EC475, LSE, Student Survey Responses about Human and AI Feedback
Description
Data were gathered through survey responses by students in EC475 during Autumn Term 2025 at the London School of Economics. Students had written referee reports on published papers, then received two blinded sources of feedback, human (source A) and AI (source B). The survey asked questions about the quality of the feedbacks. Variables in raw_data.csv are as follows. timestamp - the time and date of survey completion whichrefereereportareyouconsider - The name of the author(s) of the paper they had written a referee report on and were receiving feedback. The five papers are Waldiner (2022), Christensen et al. (2024), Ashenfelter and Jurajda (2022), Chen et al. (2013), and Mirenda et al. (2022). howusefulwasfeedbacka - a 1-5 response to "How useful was Feedback A?" howusefulwasfeedbackb - a 1-5 response to "How useful was Feedback B?" whatisthemostusefulaspectoffeedb - a 1-5 response to "What was the most useful aspect of feedback A?" whatistheleastusefulaspectoffee - a 1-5 response to "What was the least useful aspect of feedback A?" v7 - a 1-5 response to "What was the most useful aspect of feedback B?" v8 - a 1-5 response to "What was the least useful aspect of feedback B?" howmuchdoyouagreewiththestatemen - a 1-5 response to "How much do you agree with the statement, "Feedback A and B are much more useful together compared to if I had received just one of the two (received whichever you thought was better)" whichfeedbackbetterhelpedyouunde - a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Which feedback better helped you understand/learn econometric concepts?" whichfeedbackwaseasiertoundersta - a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Which feedback was easier to understand?" v12- a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Which feedback better helped you understand what makes a good referee report?" whichfeedbackbetterenhancedyourc - a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Which feedback better enhanced your critical thinking about the paper you reviewed?" whichfeedbackfeltmorepersonalize- a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Which feedback felt more personalized to your specific report?" didfeedbacaorbseemmorecriteriab- a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Did feedback A or B seem more criteria-based?" didfeedbackaorbhavebetterclarity - a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Did feedback A or B have better clarity or directions of improvement?" didfeedbackaorbhaveamoresuppori - a response of "Feedback A", "Feedback B" or "Roughly Comparable" to "Did feedback A or B have a more supportive tone?" pleaseshareanyotherthoughtsyouha - an open-ended response to "Please share any other thoughts you have." analysis_3.do is Stat code that performs cleans data, performs t-tests, makes graphs, performs some regressions, and implements other analysis.
Files
Steps to reproduce
Data were gathered through anonymous survey responses by students in EC475 during Autumn Term 2025 at the London School of Economics.
Institutions
- London School of Economics and Political ScienceEngland, London