Reflective Activity 2

Case Study: Inappropriate Use of Surveys In 2018, Cambridge Analytica was in the news in the United Kingdom and the USA (Confessore, 2018) for obtaining and sharing data obtained from millions of Facebook users. 

Consider how exactly this happened and why it was used. Find one or two further examples of inappropriate use of surveys and highlight the impact of all these examples from the various ethical, social, legal and professional standpoints that apply. Record your findings in your e-Portfolio. You can also submit your findings to your tutor for formative feedback.

What happened? 
Cambridge researcher Aleksandr Kogan designed an application that featured a personality quiz, and Cambridge Analytica (CA) paid individuals to take it (Rehman, 2024). The results of this quiz were used for the 2016 election campaign of American ex-president Donald Trump, which generated much criticism as it could take advantage of voters with vulnerable profiles.   

The collateral information taken from each account (friends, liked content) affected more than 80 million profiles.    

It is claimed that data was collected and used without consent, which represented Facebook's worst known data leak, which originated an investigation from the Federal Trade Commission for supposedly not safeguarding the data security agreements or regulations such as GDPR.   

The data is blamed for being collected without consent. According to (Bareebe, 2022), only 270,000 users consented to the app using their data, however, the information of 87 million Facebook profiles were accessed by data scientists through their friends' feeds. Facebook on his defence claimed that the participants consented the collection of data.   

The issues
The lack of explicit consent by the users regarding what part of their information was going to be collected and used presents an ethical and legal problem.    

As the collected data was used to build psychological profiles, to use it in personalised advertisements for political campaigns means that most 'vulnerable' people could be targeted to receive ads built to convince voters, like in the case of the American 2016 elections (Rehman, 2019).    

CA came into possession of Facebook's private data, which consisted of tens of millions of users (The New York Times, 2018). This must have gotten Facebook's attention on the security measures that were applied to protect their users' privacy. So the problem was not just the way CA obtained data through the quizzes and then used it to build profiles but also the way that Facebook 'allowed' this to happen. 

Similar case
Also, in 2018, another scandal occurred when Statistics Canada (StatCan) encountered a massive problem when collecting sample data of 500,000 people, following this by requesting financial data from bank entities. Several media reports revealed that StatCan requested several banks to provide them with lists of financial transaction information of hundreds of thousands of Canadians (Zimonjic, 2018).    

In this case, StatCan justified the financial data collection by stating that traditional statistical methods are not sufficient to measure the Canadian economy accurately. Also, StatCan tried to ensure worried participants that their data would be secure by ensuring the highest level of privacy and confidentiality (Zimonjic, 2018).    

The participants were surprised and worried when they discovered that StatCan was requesting financial information from their banks, which revealed that StatCan failed to communicate effectively with the survey participants. They had to clearly state the type of data they would be collecting, the purpose of it, and the scope of the research.   

References
Bareebe, R. (2022) The Cambridge Analytica Scandal and Its Impact on Meta. DOI: http://dx.doi.org/10.13140/RG.2.2.19583.69285

Rehman, I. (2019) Facebook-Cambridge Analytica data harvesting: What you need to know. Available from: https://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=5833&context=libphilprac [Accessed 10 April 2024].

The New York Times (2018) Cambridge Analytica and Facebook: The Scandal and the Fallout So Far. Available from: https://www.nytimes.com/2018/04/04/us/politics/cambridge-analytica-scandal-fallout.html [Accessed 10 April 2024].

Zimonjic, P. (2018) Privacy commissioner launches probe into StatsCan over collection of financial data . Available from: https://www.cbc.ca/news/politics/personal-financial-information-statistics-canada-1.4885945#:~:text=Trudeau%20confirms%20investigation%20into%20data%20gathering&text=PM%20Justin%20Trudeau%20confirms%20an,hundreds%20of%20thousands%20of%20Canadians [Accessed 10 April 2024].

---

Lecture cast notes: Questionnaire Design

Steps

Decide the information required Define target respondents Choose the method to reach target respondents Decide the content of the questions Develop the question wording Put questions into meaningful order and format Define the length of a questionnaire Pre-test the questionnaire Develop the final survey form

Questionnaires are cheap, but the return rate is 10%.

We need to be aware of the likelihood of respondents giving false answers.

Closed question: A list of alternatives to choose from. The obtained information can be converted into quantitative data. Likely to obtain false answers that don't express the actual participant's feelings.

Open Question: To obtain more detailed answers. Elaborated qualitative data. It assumes the respondent has better writing skills to express their feelings.

Questionnaire design

Questions should refer to the aim of the research and be clear and short. Long questionnaires can be easily abandoned. Logically progress from general to specific.  The answer to a question should not affect the answer to another question. The language of the questions should match the language of the responders.

Ethical issues

The researcher must protect the confidentiality of the respondents.

Seven Deadly Survey Questions

• The first question will influence the way the respondents answer the other questions. It is a good idea to provide an "other' options in the list of answers.
• Do not assume anything about the respondents. This is in terms of habits, nationality, etc. 
• Like the leading question, there are others that can seem to push the respondent to answer in a certain way.
• Avoid double negative and negative questions. Make them as logical as possible.
• Be prepared for the respondent to pick a random question.
• Put only one issue in one question to avoid asking for more than one answer.
• A pilot survey can help avoid ambiguous questions.