Back to History

Fields Changed

Registration

Field Before After
Abstract An extensive literature indicates that recipients’ characteristics are salient in fostering others’ prosocial behavior. However, early researches remain uninformative about these influences in a general and eastern situation. The current work addresses this ambiguity by specifically exploring whether recipients' beauty, local identity and gender impact receiving help in daily life in China. We will conduct a randomized controlled trial to detect such bias. We use lost-resumes and randomly deliver the personal resumes with file bags to shared-bikes in public within cities. A wide range of unpaid passersby might notice and make decisions accordingly. We intervene by modifying the resumes in one's traits: beauty, local identity and gender, and therefore we have eight treatments that are consistent in other factors. Our interest is in the frequency with which people finding our file bags accidentally are willing to contact us by calls or emails, and whether it varies based on their preferences for the three controlled factors. Therefore, we can compare the spontaneous response rate between treatments, to examine the effects of beauty, local identity and gender on Chinese helping behavior. Our study casts new light on the role of relevant individual characteristics in fueling prosocial behavior and improving well-being of people. An extensive literature indicates that recipients’ characteristics are salient in fostering others’ prosocial behavior. However, early researches remain uninformative about these influences in a general and eastern situation. The current work addresses this ambiguity by specifically exploring whether recipients' beauty, local identity and gender impact receiving help in daily life in China. We will conduct a randomized controlled trial to detect such bias. We use lost-resumes and randomly deliver the personal resumes with file bags to shared-bikes in public within cities. A wide range of unpaid passersby might notice and make decisions accordingly. We intervene by modifying the resumes in one's traits: beauty, local identity and gender, and therefore we have eight treatments that are consistent in other factors. Our interest is in the frequency with which people finding our file bags accidentally are willing to contact us by calls or emails, and whether it varies based on their preferences for the three controlled factors. Therefore, we can compare the spontaneous response rate between treatments, to examine the effects of beauty, local identity and gender on Chinese helping behavior. We predict nonlocal candidate is helped more than local candidate, plain candidate is helped more than attractive candidate, female candidate is helped more than male candidate. As a consequence, we anticipate nonlocal plain woman is helped the most, local attractive man is helped the least. Our study casts new light on the role of relevant individual characteristics in fueling prosocial behavior and improving well-being of people.
Trial Start Date May 26, 2026 June 16, 2026
Last Published May 23, 2026 11:32 PM June 15, 2026 08:06 AM
Intervention Start Date May 26, 2026 June 16, 2026
Intervention End Date September 30, 2026 December 30, 2027
Primary Outcomes (End Points) Whether and how soon people are willing to call to contact us: 1. Total calls: the number of the first call received for each resume in each treatment. 2. Distribution of calls received over time. 3. Hazard rate: time duration between the time a resume is delivered and the first call received. 4. Whether a helper contacts and communicates with us more than once (through phone call or text message) for one resume. Whether and how soon people are willing to call to contact us: 1. Call rate: the number of the first call received for each resume relative to the number of resume delivered in each treatment. 2. Distribution of calls received over time. 3. Hazard rate. 4. Whether a helper contacts and communicates with us more than once (through phone call or text message) for one resume.
Sample size (or number of clusters) by treatment arms We plan to deliver 6000 resumes in total which will be assigned to the eight treatments as shown in the table below. -----------------------------------Male----------Female Beauty---------Local----------750-------------750 -----------------Nonlocal-------750-------------750 Plainness-----Local----------750--------------750 -----------------Nonlocal-------750-------------750 We plan to deliver 5200 resumes in total which will be assigned to the eight treatments as shown in the table below. -----------------------------------Male----------Female Beauty---------Local----------650-------------650 -----------------Nonlocal-------650-------------650 Plainness-----Local----------650--------------650 -----------------Nonlocal-------650-------------650
Power calculation: Minimum Detectable Effect Size for Main Outcomes Given the sample size, we calculate the minimum detectable effect size using the G Power, as shown below: t tests - Means: Difference between two independent means (two groups) Analysis: Sensitivity: compute required effect size. Input: Tail(s) = Two α err prob = 0.05 Power (1-β err prob) = 0.8 Sample size group 1 = 3000 Sample size group 2 = 3000 Output: Non-centrality parameter δ = 2.8020305 Critical t = 1.9603596 Df = 5998 Effect size d = 0.0723481 Based on the sample size of 6000 resumes, the study has 80% statistical power at a 5% significance level to detect the difference in mean values between treatments. Given the sample size, we calculate the minimum detectable effect size using the G Power, as shown below: t tests - Means: Difference between two independent means (two groups) Analysis: Sensitivity: compute required effect size. Input: Tail(s) = Two α err prob = 0.05 Power (1-β err prob) = 0.8 Sample size group 1 = 2600 Sample size group 2 = 2600 Output: Non-centrality parameter δ = 2.8020995 Critical t = 1.9604205 Df = 5198 Effect size d = 0.0777163 Based on the sample size of 5200 resumes, the study has 80% statistical power at a 5% significance level to detect the difference in mean values between treatments.
Back to top