Variable Average for treatment group Average for control group Difference t statistic for difference Read News 0.71 0.73 -0.02 0.10 Political Interest 2.59 2.50 -0.09 1.22 Education 16.21 16.00 0.21 0.88 Household Income (in thousands of) 20.20 20.09 0.11 0.40 Female 0.54 0.53 0.01 0.12
Problem 3-
ABC Challenges: Attrition, Balance and Compliance
Can television inform people about public affairs? Political scientists Bethany Albertson and Adria Lawrence (2009) conducted an experiment in which they randomly assigned people to treatment and control groups to evaluate the effect of watching TV on a person’s information level. Those assigned to the treatment group were told to watch a specific television broadcast and were later asked questions related to what they watched. Those in the control
group were not shown the TV broadcast but were asked questions related to the material in the TV broadcast.
The dataset contains the following variables:
: Dummy variable which =1 if a person reads news and 0 otherwise.
: interest in political affairs (not interested=1 to very interested=4)
: years of education
: female dummy variable (female=1; male=0)
: family income in thousands of dollars
: information level (low information level=1 to high information level=4)
=1 if the individual is assigned to watch the specific television broadcast
=0 if the individual is not assigned to watch the specific television broadcast
a. You test for balance using the variables mentioned in the table below. Based on the results do you think that the treatment and control group are balanced? If your answer is “yes” then explain why. If your answer is “no”, then explain why and mention how will you address the issue of imbalance.
For parts (b) and (c) suppose that the people in the control group who have political interest
decide to watch the TV broadcast.
b. Is this an issue of attrition or non-compliance?
c. Does still reflect the unbiased causal effect of watching the TV broadcast on Information level? If your answer is “yes” then explain why. If your answer is “no”, then explain how you will interpret now.
d. Mention any two limitations of experiments.
AI-generated content may present inaccurate or offensive content that does not represent bartleby’s views.
Unlock instant AI solutions
Tap the button
to generate a solution
