Data Presentations and Research Proposal For Your Research

Data, Presentations and Research Proposal

For Your Research n n This about something you are interested in, that might have data, and for which you can come up with a good research design. If can’t do an RCT – which most of you won’t § Pick an outcome that is important and can’t do with RCT § Mortality – unless mortality rates are high will never have enough sample with RCT § Work on questions where randomization difficult § But make sure still have some good identification. § Look at effects on really big an important policies where you want the real world estimate § Point out why RCTs may not scale up etc. § Highlight the external validity (and internal) of your results. § Historical or longer-term results can’t get with RCT 2

Back and forth process n Choosing topic is a back and forth with question and data. § Think of an area interested in § § Read review articles on area, or articles in general Read handbook chapters Look for data sources used in other articles Need to find a question others have not looked at before or that you can do in a new better way, or different outcomes. § I DON’T want replications. § Find interesting data set § Think about would could be answered § Research what has been answered n It is key to have a good research design 3

Data n Surveys § Living Standards and Measurement Surveys (LSMS) § http: //econ. worldbank. org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/EXTLSMS/0, , content. MDK: 215888 00~menu. PK: 4196952~page. PK: 64168445~pi. PK: 64168309~the. Site. PK: 3358997, 00. html § Household and facility/community data § Books/Articles on the LSMS design n Rand Family Life Surveys http: //www. rand. org/labor/FLS. html § Malaysia, Indonesia, Guatemala, Bangladesh § Mexico § : http: //www. ennvih-mxfls. org/en/mxfls. php? seccion=1&subseccion=1&session= 4

Data n Government § Statistical Bureau Websites § Mexico (INEGI): http: //www. inegi. org. mx/ § Municipality data are public § Contact about data at lower levels, or older § Other Ministries § Population data on CONAPO (Nat’l Pop Council) n n UN - WHO, UN statistics – tend to be country level World Bank Development Indicators § Country level n n IPUMS international: https: //international. ipums. org/international/ Satellite Data: deforestation, light by night 5

Pitch a Paper Presentation n n n Maximum 10 minute presentation Find a data set that you could use to answer a question Slide 1: Motivate the choice of question/data set Slide 2: What is the research question Slide 3: Describe the dataset Slide 4: Empirical Strategy: Specify the equation and discuss what variation you will use in the data to answer the question. Slide 5: Limitation: discuss what biases or confounding factors you are worried about. 6

Research Proposal n Write a proposal for research you would like to do § Maximum of 8 single spaced pages – can be shorter § Longer if are including result tables – but these are not necessary n n Data analysis is not required but is preferred if you do some first runs. Have to describe the kind of data you would need, and what variation you would need in the data. § Must be realistic n n Can be based on your pitch a paper idea but does not have to be. Purpose is to: § Have you practice thinking about a researchable question. 7 § Practice writing journal style papers

Sections of Research Proposal n n These are suggested sections, depending on your proposal it may differ Section 1: Introduction § § § n n Mimic actual introductions in journals (get some examples) Motivate your research topic Write a clear statement of what your research question is Say what you do (even if it is describing fake data) Discuss the contribution it the paper makes to the literature Section 2: Background Section 3: Data § If don’t have data describe what the data would need to be § Unit of analysis, variation you are taking advantage of. 8

Sections of Research Proposal Continued n Section 4: Empirical Strategy § Specify the equation and the research strategy § Single of double difference, IV, regression discontinuity etc. § Be clear what variation in the data you are using to identify the coefficients (i. e. get a causal estimate) § Make clear any threats to the validity of the research design § What do you worry about that may lead the coefficients to not be causal § Selection bias, reverse causality, attrition, confounding programs etc. n Section 5: Results § Really don’t need any, but if you have them then report them. 9