Measuring Political Media Slant Using Text Data
[ pdf, Web appendix]
Viz. Longer summary Speech topics Viz.

This paper explores a novel and objective measure of political media slant using the aid of natural language processing and machine learning. Implementing this approach in the institutional context of Singapore, the paper finds that The Straits Times, the flagship daily, is less accurate when quoting the parliamentary speeches of opposition politicians.

1 / 2

Substring accuracy across time

2 / 2

Effect Sizes

❮ ❯

Women's Wave or the Blue Wave: The Me Too Movement and the 2018 U.S. Midterm Elections
[ pdf, old version]
Viz. Longer summary

In this study I download all tweets containing the MeToo hashtag in the year 2018 leading up to the midterm elections in November and parse the geolocation tags of the tweeters. With the county-level density measure of the tweets in 2018, I test whether the movement can explain the historic performance of women candidates in the 2018 U.S. House elections.

1 / 3

Timeline

2 / 3

Conditional plot

3 / 3

Higher prob. of Dem. woman candidates

❮ ❯

International Trafficking in Persons and Global Health Security: Evidence from Two Modern Pandemics
[ pdf]
Abstract

Does Working from Home Work? A Natural Experiment From Open-Source Software Development and Lockdowns
[ Available on request] [ Online appendix]
Abstract Viz.

1 / 4

Event study

2 / 4

Event study w user controls

3 / 4

Decomposition

4 / 4

Treatment arms

❮ ❯

Neighborhood Mismatch and Visits in a Small City
[ Available on request]
(with LIM Jing Zhi)
Abstract Viz.

1 / 4

Scale & borders

2 / 4

Coeffs

3 / 4

Two neighbourhoods (Mount Pleasant & Toa Payoh Central)

4 / 4

❮ ❯

Social Exposure Zones in a Small City
(with LIM Jing Zhi)

Income and Productivity Trends of Guangdong: A Data-Driven Case Study of the Greater Bay Area
[ Available on request]
(with LI Jingwei and ZHANG Xuyao)

Segregation Across Neighborhoods in a Small City
[ pdf]
(with LEE Shu En, LIM Jing Zhi)
[*Superceded by Neighborhood Mismatch and Visits in a Small City]
Abstract

Unexpected Shocks to Movement and Job Search: Evidence from COVID-19 Policies in Singapore Using Google Data
[ pdf]
Abstract Viz.

1 / 3

Early changes (anomalies) in movement patterns

2 / 3

Lockdown announcement effect

3 / 3

Early changes in job searches

❮ ❯

LexicalRichness: A module to compute textual lexical richness
LexicalRichness computes measures of lexical richness (aka lexical diversity or language sophistication) from textual data. Lexical richness refers to the ratio of different unique words to the total number of words in a text. Measures include MTLD and HD-D, which are resilient to varying text lengths.
Src PyPI ReadTheDocs DOI

runPyNB: Run Jupyter notebooks in command-line and Makefiles
runPyNB offers a command-line utility that runs (and time) Jupyter notebook files (.ipynb) without launching the IDEs (Jupyter notebooks or JupyterLab).
Src PyPI ReadTheDocs DOI

Forestplot: Make publication-ready but customizable forest plots
Forestplot makes publication-ready forest plots easy to make out-of-the-box. Other names for forest plots are: coefplots, coefficient plots, meta-analysis plots, dot-and-whisker plots, blobbograms, margins plots, regression plots, and ropeladder plots. Users provide a dataframe (e.g. from a spreadsheet) where rows correspond to a variable/study with columns including estimates, variable labels, and confidence interval limits.
Src PyPI ReadTheDocs DOI