@MMandySha profile picture
Mandy Sha @MMandySha
Joined May 2014
  • Tweets 986
  • Following 314
  • Followers 437
702 Photos and videos
Book launch social starting 6:00PM May 13, 2026 at the Los Angeles @thebonaventure πŸ†•πŸ“– 𝘊𝘳𝘰𝘴𝘴-𝘊𝘢𝘭𝘡𝘢𝘳𝘒𝘭 π˜™π˜¦π˜΄π˜¦π˜’π˜³π˜€π˜© π˜ͺ𝘯 π˜—π˜³π˜’π˜€π˜΅π˜ͺ𝘀𝘦 of 1⃣9⃣ chapters for improving research quality across cultures, languages, countries πŸ‘‡πŸ» amazon.com/Cross-Cultural-Re…
1
101
#ThematicAnalysis #QualitativeResearch ok to 'blend' codebook reflexive elements? MariekeHaan: πŸ‘ when deliberate/justified βœ”οΈ no single correct way, must fit aim βœ”οΈ transparent in analytical choices➑️ quality βœ”οΈ reflexive positionality (how rolesπŸ€” shape analysis, decisions)
1
218
Besides in-person surveys, what're remote interviewing modes? @NSSRNews M.Schober Text-only web survey πŸ“±β˜ŽοΈ Live video Play pre-recorded video of interviewers in web survey "On average, ppl are as uncomfortable in live video as in person, > uncomfortable than in other modes"
1
57
What is it about live video πŸ†š other survey modes for respondent discomfort? @NSSRNews M.Schober "social presence" πŸ‘€, being πŸ‘€ perceived usefulness enjoyment relative😱🫦 "mode-specific discomfort may play a larger role in survey (non)response than we have understood so far"
32
Is #AI standard speed or trust alone? #AAPOR Webinar @MLink01 and D.Graff-Radford @discuss_io formerly @Voxco: No, "it's faster 𝘒𝘯π˜₯ trustworthy. And 𝘸𝘦 π˜₯𝘦𝘀π˜ͺπ˜₯𝘦 what that means." GOOD RESEARCH is "what we can explain, evaluate, defend-and question when it looks wrong"
1
70
How to expect/verify "good" #AI use? @MLink01 #AAPOR Webinar 1⃣ transparent: HOW produced 2⃣ researcher-in-the-➿ 3⃣ reproducible 4⃣ validate w accuracy evidence: compareπŸ§‘πŸΏβ€πŸ€β€πŸ§‘πŸ», subgroupπŸ” 5⃣ compliant: handling/controls appropriate for your environment #TotalSurveyError? stillπŸ‘‘
1
55
Can respondents give qualityπŸ—£οΈ on πŸ“³? @chrisantoun01 #FedCASIC: yes but πŸ”Ž UI choices modular design break long instrument but risk breakoff (need recontact method low-friction return) πŸ€”in-the-moment measurement ➑️ event-specific info but recruitment/coverage/activity matter
34
πŸ“³ are today's #communication technology πŸ†•#data types (apps, receipt scanning, in-store buys), what design issues πŸ€”researchers? @chrisantoun01 #FedCASIC 1⃣ device effects 2⃣ user interface πŸ”€ input error #usability 3⃣ survey burden 4⃣ real-time measurement πŸ†š retrospective
36
Can #AI replace experts? TingYan @NORCNews #FedCASIC keynote #SurveyResearch future is "methodologist-driven w/ AI in service of goals" because βœ”οΈ error-cost tradeoff reasoning βœ”οΈ validity checks βœ”οΈ human subject protections AI🚫 error/qual theory, πŸ—οΈintent, accountable/ethicπŸ€”
1
1
144
How's #AI use framed in survey research? Rothschild et al (2025): research colleague, interviewer, respondent, labeler, modeler, briefer ‼️ useful descriptors, not decision rules πŸ™ Ting Yan #FedCASIC keynote
2
43
What's #ResponsibleAI? Ting Yan @NORCNews #FedCASIC keynote 1⃣ intentional 2⃣ fit for purpose 3⃣ human-in-the-loop 4⃣ Total Survey Error (TSE)-guided validation/eval 5⃣ Transparent πŸ€” within the survey framework #AI roles ➑️ Survey needs What can AI do ➑️ what survey requires
1
1
136
πŸ†•πŸ“– 𝘊𝘳𝘰𝘴𝘴-𝘊𝘢𝘭𝘡𝘢𝘳𝘒𝘭 π˜™π˜¦π˜΄π˜¦π˜’π˜³π˜€π˜© π˜ͺ𝘯 π˜—π˜³π˜’π˜€π˜΅π˜ͺ𝘀𝘦 presents 19 chapters on improving research design & quality across cultures, languages, countries πŸ‘‡πŸ» amazon.com/Cross-Cultural-Re… Ch1⃣7⃣ uses Demographic and Health Surveys DHS data to study 🌍 coverage structure
48
Check out πŸ†•πŸ“– 𝘊𝘳𝘰𝘴𝘴-𝘊𝘢𝘭𝘡𝘢𝘳𝘒𝘭 π˜™π˜¦π˜΄π˜¦π˜’π˜³π˜€π˜© π˜ͺ𝘯 π˜—π˜³π˜’π˜€π˜΅π˜ͺ𝘀𝘦! Presenting 19 chapters of useful and timely topics, methods & solutions to improve research quality across cultures, languages, countries πŸ‘‡πŸ» amazon.com/Cross-Cultural-Re…
1
74
πŸ™€ survey cost, use #SyntheticRespondents? @serinachang5 @UCBerkeley @berkeley_ai πŸ‘ natural lang flexible inputs/outputs, pretraining knowledge stored in billions param ❗️out-of-the-box LLM biased toward groups, πŸ”½estimate w/in-grp variance So, finetune #LLM predict πŸ§‘β€πŸ¦°response
104
Can #AI chatbots collect better survey data, not worsening experience? S.Barari #Likert symposium @MichPSM "#LLM can provide useful survey data [accurate coding @BLS_gov occu category, more specific/relevant responses] but too much may create a lower quality survey experience"
73
Postal push2web πŸ†š dual-frame RDD @pewresearch study in Sweden, what to know? @MoynihanPatrick βœ”οΈ frame=Swedish State Personal Address Registrar (SPAR) βœ”οΈ (un)conditional incentive experiment βœ”οΈ Yield=24%, πŸ›œ71% πŸ“°29% ⏭️steps: πŸ”Ž@ESS_Survey, πŸ“«delivery, πŸ’Άavail host @LAPOP_Lab
2
63
Check out Chapter6⃣ led by @NORCNews scientist on #MachineLearning for multilingual survey evaluation πŸ†•πŸ“–Cross-Cultural Research in Practice presents useful timely topics, methods, solutions to πŸ“ˆ research quality across cultures, languages, countries πŸ‘‡πŸ» amazon.com/Cross-Cultural-Re…
3
70
#ISWGHS @UNStats handbook #survey on households individuals foundations, what're 3⃣ applied frameworks for quex design #translation instrumentation? 1⃣ cognitive response process model 2⃣ QDET ques design, dev, eval, testing 3⃣ repsondent-centered design: ‡️ burden, UX-informed
1
60
#ISWGHS @UNStats handbook #survey on households individuals foundations, what're emerging approaches? βœ”οΈ machine translation post-edit βœ”οΈ #GenAI draft/eval items, coding open-ends βœ”οΈ Tools for workflow (DataCTRL), data capture, documentation βœ”οΈ CAVI πŸ—£οΈ, web probing, crowdsource
46
Check out Chapter2⃣ by @NORCNews researchers in πŸ†•πŸ“– 𝘊𝘳𝘰𝘴𝘴-𝘊𝘢𝘭𝘡𝘢𝘳𝘒𝘭 π˜™π˜¦π˜΄π˜¦π˜’π˜³π˜€π˜© π˜ͺ𝘯 π˜—π˜³π˜’π˜€π˜΅π˜ͺ𝘀𝘦 πŸ‘‡πŸ» amazon.com/Cross-Cultural-Re… The volume presents useful timely topics, methods, solutions to πŸ“ˆ research quality & design across cultures, languages, countries.
1
3
126
Load more