“Will” It Be Seen? Using Eye Tracking to Re-examine the Effect of Future Tense in Intertemporal Choice

How to cite

Hu, Jiang Shiang; Chen, Josie I; Chen, Wei James. 2022. "“Will” It Be Seen? Using Eye Tracking to Re-examine the Effect of Future Tense in Intertemporal Choice." AEA RCT Registry. AEARCTR-0009332. https://doi.org/10.1257/rct.9332-1.0

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

Jiang Shiang Hu* , Josie I Chen , Wei James Chen

* Primary investigator

Registry ID
AEARCTR-0009332
DOI
https://doi.org/10.1257/rct.9332-1.0
First registered
Last updated
Status
In development
Trial dates
2022-05-02 – 2023-05-30
Countries
Taiwan, Province of China ()
Registry record
http://www.socialscienceregistry.org/trials/9332

Abstract

The linguistic-savings hypothesis (LSH) indicates that languages with future tense reference would distance people’s perception of future events and causally reduce future-oriented behaviors. Chen, He, and Riyanto (2019) examined this hypothesis taking an experimental approach based on a Chinese linguistic feature: the omission of the future tense is allowable in grammar. To be specific, the auxiliary word ‘will’ was arranged in the treatment group and was omitted in the control group. This manipulation was expected to drive subjects to be more impatient in the treatment group. Nevertheless, no supporting evidence for the LSH was found. It is essential to have a stronger intervention and investigate the cognitive mechanism of the LSH. Therefore, we replicate Chen et al. (2019) and modify it in several dimensions. First, instead of using a price list to elicit preference, we use a binary choice design to strengthen the treatment effect. Secondly, we apply an eye-tracker to investigate the visual pattern and measure the percentage of fixation on the key word “will”. Lastly, we estimate parameters in both discounting models and a drift-diffusion model at an individual level. These modifications lead to a more comprehensive exploration of the cognitive process of the LSH.

Experimental design

**Treatment and Prediction**

The experiment aims to examine whether the future tense reference influences people’s preference for intertemporal choice. The future payment in the treatment group will be described as “After Y weeks will receive Z tokens.”; while that in the control group will be described as “After Y weeks, receive Z tokens.” The only difference between the two descriptions is that “will” was added in the treatment group and was replaced by a “,” (comma) in the control group. Importantly, in Chinese characters, “will” and a comma both occupy one-word space. As the linguistic-savings hypothesis (LSH) predicts, the subjects in the treatment group (future-tense condition) should exhibit less patience than those in the control group (present-tense condition).

**Binary Choice Design**

Instead of the popular multiple price lists method, this study adopted a binary choice design. Subjects were in a scenario in which they decide to receive payment either today or in the future. The scenario was composed of 72 questions and was displayed sequentially. In each question, a subject had to choose between an immediate payment and a delayed payment.

**Screen Displayed**

To apply eye-tracking in the study, we simplified the screen as much as possible. First, the amount of immediate payment “X” was displayed in the left-upper corner of the screen. Second, the description of future payment “After Y weeks (will/,) receive Z tokens” was displayed at the upper center of the screen. Lastly, two rectangles indicating “today” and “future” were displayed at the bottom of the screen, and which one is on the left is randomly decided.

**Experiment Procedure**

College subjects will be recruited through Taiwan Social Sciences Experimental Laboratory (TASSEL) at National Taiwan University. Experiments will be conducted in person and programmed through MATLAB. To start the experiment, the computer will randomly assign each subject to either the control group or the treatment group (between-subjects design). Also, the experimenter will not monitor the subjects’ responses throughout the experiment (double-blind design).

The experiment is divided into two stages. At the beginning of a stage, a subject will be informed the immediate payment was 100 or 120 tokens. Then, he will compare this immediate payment to a future payment which has nine different amounts (110, 120, 130, …, 190 tokens) and four different durations (one, two, four, or eight weeks). To control the order effect, two stages with different immediate payments are in random order. Also, nine different amounts and four different durations of future payments are randomly decided.

After all intertemporal choices in the two stages are made, the computer program will randomly draw one decision from the first stage and the other from the second stage, to realize the payoff. The tokens will be exchanged for real money with the conversion rate of 10 tokens = $12 New Taiwan Dollar (NTD). The money will be paid by a wire transfer, regardless the paid time is today or several weeks later. Finally, subjects are asked to complete a questionnaire with demographic characteristics and paid $100 NTD in cash for their participation.

Design details

Randomization method
The treatment randomization was done by computer programming (MATLAB).
Randomization unit
Individual
Clusters
N/A
Observations
102 student subjects
Arms
51 subjects for the control group and 51 subjects for the treatment group
Primary outcomes
1. The total number of future rewards is chosen for each subject. 2. Whether the future reward is chosen in each intertemporal choice. 3. Percentage of dwell time / Number of fixations of the regions of interest (ROI). 4. Estimated parameters (λ,δ) of the exponential discounting model. 5. Estimated parameters (λ,κ) of the hyperbolic discounting model. 6. Estimated parameters of the drift-diffusion model. - Boundaries: +a for delayed reward is chosen and -a for immediate reward is chosen. - Response bias B_R (B_R < 0 if Immediate reward is favored at the beginning) - Evaluation bias B_E (B_E <0 if Immediate reward is favored during the process) - w_R: attribute weights for monetary amounts. - w_T: attribute weights for time delays. - Non-decisional time t_ND

Cited by

No citation data available for this registration yet.

Keywords