable data including their address, phone number, likely income and socio demographic data.

To limit issues where the end client may have access to data linked directly to its source, researchers are required to inform the client that the information being provided is for research purposes only and have contractual terms in place requiring clients to observe the ICC/ESOMAR Code and the provisions of this guideline. Alternatively, they must present such services as ‘social media data services’ and ensure that what they are doing is clearly differentiated from their research activities when dealing with respondents or clients (see section 2.1).

2.3. Ensuring no harm

Another key principle of the Code is that the rights of respondents as private individuals shall be respected and that they shall not be harmed or adversely affected as the direct result of participating in research.

The greatest risk in social media research relates to inadvertently revealing the identities of participants who would not expect to be identified. Again to ensure respondents are not harmed by research activities, the abiding principle must be one of caution, removing any personal identifiers in data (unless it is collected from a public space) as soon as possible taking into account necessary quality controls. 

2.4.  Conforming to the law

Researchers must conform to all relevant national and international laws. There are two main legal aspects to consider here.

First social media research must comply with national and international data privacy legislation and relevant requirements for notice, consent, accuracy, security and access when personally identifiable data is collected and stored. There are also legal issues and measures (e.g. Safe Harbour requirements) relating to international transfer of personal data where data from which personal identifiers have not been removed or cloaked is transmitted across national borders.

The second legal issue is that by accessing virtually all online services, researchers will be subject to the service owners’ Terms of Use (ToU). Most ToU have intellectual property rights clauses that explicitly forbid the unauthorised copying of material. Many go further to bar all forms of scraping.

Subject to ‘fair use’ exceptions in certain countries, such ToU could prevent a researcher from even copying material to their computer for further analysis and forbid any form of selling on of that information to their clients, without permission.

For example, the following popular social media ToU can be located on the web at:

Facebook: http://www.facebook.com/terms.php

Twitter: http://twitter.com/tos

LinkedIn: http://www.linkedin.com/static?key=user_agreement

Researchers should therefore check what conditions apply to the content they use from social media and respect any requests for privacy (including robot.txt file requests). They are advised to seek permission to scrape content from any source where this might breach the ToU and to abide by that service’s ruling. Where permission is not granted, reading of such information and summarising the issues without copying anything is permissible subject to the guidelines in section 4. 

Where researchers use third party aggregators for scraping services, the onus is on the researcher to check with their supplier that permissions have been obtained and the data has been sourced lawfully (see Appendix 3).

2.5. Consent

The ICC/ESOMAR Code states that respondents’ co-operation must be based on adequate information about the purpose and nature of the project and their agreement to participation obtained.

Although obtaining consent is potentially easy for members of market, social and opinion research communities, it poses more issues for other social media where respondents may not have been informed in advance or have consented to its use for research. 

Research data must be anonymised or cloaked before being passed on to clients (for example in reports) except when:

1.    Permission has been obtained from the respondent, or

2.    The data derives from a public social media space as defined in 3.1, or

3.    The data is from a semi public space as defined in section 3.2 but is so inconsequential that it does not need cloaking (see guidance in Appendix 2).

If the data is passed on to another researcher in a form whereby respondents could be identified (e.g. for text analytics), the research data provider must have a contract with the recipient of the data which requires them to observe the ICC/ESOMAR Code and the provisions of this guideline.

When researchers, or their automated agents, take a more active role within social media, all opportunities must be taken to inform people they interact with about their purpose, role and how they will use any comments. They must also seek permission from respondents and the service owners or their representatives to conduct their work. Furthermore when taking part, researchers must ensure they do not misrepresent themselves as a genuine member of that social media space.

When researchers, or their automated agents, are working in social media spaces directly, they must provide an email address and a telephone number and/or mail address to facilitate contact and verification. For additional transparency, and to meet data collection notice requirements, researchers must publish a privacy policy on their website which explains how any personal data they collect is handled.

2.6. Children

Researchers must take special care when carrying out research among children and young people (see ESOMAR Guideline on Interviewing Children and Young People).

This is a particular issue with many social media platforms as their users can include children. Where data is likely to be from a child, researchers must take particular care in cloaking responses to ensure that the respondent cannot be identified.

Where a market research online community (MROC) is being conducted with young people, permission must be obtained from a parent or legal guardian (see ESOMAR Guideline for Online Research).

2.7. Reputation of the industry

Researchers must not do anything that might damage the reputation of market, social and opinion research.

Given its specific nature, working with social media requires additional care since any mistakes or misunderstandings can be spread virally within minutes across the network. Social media researchers must therefore be mindful of the core principles of the ICC/ESOMAR Code in the work they and their companies conduct and avoid activities and practices which could undermine public confidence in market, social and opinion research.

2.8. Reporting

The Code requires that projects are reported and documented accurately, transparently and objectively.

As social media research is a relatively new area and data collected cannot be assumed to be representative of a clearly defined universe, care needs to be taken by researchers to explain the impact that this sample source may have had on the results and their validity and reliability. This is necessary to ensure transparency for all involved and to educate users who may not be familiar with this type of research data rather than to devalue the information obtained.

3. DEFINING THE SOCIAL MEDIA SPACE

Before deciding on any specific recommendations for working within the social media space, it is sensible to distinguish different areas of it. We believe there are currently four areas in which market, social and opinion research could be conducted:

3.1. Public space

A place where content is contributed with the assumption that it could be read by anyone in the public and where contributors could not be surprised that it is linked to/copied/cited. Examples include many public blogs or comments left on news websites.

3.2. Semi-public space

A place where people contribute content which although technically open to all to read, many would not expect it to be read or used by people not involved in the specific topic or conversation.

The boundary between public and semi-public space is open to interpretation and researchers are encouraged to act with caution and regard sites as ‘semi-public’ if they have doubts. Examples include Facebook home pages and many niche (but open) forums/communities. It also includes all forms of open synchronous online activity e.g. open chatrooms, Twitter, etc.

3.3. Private space

A place where most users would expect their comments to be private and which is available only to genuine community members. They are often called ‘walled gardens’ as they can only be accessed after the user has obtained a login and/or password, even if entry is automatic. Examples include many private forums, communities and chatrooms as well as instant messaging.

3.4. Specific market, social and opinion research space

An online place specifically created for market, social and opinion research purposes where users have been informed of its function and the use to which their comments might be put. Typically (but not always) these are also private spaces. Examples include Market Research Online Communities (MROC’s), certain online ethnographic and co-creational techniques which utilise social media platforms.

4. RECOMMENDATIONS FOR RESEARCHING EACH SOCIAL SPACE AREA

4.1. Research in the public space

Where people publish content with an expectation that it will be read and spread to others, it is reasonable to define this as a public space and normal public conventions for that should apply.

Social media researchers can potentially make full use of this information in providing research and copy, scrape or use content subject to ToU policies and applicable data privacy laws.

Although quoting and passing on identifiable comments is possible from this area, cloaking may be required when potential harm might come to the participants. Key considerations for judging this include:

·       Sensitivity of subject

·       Inclusion of abusive language

·       Discussion of anything unlawful, embarrassing or likely to impact career opportunities

·       Inclusion of personal data about the respondent or others (except when this is about a well known person in the public domain - provided it is not libellous).

See Appendix 2 for details on how to cloak such comments.

If comments are quoted, researchers need to ensure they comply with relevant national laws.

4.2.  Research in the semi-public space

The definition of a semi-public space is not always clear-cut, as has been noted in section 3.2. However it is intended to refer to those spaces where most people would not expect intentional eavesdropping or want their content to be distributed more widely than the audience for which it was created.

Social media researchers have unrestricted access to content in the semi-public space provided it is permitted by the service ToU and/or with the service provider’s agreement. However, the key difference to a public space is that researchers should not pass on the identity of information derived from it in reports.

Furthermore, where potential harm might come to participants by quoting them (see Section 4.1), responses should be cloaked. The degree of cloaking required depends on the potential risk or harm that might result to the participant if identified and is likely to be greater than that needed for information derived from a public space (see Appendix 2).

As an alternative, researchers can seek permission from “posters” to quote their comments and in doing so, researchers should be sensitive to the respondent’s concerns about being observed and explain clearly and honestly the purpose of their work. In such circumstances, posters should be given three options:

·       Quote and identify me

·       Cloak comment, or

·       Not use comment at all.

The respondent should be given the opportunity to check the bona fides of the researcher, if they so wish, before deciding what to do. Furthermore researchers should be respectful in approaching individuals for permissions.

For some small groups, and especially in a B2B context, it may be impossible to cloak the comment sufficiently to avoid recognition of an individual. In such circumstances no attempt should be made to pass on verbatim comments without permission.

In the case of pictures or videos, consideration should be given to techniques such as pixelation of faces where cloaking is required.

Where researchers use services to provide demographic profiles (e.g. PeekYou) to enhance their data, they should only use this information for classification purposes.  Since such services usually provide identifiable personal information based on a username or real name, such as a phone number, address or email in addition to providing demographic profiles, they allow unintentional linkage of research data to personal data. Researchers should therefore ensure that databases which contain the resulting personal data are kept secure and should only hold the personal information as long as necessary to execute the project. Under no circumstances should personal data be transmitted to another organisation or the client without first obtaining respondents’ permission.

If researchers interact with people in semi-public spaces, they must seek permission and follow section 2.5 to ensure that their purpose, role and the use to which they will put any comments are clear.

4.3. Research in the private space

Private spaces are often called ‘walled gardens’ and are defined as places you can only access with a username and/or login.

Social media researchers can only access these areas with the permission of the service operator or their agents. They should make it clear in their profile and preferably also in their avatar and/or username that they are a researcher, who they work for and their purpose. In interacting with members of the private space they should include a reference to their role, so that members are left in no doubt who they are talking to.

As a general rule, researchers should not copy or scrape content within private spaces, even if they have permission of the site owner. If researchers do so, it should be made clear to all participants that this is happening and they should provide individuals with a process to be excluded from such scraping.

In a private space, researchers must seek permission from posters to quote their comments. They must ‘cloak’ comments before they are passed on to clients, unless the creator has given explicit permission not to do so. The degree of cloaking required depends on the potential risk or harm that might result to the participant if they were identified (see Appendix 2).

Researchers must observe great sensitivity interacting with people in private spaces. When this needs to be done, they should follow the guidelines in section 2.4 and ensure that their purpose, role and use to which they will put any comments are clear before people choose to interact with them. The respondent should be given the opportunity to check the bona fides of the researcher, if they so wish.

4.4. Research in the specific market, social and opinion research space

Market, social and opinion research spaces are normally private walled gardens where members must agree the purpose for which data is being collected and the terms and conditions for participation before they sign up to take part in the community or co-creational project.

These terms must be simple, clearly worded and easy to understand. Members must be fully aware of:

·       The purpose of the space - that it is for research, but in the case of a MROC, that they may be exposed to marketing information

·       That all data may be shared with the client

·       How it could be used 

·       The rules for interacting (i.e. no cyber-bullying, defamatory comments etc)

·       The site privacy policy.

Content can be copied and scraped and utilised for any research purpose, subject to members being fully aware of these applications. However the personal identity of those making comments must be protected. Some research communities offer end-users opportunities to meet members and interact with them directly but this must be only with the consent of members.

Where spaces are communal, members should be warned about contributing personally identifiable information. For example, members in a community should be given the option of using a pseudonym and uploading a photo of an avatar, a pet or inanimate object, rather than requiring that they use their real name and a photo of themselves. In addition on sign up, researchers should provide reassurance to community members that they will never ask for information that could create a risk if lost, misused or disclosed to an unauthorised party, such as credit card numbers, social insurance/social security numbers or bank account details.

The role of moderators and clients should be unambiguously identified as such in all their interactions and communications with members. However care needs to be taken to protect them from being easily contacted outside the project, so it may be necessary to withhold full names and/or emails of such people and to utilise secure email systems within the relevant software.

Where MROC’s are used to test products, advertising and/or communication of messages it is important that respondents are made aware of this. If respondents are taking part in a simulated sales test, it must be clear that they are helping in a research project and not in some form of direct marketing or sales exercise. No personal data collected during the course of an MROC may be used for any non-research purpose such as subsequent direct marketing or promotion to the individuals taking part.

5.  SUMMARY OF SCRAPING OPTIONS IN THE SOCIAL MEDIA SPACE FOR RESEARCH PURPOSES

Space

Available for researchers

Identifiable in reports

Cloaking of verbatim quotes in reports
(see Appendix 2)

1 Public space

Yes, subject to service ToU

Yes, except if might cause harm

No, only required if it might cause harm

2 Semi-public space

Yes, subject to service ToU

No, except with user permission

More likely to be required than a public space and essential if it might cause harm

3 Private space

Only with permission of service

No, except with user permission

Essential unless user gives permission to cite

4 Market research space

Yes

Possible, subject to sign up agreement

No, only required if it might cause harm

APPENDIX 1: KEY FUNDAMENTALS OF THE ICC/ESOMAR CODE

1.    Market researchers shall conform to all relevant national and international laws.

2.    Market researchers shall behave ethically and shall not do anything which might damage the reputation of market research.

3.    Market researchers shall take special care when carrying out research among children and young people.

4.    Respondents’ cooperation is voluntary and must be based on adequate, and not misleading, information about the general purpose and nature of the project when their agreement to participate is being obtained and all such statements shall be honoured.

5.    The rights of respondents as private individuals shall be respected by market researchers and they shall not be harmed or adversely affected as the direct result of cooperating in a market research project.

6.    Market researchers shall never allow personal data they collect in a market research project to be used for any purpose other than market research.

7.    Market researchers shall ensure that projects and activities are designed, carried out, reported and documented accurately, transparently and objectively.

8.    Market researchers shall conform to the accepted principles of fair competition.

 

APPENDIX 2: ADVICE ON CLOAKING

Cloaking is a masking technique whereby raw data is so changed that it becomes very difficult for others to find the data online with a search engine and thereby identify the person who made it. It is a useful technique to ensure that the anonymity of people making comments is preserved where;

1.    The researcher has not sought their permission 

2.    The researcher believes there could be reason for them not to want their identity revealed or where harm could be caused by doing so.

 

It can be applied in varying degrees ranging from just changing the odd word through altering key features of a comment to précising. It is the responsibility of the researcher to decide which is most appropriate to be used which depends on the degree of the following factors:

§       If the topic being discussed is sensitive or personal

§       If abusive language is used

§       If it includes anything against the law

§       If it includes anything embarrassing or is likely to impact career opportunities

§       If it includes any personally identifiable information

§       If it includes and data about others that is not already public.

 

Level 0: Safe verbatims – no cloaking of comment required

Raw examples

Cloaked examples

I think people should lay off drinking Starbucks as the caffeine level is really high

(No cloaking required)

How do I watch YouTube on my blackberry world edition?

(No cloaking required)

 

Level 1: Possibly harmful verbatims that need some cloaking

Raw examples

Cloaked examples

Eating McDonalds turns you into a homo fag

Eating McDonalds makes you a homosexual

Wal-Mart is meeting its ambitious environmental goals. Suck it Gore, you seabass killin' hypocrite

Walmart will meet its tough environment goals. Screw you Gore, you fish killing hypocrite

 

Level 2: Harmful verbatims that must be fully cloaked

Raw examples

Cloaked examples

RE: Charles Spencer, University Of Utah – Having worked with him, I think Charles is an insane bastard. He encourages the abuse of the mice and rats by making his technicians like me transfer mice from one cage to another at unnecessary fast speed (the mice squeak in pain as the mice are grasped by the skin behind the neck-nape- or just the neck). This is inhumane and shows a lack of care for the animals and his staff.

RE: University animal lab technician – Abuse term about their boss. Provides evidence of abuse of mice and other animals under instruction of his boss. Says he is inhumane and not concerned about his team or the animals.

 
APPENDIX 3: CONTRACT/POLICY ADVICE WITH SUB-CONTRACTORS/THIRD PARTY SUPPLIERS OF SMR

When sub-contractors are used, ESOMAR recommends the researcher checks if they follow appropriate practices and procedures, especially with respect to observing any legal requirements and also about the privacy and protection of identifiable respondent data, including:

·       Performance of due diligence when identifying and selecting sub-contractors;

·       Execution of written Non-Disclosure agreements;

·       Execution of written contracts that outline duties, obligations, and responsibilities of the sub-contractors that address all parts of the research process, especially privacy and data protection; parties involved and address non-disclosure requirements; and

·       Engagement in on-going oversight of sub-contractors and their activities

·       Ensuring any data provided to them by sub-contractors is provided legally and in accordance to service ToU.

Policies and contracts relating to the research process and privacy are available from CASRO (http://www.casro.org ) through the CASRO Privacy Protection Program (CASRO 3P).  The CASRO 3P program has been designed to address the needs of various geographies, including the US and the EU.

Model contracts for transfer of personal data from the EU are available from the European Commission (http://ec.europa.eu/justice/policies/privacy/modelcontracts/index_en.htm)

 

Project Team

o     Adam Phillips, Committee Chair, Managing Director, Real Research and Chair of ESOMAR Professional Standards and Legal Committees

o     Ulf Andersen, MD of Synovate Scandinavia

o     Pete Comley, Chairman, Virtual Surveys

o     Ed Keller, CEO, Keller Fay Group, co-founder of WOMMA Word-of-Mouth Marketing Association

o     Peter Milla, Member of CASRO Task Force

o     Annie Petitt, Chief Research Officer, Conversition

o     Niels Schillewaert, Managing Partner of InSites Consulting

o     David Stark, VP, Compliance and Privacy, GfK, member of ESOMAR Professional Standards and Legal Committees

o     Kristin Sharp, CEO of Ipsos Understanding UnLtd

 

Опубликовано на сайте ESOMAR