Education

The Balsakhi program had greater effects on students in Vadodara in: a) Year 1 b) Year 2

b) Year 2. The Balsakhi program had a positive effect on students' test scores in Vadodara in both Year 1 and Year 2, but students in Year 2 showed more of a difference in test scores between treatment and control schools.

What percentage of students complete their education all the way through the 12th grade in developing countries? a) Everyone completes 12th grade as mandated by law. b) Over 80% c) 20% or less

c) 20% or less. In many developing countries less than 20% of students complete the 12th grade. Teacher and parents observe this and therefore form strong beliefs that not everyone will complete the 12th grade. Therefore teachers have an incentive to allocate their time spent teaching in the classroom to the "winners" who have the best chance of completing a full education. Thus, we see teachers teaching to the top of the class.

Enter a value. Referring to the bar graph below, the "Uniform" treatment intervention decreased pregnancy rates by how much? (Round to the nearest 0.01).

0.03. To determine the decrease in pregnancy rates for the "Uniforms" treatment group, we look at the control group and subtract the difference between the control group and the group treated with uniforms. The pregnancy rate for the control group is approximately 0.28, while that for the uniform treatment group is approximately 0.25. The difference, 0.28 - 0.25 = 0.03.

True or False: The Millennium Development Goals specified that every child should receive at least a basic education (basic is defined as nine years of education).

True. The Millennium Development Goals (MDGs) include eight international development goals agreed upon by 189 United Nations states in 2000. One of these goals included universal primary education. Nations around the globe agreed to try to achieve these goals by 2015.

In 2011, there were 640 districts in India. In how many of these districts does Pratham work? a) Around 600 b) Around 300 c) Exactly 130 d) Exactly 28

a) Around 600. Pratham works in over 600 districts in all of India. The organization has reached 20 million children in their quest to boost reading and learning outcomes (not just enrollment) throughout the country.

Based on Qian (2008), would we expect the female to male ratio to rise, fall, or remain unchanged in the following scenario? The price of orchard fruits rises and the change is known to be temporary, lasting only one agricultural season. a) Remain unchanged b) Fall c) Rise

a) Remain unchanged. Qian (2008) argues that the gender ratio responds to expected returns. If price changes are only temporary, then long-term returns and gender ratios should be unaffected. You can read the full study by Qian here.

Many of the first rigorous experiments aimed at boosting educational outcomes by increasing or improving ________ have proven unsuccessful. It seems that the key to making progress in educational attainment might instead be with ______ and ______. a) inputs; pedagogy; incentives b) pedagogy; inputs; incentives c) incentives; inputs; pedagogy d) randomized experiments; different research designs; other experimentation methods

a) inputs; pedagogy; incentives. Michael Kremer's textbook study sparked a series of studies investigating the effects of inputs such as more textbooks, teachers, or flipcharts on educational attainment. These studies have been met with disappointing results. However, few studies tested the effect of various pedagogies or incentives on learning, both key elements of education. By pedagogy we refer to the style and method of teaching and learning. Answer A is correct, since inputs have been largely unsuccessful, while pedagogy and incentives might be more fruitful reforms. Answer D is incorrect since it is not the fault of randomized trials that inputs have failed. Randomized trials are just a tool used to evaluate various interventions. This is a bit tricky since Prof. Duflo mentions that the first series of randomized trials implemented in the developing world were met with unsuccessful results because they happened to evaluate inputs.

In India, surveys have found that ____ % of teachers may be absent and _____ % of students may be absent from school on any given day. a) 25; 25 b) 25; 30-50 c) 30-50; 25 d) 30-50; 30-50

b) 25; 30-50. World Bank surveys have found that on any given day, 30-50% of children are absent from school. This trend seems to be true for most children, not just a select few. This implies that it is not 30-50% of students missing school all the time with the remainder attending every day, but rather that each child misses over a 1/3 of the school year. A survey also found that 25% of teachers are absent from school on any given day. That means that even if a child attends school, they may not be learning if their teacher is not present and actively teaching.

Based on studies from Burkina Faso, Colombia, and India, parents will invest in the education of their children _____________. a) Unequally: they will invest more in the oldest child's education b) Unequally: they will invest more in boys' education c) Unequally: they will pick the "smartest," highest-return child and invest more in that child d) Equally: they do not invest different amounts in different children

c) Unequally: they will pick the "smartest," highest-return child and invest more in that child. In some contexts, we observe a tendency among parents to "pick winners" and invest more in the children they believe will benefit the most from education. Pressure to fund only the most gifted child is not necessarily to the detriment of girls: in fact, in a 2012 paper by Robert Jensen (Jensen 2012), labor market opportunities for girls led to parents shifting their limited school budgets over to girls.

The evaluation of the Read India program in Jaunpur, Bihar, and Uttarakhand indicates which of the following? a) Teachers are never effective at boosting student learning. b) Pedagogy matters more than incentives for teaching. c) When teachers and volunteers choose to teach well, they are very effective. d) Balsakhi volunteers are always more effective instructors than traditional teachers.

c) When teachers and volunteers choose to teach well, they are very effective. The results of the experiment showed when teaching to the test and invested in teaching, teachers perform just as well as the Balsakhi volunteers in the studies mentioned earlier in the lecture sequence. Thus, answers D and A are incorrect. That said, without accompanying incentives and when teaching curriculum that is not at the right level, teachers alone have no effect. It seems that both pedagogy and incentives matter, and if anything, incentives matter most. The key takeaway from this study is that we have the "technology" to teach, meaning that when teachers choose to be effective they are. Often systemic failures and broken incentive structures result in instructors choosing not to teach to the best of their ability. Policy makers and governments need to create the right environment which will provide incentives for teachers to teach to the best of their ability. Thus, answer C is correct.

A 2005 ASER study found that ____ % of Indian students between the ages of 7 and 14 could not read a paragraph rated for a first grade level and that ______ % could not read a story rated at a second grade level. a. 45; 55 b. 60; 30 c. 35; 60 d. 70; 40

c. 35; 60. A 2005 ASER survey run by Pratham across India found that a large number of students in primary schools were unable to read even at a very basic level: 35% of students could not read a short paragraph designed for first graders ("standard I level"), and 60% of students could not read a story designed for second graders ("standard II level"). Although primary school enrollment was very high, it became clear that students were not actually learning. Similar results to the ASER survey have been found in a number of other developing countries, indicating that the issue of high enrollment and low achievement is not limited to India alone. Moreover, Pratham has conducted ASER surveys every year in India since 2005, and has observed no improvement in these statistics since that time.

According to this lecture, Pratham serves around ___ children in India. a) 38 thousand b) 20 million c) 3.8 million d) 38 million

d) 38 million. Pratham now serves around 38 million children. To give you a sense of how substantial a population this is, that is equivalent to half the population of France. By many accounts, Pratham is one of the largest NGOs in the world.

In the Kenya experiment by Duflo, Dupas, and Kremer, we see that when you split classes in half by ability, assigning one half of the class (high or low performing) to the original teacher and the other half to the extra teacher: a) Only the bottom half of the original class benefits. b) Only the top half of the original class benefits. c) Only the top half in both of the new cohorts benefits. d) Everyone benefits.

d) Everyone benefits. Duflo, Dupas, Kremer (2011) find that tracking (splitting classrooms by ability) leads to better learning outcomes for everyone. The half of the class that was previously considered the bottom half is now concentrated. Therefore, the top of what was previously the bottom half of the class now becomes the top of the class, so they get significantly more attention and access to effective teaching compared to when they were middle of the back in terms of performance. At the top of the class, reading may not change, but students now benefit from a strong group of peers. Thus, we see that the level at which a teacher teaches is highly important.

The table below shows the effect of the South African pension reform on child height for age. Use the table to answer the question. In the table, girls born 01/92 or later are the "Treatment Group" while those born before 01/92 are the "Control Group." How much did height change between the treatment and control groups for households where a woman received the pension? (Round to the nearest 0.1).

0.9. Column 1 of Panel B reveals that in households in which a woman received a pension, girls born 01/92 or later had a height-for-age measure that was -0.57, while girls born before 01/92 had a height-for-age measure that was -1.47. If we calculate the difference between the two, we find that girls born 01/92 or later have a 0.9 greater height-for-age measure than that of girls born before 01/92.

If 96% of Chinese couples have one child (which is female 50% of the time), and the remaining 4% of Chinese couples engage in sex-selective abortion until they have a male child (also their first and only child), what would be the number of females per 100 males among the children (to the nearest whole number)?

92. The percentage of children born female can be calculated as half of the 96% of non-sex selective births plus 0% of the sex-selective births, in other words, (0.96*0.5) + (0.04*0) = 0.48. So 48% of children born are female, implying that 52% of children are born male. This, in turn, implies that the ratio of females to males would be 48/52, or about 92 females per 100 males.

Using the below diagram and based on what Prof. Duflo said in class, please explain what was an advantage of the design of the Balsakhi evaluation in Vadodara, India? a) It enabled Pratham to be involved with every school, which made data collection easier. b) It allowed researchers to compare the effects of the Balsakhi program on students who were exposed for one year and on students who were exposed for two years. c) It prevented schools from reallocating teachers to other grades after they received a Balsakhi.

A and B are correct. In the original evaluation of Pratham's Balsakhi program, half of the schools surveyed received a Balsakhi for students in third grade, while the other half received a Balsakhi for students in fourth grade. Thus, every school in Vadodara received a Balsakhi from Pratham, allowing the NGO and researchers to be involved with each school. In the second year, each school switched the grade to which the Balsakhi was assigned, creating groups of students who were exposed to the program for zero, one, or two years. A hypothetical disadvantage of the design was that schools could have reallocated teachers to other grades; if this had occurred, researchers would have been unable to disentangle the effects of the Balsakhi on students from the effects of the teacher reallocation. In Vadodara, however, this was not an issue since each school only assigned one teacher per grade as a matter of policy.

If you believe that education is a lottery ticket and that you have to get all the way through high school in order to get richer in the future, parents will discriminate against their kids such that a) They spend money on educating fewer children for a longer period of time. b) They are biased to invest in certain children based on early yet unreliable predictors of educational success. c) They choose to invest in their children randomly. d) They spend money on educating all of their children equally.

A and B are correct. They spend money on educating fewer children for a longer period of time. They are biased to invest in certain children based on early yet unreliable predictors of educational success. They choose to invest in their children randomly. They spend money on educating all of their children equally.

Sen also argues that underdevelopment alone cannot explain the missing women phenomenon. Which of the following claims are used to support this argument? a) Sub-Saharan Africa has more women than men correct b) Cash transfer programs in poor countries have not significantly affected gender ratios. c) Male-to-female gender ratios in India and China have remained the same or become more skewed as these countries' economies have developed. correct d) As countries become wealthier, ultrasound machines become more readily available and this facilitates sex-selective abortion.

A and C are correct. Both A and C are cited by Sen as evidence against the hypothesis that unequal treatment of women is simply a consequence of poverty. B and D are not claims that are discussed in the article.

Which of the following stories could explain this graph? a) When people become richer, they have fewer children b) When people have fewer children, they have more money to invest in each child and those children become more educated and earn more. c) Countries that have policies encouraging low fertility also tend to have high-performing economies. d) Countries in colder regions have lower fertility rates and also have worse-performing economies.

A, B, and C are correct. A, B, and C are all possibilities. If fertility rates and GDP are negatively correlated as shown in the graph, one may have a negative causal impact on the other, and therefore explanations A and B are both possible. Alternatively, a third variable (such as institutional quality) may simply be positively correlated with one and negatively correlated with the other, which is the explanation provided in C. D implies a positive relationship between fertility and GDP, which is not the relationship depicted in the graph.

In his article "More than 100 Million Women are Missing," Amartya Sen argues that East-West differences alone cannot explain the "missing women" phenomenon. His argument is based on which of the following? a. Countries such as Thailand and Indonesia have significantly more women than men. b. Countries such as Japan have seen female-to-male gender ratios increase historically as economic development has occurred. c. South Asian countries have been pioneers in electing female leaders. d. Western countries such as Canada and Mexico have significantly fewer women than men.

A, B, and C are correct. Sen cites each of A, B, and C as evidence that the "missing women" problem is not simply a result of Eastern civilization being more sexist than Western civilization. D is not true (and, in any case, is not mentioned by Sen).

What were some of the issues that came up when implementing the evaluation of the Balsakhi Program? a) Student attendance was low, so test results did not represent the entire student body. b) Teachers cheated, which biased their students' results. c) The test was too hard to detect any differences between treatment and control groups. d) The test was too easy as to detect any differences between treatment and control group.

A, B, and C are correct. Since absenteeism rates are quite high in India, a significant portion of the student body might not be present on test day, meaning that the results of the test would not represent the complete treatment and control groups. Furthermore, since different student do not attend every day (as opposed to a consistent group of non-attending students) there were many students that only had one test score, instead of before and after scores. This made it difficult to measure improvement over time. To address this, the research team found students at home and administered the test at home. Even so, answer A is correct. In addition, when the test was implemented, many of the teachers cheated, largely because the test was too difficult. If the test was too hard, then everyone would get a zero, and it would be impossible to detect differences between treatment and control groups. So answers B and C are also correct. To address this, the researchers made the test easier and had independent personnel from Pratham administer the test. Answer D is tricky since it is also true that if the test is too easy and everyone gets 100%, then it would also be impossible to detect differences. However, in the evaluation of the Balsakhi program, the test was too difficult, not too easy, so answer D is incorrect.

Which of the following does Rukmini Banerji discuss as a key issue in addressing education in India? a) Enrollment in schools in India is very high, but few children are actually learning. b) The Indian educational system has done an excellent job teaching students basic reading skills. c) Teachers often teach the state curriculum without paying attention to the level of their students. d) In order to ensure that students are able to read, teachers must be able to devote time to those students who are falling behind.

A, C, and D are correct. Even though rates of primary school enrollment in India are quite high, research has shown that few children in these schools acquire basic math or literacy skills. In her talk, Rukmini Banerji partly attributes this skill deficit to the fact that teachers in Indian schools teach the state curriculum without regard to individual students' educational levels, and to the fact that teachers do not devote significant time to those students who lag behind their peers. As we will see in the following lecture, Pratham (the NGO which Rukmini Banerji leads with Madhav Chavan) has partnered with researchers from the Abdul Latif Jameel Poverty Lab (J-PAL) to discover the most effective ways of addressing Indian students' poor educational attainment.

The graphs above present the results of the INPRES program. The graphs support the claim that: a) Schools caused an increase in education b) Schools had no effect on wages c) Schools caused an increase in wages d) An increase in wages is likely due to the increase in education

A, C, and D are correct. Indeed these graphs indicated that when a sudden burst of schools were built (62,000) in Indonesia, younger students received more education than older students in regions where more schools were built. Thus, this top-down supply-side intervention had an effect on educational attainment. In addition, we see that the wages increased for the same cohorts who benefitted from this intervention as well, supporting the notion that higher levels of education contributed directly to increased wages.

Primary school enrollment around the world has ____ dramatically between 1999 and 2006. (A: increased, B: decreased, C: stayed the same)

A: increased. Primary school enrollment rates have indeed skyrocketed across the globe. In sub-Saharan Africa, enrollment rates increased from 54% to 70%. In East and Southeast Asia, enrollment increased from 75% to 88%. While universal enrollment has not yet been achieved, there has been a substantial increase in primary school enrollment and an increase in secondary school enrollment, though secondary school enrollment has increased to a lesser extent.

In Indonesia, the INPRES program focused on building tens of thousands of ____. (A: schools, B: oil rigs, C: man-made lakes, D: teacher training programs)

A: schools. From 1974-1978, Indonesia used much of its oil money to build 62,000 schools across Indonesia. This new policy provides a great tool to study the effects of introducing a top-down supply intervention in education, since suddenly the government increased the supply of education dramatically by building tens of thousands of schools.

What are some reasons Balsakhis (young 18-year old women in India) might be a particularly good demographic of teachers? a) They are available, since in most cases they can only volunteer. b) They are young, energetic and motivated. c) They are local and can relate to students in the region. d) They are cheap to hire.

All answers are correct. Balsakhis provide a strong demographic group to perform as teachers for all of the reasons mentioned above. They tend to be available to become extra teachers, since at young age it is considered unattractive to work and parents often do not allow it. However, teaching is considered volunteer work that takes place in a safe environment (a school), so Balsakhis are eager to teach and see teaching as a productive use of their time. Balsakhis are often motivated and energetic. They are young and they have chosen to teach as their occupation, while at the same time they only intend to stay for a short period of time and thus push themselves very hard. In addition Balsakhis tend to be local and therefore know the local dialect and cultural context very well, which enables them to connect and work well with local students. Finally, Balsakhis are inexpensive to hire since they are largely unqualified (in the sense that they don't carry numerous degrees, even if they might be better teachers) and are mostly unmarried (and hence often don't have families which employers know they need to support financially).

Given the examples from Indonesia's INPRES program, and Taiwan's compulsory education policy, what are some impacts of education? (A: Higher divorce rate; lower consumption of food, B: Higher wages; lower infant mortality rates, C: Higher consumption of books; lower consumption of food, D: Greater political activity in more educated areas)

B: Higher wages; lower infant mortality rates. The INPRES and Taiwanese examples allow us to observe comparison groups: areas that had a greater investment in education, and areas that did not. In the INPRES example, we observed increases in the wage rates in areas that had more schools built in the 1970s. In the Taiwanese example, we observed falls in infant mortality rates in education-heavy areas. This implies that there may be benefits, even non-educational benefits, to increased educational investment. These results indicate that despite the various reasons why we cannot interpret the observational correlation between higher education and higher earnings as proof of a causal relationship, we do see a similarly positive causal relationship between these variables even when we employ more rigorous research designs which control for some of the biases present in the graphs presented at the beginning of this lecture series.

Which of these scenarios is inconsistent with a Pareto efficient household? a) A husband and a wife are both shopkeepers and decide to spend money buying inputs in whichever shop is more profitable at the time. b) A husband and wife decide the wife should go back to school so that she can earn more for the household in the future. c) A wife is willing to earn less while working for a particular farmer because that farmer is not friends with her husband and she can privately save some of her earnings. d) A husband prefers to spend more on his own crops because he thinks he knows better what to do with the profits.

C and D are correct. In a Pareto efficient household, members maximize income. Both (a) and (b) are consistent with income maximization, while (c) and (d) suggest households are making choices that reduce total earnings below the possible maximum.

In his book The Elusive Quest for Growth, William Easterly makes an argument that - when comparing the difference in average educational investment and national income, we see no influence. His conclusion: investing in education does not lead to growth, and may be a waste of foreign aid. What is one reason why we may hesitate to reach that conclusion? (A: Education is important in and of itself, B: It's not an issue of investment in education, it's a matter of teacher quality. If the teachers were better, education would work., C: There's no credible counterfactual: we don't know what would have happened in countries that invested a lot in education in the absence of those investments.)

C: There's no credible counterfactual: we don't know what would have happened in countries that invested a lot in education in the absence of those investments. Given that we cannot observe what happened in the absence of that investment (or, for that matter, the absence of foreign aid, or the absence of civil wars in Africa), we cannot distinguish which variable influenced another variable and how. For example, it may have been that, in countries that chose to invest a lot in education, had investment been lower, national incomes would have been even lower.

Does the graph below show the causal effect of years of schooling on economic growth? (A: Yes because there is a strong correlation, B: No because it is a graph using log output per worker, C: Yes because countries with more years of schooling are also richer, D: No because it is not a causal relationship)

D: No because it is not a causal relationship. While it is true that there is a strong correlation between years of schooling and income of the country, it is not clear that there is a causal effect of levels of educational attainment on economic growth. In contrast, the relationship could run the other way, where growth drives education. Richer countries tend to be able to build more schools and employ a more consistent teaching staff, yielding higher educational attainment. In addition, if you anticipate growth, an education is more valuable, so you may invest in additional years of education. There might also be another factor that drives both growth and education levels. For example, a great health care system might result in more students being able to attend school and becoming more productive citizens who can earn more. Thus good health would result in higher education and GDP levels.

Based on Qian (2008), would we expect the female to male ratio to rise, fall, or remain unchanged in the following scenario? The price of tea rises and the change is known to be permanent? (Hint: we assume that females are better at picking tea.) a) Remain unchanged b) Fall c) Rise

Qian (2008) argues that the gender ratio responds to expected returns. Consequently, a permanent increase in tea prices will increase the relative return to females, and thus families will want to have more girls than boys. This would lead to an increase in the female to male ratio.

The Balsakhi program had a greater effect on which group of students? a) Lower-performing students b) Higher-performing students

a) Lower-performing students. Students in the bottom third of their classes showed the greatest improvement in test scores when a Balsakhi was added to their classes. This indicates that the increase in average test scores was due to an increase in the scores of weaker students, those who were sent to see the Balsakhi. Stronger students, on the other hand, showed no significant improvement even though the size of their classes was reduced when weaker students were removed to work with the Balsakhi. This finding suggests that teachers did not take advantage of this reduced class size to teach to the high-achievers, and that the program was effective in reaching lower-achieving students.

What seems to be one of the cheapest ways of improving parents' investment in education? a) Providing information on the returns to schooling. b) Providing free textbooks. c) Providing free schools uniforms. d) Conditional cash transfers to parents (i.e. "For every year your child is in school, I will pay you X...")

a) Providing information on the returns to schooling. As we have seen in this lecture sequence, parents sometimes undervalue certain years of schooling and overvalue others. Specifically, parents often undervalue primary education and overvalue secondary education, meaning they believe there are increasing returns to education. Providing information to parents on the returns to education is a cheap and effective way of getting students in school. In particular, informing parents that sending their children to school for even 3 or 5 years has positive returns challenged the assumption that schooling is an "all or nothing" investment and encouraged parents to send their children to school. Thus, informing parents about the returns to schooling at each level turned out to be a very effective way to boost attendance. After all, talk is cheap - almost free! A cost-effectiveness analysis by the Abdul Latif Jameel Poverty Action Lab (J-PAL) compared a variety of programs and found that information was one of the most cost-effective way to boost investments in education (with deworming a close second).

Based on Qian (2008), would we expect the female to male ratio to rise, fall, or remain unchanged in the following scenario? The price of tea rises and the change is known to be temporary, lasting only one agricultural season. a) Remain unchanged b) Fall c) Rise

a) Remain unchanged. Qian (2008) argues that the gender ratio responds to long-term expected returns. If price changes are only temporary, then long-term returns and gender ratios should be unaffected.

In research about education across countries, researchers have used the number of completed years of education as a measure of basic educational attainment, what are some weaknesses of this measure of educational attainment? a) Years of completed schooling do not necessarily reflect a person's cognitive skills and knowledge. b) Quality of schooling is even across countries.

a) Years of completed schooling do not necessarily reflect a person's cognitive skills and knowledge. Years of education completed may be an imperfect measure for educational attainment for many reasons. Quality of schooling available varies greatly across the globe and within countries. Many schools lack resources or trained teachers, and experience high rates of teacher and student absenteeism. For a given year of enrollment, students in these schools may learn less than their counterparts in well-funded schools with highly qualified teachers. Thus, A is correct while B is incorrect.

Based on Qian (2008), would we expect the female to male ratio to rise, fall, or remain unchanged in the following scenario? The price of orchard fruits rises and the change is known to be permanent? (Hint: we assume that males are better at orchard work.) a) Remain unchanged b) Fall c) Rise

b) Fall. Qian (2008) argues that the gender ratio responds to expected returns. Consequently, a permanent price increase for orchard fruits will reduce the relative returns to girls since boys are more suited to this type of work. As a result, families may want to have more boys than girls, which would decrease the female to male ratio.

In the Balsakhi evaluation, if the following conditions held: Only students who were present in school took the final test, The children who were absent tended to be those who understood the school curriculum the least Absence rates went down in treatment schools (say because students were more motivated) then the actual effect of the Balsakhi program would be ________ the effect observed by researchers. a) Smaller than b) Larger than c) The same as

b) Larger than. Given the high rates of student absenteeism in Indian primary schools, attrition was a major issue in the Balsakhi program evaluation. If weaker students didn't show up for the exams in control schools but were more likely to show up in the treatment school, then control schools would look like they had higher average test scores (since the weaker students had stayed home). Thus, if we compared the group of students present in each school, we would be comparing the strongest students in the control group and both the stronger and weaker students in the treatment group. The observed effect would look relatively small, while the real effect would be larger.

In Hanna and Linden (2009), surprisingly, ______________ teachers discriminated against ____________ students. a) Male; female b) Low-caste; low-caste c) High-caste; low-caste d) High-caste; high-caste

b) Low-caste; low-caste. Hanna and Linden (2009) found that in Udaipur, teachers gave lower grades to low-caste students when they knew they were low-caste compared to when the exams were blind to caste. Surprisingly, this effect was driven by low-caste teachers, suggesting that caste discrimination was at least partially internalized. This study accompanies others that show that beliefs about ones abilities can be self-fulfilling. Another study by Karla Hoff and Priyanka Pandey showed that when low-caste and high-caste children were lumped together without revealing their caste, both groups performed equally well on a cognitive maze activity. However, in another scenario when the groups publicly revealed their caste, the low-caste students did worse, exhibiting self-discrimination as a result of revealed preconceived notions.

After multiple experiments in Kenya, Kremer concludes that providing children with textbooks had which of the following impacts on learning? a) Providing textbooks had no effect. b) Textbooks only had an effect on students with a high enough ability at baseline. c) The sample size was too small to be conclusive. d) The test was too hard to capture the marginal effect of the textbooks provided.

b) Textbooks only had an effect on students with a high enough ability at baseline. Answers C and D are incorrect since they refer to the first cohort of experiments Kremer ran. In this first cohort, the sample size was too small to detect the effect of textbooks on learning. Next, he increased his sample to 100 schools and again found no effect, even in a large sample, perhaps because the test was too difficult for most students to capture any effect. Next, he measured learning using a more level-appropriate test which would pick up textbook-related learning improvements. Again he found no effect. Answers C and D refer to problems he identified with his initial experiments. However, he ran additional experiments to address these issues. Once correcting for these issues, Kremer separates out the effect of textbooks based on students' ability at baseline. He finds that students with higher starting ability did see improvements in learning as measured by test scores, while other students with lower initial performance did not the same improvements. The explanation for this finding is that the textbooks were in English (as was the school curriculum) but only the highest achieving students knew English well enough to make use of the textbooks. Thus answer B is correct, and A is incorrect.

The study by Ashraf, Field and Lee found that: a) Women who experienced reduced pregnancy rates as a result of the experiment continued to do so after the experiment ended. b) Vouchers reduced pregnancy rates more when given to the wife alone. c) Vouchers reduced pregnancy rates more when given to both the wife and husband by providing new information to the husband. d) Vouchers for family planning appointments were equally effective when the wife alone was given the voucher and when the wife and husband were given the voucher together.

b) Vouchers reduced pregnancy rates more when given to the wife alone. The importance of private information (whether the wife alone received the voucher) provides evidence against the unitary household hypothesis. Indeed, voucher use rates were higher when the women alone received the voucher, and negative impacts on unwanted pregnancies (for women receiving the voucher alone relative to women receiving the voucher with husband present) proved only temporary. You can read the full study here.

Fill in the blanks: The following graph implies that the "Teacher Training" intervention ________ pregnancy rates. a) Decreased b) Increased

b) increased. By comparing the pregnancy rates of those in the "Teacher Training" group to those in the control, we see that the pregnancy rates of the "Teaching Training" group is higher than those in the control. This suggests that the "Teaching Training" intervention increased pregnancy rates.

Intention To Treat (ITT) estimates will tend to ____ the actual effect of the Balsakhi intervention compared to measuring only the effect of schools that participated. a) overestimate b) underestimate c) perfectly estimate

b) underestimate. ITT estimates tend to underestimate the effect of an intervention. This is because they measure the effect of offering a program, not the effect of the program itself. ITT estimates treat groups that drop out or opt out from the program as if they participated. Since groups that attrite are generally lower performing and less motivated, including these groups in a sample most likely brings down average outcomes. In contrast, excluding them from the sample would bring the average effect up, since the measurements would reflect the higher-achieving or more motivated groups that self-selected into the program. Thus, by including those who attrite in the treated sample through ITT estimates, estimates often understate the impact of a program, making them more conservative. Researchers employ the ITT approach since if an effective is found even using this conservative strategy, then the true effect must be even larger, which lends further supports that a program was effective.

____ of students who were offered participation in the Read India Program and could not initially read actually attended. Out of these students ____ of students were able to read basic letters by the end of the Read India Program. a) 100%; 100% b) 100%; 13% c) 13%; 100% d) A majority; 100%

c) 13%; 100%. While the initial effects of the Read India Program seemed only slightly higher than the control group average (about 0.1 of a standard deviation higher), this result was in fact very significant. In large part this is because the 0.1 effect was an ITT estimate, which meant that students who did not participate in the program but were offered it were effectively counted as treated. However, only 13% of students ended up participating, such that the 0.1 standard deviation effect was very conservative; the group of students that did participate in the program reached basic letter-reading proficiency as a result of the program. Answer (c) is correct. Answer (d) is a bit tricky, but as mentioned above, a minority not a majority of students participated in the Read India Program. Given its effectiveness, perhaps a majority should have participated!

Based on the Deaton and Subramanian (1996) approach, which of the following statements should be true if girls are given less to eat than boys? a) Consumption of adult goods falls the same after the birth of a female child as after the birth of a male child. b) Consumption of adult goods falls more after the birth of a female child than after the birth of a male child. c) Consumption of adult goods falls more after the birth of a male child than after the birth of a female child. d) Consumption of adult goods rises after the birth of a child

c) Consumption of adult goods falls more after the birth of a male child than after the birth of a female child. Deaton and Subramanian (1996) argue that spending on adult goods should fall more after the birth of a boy than after the birth of a girl if boys are fed more than girls. This is based on the idea that adults pay for food for children by reducing spending on adult goods. Given this, "Consumption of adult goods falls more after the birth of a male child than after the birth of a female child." is the correct answer. You can read the full paper here.

The table below shows the effect of the South African pension reform on child height for age. Use the table to fill in the blanks. In the table, girls born 01/92 or later are the "Treatment Group" while those born before 01/92 are the "Control Group." The difference between these groups implies that women receiving pensions ________ height for girls. a) Had no effect on b) Decreased c) Increased

c) Increased. Column 1 of Panel B reveals that, in households in which a woman received a pension, girls born 01/92 or later had a height-for-age measure that was -0.57, while girls born before 01/92 had a height-for-age measure that was -1.47. If we calculate the difference between the two, we find that girls born 01/92 or later have a 0.9 greater height-for-age measure than that of girls born before 01/92. Therefore, this implies that women receiving pensions led to an increase in height for girls.

What was one of the surprising findings about returns to education in the USA and Indonesia? a) The returns to education were negative in the US due to high tuition costs. b) The returns to education were higher in Indonesia than in the US. c) The returns to education were roughly the same in both countries. d) The returns to education were higher in the US than in Indonesia.

c) The returns to education were roughly the same in both countries. We would expect the returns to education in these two contexts to be quite different given the different labor force needs and the different educational systems. However, studies from both countries indicate that the returns are quite similar: a proportional increase in wages of about 7%-10% for every completed year of education.

How did parents in Madagascar perceive returns to schooling? a) The first years of school are really important - but once the fundamentals (reading, writing, basic math) are covered, the per-year value falls rapidly. b) Each year of education counts more or less equally. c) The returns to grade school are low, but graduating from high school is key.

c) The returns to grade school are low, but graduating from high school is key. When parents believe in "all or nothing" education - that is, it's either a high school diploma or nothing at all - they tend to undervalue the earlier years of education. This may lead to underinvestment, where parents pull children out of school earlier than necessary, thinking that the early years will have little benefit if their child will not reach the top level of education. Actual returns, however, appear fairly consistent each year: each additional year counts for the same proportional boost in future income. In fact, it is easy to see why just 3 years of education might be useful. For example, 3 years of education may enable a student to learn to read. If reading is useful for understanding written instructions that can be used in production, for example reading fertilizer instructions, then the returns to those first three years of education might be quite high.

The Balsakhi intervention _____ average test scores. Moreover, it ______ test scores for the lowest achieving group of students (who were also the most likely to participate in the Balsakhi program). a) increased; decreased b) decreased; increased c) increased; increased d) had no effect on; increased

c) increased; increased. The Balsakhi program increased average test scores by over 0.2 of a standard deviation in the second year. This effect is quite large given the current literature on education. We might be concerned that the program was only effective because the Balsakhis took out the most disruptive students from class. According to this line of thinking, it could be that the highest achieving students benefitted while the lower-achieving students still learned nothing. However, the results show that the lowest-achieving students learned the most (an effect of 0.5 of a standard deviation). This indicates that the Balsakhi program might indeed be a great investment to boost academic outcomes for the students who need it most. Furthermore, the effect of 0.5 of a standard deviation which measures ITT estimates for a class assigned a Balsakhi, was driven by just 20% of students, those same 20% who worked with the Balsakhi. The students who remained in the classroom did not benefit from a smaller less-disruptive classroom. This provides support that Balsakhis are especially effective instructors and that teachers did not take advantage of a smaller more homogenous classroom to teach to the highest-achieving students. To understand why we analyze the impact of education in standard deviations, refer to the Supplementary Stats and Econ section of the course.

We see a vicious cycle of false beliefs emerging when teachers teach to the ___ half of the class. When this occurs, students in the _____ half of the class get discouraged, stop trying, receive less attention, and ultimately perform worse. Teachers then feel justified in their original beliefs, even though the low-performing children may be performing worse partially as a result of that original belief. Accordingly, teachers continue to teach to the ____ half of the class. a) bottom; bottom; top b) bottom; top; bottom c) top; top; bottom d) top; bottom; top

d) top; bottom; top. While it is possible that the bottom half of the class is indeed less likely to complete 12 grades of education, teachers' beliefs that (1) these students must complete 12 grades in order to get anything out of an education, and (2) are least likely to finish their education, means that teachers devote fewer resources to them, which causes them to perform more poorly. This in turn justifies teachers' initial decisions since they see certain students performing poorly as they had anticipated. For that reason, the teacher may then continue to devote resources and energy to teaching the top of the class. This generates a vicious cycle based on potentially false beliefs which penalizes the students who need and education most. Indeed, the Pratham experiment illuminates the dangers of these false beliefs, since we saw the same teachers who produce no improvements during school, do in fact boost student outcomes dramatically when teaching over the summer. Since these outcomes come from the same teacher, it cannot be a reflection that the teacher is ineffective. It could be that during the school year, the teacher is trained to teach to the top of the class, while during the summer the teacher intends to focus on remedial students (the bottom half) and therefore has a different set of goals and beliefs for those students.