Who are the "elite"? Dimensions of class-consciousness in Japan

How to cite

Nakabayashi, Masaki; Kawata, Keisuke; Mcelwain, Kenneth M.; Fujihara, Sho. 2021. "Who are the "elite"? Dimensions of class-consciousness in Japan." AEA RCT Registry. AEARCTR-0008322. https://doi.org/10.1257/rct.8322-1.2000000000000002

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

Masaki Nakabayashi* , Keisuke Kawata , Kenneth M. Mcelwain , Sho Fujihara

* Primary investigator

Registry ID
AEARCTR-0008322
DOI
https://doi.org/10.1257/rct.8322-1.2000000000000002
First registered
Last updated
Status
Completed
Trial dates
2021-11-08 – 2021-11-24
Countries
Japan ()
JEL codes
D6; I24; I31
Registry record
http://www.socialscienceregistry.org/trials/8322

Abstract

Societies are more or less stratified but often in different dimensions. For example, income is more skewed in the United States than in Japan, but universities are more stratified in Japan than in the United States. The skewness of income in Japan, by itself, does not necessarily mean that Japanese society is more equal than American society.

In this study, we will investigate which factors Japanese individuals count consider when identifying their "elite" people among them by an experimental survey. We will extract these factors in two stages. First, we will ask respondents to assume that Japanese society is stratified into 10 tiers and to write any word(s) to describe people classified as the most elite to the least elite or at tiers 1 (highest), 3, 5, 7, 9, or 10. Tier 1, 3, 5, 7, 9, or 10 is randomly drawn and shown to each respondent. Next, we will ask respondents to evaluate a) education, b) income, c) total household income, d) occupation, e) wealth, f) family origin, g) gender, and h) social reputation as measures with which to classify people between 1 (the most important) and 5 (the least important).

We will also collect information about respondents' background characteristics such as age, gender, prefecture, and education; working status, if employed, size of the employer and job title; income; total household income; whether he or she reads a newspaper, internet news, or books; political preferences, partisanship; and self-perception of social class.

By the design, we will attempt to identify factors shared by Japanese people in recognizing social class.

Intervention

We will adopt a factorial design with two stages.

I. Respondents will be shown that we assume that Japanese society can be classified into 10 classes from 1 (the highest) to 10 (the lowest).
II-1 During the first stage of our survey, we will ask the respondents to write any word(s) to describe class 1, 3, 5, 7, 9, or 10. A class will be randomly drawn from 1, 3, 5, 7, 9, or 10 and shown to each respondent.
II-2 During the second stage of our survey, we will ask the respondents to evaluate a) education, b) income, c) household income, d) occupation, e) wealth, f) family origin, g) gender, and h) social reputation as measures with which to classify people from 1 (the most important) to 5 (the least important).

Experimental design

We will identify factors used to classify fellow Japanese citizens by soliciting

I-1 respondents' free descriptions of a randomly drawn social class

and

II-2 respondents' evaluation of education, income, total household income, occupation, wealth, family origin, gender, and social reputation between 1 (the most important) and 5 (the least important).

We will analyze the results of II-1 by a machine learning algorithm for text mining. We use the results of II-2 such that our interpretation of the results of I-1 is consistent with our interpretation of the results of II-2 by a machine learning algorithm.

We will also collect information about respondents' background characteristics such as age, gender, prefecture, and education; working status, if employed, size of the employer and job title; income, total household income; whether she or he reads a newspaper, internet news, or books; political preferences; partisanship; and self-perception of social class. To investigate possible heterogeneity of the classification perception across the background characteristics, we will deploy a generalized random forest algorithm.

Design details

Randomization method
Randomization in process I-1 is to be conducted by a computer in a survey company.
Randomization unit
Individual.
Clusters
15,000 individuals
Observations
15,000.
Arms
15,000.
Primary outcomes
We will inductively identify factors that Japanese people use to classify fellow citizens.

Cited by

No citation data available for this registration yet.

Keywords