Education and Economics Development

EDUCATION AND ECONOMICS DEVELOPMENT
Learner's Name
Institution's name
Course Code and Name
Instructor Name
Date
(10 points) Multiple choice questions
1 You cannot apply a regression discontinuity method if
1 The discontinuity in the running variable (say test scores, etc.) at the cutoff is fuzzy instead of being strict
2 You do not know the exact location of the cutoff
3 Other covariates (say gender, race, poverty, etc.) are discontinuous at the cutoff.
4 Any of the above
2 The main reasons why Muralidharan and Sundaraman (Aggregate Effect of School Choice) conduct a 2-stage experiment (randomization among villages followed by randomization at the student level) are to
5 Ensure that the RCT involve as many students as possible
6 Get the most significant possible treatment group in terms of number of students, and increase the statistical power
7 Avoid the problem of a contaminated control group and estimate spillover effects
8 Serve as a template for national-level implementation
3 To conduct a falsification test, Barrera-Osorio and Raju (Evaluating the impact of public student subsidies)
9 use raw student test score data to calculate their own school SLQAT pass rates
10 estimate RD impact estimates at two arbitrarily-selected false cutoffs: 57 per cent and 77 per cent
create two alternative bandwidths around the optimal bandwidth, at 75 per cent and 125 per cent of the optimal bandwidth
administer a field survey to schools with SLQAT pass rates between +/- 15 percentage points of the 67 percent cut-off
4 Which of the following are caveats that Kazianga et al. (The Effects of "Girl-Friendly" Schools) mention with respect to calculating the cost-effectiveness of the BRIGHT school construction program in Burkina Faso?
Many of the interventions included in the BRIGHT schools could have affected outcomes other than enrollment and test scores.
Though the authors have detailed cost information on the BRIGHT schools, their estimates of the cost of the government schools are more tentative.
They estimate the cost-effectiveness of having a BRIGHT school and then add to it the costs that would be incurred in having the unique BRIGHT amenities in a planned government school.
All of the above.
5 Which of the following is not true of meta-analysis?
It converts the results of all the included studies to a common metric – then estimates the average effect of different types of intervention
By pooling across multiple studies, it provides a stronger statistical power compared to what can be achieved by individual studies
It allows the researcher to control for the quality of studies
It allows the researcher to include all studies of high quality, regardless of the exact variables and data examined and reported
Short essay questions
(25 points) The Basics of Program Evaluation
1 (2 points) What is the problem of causal inference?
Casual inference tends to assume that the effect of treatment assignment is constant in all individuals being examined and, therefore, the impossibility of observing the causal impact on a single unit.
2 (2 points) What is selection bias?
Selection bias is when you don't randomly select data for statistical analysis. As a researcher, you fail to choose specific data because of a particular attribute you think is not significant in your research.
3 (2 points) How do RCTs solve the problem of selection bias?
RCTs randomly assign participants to groups. By randomly selecting the participants, there is no interference with the outcome; therefore, the result achieved shall be considered much more credible and reliable CITATION Ban07 \l 1033 (Banerjee, Cole, Duflo, & Linden, 2007).
4 (4 points) What are some of the other strategies you can use to solve the problem of selection bias?
Multiple ways can be used to solve the problem of selection bias. One is allowing equal opportunity for the targeted information/people to be of the sample group. Second, avoid making assumptions about the response you are likely to get from a specific sample group compared to the other. Third, widen your sample group search and ensure it is as random as possible.
5 (4 points) Briefly describe the pros and cons of RCTs. Also, discuss the qualifications/caveats to RCTs on education in developing countries that Kremer and Holla point out in their literature review.
Some of the significant pros of RCTs are that there is a clear causal linkage between the interventions and outcomes and that in RCTs, you can as the question you want. The disadvantages include that it is resource intensive, time-consuming, and restricts the range of questions. In their literature review, Kremer and Holla point out that RCTs are in resource demand which will broadly impact how the RCTs are implementedCITATION Kre09 \l 1033 (Kremer, Michael, & Holla, 2009).
6 (3 points) Mention one advantage of RCTs over regression discontinuity and one disadvantage.
RCT overwhelms RD because it does not depend on strict cutoffs determining the assignment. On the other hand, the RD has an advantage over RCT because RD doesn't require an elaborate experiment, which is expensive compared to RCT.
7 (4 points) Recall the Muralidharan and Prakash paper. Assume bicycles were only given to households whose income fell below a threshold, say, x dollars per month. What strategy could you use here instead of the difference-in-differences method used by Muralidharan and Prakash? What assumptions would you need for a causal inference? Would this necessarily be an improvement?
In the case of Muralidharan and Prakash, I would use RCTs. The assumption I would make for the causal inference is that there will be no bias in the selection of the sample group. This won't necessarily be an improvement, although it might be a better strategy depending on how unbiased the selection is.
8 (4 points) Compare the Muralidharan and Prakash paper with the more recent Prakash et al. paper "Wheels of Change". What are the outcome variables in each case? Give one reason why you might prefer the evidence from Bihar, India – and give one reason why you might choose the evidence from Zambia.
In India, Prakash used the RCTs strategy in selecting the sample group, while in Zambia, the section used was different-in-differences—the differences in age in both cases. In India, the issue is for girls aged 1-17, while in Zambia, it is for girls aged 14-15. Evidence from RCTs, which is from Prakash in India, is not biased, making it dependable on the research. The difference-in-differences strategy used in Zambia, as indicated on Wheels of Change, is more accurate as the selection is made based on the data of the girls on the distance from the home to school CITATION Ban20 \l 1033 (Banerjee A. V., 2020).
(14 points) Examining External Validity (Generalizability)
1 (2 points) What is internal validity?
Internal validity is defined s the level of truthfulness likely to be obtained from the population being studied without the impact of the methodologies used.
2 (2 points) What is external validity?
External validity refers to the extent to which you can use the results obtained in a study for another situation or measure. It can also be referred to as the broader applicability of research results.
3 (4 points) Mention 2 program evaluation methods that are (relatively speaking) strong on internal validity but weak on external validity, and 2 program evaluation methods which are (relatively speaking) weak on internal validity but strong on external validity.
Experiments and math proofs are considered to be strong on internal validity but weak on external validity evaluation methods. On the other side Case Studies and field, Surveys are considered to be strong on external validity while weak on internal validity.
4 (3 points) Briefly discuss how external validity is a big concern for studies employing regression discontinuity.
Studies employing regression discontinuity inferences and conclusions are said to be valid for the study's population and thus can be generalized to fit the entire setting or the sample group. Additionally, selection bias, experiment effect, testing effect, and situation effect impact the study significantly.
5 (3 points) Discuss how the Lucas & Mbiti paper on elite secondary schools in Kenya use district-specific admissions thresholds to improve the external validity of their findings.
Each national school has multiple district-specific quotas that enhance and promote unity and address concerns over regional inequality. Therefore teachers target instruction to students near the medi...