Secondary Outcomes (explanation)
We conduct four different cognition tests:
1. Digit span forward:
The forward digit-span assesses short-term auditory memory and simple verbal expression53. It is the only test that does not directly involve executive functions. The child is asked to repeat a series of numbers immediately after the enumerator red the series out. The number of digits is continuously increased and the longest list of numbers the child could remember was defined as the digit-span. The test is the same as in Malin’s Intelligence Scale for Indian Children.
2. Digit span backward:
The backward digit-span test measures the ability to store, use and manipulate new information. Backward digit-span also involves attention, impulse control and shifting from a forward to backward sequence. All of these abilities are commonly considered to be a part of the group of executive functions53–55. The administration of the backward digit-span test is the same as in forward digit-span test; however, the child is requested to repeat the digit sequence in its reverse order. The test is the same as in Malin’s Intelligence Scale for Indian Children.
3. Block Design:
The Block design test assesses planning and organizing56. In this test, children are asked to arrange red and white colored blocks in a way that they match a pattern of a picture. Children receive two points if they correctly arranged the blocks on the first try, one point, if they correctly arranged the blocks on the second try, after the test administrator had shown the correct solution to the child, and zero otherwise. They were asked to arrange four different pictures in the baseline survey that increased in difficulty. To account for a general increase in cognitive ability at the endline survey, two more pictures were added. The test is the same version as in Malin’s Intelligence Scale for Indian Children.
4. Colored Progressive Matrices
We use an abbreviated version of Raven’s Colored Progressive Matrices (RCPM)57 that measures abstract reasoning and the capacity to simultaneously solve several problems involving new information58. There is some evidence that the RCPM test measures executive functions59,60.18 We use six matrices from set A and six matrices from set AB. In this test, the child is shown an array of pictures with one missing box. Out of the six options, they are to select the picture that fits the missing box. The pictures progressively increase in complexity and abstraction. We score each correct answer with one point, hence the scale for RCPM ranges from 0 to 12. The RCPM are designed for children between 5 and 11 years old.
5. Stroop test:
The Stroop-like day-and-night test (Gerstadt et al. 1994) assesses the ability of inhibition (suppressing a habitual response), which is also considered to be a classical executive function (Anderson 2001, Carlson 2005). We used an extended version of this test, where six pairs of cards, which show pictures of opposites (day–night, boy–girl, large–small, up–down, warm-cold and young-old), were presented to the child. After shuffling the pictures, they were presented to the child one after the other and the child was asked to say the opposite of what they were seeing on the card. The scale for the day-and-night test ranges from 0 to 12. Initial errors that were self-corrected by the child were scored as a half point. Apart from inhibition, this test also requires memorizing two rules simultaneously. First, what the picture on the cards represent and second, to always say the opposite.
The Stroop like “Day-Night” test by Gerstadt et al. (1994) used in the first two survey waves to measure the ability of inhibition control is no longer appropriate for the new age group of the children in the endline. That is why, we use an adapted version of the fruit/vegetable Stroop by Röthlisberger et al. (2010). This test has been conducted with the help of the Institute for Psychology, University of Hildesheim.
For educational outcomes we measured scores in a reading and math test and attendance. For examining the reading and math skills the Annual Status of Education Report [ASER] reading assessment tool and the ASER Math tool are used that are developed by the Indian Non-Governmental Organization Pratham (ASER Centre, 2014). However, the conduction of the test has been modified to accommodate the setting.
School attendance is recorded from the official record books of the schools. The focus lies in particular on the past 12 months before the survey. The official records of many schools go only back for 12 months or to the start of the school year. The baseline includes attendance data for the time between November 2013 and October 2014, the midline for time between August 2015 and July 2016, and the endline for the time between January and December 2018. We calculate school attendance as the rate of the total number of days present of a child and the total number of days school is open for each time period.
For the analysis all secondary outcomes are normalized.