Meta & Children's Online SafetyChat: the nielsen study does show causal impact on social comparison 🙁
Thread View
████ name redacted. Different bars may be the same person or different people.
quoted in an excerpt in the archive.
Friday 26 February 2021
07:52████Hiiii -- I mean I would love to do more here and it's something we should absolutely think about what could we do differently here to be more convincing
07:52████cause we should totally integrate it into your research plans even if we don't have the data right now
08:50████I also think we need to work with Responsible Innovation more I think they actually were ENCOURAGING of measuring emotions... like seeing it as a good thing!
08:50████we've had several meetings with them but they are probably a way for us to scale ourselves if we can integrate with them because more and more people will go to them
08:52████yeah, I met with someone from their team last half as well. i'll reach out and try to understand the rationale behind their perspective here, and begin to share ours. also just build those relationships
08:53████Maybe they can also partake in the white paper process about our role
08:54████would for you to spend some time though thinking how we could be more convincing to these stakeholders, because as said this will keep coming up.
08:59████at some point want to debrief feedback? want to keep moving that fwd in anticipation of needing to send this for leadership steer end of next week.
08:59████yes to be honest, I did not fully understand her feedback.
08:59████but yes maybe between the two of us we can figure it out :)
09:01████I'm around now if you want to chat for a few
09:58████hey there - for the exercise for WB, should I take a stab at filling out the table assuming we're going with our option a) Feelings over time?
10:01████yes I'm actually netting that we just define this
10:01████and that's our definition
10:04████yeah in my original thinking and the way we've framed it as "over time", it was selected specifically because it hits all the bases of addressing some of the important narratives, 2) consistent with external research (on emotions and affect), and 3) ladders up to Mark's original ask about energy and about making sure we're assessing both positive and negative impacts
12:17████did you hear the two use cases, I missed them
12:18████metric to ENG to measure prevalence (how many people feel angry? how many posts make people feel angry?)
12:19████and identifying content that makes people feel angry
12:20████does feeling irritated make people think FB is bad? make them want to use it less? make them actually use it less?
12:23████Pratiti is not on our side here
12:23████nope
12:23████this is a surprise
12:24████It sounds like there's a lot more discussion that's happened with mark about this
12:24████and he's pushing hard
12:28████maybe we can help here
12:29████yup suggest to we workshop that table together?
12:36████that was kind of a weird content about policy cause I didn't get that vibe from policy
12:37████1 person was into it; the other had highlighted a bunch of risks
12:37████yeah I felt policy risks were understated
12:37████also clearly more due dligigence needed there
12:38████that policy guy hadn't even looked at this content before that 30 minute meeting the other day! his responses were totally off the cuff
12:38████also, so much for being a fly on the wall in that mtg
12:38████like, Can you just make up a question for us on the spot?
12:38████OMG
12:38████I could not stay silent
12:39████like I was the only one in the room fighting for our stance
12:41████I really wanted to say something about the "5% of people feel irritated when they end a session". do we know that's bad?
12:41████like, my kid leaves basketball practice feeling irritated or even fully angry about 30% of the time
12:41████but mine LOVES basketball and can't wait for the next practice
12:56████I know dude even just listening to them speak, makes me sick to my stomach
12:59████do people really think that a meaningful product that plays an important role in people's lives should only make people feel good/delighted/inspired, all the time??
12:59████are we ice cream??
13:29████Trying to merge several discussions happening at once I put together this document would love your eyes on it: https://docs.google.com/document/d/1s90NZjFQ-ZbEiOxHugrszIeJmPTzWVF6v-ONP1AaJbk/edit
13:30████This would go to Pratiti where she would use it to facilitate/escalate a convo with leadership about IG alignment
16:04████do you think life satisfactionwe should fold into Feelings Over Time?
16:04████I would say no
16:05████that's more of a self-attitude/beliefs category
16:05████and loneliness is not considered, correct?
16:05████loneliness we're doing through connection workstream
16:05████agreed just want to make sure we're aligned
16:05████great just confirming!
16:05████it actually probably would be great to have an excludes somewhere
16:05████like a footnote
16:12████I added in alignment doc (feel free to move). i will aslo put the causal study citations in the appendix
16:13████the nielsen study does show causal impact on social comparison
16:14████I know - I remember that. specifically remembering saying that's teh outcome he stands behind
16:14████During the last 7 days, how often did you compare your own accomplishments to those of other people?
16:14████During the last 7 days, how often did you feel worse about yourself after comparing yourself to someone else?
16:15████and I think both held
16:25████count o causal studies
16:28████for "emotions"?
16:29████for emotions and affect
16:29████do you want just run the DVs by me to make sure we're aligned?
16:30████https.//docs.google.com/spreadsheets/d/1ufszwWYWEDvjsiKdSIJgWOKR-7GQqSLQSv9vbcivveA/edit#gid=257067240
16:30████in Causal Impact tab
16:30████first 3 columns
16:31████gah sorry I'm confused
16:31████also I just realized there are two versions of this floating around
16:31████since there's one in my tab too
16:32████not sure which is the most up-to-date?
16:32████but either way how do I see the DVs you're currently counting as "emotions"?
16:34████so coulmn 1 is tx effect
16:35████coumn 2 is whether or not DV aligns with our definition
16:35████3rd column is how DV was operationalized in the study
16:36████so is column V wildly off from the instruments?
16:36████for example row 3 says "quality of day"
16:37████I don't know whta that column is, but i'm using the operationalization of the DV
16:37████I also think there's 1 causal study for social comparison
16:37████the Yuen study
16:38████is Row 24 correctly labeled as Y?
16:38████I hadn't counted "affective well-being" as emotions, but could be convinced
16:39████I want to include positive and negative affect in our definition. these are measures of emotional response over time
16:39████they include emotions and mood
16:39████I also had not included "absorbed" as an emotion
16:39████which row?
16:39████or "Aggregation of well-being index"
16:39████Row 26
16:39████Row 31
16:40████those are the only 3 where I think I had difference
16:40████I'd count both of those, but absorbed could go either way
16:41████but doesn't change the overall # of studies
16:41████but 24 is incorrect right?
16:41████Well, it's asking people about anger
16:42████interesting I was definitely going by the DV label rather than the instrument
16:42████So that's probably why we're off
16:43████yup
16:43████but we should go with the instrument
16:44████hard to say -- I think it's a lot less feasible to have us evaluate the instruments to determine what the study was about
16:44████for the causal studies?
16:44████there aren't that many...
16:45████it's probably more manageable through that lens, but I worry about using different processes for the different methods
16:45████where in 1 we determine what the study was about based on the instrument, and the others we determine what the study is about based on how the academic labeled it
16:45████I'm fine with either, just recommending we should be consistent
16:48████when have we used the shorthand for what the author called the DV as opposed to the instrument/what they actually measured?
16:49████we only used the DV label from the study for generating well-being topics
16:49████hmmmmm
16:49████we stuck to the author's interpretations and labels to avoid getting into an interpretation battle
16:50████for this part of the process, where we're really digging in to what the topic is, i think we should go with the operationsalization
16:52████so what's your assessment now to determine whether something is an emotion in the instrument?
16:53████>Like, what is this one: "Aggregation of well-being index"
16:53████if they use an emotion word in the instrument that qualifies? Is that the rule?
16:56████no Its if an instrument asks about a person's emotional response, or positive/negative affect. NOT if it's a scale used to assess symptoms of a diagnosable mental health disorder (e.g., CESD depressionscale)
16:59████I think row 24 is prob incorrectly labelled
16:59████got it happy to have you do whatever you prefer, let's just make sure the decision is documented somewhere and try to be consistent in how we do this across different exercises
16:59████like if we're going to do anything about any of the metanalyses or cross-sectional work
16:59████sounds good.
17:11████On Monday, can you just update the table in the IG doc with where you land on the numbers and studies?
17:12████(for both social comparison and emotions)
17:12████will do first thing
18:41████@silent thanks so much for great collaboration and communication this week. So glad to be working together. Have a great weekend!❤❤
18:45████@silent same!!! Can't wait for more! We are going to do some incredible things together and excited to trailblaze with you through this!