background-color: #094e0d class: inverse, middle background-image: url(https://textvulture.github.io/portrait_July2024.png) background-position: 23cm 2.5cm background-size: 20% # .large[Inside a Quantitative Researcher’s <br> Workflow: .yellow[A Sneak Peek]] <br><br> ---- ## .right[Byunghwan Ben Son <br> .tiny[(Associate Professor, GLOA, GMU)]] ### .right[GLOA 491 (MK) Guest Lecture] --- # Before we start: ### The slides are completely .pink[online] and you can easily access it. <img src="qr.png" width="400px" style="display: block; margin: auto;" /> --- # An .pink[example] to work with: ![](https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/bkv_screen.png){width="80%"} --- class: inverse background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/waveBackground2.jpg?raw=true) background-size: 120% -- <br><br> ## But This won't be a research presentation. - .Large[You've already read the paper
.] -- ## Instead, we will take a look at how a paper like that is .yellow[conceived] and then .bluey[produced] (hence the title). -- - .Large[If you're interested in how this project was .pink[presented] in a typical research talk, here's the link to the slides: (Presented at MPSA 2021)] -- <img src="qr2.png" width="150px" style="display: block; margin: auto 0 auto auto;" /> --- class: inverse, left, top background-image: url(https://media1.tenor.com/m/UQsZGD3pinAAAAAd/shall-we-begin-star-trek.gif) background-size: 110% # Let's start with some method talks. -- # I am sure you must've <br> talked about this <br> .red[a lot] already in this class. -- # But we need to first talk about <br> .pink[where this research belongs.] --- <img src="https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/methods_tree_1.png" width="75%" style="display: block; margin: auto;" /> --- <img src="https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/methods_tree_3.png" width="75%" style="display: block; margin: auto;" /> --- <img src="https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/methods_tree_4.png" width="75%" style="display: block; margin: auto;" /> --- <img src="https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/methods_tree_5.png" width="75%" style="display: block; margin: auto;" /> --- <img src="https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/methods_tree_6.png" width="75%" style="display: block; margin: auto;" /> --- class: inverse, middle background-color: black background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/forestBackground2.jpg?raw=true) background-size: 100% # Okay, we now know what kind of .yellow[method] we are talking about. -- # Now let's get on to the .Blue[project]: ###
"Consequences of Democratic Backsliding on Popular Culture: Evidence from Blacklist in South Korea" *Democratization*, Forthcoming. --- class: inverse background-color: black background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/blacklist_eg.jpeg?raw=true) background-position: right #
Blacklists, literally -- ## - names of individuals and groups ## - 'anti-government' or 'lefty'
## - mostly using existing lists <br> (open letter signatories) -- ## - exclusion and disadvantage ## - Govt: NIS, Blue House, MHST ## - discrete operation --- background-image: url(https://img.khan.co.kr/news/2015/11/04/l_2015110401000555600049501.jpg) background-position: right background-size: 40% # How to Get on the List .pull-left[ <table class="table table table-striped" style="width: auto !important; margin-left: auto; margin-right: auto; width: auto !important; margin-left: auto; margin-right: auto;"> <thead> <tr> <th style="text-align:left;"> List </th> <th style="text-align:left;"> Based </th> <th style="text-align:left;"> Pres </th> <th style="text-align:left;"> Reasons </th> </tr> </thead> <tbody> <tr> <td style="text-align:left;"> 1 </td> <td style="text-align:left;"> 2008 </td> <td style="text-align:left;"> Lee </td> <td style="text-align:left;"> critical of the govert; 'left-leaning' -- only 60 movie workers (but more, unidentified) </td> </tr> <tr> <td style="text-align:left;"> 2 </td> <td style="text-align:left;"> 2012 </td> <td style="text-align:left;"> Park </td> <td style="text-align:left;"> supporting the opposition presidential leader </td> </tr> <tr> <td style="text-align:left;"> 3 </td> <td style="text-align:left;"> 2014 </td> <td style="text-align:left;"> Park </td> <td style="text-align:left;"> participants of the open letter regarding the Sewol Ferry sinking; supporting the opposition candidate for Seoul Mayoral election </td> </tr> <tr> <td style="text-align:left;"> 4 </td> <td style="text-align:left;"> 2015 </td> <td style="text-align:left;"> Park </td> <td style="text-align:left;"> participants of the open letter criticising government handling of the Sewol Ferry sinking </td> </tr> </tbody> <tfoot> <tr> <td style = 'padding: 0; border:0;' colspan='100%'><sup>*</sup> 'based' is the year in which the event that the list is based on occurred.</td> </tr> </tfoot> </table> ] --- class: inverse, middle background-image: url(https://img.seoul.co.kr/img/upload/2016/12/08/SSI_20161208102613_O2.jpg) background-position: right background-color: black background-size: contain # "*Why have you not .red[sanctioned] <br> these [anti-government] <br> movie producers yet?*" <br><br> ## **Kim Ki-choon**, <br> Park Geun-hye's Chief Secretary, <br> after watching *The Attorney* <br> to **Yoo Jin-ryong**, <br> Minister of Culture, Sports and Tourism. --- background-image: url(https://pbs.twimg.com/media/ENzTfM7VUAEyaEI.jpg) background-position: right background-size: contain # **Motivation** .pink[.small[.tiny[(why chose the particular method)]]] -- - This is upsetting
. -- - But .pink[what] exactly am I upset about? -- - Suppression of freedom of speech? Yes, but more. -- - .green[personal] to me: - *sleazy people ruining cool people's lives.*
-- - Okay, but how can I demonstrate this empirically? -- - Interviews? Fieldwork? -- - **Only** if the victims were really aware of things. --- .full-width-tinted[ # Causal Inference Approach ] ### Again, I want to argue: *sleazy people ruining cool people's lives.* - 'sleazy people' getting power: **cause** - 'lives ruined': **effect** -- ### The argument has a good theoretical framework to work with: .orange[democratic backsliding] - will come back to this later -- ## .content-box-green[The Hypotheses .tiny[(translation of the argument into a relationship between the .red[observables])]] ### `\(H_1\)`. Being blacklisted .red[decreases] a movie worker's chances of securing a movie job. ### `\(H_2\)`. The negative career effect of Blacklist is larger for .red[non-actor] movie workers. --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br> *'A downward career trajectory <br> for the blacklisted, <br> compared to those who weren't.'* --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br> *'A downward career trajectory <br> for .content-box-green[the blacklisted], <br> compared to those who weren't.'* --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br>*'A downward career trajectory <br> for .content-box-green[the blacklisted], <br> compared to those who weren't.'* - .Large[.content-box-green[treated]] --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br> *'A downward career trajectory <br> for .content-box-green[the blacklisted], <br> compared to .content-box-yellow[those who weren't.]'* - .Large[.content-box-green[treated]] --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br> *'A downward career trajectory <br> for .content-box-green[the blacklisted], <br> compared to .content-box-yellow[those who weren't.]'* - .Large[.content-box-green[treated]] - .Large[.content-box-yellow[controlled]] --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br> *'A .content-box-red[downward career trajectory] <br> for .content-box-green[the blacklisted], <br> compared to .content-box-yellow[those who weren't.]'* - .Large[.content-box-green[treated]] - .Large[.content-box-yellow[controlled]] --- background-image: url(https://images.stockcake.com/public/f/5/6/f56d2eac-113f-44e3-9929-d5c3e24c3085_medium/rustic-direction-sign-stockcake.jpg) background-position: right background-size: contain # In a .Yellow[causal inference framework], <br> I need to show <br> *'A .content-box-red[downward career trajectory] <br> for .content-box-green[the blacklisted], <br> compared to .content-box-yellow[those who weren't.]'* - .Large[.content-box-green[treated]] - .Large[.content-box-yellow[controlled]] - .Large[.content-box-red[outcome]] --- class: center, middle # If only there was .orange[a complete database] that contains .pink[all credit information] for each of the movies in Korea ... --- class: center, middle background-image: url(https://raw.githubusercontent.com/textvulture/textvulture.github.io/refs/heads/master/images/kobis.png) background-size: 100% # If only there was .orange[a complete database] that contains .Blue[all credit information] for each of the movies in Korea .red[!!!] -- # **.LARGE[.pink[Ko]rea .pink[B]ox Office .pink[I]nformation .pink[S]ystem (.pink[KOBIS])]** --- # **KOBIS**, API (application programming interface) -- ![](https://github.com/textvulture/textvulture.github.io/blob/master/images/kobis_api.png?raw=true) --- # Getting the data using API .tiny[(there was a daily limit)] ``` r # Scraping the simple information of Korean movie workers library(tidyverse) library(jsonlite) masterlist1 <- data.frame() for (I in 1:500){ url_jsonlist <- paste0("http://kobis.or.kr/kobisopenapi/webservice/rest/people/searchPeopleList.json?key=[API Key Here]=", I ,"") temp1 <- as.data.frame(fromJSON(url_jsonlist)) masterlist1 <- plyr::rbind.fill(masterlist1, temp1) Sys.sleep(1) } ... ``` ### After quite a bit of data cleaning
, a complete list of movie workers' participation in movie projects on .pink[a yearly] basis. --- class: inverse, center, middle background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/blackboardBackground.jpg?raw=true) background-size: 110% # We now have some good data to work on, but it is equally important to identify the .yellow[theoretical context] of our empirics. -- # *Where in the sea of knowledge can the findings be plugged in.* -- # .pink[Democratic Backsliding] literature --- # **Democratic Backslidng,** .red[the concept] .center[ ![](https://github.com/textvulture/textvulture.github.io/blob/master/images/backsliding_concept.jpg?raw=true) ] --- ## .orange[Backsliding] in S. Korea, .small[compromised .green[Freedom of Speech], in particular (V-dem)] <img src="GuestLecture_10292024_files/figure-html/unnamed-chunk-11-1.png" style="display: block; margin: auto;" /> --- ## .content-box-red[Match The Data] ###- Match the KOBIS data with Blacklist ###- The total number of movies where an worker's name appeared each year. --- # .content-box-green[Showing some descriptive data is always important:] ### Number (L) and Portion (R) of the .red[blacklisted] movieworkers (.violet[employed]) .pull-left[ ![](https://github.com/textvulture/textvulture.github.io/blob/master/images/bkv_count.jpeg?raw=true) ] .pull-right[ ![](https://github.com/textvulture/textvulture.github.io/blob/master/images/bkv_count2.jpeg?raw=true) ] -- ### .center[But it ain't give you no real answer.
What about .red[not] working for a movie?] --- ## .content-box-yellow[How can I **compare** the treated .tiny[(blacklisted)] & the control .tiny[(not blacklisted)]?] -- .large[The K-movie industry has evolved so dramatically. There are so many factors in play!
] -- .large[Solution: .red[D]ifference-.red[i]n-.red[D]ifference (.red[DiD])] -- .pull-left[ ![](https://thetarzan.wordpress.com/wp-content/uploads/2011/06/d-i-d.png) .tiny[source: [https://thetarzan.wordpress.com/2011/06/20/differences-in-differences-estimation-in-r-and-stata/](https://thetarzan.wordpress.com/2011/06/20/differences-in-differences-estimation-in-r-and-stata/)] ] -- .pull-right[ .center[ <div style="width:260px;max-width:100%;"><div style="height:0;padding-bottom:150.77%;position:relative;"><iframe width="260" height="392" style="position:absolute;top:0;left:0;width:100%;height:100%;" frameBorder="0" src="https://imgflip.com/embed/982w1b"></iframe> ] ] --- # Turn the Data into a .pink[panel] structure. -- ### Filling in .red[zeros] for the years of no movie work. ``` r df %>% head(., 7) ``` ``` ## # A tibble: 7 × 9 ## # Groups: id [1] ## lncount treatment shock2012 year id role cumu lncareer actor ## <dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <dbl> ## 1 0 0 0 1991 10000006 배우 0 0.693 1 ## 2 0 0 0 1992 10000006 배우 0.693 1.10 1 ## 3 0 0 0 1993 10000006 배우 1.10 1.39 1 ## 4 0 0 0 1994 10000006 배우 1.39 1.61 1 ## 5 0 0 0 1995 10000006 배우 1.61 1.79 1 ## 6 0 0 0 1996 10000006 배우 1.79 1.95 1 ## 7 0 0 0 1997 10000006 배우 1.95 2.08 1 ``` -- ### N=1.6M, unbalanced (1991-2022); .red[2012] = treatment year --- # .content-box-yellow[The model and *estimand*]: .tiny[articulate what exactly you're testing and how] .large[ `$$ln(movies + 1) = \beta_0 + \color{red}{\beta_1}Blacklisted_{it} \times Treatment_{t} + ... + FE(unit, year, role) + \epsilon_{it}$$` ] -- ### Recall our `\(H_1\)` and `\(H_2\)`: -- ### `\(H_1\)`. Being blacklisted .red[decreases] a movie worker's chances of securing a movie job. ### `\(H_2\)`. The negative career effect of Blacklist is larger for .red[non-actor] movie workers. --- ## **DiD** Result -- <iframe src="table.html" width="100%" height="550px"></iframe> --- ### .Large[**Event Study**] -- .pull-left[.large[Benchmark] <img src="https://github.com/textvulture/textvulture.github.io/blob/master/images/bkv1.JPG?raw=true" height="90%" style="display: block; margin: auto;" /> ] -- .pull-right[.large[.red[Non]-actors] <img src="https://github.com/textvulture/textvulture.github.io/blob/master/images/bkv2.JPG?raw=true" height="90%" style="display: block; margin: auto;" /> ] --- # Endless Robustness Check: .pink['But can we trust you?'] <iframe src="https://textvulture.github.io/mypage/Son2024a/OnlineAppendix.pdf" width=1200 height=450> </iframe> --- class: inverse background-image: url(https://img2.yna.co.kr/photo/yna/YH/2017/06/14/PYH2017061410710001300_P2.jpg) background-position: right background-size: contain background-color: black # .large[**Conclusion**] .tiny[(summarize and offer .Green[implications])] ##
backsliding messes with culture by <br> suppressing .yellow[freedom of speech] -- ##
often unknown as operating .red[discretely] -- ##
authoritarian toolkits .green[recycled] -- ##
another reason backsliding is .orange[undesirable] -- ##
democracy eventually worked .violet[this time] --- class: middle background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/background_White2.jpg?raw=true) background-size: 120% # Maybe .red[not] interested in quanitative methods per se, but still want to use some data in your project? # Here are some great sources that offer visualization. No coding skills needed. .Yellow[Drop-down menus] only. --- class: right, top background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/vdem1.JPG?raw=true) # .content-box-blue[**Variety of Democracy**: [Click](https://v-dem.net/graphing/graphing-tools/)] --- class: right, top background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/wvs1.JPG?raw=true) background-size: cover # .content-box-yellow[**World Value Survey**: [Click](https://www.worldvaluessurvey.org/WVSOnline.jsp)] --- class: right, top background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/WDI1.JPG?raw=true) # .content-box-red[**World Development Indicators**: [Click](https://data.worldbank.org/)] --- class: right, top background-image: url(https://github.com/textvulture/textvulture.github.io/blob/master/images/aiddata.png?raw=true) background-size: 100% # .content-box-army[**Chinese Development Finance** [Click](https://china.aiddata.org/)]