Total Recall: Measurement Experiment

How to cite

Subramanian, Nivedhitha; Field, Erica; Garlick, Robert; Vyborny, Kate. 2023. "Total Recall: Measurement Experiment." AEA RCT Registry. AEARCTR-0011110. https://doi.org/10.1257/rct.11110-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Nivedhitha Subramanian* , Erica Field , Robert Garlick , Kate Vyborny

* Primary investigator

Registry ID
AEARCTR-0011110
DOI
https://doi.org/10.1257/rct.11110-1.1
First registered
Last updated
Status
On going
Trial dates
2021-06-01 – 2025-12-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/11110

Abstract

There is substantial variation across developing country surveys in labor market measurement. Different surveys use different recall periods, definitions of employment and search, and levels of detail in prompts. This variation may reflect open questions about how best to collect labor market data in these settings. We conduct a series of survey methods experiments in the context of a large panel study in Lahore, Pakistan. We will use these experiments to understand how variations in survey timing, recall periods, and questionnaire design influence measures of job search and employment. We benchmark the survey data against administrative data via a job search platform.

Experimental design

We conduct a series of experiments related to measurement with a sample of job platform users. Over a series of surveys, we randomize the length of the recall period, level of detail in a question related to search activity, and the timing of the survey relative to when respondents receive notifications about open job postings through the platform. We also examine how these reporting patterns differ for more versus less active users of the platform, using a randomized encouragement design from another experiment.

Design details

Randomization method
Randomization is conducted using Stata.
Randomization unit
For the recall period, we randomize at the level of the individual. For the other randomizations, we randomize within person through time.
Clusters
These experiments are embedded in a broader survey operation targeting 15,000 individuals. The randomizations are implemented during phone calls, so the effective sample size depends on the number of respondents who start surveys. We will only know this number after data collection finishes, but it will be at most 15,000 respondents.
Observations
See description above.
Arms
As noted above, the final sample size by treatment arm will depend on the number of respondents who complete these surveys. We randomize half of the units into each treatment arm.
Primary outcomes
Reporting patterns in surveys about Job search and employment
Secondary outcomes
On-platform job search

Cited by

No citation data available for this registration yet.

Keywords