“Ask this from the person who has private stuff”: Privacy Perceptions, Behaviors and Beliefs Beyond W.E.I.R.D

In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems (CHI '22). Association for Computing Machinery, New York, NY, USA, Article 105, 1–17. https://doi.org/10.1145/3491102.3501883

In Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems (CHI '22). Association for Computing Machinery, New York, NY, USA, Article 105, 1–17. https://doi.org/10.1145/3491102.3501883

Publication

Presentation Video

the big question?

How is privacy enacted and what strategies are employed by low-literate men and women to navigate the norms and perform privacy?

the goal

Understand privacy perceptions and practices among low-literate, low-income Pakistani users.

Explore the impact of cultural expectations arising from a patriarchal context on privacy beliefs.

Facilitate privacy preservation amongst low-literate low-income users given the cultural restrictions.

Understanding the existing landscape before stepping into the field

Although I did understand the Pakistani cultural landscape having spent 20 years (my entire life) in the country, I did not belong to the demographic I was going to be interacting with. To develop a better understanding of the landscape, I combed through existing literature and prior ICTD work done with similar demographics and read through 120+ academic articles published at reputable venues. 

Although I did understand the Pakistani cultural landscape having spent 20 years (my entire life) in the country, I did not belong to the demographic I was going to be interacting with. To develop a better understanding of the landscape, I combed through existing literature and prior ICTD work done with similar demographics and read through 120+ academic articles published at reputable venues. 

Stepping into the field

We conducted 40 semi-structured interviews (23 female, 17 male) with participants belonging to the low-literate low-income community. 

Matching bandwidth with the participants

It was essential that the interview language and location were set to ensure participants' comfort.

It was essential that the interview language and location were set to ensure participants' comfort.

Punjabi and Urdu (local languages) as interview languages

Primary language: Urdu, as it is the national language of Pakistan.

Language flexibility: We adapted to Punjabi, the regional language of Punjab, if:

  • Participants responded in Punjabi.

  • Participants seemed to struggle with Urdu.

This approach ensured participants were comfortable and could communicate clearly during the interviews.

Choosing a neutral location to ensure participant comfort

Choosing a neutral location to ensure participant comfort

Choosing a neutral location to ensure participant comfort

Interview location: A school in the participants' local area.

  • Provided a familiar and secure environment for participants.

  • Maintained a good distance from their homes to encourage honest and open responses, minimizing any hesitation they might have felt in a more personal setting.

Gender sensitive interviews

Gender sensitive interviews

Initial approach:

  • Both my co-researcher (male) and I (female) conducted interviews together.

  • One person focused on the interview, while the other took notes.

Observation after first 4 interviews (2 male, 2 female):

  • The presence of other genders noticeably restricted participants' responses.

Revised approach:

  • I (female) conducted interviews with female participants.

  • My co-researcher (male) conducted interviews with male participants.

This adjustment improved the openness and quality of participant responses.

~ female participants hesitated to mention things such as intimate health etc. in front of my male co-researcher and male participants hesitated to express their views in front of me.

Real-time reframing of the interview protocol

The interview protocol evolved as I made discoveries about our participants in the initial interviews

The interview protocol evolved as I made discoveries about our participants in the initial interviews

  1. Making the interview protocol non-scandalous and culturally sensitive

Goal

Understand if participants were employing any privacy-preserving measures on shared phones to guard their messages, search history, etc.

Problem

Female participants were reluctant to mention that they hid something on their phones from their husbands

Resolution

I modified the interview protocol in real-time to ask them if they hid something on their phones from their children; this gave me insights into if they knew how to use a certain hiding feature or not.

  1. Resorting to simpler terms: read receipts to blue ticks

Goal

Assess participants' understanding of WhatsApp features, specifically regarding privacy settings like read receipts.

Problem

Participants had low literacy levels and struggled to comprehend terms such as "read receipts," which led to confusion when I inquired if they had enabled that feature.

Resolution

I adjusted my interview approach in real-time by simplifying the terminology, rephrasing questions to specify "the first delete option" and "the delete option below it," which helped clarify the features and gather accurate insights.

  1. Transition to a visual interview protocol

Goal

Evaluate participants' awareness and usage of specific app features during the interviews.

Problem

Participants struggled to understand my verbal references to certain app features, making it difficult to gauge their familiarity and usage.

Resolution

I adapted my interview technique by demonstrating the features directly on my phone, asking participants to point out the options they recognized, which facilitated clearer communication and more accurate responses.

Data Analysis

After each interview session, we dumped our insights in the form of sticky notes. At this point, the most random and unstructured insights were also drafted to ensure we did not miss out on any data and thoughts we had during interviews. We grouped these insights into sections. The sections were labeled according to whatever name we saw fit at that time.

We then refined and relabelled the insights in order to analyze them better.

Swapping roles for thematic analysis to ensure unbiased analysis

The insights we gathered provided a solid foundation for beginning our thematic analysis as they helped in getting the initial codebook in place. My co-researcher and I coded the transcripts using open coding on Deedose. To ensure a fresh perspective, we swapped roles during the coding process - I focused on coding the male interviews, while he took on the coding of the female interviews. This approach enhanced the breadth of our analysis as I was not a part of the male interviews and he was not a part of the female interviews. This approach therefore lead to unbiased analysis. 

The insights we gathered provided a solid foundation for beginning our thematic analysis as they helped in getting the initial codebook in place. My co-researcher and I coded the transcripts using open coding on Deedose. To ensure a fresh perspective, we swapped roles during the coding process - I focused on coding the male interviews, while he took on the coding of the female interviews. This approach enhanced the breadth of our analysis as I was not a part of the male interviews and he was not a part of the female interviews. This approach therefore lead to unbiased analysis. 

Findings

  1. Enacting Privacy as Scandal for Women

Female participants associated enacting privacy and the desire for privacy with ‘scandalous’ and inappropriate content.  Our female participants were particularly concerned with the negative connotation of privacy - i.e. if you have not done anything ‘wrong’, you do not need privacy:

Female participants associated enacting privacy and the desire for privacy with ‘scandalous’ and inappropriate content.  Our female participants were particularly concerned with the negative connotation of privacy - i.e. if you have not done anything ‘wrong’, you do not need privacy:

~ " i think when you have done something mysterious or wrong only then you will feel the need to hide it from your family members. My kids are very good and my husband and I have nothing to hide from one another. We have everything in the open"

Another female participant felt that as long as she is not doing anything ‘wrong’ she does not need a phone lock. She noted that her phone had nothing private and was left unlocked on a table when she was busy with household chores: 

Another female participant felt that as long as she is not doing anything ‘wrong’ she does not need a phone lock. She noted that her phone had nothing private and was left unlocked on a table when she was busy with household chores: 

~ " i do not need a phone lock and i never will"

She further explained that a phone lock would be synonymous with hiding things from her husband and it is important for her to ensure he had ready access to her phone.

She further explained that a phone lock would be synonymous with hiding things from her husband and it is important for her to ensure he had ready access to her phone.

However, In exploring women’s privacy practices we found that women employ specific mechanisms to maintain privacy on digital devices without overtly signaling their actions. For example, they delete WhatsApp messages after sending them to prevent others from viewing them, though they worry about the "this message was deleted" notification. Additionally, they hesitate to discuss sensitive topics on their phones for fear that others may read their conversations.

However, In exploring women’s privacy practices we found that women employ specific mechanisms to maintain privacy on digital devices without overtly signaling their actions. For example, they delete WhatsApp messages after sending them to prevent others from viewing them, though they worry about the "this message was deleted" notification. Additionally, they hesitate to discuss sensitive topics on their phones for fear that others may read their conversations.

this leads to 2 user needs

the desire to have closed conversations about sensitive topics without the fear of some one else accessing them

traces being left behind of private conversations and the inability to remove these traces.

design recommendation:
Private Conversation Mode

No trace of conversations or media left on the device.

Content shared is restricted to the session only.

Interactions are organized into sessions.

Messages and media are inaccessible outside the session after it ends.

No session log is maintained.

Once the session is closed, users cannot access the previous conversation.

  1. Aftermarket applications are a threat to privacy

Many participants used after market applications/modded app versions such as GB WhatsApp and FM WhatsApp to supplement their privacy needs. These applications provide additional features over standard WhatsApp, some of these include: 1) viewing WhatsApp story/status without notifying the poster, 2) reading messages without them being marked as read (for specific participants), 3) being able to view deleted stories and messages, and 4) being able to freeze your last seen online time stamp. 

Many participants used after market applications/modded app versions such as GB WhatsApp and FM WhatsApp to supplement their privacy needs. These applications provide additional features over standard WhatsApp, some of these include: 1) viewing WhatsApp story/status without notifying the poster, 2) reading messages without them being marked as read (for specific participants), 3) being able to view deleted stories and messages, and 4) being able to freeze your last seen online time stamp. 

 The contacts of a person using a modded version of WhatsApp are unaware of it’s use - on their end the application looks and functions exactly like the official application. however , this leads to compromised the privacy of others by being able to view their deleted messages and stories.

 The contacts of a person using a modded version of WhatsApp are unaware of it’s use - on their end the application looks and functions exactly like the official application. however , this leads to compromised the privacy of others by being able to view their deleted messages and stories.

~ " there is a sort of privacy where even when you see their story (referring to WhatsApp status) they don’t get to know about it (referring to GB and FM WhatsApp). Similarly status can be hidden from specific people as well"

this leads to 2 challenges

Users may unknowingly share information they believe they can later delete for everyone.

GB or FM WhatsApp users might view their content, such as WhatsApp Status, without their knowledge.

design recommendation:
Notification for modded apps

Users should be notified if a contact is using a modified version of WhatsApp.

User should get the option to agree or refuse to continue conversation with the person using modded app.

Conclusive Thoughts

Conclusive Thoughts

How did i grow as a researcher?

Culturally and Contextually Sensitive Research Questions

I learned to craft research questions that are sensitive to cultural nuances and contextual factors as one question style does not fit all.

Real-Time Protocol Adjustments

I honed the skill of modifying interview protocols dynamically to encourage candid responses, particularly on sensitive topics. This flexibility allowed me to uncover insights that participants might otherwise have been hesitant to share.

Dumping raw insights

Dumping raw insights immediately after interviews proved invaluable. It facilitated the synthesis of concrete findings later on, guiding the development of actionable recommendations.

What would i do differently?

Using a different tool for thematic analysis

I found that using Deedoose for thematic analysis posed challenges and impacted efficiency. In the future, I would opt for more user-friendly and efficient tools to streamline the analysis process.

Do more focused interviews

I would conduct more interviews but with a narrower focus . Each individual set of interviews would be about particular topic - modded apps, WhatsApp

Maintain a process book

While I do have all the materials of the project in a dedicated Google Drive folder, I would like to maintain a process book in Notion or a similar tool that details every part of the process - this would serve as a one stop for anything related to the project arranged in a chronological order.

What would i do with the results?

Discuss implementation feasibility with product teams

While we do have the results in the form of actionable design recommendations, I would want to discuss the feasibility of their implementation with the product teams at WhatsApp and see if they can be implemented

Do usability testing with the prototypes

I would make prototypes of the design recommendations and test them with the same population to further validate the findings.

Participatory design

Alternatively, I would take the results back to this population and design solutions with them.