Retelling and Memory

How to cite

Laubel, Alexander. 2024. "Retelling and Memory." AEA RCT Registry. AEARCTR-0015038. https://doi.org/10.1257/rct.15038-1.1

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

Alexander Laubel*

* Primary investigator

Registry ID
AEARCTR-0015038
DOI
https://doi.org/10.1257/rct.15038-1.1
First registered
Last updated
Status
Completed
Trial dates
2024-12-16 – 2024-12-31
Countries
United States of America ()
JEL codes
D91, C91, D83
Registry record
http://www.socialscienceregistry.org/trials/15038

Abstract

Online experiment to study the effect of whether biased retelling makes recall of an original event biased as well.

Intervention

Online experiment to study the effect of whether biased retelling makes recall of an original event biased as well.

For more details, see attached pre-analysis plan.

Experimental design

I. DATA COLLECTION
This study consists of two experiments: a main experiment and a robustness experiment. Both experiments will be conducted on Prolific in December 2024. Both experiments will pre-screen participants in the following way: first language “English”, fluent languages “English”, Sex: 50 % male, 50 % female.

Main Experiment
The main experiment will be conducted over two consecutive days. The timeline is as follows:
Day 1:
1.1 Subjects see descriptions of three hypothetical products and provide a subjective rating for each of them (incentives: match the average rating among all subjects).
1.2 Subjects provide an audio recording of themselves retelling each product description. Treatment variation happens on the subject level and concerns whether a subject is incentivized to retell accurately or overly optimistically.
Day 2:
2. The same subjects are invited again and are asked to recall their rating from 1.1 and provide another recording. This time, the incentives are always accuracy.

There are two sets of three distinct products each (set 1: DreamGlow, EchoScent, WonderWeave; set 2: VibeVest, ZenPod, BrainBoost) and it will be randomized on the subject level which product set a subject encounters.

Robustness Experiment
The robustness experiment will be conducted with an entirely different set of subjects.
The timeline is as follows:
1. Subjects see the description of just one of the hypothetical product and provide a subjective rating (incentives: match the average rating among all subjects).
2. Subjects provide an audio recording of themselves retelling the product description. Treatment variation concerns whether a subject is incentivized to retell accurately or overly optimistically.
3. Subjects are asked to recall the rating provided in 1.
The product a subject encounters is from the same set of in total six products of the main experiment.
 
Additional Data: Human Coding of Recordings
Subjects’ recordings from the main experiment will be coded by two human coders (research assistants at Institute for Applied Microeconomics, Bonn University).

For each recording, the coders assess:
1. Is the audio quality of the recording sufficiently good to understand it?
2. Is there sufficient information in the audio recording to answer 3. and 4.?
Only if the answers to 1. and 2. are “Yes”:
3. Coders assess whether the recording was overly optimistic or accurate.
4. Coders provide a (subjective) rating based on the audio recording.

II. TREATMENT CONDITIONS
There are two treatment conditions each in the main experiment and in the robustness experiment:
Treated: subjects are asked to record an overly optimistic retelling.
Control: subjects are asked to record an accurate retelling.
Note: For the main experiment, treatment variation only concerns the recording on day 1. On day 2, all subjects are asked to provide an accurate retelling.

Design details

Randomization method
Randomization done by survey provider (Prolific).
Randomization unit
Subject
Clusters
For the main experiment, “one observation” requires that a subject shows up on day 2. Based on pilots, this applies to approx. 80 % of all subjects who complete day 1. For procedural reasons, I pre-register the day-1 sample size for the main experiment. I pre-register 550 complete day-1 responses for the main experiment and 420 complete responses for the robustness experiment. Note: “complete” only excludes subjects who did not pass the attention checks, but it includes subjects (a) who did not show up on day 2 (only applies to the main experiment), (b) who will be excluded from data analysis because they indicated that they used notes, or (c) – only applies to “Hypothesis Retelling” – whose audio recordings are assessed by the human coders to have not sufficiently good audio quality or to provide not sufficiently much information to infer a rating (only applies to the main experiment).
Observations
For the main experiment, there are 3 observations per subject. For the robustness experiment, there is 1 observation per subject.
Arms
50:50 by survey provider during data collection (precise numbers on data collected depend on attention checks, attrition etc.)
Primary outcomes
Main Experiment: Day 2 Rating, Day 2 Retelling (as assessed by human coders)
Secondary outcomes
Robustness Experiment: Rating Recall

Cited by

No citation data available for this registration yet.

Keywords