class: center, middle, inverse, title-slide .title[ # Sociological Projectdesign ] .subtitle[ ## Quantitative MA Projects ] .author[ ### Merlin Schaeffer<br> Department of Sociology ] .date[ ### 2022-11-07 ] --- class: center middle background-image: url(Figures/erbsen_zaehlen.jpg) background-size: cover # How to go about <br> a quant speciale in four steps --- layout: true class: clear # 1. A comparative yes/no research question .left-column[ *1. You need more than a topic/phenomenon of interest.* **By the end of this course:** <br> You should have formulated a **research question** that you will set out to answer, and a **theoretically-informed claim** of what you think is most likely the answer to it. <br> <br> Contact potential supervisors; I have a folder with vague potential topics! ] --- .right-column[ - The working title of your MA project should be that research question - ideally a yes/no question. + .backgrnote[ **Inintial research question:** Are immigrants and their descendants more happy in mainstream middle-class neighborhoods *than* in ethnic enclaves?] ] --- .right-column[ - Your claim/answer should be based on sound sociological reasoning/theorizing and established theory. Your claim is, of course, only one part of the overall answer. - .backgrnote[ **Claim:** No, because ethnic enclaves offer dense ethnic infrastructures (ethnic associations, shops, and sites of worship), which are a source of life satisfaction to minorities.] .push-left[ - A really stellar thesis makes a novel claim and argument; that's really difficult. ] .push-right[ - If your claim is not novel, you need to apply it to a novel context or empirically test it in a better/innovative way. ] <img src="Figures/Segregation.jpg" width="90%" style="display: block; margin: auto;" /> ] --- layout: true # 2. What is the central comparison? .left-column[ *2. How can you provide a convincing test of your theoretically-informed claim?* **By the end of this course:** <br> You should have devised a **research design.** <img src="Figures/Wellmob_1.png" width="100%" style="display: block; margin: auto;" /> ] --- --- .right-column[ - Draw a conceptual/causal diagram of your idea. <img src="Figures/WELLMOB.png" width="88%" style="display: block; margin: auto;" /> ] --- .right-column[ - What .alert[comparison] would provide an empirical test of your claim? Also check out my [slides on good comparisons](https://merlin-guest-lectures.netlify.app/2022/02/22/apples-and-oranges-comparisons-in-social-science-ressearch/). + .backgrnote[Immigrants in neighborhoods with no versus very dense ethnic infrastructures.] - What could bias that comparison? + .backgrnote[ Neighborhoods with dense ethnic infrastructures are often poor inner-city neighborhoods and the others are often mainstream suburbs.] + .backgrnote[ Well-off immigrants move to suburb, marginalized less-integrated live close to ethnic infrastructure.] - Sketch out what kind of data would be *ideal* to test your idea. + .backgrnote[ Randomized assignment of immigrants to neighborhoods differing in ethnic infrastructure only.] ] --- .right-column[ - It is difficult, time-consuming, and (usually) expensive to collect high-quality data. Existing data contain qualities that you are not able to match: - measures over time, - across countries, - post-stratification weights. - Consider what data are already our there, and which come closest to your ideal setup! - [Afrobarometer](https://www.afrobarometer.org/) - [Party Manifesto Data](https://manifesto-project.wzb.eu/) - [Danish Data Archive](http://dda.dk/simple-search) - .backgrnote[We use the German Socio-Ecomic Panel 1984-2018. Allows us to compare persons of immigrant origin who moved between neighborhoods.] + Newspapers + Wikipedia + Google search data + .backgrnote[We use Google maps to scrape and geo-locate ethnic infrastructures.] ] --- layout: false # 3. Feasibility! .left-column[ <img src="https://www.businessstudynotes.com/wp-content/uploads/2016/12/Feasibility-Study.jpg" width="100%" style="display: block; margin: auto;" /> ] .right-column[ - Check **feasibility!** That is, download the data and check: - Does it contain all variables you need? Make a histogram/bar chart of each variable you want to use. - Does it contain the populations you are interested in? How large are the case numbers in the sub-populations you want to study (e.g., unemployed, foreign-born women). - How about missing values; are there enough observations to estimate associations? - .alert[Whether you have a significant result is **irrelevant**!] ] --- layout: false # 4. Modularize! .left-column[ <img src="Figures/Rainbow-Cake760x580.jpg" width="100%" style="display: block; margin: auto;" /> <img src="Figures/bubblegum-rainbow-cake.jpeg" width="100%" style="display: block; margin: auto;" /> ] .right-column[ - **Modularize your envisioned analysis!** That is, structure your analysis as a layered cake with potential icing and cherry on top. - The core analysis should be feasible and earn you a 7. - Only then add more complex analyses and work your way up to a 10 or even 12. ] --- class: inverse middle center **By the end of this course:** Discuss your expose with potential supervisors --- class: clear # Collecting your own data .font60[**I strongly advise against it!**] .left-column[ *The trade-off associated with collecting you own data:* - **Novel data (yay!)** .center[*versus*] - **A strong tendency for mediocre quality** - **Mediocre statistical analyses** + `\(\rightarrow\)` Little opportunity to showcast your skills. ] -- .right-column[ .center[*Emphasize the novely of your data!*] - If you want to do a survey: + Explain well why there is no existing data on the population of interest yet. + If there is data, explain well why the variables contained are insufficient and can't even be used as proxies. + Use novel ways to collect data, such as wiki surveys (Salganik and Levy, 2015). + Consider implementing a survey experiment (Auspurg and Hinz, 2014). - Consider collecting non-survey data. - Consider implementing a field experiment. ] --- --- # References .font80[ Auspurg, K. and T. Hinz (2014). _Factorial Survey Experiments_. Newbury Park: Sage. Salganik, M. J. and K. E. C. Levy (2015). "Wiki Surveys: Open and Quantifiable Social Data Collection". In: _PLOS ONE_ 10.5, p. e0123483. ]