Hackpads are smart collaborative documents. .

Joey Savoie

1131 days ago
Unfiled. Edited by Joey Savoie , Ozzie Gooen 1131 days ago
  • Receivers / Receiving Projects
 
The first 10 people who create Gittip pages will receive a $7 weekly donation from Ozzie for 3 weeks (that's $21 free).  You can do this on behalf of an individual or organization.
To qualify for this, you will need to do the following:
 
Joey S
 
991 days ago
Unfiled. Edited by Joey Savoie 991 days ago
  • Cats
Joey S
  • nich marketing (e.g. targeting poker players)
 
1191 days ago
Unfiled. Edited by Peter Hurford , Joey Savoie 1191 days ago
Peter H Conversation with Joey Savoie and Xio Kikauka, 18 June 2014
 
  • Goal
  • Figure out how to make progress on animal rights research.
 
 
  • Participants
Xio Kikauka
Peter Hurford
 
 
  • Problems with Current Studies
  • Current studies are statistically underpowered.
  • There might not be enough money to fund a study.
 
The Problem of Statistical Power
  • <What statistical power is>
  • This study is underpowered <elab>
  • We need a larger sample <elab>
  • Larger sample would be expensive <elab> (like ~$30K?)
 
There are no obvious solutions to fixing some of these methodological problems
  • It was easy enough to fix the problem of control groups, but it's entirely unclear what could be done about the potential non-reliability of self-reports
  • Reduce meat consumption in a cafeteria at a university?
  • Too small of an effect to track among month-to-month changes
  • Have people track their food day-to-day (Instagram)
  • Hard to get a large sample size to be this thourough
 
Is there enough money to fund a large study?
  • Peter recalls at least five people said they'd be willing to fund "one large study" on leafleting if the methodology was good enough.
  • How much money this would total to, or whether or not the methodology could ever be good enough remains to be seen.
  • It's possible people would be really responsive to a nice Kickstarter campaign with clear goals and thorough reasoning.
Joey S
  • Self reports might never provide enough evidence 
 
 
Peter H
  • Would a large study be worth the cost?
  • Xio thinks that very few people seem motivated by the studies.
  • Who?
Joey S
  • EA's 
  • Few are both empirical and AR interested
  • AR people 
  • Mainly focus on what they intuitively think will work
  • Peter agrees that he's found many AR people who aren't very responsive to studies or are quick to go with anecdotes from leafleters.
  • Empirical EA's (would need tons of studies to change) 
  • 3+ positive large RCTs?
Peter H
  • Peter notes that it would only take about one year from one person doing earning-to-give to move enough money to make the study worth it.
 
  • Joey thinks that even if we worked very hard to create a great $30K study, it would still have a high enough probability of being flawed enough to be ambiguous
  • Joey and Xio's study was already supposed to be good enough to put the issue to rest, but didn't.
  • Joey and Xio's study was supposed to correct the obvious flaw of the lack of a control group
Joey S
  • Clear flaws, better control, larger sample, less questions (huge cost) 
 
 
Peter H
  • Would it be possible to aggregate many small studies?
  • It's possible that a lot of converging evidence could be just as persuasive as a large study, without as much risk.
  • Even if multiple studies are all flawed, as long as they are not flawed in the same way, a collective effect could be persuasive.
  • Though if each study is very underpowered, many studies might not succeed in detecting the effect solely by chance.
Joey S
  • Full cost of running enough small studies to be powerful? 
  • 3 strong studies (20k each)
  • 3 weak studies (5k each)
  • net cost 75k cost
Peter H
  • It seems decently easy to create studies that solve one of the problems in isolation, but not others
  • Doing a survey of students at a university could produce a reliable panel study to track differences in consumption
 
 
  • Is Doing this Research Worth *Our* Time in Particular?
  • Pretty likely that we're biased to be in favor of animal rights when [elab]
  • Do we have a strong theoretical reason to think that animal rights would be particularly high impact?
  • possible reasons for
  • anecdotes of leafleters
  • possible reasons exist
  • other history of advertising
  • no one really leaflets any more (for businesses or other public health change)
  • Potential AP: Look into this further.
  • vegetarian rates nationwide appear flat
  • though the combined aggregate effect of vegan outreach might not be enough to move the surveys within their margin of error
  • ?: What are the trends in veg rates?
  • ?: What is the margin of error in these rates?
  • ?: How many people would have to be changed to make a difference in this macro-level data?
  • ?: How many persuasive attempts have been made? (e.g., flyers + video views + etc.)
  • Potential AP: Do research to try and answer these questions.
Joey S
  • How does it compare to other unstudied areas (e.g. labor mobility)?
 
Peter H What else would we be doing with our time?
  • J&X: Doing AR research would take away from Charity Science.
  • P: Will be doing earning-to-give, but don't have a clear alternative to do in my non-work time (~20hrs/wk).
  • Though maybe Peter should be doing something else, like networking/fundraising in Chicago.
  • AP: Work out higher return opportunities for Peter's work.
 
 
  • Is it possible to advise other people to do AR research?
Joey S
  • Help something like THL labs get much bigger and more rigorous?
Peter H
  • Idea: Become an advisor to Nick Cooney
  • THL seems to have already moved from "testing to see what works" to "testing what works best"
  • ?: How does THL decide which studies to run and how to run them?
  • AP: Skype with Nick Cooney (or some THL representative)
 
Joey S
  • Help ACE do the research?
...

Contact Support



Please check out our How-to Guide and FAQ first to see if your question is already answered! :)

If you have a feature request, please add it to this pad. Thanks!


Log in