Does thinking about the lockdown affect perseverance and cognitive skills? Evidence from El Salvador

2020-11-09

2020-11-29

Cognitive skills and perseverance

The first primary outcome is an index of cognitive function constructed through the inverse covariance weighting method in Anderson 2008. The index is based on the scores of the two cognitive tests administered: the Raven Matrices (10 trials) and the Spatial Stroop Test (20 trials). The score on each test is computed by summing up the number of correct trials.

The second primary outcome was measured only for children. Following the two tests, children were asked if they would prefer to complete another easy or difficult Raven trial, thus the secondary outcome is a dummy variable. The task was incentivized, solving correctly the easy/hard trial implies winning 1/3 prizes.

Positive and affective states, time on task, endurance

Positive and affective states - these variables are the manipulation checks and are measured just after the priming. Four positive and four negative affective states are listed and respondents report to what extent they feel that at the moment (from 1 “Not at al”l to 5 “Extremely”). The final scores are computed by summing up the items on the respective subscales.
Time on task is a proxy of noncognitive skills derived from task behavior. We will investigate treatment effects on the response time for each individual trial in the two tests. The time on task in the Raven test will represent a proxy of effort and motivation, while for the Stroop, we will investigate if response times on the trials (mean and median) are in line with the main treatment effects (e.g. higher stroop score and faster response time)
Endurace is another proxy of noncognitive skills derived from task behavior. Given the randomized order of the test, we will investigate if treatment effects are larger for the cognitive test administered second. In addition, for each test, we will explore if treatment effects increase in the second half of the trails (last 5 for Raven, last 10 for Stroop).

The survey experiment implies that respondents are randomly allocated to one of the following two primes at the beginning of the survey. In particular the following

Lockdown scenario: “I will ask you about your life next year. Please, imagine that at the beginning of the school year (on Feb) schools remain closed and children will study from home. Similarly, many shops will remain closed and people will need to stay home and will be allowed to get out only for urgent matters.
No Lockdown scenario: “I will ask you about your life next year. Please, imagine that at the beginning of the school year (on Feb) schools are open and you/your child will study in school. Similarly, many shops will be open and people will be allowed to get out and do the usual activities

In both treatment arms some follow-up questions characterizing the life in that scenarios are asked:

What would you do on a normal day in this situation? Please, list three main activities
Could you do everything you would normally like to do?
How do you imagine a schooling day under this situation? How are you/is your child going to learn new things?
Do you think learning new things is going to be harder or easier in this situation?
Will this affect your family situation? Please, list three things that are likely to change in good or bad
Do you think that the family situation will improve, worsen or remain the same?

After the randomized prime, individuals are first inquired about their affective states in that moment. This is used as a manipulation check for the prime. Second, individuals are administered two cognitive tests (Raven and Stroop) and one for measuring perseverance.

The randomization is carried out in the office by a computer. Each household is assigned to one experimental arm.

The unit of randomization is the household. The same experimental treatment applies to both parent and child belonging to the same household.

Yes

We plan to visit about 220 households.

Surveys are administered to one child and to one parent in each selected household. This leads to a potential total number of observations of 440.

Half of the sample is assigned to each of the two treatment arms. This means that we expect 110 households (220 individuals) per treatment arm.

The outcome variable for cognitive function is standardized using the control group mean and standard deviation. Assuming an intra-cluster correlation of 0.1, power equal to 0.8 and significance level of 0.05, 220 clusters with 2 individuals per cluster, the MDE is 0.28. As for perseverance, which is only calculated for children, the design implies individual level randomization with 220 individuals. Assuming power equal to 0.8 and significance level of 0.05, the MDE, considering the standardized index calculated as above, is 0.38.