Fluent Essays

I have attached a document where you will find most of the information required to answer the questions in the second document. In case you do not find the answer in the document after looking through it completely you may use the internet to find the answer but do rephrase and use your own words and include the sources used.Please stick to the maximum number of the sentences required in each question. Do not copy paste answers, rephrase them and use your interpretation. Keep your answers simple and straightforward, nothing complex. Please use simple wording that is appropriate to each question. questions.docx notes.docx Unformatted Attachment Preview • • You can use the internet and the notes in the attached document. Make sure you use your own words in answering the questions. Not something you took from the internet. You need to understand yourself what you write down. Q1. How may propensity score matching lead to problems of external validity? (max 3 sentences) [5 points] Q2. In Congo, I work with an NGO that is interested in understanding whether giving aid to very vulnerable, displaced households makes them healthier. The NGO asks me if I can do an impact evaluation to learn about the impact of aid. They are explicit, I cannot randomize. Q2a. Why do you think I am not allowed to randomize the treatment? (max 3 sentences) [5 points] Q2b. After more interaction with the NGO, I learn a lot about their program. They work as follows. The NGO arrives in a target village and evaluates all households and assigns each of them a “vulnerability score”: running from 1 to 10, where 10 are the households that are most in need, and 1 is the least in need. The NGO doesn’t have aid for everybody (it has limited funds), so those households with a score higher than 3.9 get aid, those below 3.9 do not get aid. Given that I am not allowed to randomize, what, given the information just provided, is the other obvious strategy that I will use to get at causality? (max 1 sentence) [5 points] Q2c. Using this strategy will give me a “Local Average Treatment Effect”. Why do we call it “local”, and what has it to do with being able to make causal claims. Use the example provided in your response. Hint: yep, this has something to do with the bandwidth. (max 5 sentences) [5 points] Question 3 Let’s say that we collected data in 100 villages in 2015. Then in 2017, 50 villages were hit by conflict. It is 2019 now, and you suggest to your professor to collect data in all 100 villages again to get at the impact of conflict on health. Your professor says OK. Now, of course, we cannot simply build on the 2019 data and compare health outcomes in the villages that experienced conflict with those that were not. That’s apples and oranges. We can also not compare health outcomes in the treated villages before and after the conflict. Apples and oranges again. Combining these, however, may help with getting at causality. This technique is called Difference-In-Difference (DID). Let’s say if we use the information from 2015 and 2019 and we create the figure below, where blue indicates the control villages, and red the treatment villages. The y-axis has information on the average quality of health, and the vertical line indicates when the treatment took place. Q3a. Please tell me what, following DID, the counterfactual is for the treated villages in 2019 (i.e. the red dot above “2019”)? Remember if we want to get at causality we always want to compare the units that were treated to a good counterfactual. (max 2 sentences) [5 points] Q3b. So what is the effect of conflict on health if we believe DID? Just a number (and a sign if necessary) as response is enough. [5 points] Q3c. “But in 2019 treatment villages are different than control villages!” somebody – who doesn’t know much about DID – might say. How does DID get rid of this ‘problem’? (max 5 sentences) [5 points] Q3d. Please explain how migration might be a problem for internal validity in this specific case? Use the example. (max 5 sentences) [5 points] Question 4 In developing countries, does poverty lead to war? This is a difficult question to answer because of reverse causality: poverty may cause war, but war may also cause poverty. Somebody suggests to you to use an Instrumental Variable (IV) strategy to get at causality, and suggests “rainfall” as an instrument. Let’s say you have all the data you need. Q4a. What is the proposed first-stage in this example, and is it convincing in your opinion? (max 2 sentences) [5 points] Q4b. What is the proposed independence assumption in this example, and is it convincing in your opinion? (max 2 sentences) [5 points] Q4c. What is the proposed exclusion restriction in this example, and is it convincing in your opinion? (max 2 sentences) [5 points] Q4d. In the example “Abuse busters” we discussed in class, the researchers make use of an IV strategy. Why did they do so, given that they had randomly assigned the treatments, and we know that randomization is so much better than all the other strategies? While giving your answer, please provide an answer to the question that is on the slides “Why can’t we compare those people that received separation/advice with those that were arrested?” (max 5 sentences) [5 points] Q5. You area a researcher and you are interested in testing the hypothesis that Trump leads to more polarization in the United States. You collected survey data in 2019 from a representative sample of US citizens that, among others, collected information about polarization. It is an interesting hypothesis. However, why can you not test it given this set up? (max 5 sentences) [5 points] Research Design & Causality Week 1 Intro Keep in mind: - Studies are not causal (no “internal validity”) - Badly measured outcomes (“measurement error”) - Generalizes too far (no “external validity”) - Selective selection of:analysis, results, etc. (“data mining”, “p-hacking”, etc.) - Wrong interpretation of results - And even: fake data, fake universities, etc. Research Process: Research Design: - Research question Theory Hypotheses Discussion of how to test hypotheses Data collection Data analysis And presentation of results Research Design Week 2 Hypotheses/Research Research question: - Over-arching umbrella questions that address your topic. Include key terms that you can use to help you research your topic. Generally, research questions are questions you do not know the answer to. The idea is you are using your research to gain new knowledge. How to find your research question? - Read the previous literature. What is the research frontier? What excites you? - Talk to mentor, others knowledgeable about research area - An experience you or somebody had - Read from other literatures and apply to your own field - Imposed on you/ suggested to you. - Strategic: will it get published? - What is possible: E.g. T2+ - Etc… How to get to a bad research question How to get to a good research question For a good research project: - Anticipate the results before doing the study Pick an area on the basis of the interest of the outcome Look for an under-occupied niche that has potential Go to talks and read papers outside your area of interest Hypotheses: Hypotheses are educated guesses A good hypothesis has a number of characteristics: 1. A statement, not a question. Prediction about what will happen. 2. Theoretically grounded: based upon relevant literature. Plausible. 3. Specifies the relationship between the values of two or more variables, including connection and tendency 4. Should be clear. Clear and simple language. 5. Keep the variables in mind. Variables in easy-to-measure terms. 6. Makes it “testable” using empirical data. This means that the data collected can disprove the hypothesis. Should be able to repeat your analysis over and over again. Falsifiable(Karl Popper). “all swans are white” Types of research questions - What is the effect of X on Y? How does X lead to Y? What determines Y? Also: - What is the effect of X1versus X2on Y? When does X have an impact on Y? Three ways of writing the question: 1. Can X lead to Y a. Hypothesis will be X leads to Y 2. What causes Y a. Hypothesis will be all the possible Xs that determine Y, (X1 leads to Y, X2 leads to Y) 3. How/why X leads to Y a. Hypothesis will be X leads to Y via Z Final note on hypotheses Week 2 Ethics “… most current published research findings are false.” Ioannidis (2005) Why we worry: - Results not replicable Results not stable Results not credible Results do not cumulate Etc. Publication bias = the outcome that occurs when, for whatever reason, publication practices lead to bias in the published parameter estimates. Why publication bias? 1. Editors and reviewers may prefer significant results (and reject methodologically sound articles that do not achieve certain statistical significance thresholds); 2. Scholars may only submit studies with statistically significant results to journals and place the rest in “file drawers”; 3. Investigators may adjust sample sizes after observing that results narrowly fail tests of significance; and 4. Researchers may engage in data mining/ massaging/ torturing? Why is it a problem? Publication bias just allows people/readers to assume that the are a directly correlated relationship whereas it could have solely been a coincidence How to “data mine” for results - Use certain variables over others variables Cut part of the data Look at only certain subgroups Add certain variables to your equation You can also… just make up your data What are possible solutions? 1. Transparency: see for more information about transparency in research: http://cegablog.org/transparency-series/ 2. Change when work gets reviewed 3. Replication 4. Pre-registration When does work get reviewed? Replication “Reproducibility is just collaboration with people you don’t know, including yourself next week”—Philip Stark, UC Berkeley “Economists treat replication the way teenagers treat chastity - as an ideal to be professed but not to be practised.”—Daniel Hamermesh, UT Austin But also replication is not without its problems “Who checks the replicator?” - Most of the research are subjective - The replicator wants to find mistakes, and to gain attention. - Replication is a waste of money and time. Summer of 2015 - Study on the educational benefits of deworming in Western Kenya (Miguel and Kremer 2004). Epidemiologists (Aiken et al, 2015; Davey et al 2015) replicated analysis. Critique Of the seminal paper Start of the worm wars Pre-registration What do medical journals do? Pre-registration - https://clinicaltrials.gov/ - http://apps.who.int/trialsearch/ http://www.trialregister.nl Key elements: e.g. 13) Description of interventions 14) Key inclusion and exclusion criteria 17) Target sample size 18) Recruitment status 19) Description of primary outcome 20) Description of secondary outcomes But not much detail on method of analysis - - Implement pre-registration, but with uneven quality Mathieu et al. (2009): - 46\% of 323 trials registered satisfactorily. Others: - missing outright (28\%) - registered after the study completion (14\%) - incomplete description of the primary outcomes (12\%). Of the 46\%, 31\% had evidence of deviations from research plans. Generally these went towards finding significant results. What do social scientists do? - Almost nothing, but things are changing Increased use of existing registries in social science: http://www.3ieimpact.org/evaluation/ridie/ https://www.socialscienceregistry.org/ http://e-gap.org/design-registration/ Have a look at some of my registered pre-analysis plans: - http://egap.org/search/node/van\%20der\%20windt\%20type\%3Aregistration Benefits of registration: 1. 2. 3. 4. Informative signal Tie hands Ex ante check of instruments Database of studies Costs of registration: 1. Benefits of waiting 2. Costs a lot of time and effort 3. Isaac Asimov: “The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! but Thats funny...’” Week 3 Theory, Concept and measurements Theory construction and testing: ● What is a “theory”, and why is it necessary? What does it do for us? ● Testing our theories: from theory to concept to measure Description vs Causation: ● Which one is more interesting in your opinion? ● Which one is more important in your opinion? 90\% of the published articles/journals are causal papers 1. What is democracy? How to measure it? 2. Does democracy enhance the prospect of peaceful coexistence? What is a theory? ● “A statement of relationships between concepts” ● “A roadmap for organizing ideas and knowledge about the social world” ● Occams razor: ○ 14th century logician and Franciscan friar; William of Occam. ○ Pluralitas non est ponenda sine neccesitate“ ■ Shave off what is unnecessary, keep it as easy as possible. ○ When you have two competing theories which make exactly the same predictions, the one that is simpler is the better. Why do we need theory? ● Theories allow us to: 1. Explain: Explain how aspects of social reality ‘work’. 2. Intervene: We can use them to do something 3. Predict: Allow us to generalize from known to unknown ● “Theories are the crown of science, for in them our understanding of the world is expressed. The function of theories is to explain.” Rom Harre, The Philosophies of Science, 1985 ● “It is clear that, infact, the power to explain involves the power of insight and anticipation, and that this is very valuable as a kind of distance-receptor in time, which enables organisms to adapt themselves to situations which are about to arise.” Kenneth Craik, The Nature of Explanation (1943) ● “If I find sand on my desk, I shall want to find out the underlying reason for this circumstance. I make this inquiry not because of idle curiosity, but because only if I refer this relatively insignificant offshoot event to an underlying core event will I attain a stable environment and have the possibility of controlling it.” Fritz Heider, The Psychology of Interpersonal Relations (1958) Parts of a theory: 1. Concepts 2. Relationships 3. Can be represented by a model Concepts ● “The choice of terms and definitions” ● “The linguistic containers we use to carve up the empirical world” ● Karl Popper: “Never let yourself be goaded into taking serious problems about words and their meanings. What must be taken seriously are questions of fact, and assertions about facts, theories, and hypotheses; the problems they solve; and the problems they raise.” Defining concepts Democracy ≡ ● “Having free and fair elections” (Minimum) ● “Free and fair elections, executive is elected, universal male and female suffrage, executive constraints, and media freedom.” (Maximum) Use previous studies! If not necessary, don’t invent the wheel again. From concept to measures ● Many social science concepts are not directly observable. Latent. ● Examples: democracy, corruption, etc. ● Criteria of good measurement: reliability(extent to which assessments are consistent) and validity(“accuracy of an assessment; whether or not it measures what it is supposed to measure”) ○ E.g. “democracy” ■ Adam Przeworski: Democracy ≡ a) chief executive and legislature must be elected, 2) there must be more than a single party competing in the elections, and 3) at least one alternation in power under identical electoral rules must take place. ■ Is this a valid measure? ■ E.g. “Botswana problem” (a single party but would secede in elections) How to measure women empowerment? See Van der Windt(2018) ● Community level: ○ Presence at discussion ○ Participated in discussion ○ Committee membership ● Household: ○ Time-use ○ Domestic violence ● Project selection ● Attitudes: “if a man mistreats his wife she has a right to complain” Week 4 Literature Review Why? ● Makes you understand the research in your field ● Distinguishing what has been done from what needs to be done (i.e. what is your contribution) ● Identifying relationships between ideas and practices ● Establishing the context of the topic or problem ● Rationalizing the significance of the problem ● Etc. What (not) to include: ● Know what the review needs to contain: ○ First have your RQ (so you know your concepts) ○ Only that what is related to your research question (I’ll give examples later) ● Don’t include too much! ● Unfortunately: ○ Include people because they are possible reviewers Where to find literature: ● Google scholar ● Start with: ○ Top journal ○ Top people ○ Most recent (working papers, conferences, etc). ● And use references from there ● My strategy: ○ I collect first 3-7 “key” studies, read and summarize ○ I organize things in Mendeley ○ surnameYEARkeywords How to structure you review: 4 examples How do you build on previous studies? 1. You do the same but you do it better A very short example: “Young (2017) found that on the job training programs had a 0.12 standard deviation impact on health in the United States. Edwards and Miguel (2012) found a very similar result in Germany. These studies, however, are based on surveys, which are vulnerable to response bias. In this study we make an important contribution to this literature. We collect data from blood samples to learn about health outcomes.” 2. You do the same but you do it in a different location A very short example: “Young (2017) found that on the job training programs had a 0.12 standard deviation impact on health in the United States. Edwards and Miguel (2012) found a very similar result in Germany. These studies, however, are based on rich, Western context. To date, we know little about the impact of such programs in the Middle East. I study this question in the UAE.” 3. You do the same but different dependent variable A very short example: “Young (2017) found that on the job training programs had a 0.12 standard deviation impact on health in the United States. Edwards and Miguel (2012) found that these programs improved education outcomes in Germany. An important question is: what is the impact of training programs on labor productivity? We are the first to explore this. There are good reasons to do so: A,B,C” 4. You do the same but different independent variable A very short example: “What leads to better job satisfaction among workers? Previous studies have found important factors. Young (2017) found that on the job training programs had a 0.12 standard deviation impact on job satisfaction in the United States. Edwards and Miguel (2012) found that flexible working hours is important in Germany. In this study, I explore a factor that has thus far not yet been explored: the number of foosball tables in the common room. There are good reasons why we think this is important: A,B,C” 8 tips on how (not) to write: 1. Use writing that people understand. No need to write more complicated than necessary. “I have set out to argue” can also be written as “I will argue”. 2. Be academic in your writing. Do not use words like “fascinating”, “amazing”, etc. Thats your opinion. Readers dont care about your opinion. The data will provide evidence. 3. Be precise. Avoid words like “a lot” or “in the beginning” if you can be more precise. 4. Use references! If you use a number somewhere (e.g. in Congo unemployment is 95\%), you need to add a source. 5. How to add references? Do it the same way as I do in this syllabus. In the main text it will be: “Edwards (2015) writes that...” and then has a section at the end with references. ● If there are many authors: “Edwards et al (2015) write that….” ● Do NOT write the full name of the authors, the title of their book or article, etc. Use referencing similar as how it is done in this syllabus. 6. Use the same tense throughout your document. If you start writing in the present tense, do that throughout your document. 7. Do not make value judgements. We are academics. 8. Avoid contractions: write “they’ve” as “they have”, “cant” as “cannot”, etc. Week 5 Qualitative Research (Small-N) Some features of qualitative research: ● ● ● ● ● ● ● The natural setting is a direct source and key element. Context is critical Collection is in the form of words or pict ... Purchase answer to see full attachment