Hi is there a way anyone could help me understand what control variables are in

Question

Hi is there a way anyone could help me understand what control variables are in a regression. In the following regression the independent variable is hours spent sleeping in the 24-h period from 4 AM to 4 AM. I I also do not understand what is the purpose of a benchmark for marital status, weekdays, days, no degree and white. The bottom of table 2 states that the control variables are sex, age, marital status, child <5 in the household, day of the week, education, race, year dummies and state dummies. Full sample estimations also include a dummy for salaried workers. Married was the benchmark for marital status, weekdays (Monday through Thursday) was the benchmark for days of the week, no degree was the benchmark for education and white was the benchmark for race. Thanks!

The estimated model is Si-Ao + A In (wages), + AUR, + 3Xi + 4 State, + Year, + where S, denotes sleep duration of individual i, we take the logarithm of individual wages to account for diminishing marginal returns of income, UR is the unemployment rate in the state where subject i lives and X, represents age and the squared term of age, gender marital status, indicator for the presence of a child under the age of five in the household education, race, and an indicator for days of the week. Finally, ei is the error term. For our main estimations, we divided subjects into two wage groups; subjects who received a fixed salary and those who received hourly wages. We also ran estimations for both wage groups combined. In those analyzes, the variable X additionally included an indicator of whether the subject earns hourly wages or not. This is to control for a potential measurement-error bias that stems from the fact that while information on wages for those who earn hourly wages was readily available in the dataset we needed to construct the wages per hour variable for those who earn a fixed salary. Despite this attempt, it should be kept in mind that the assumptions for a dummy variable to fix the measurement-error bias are strong and probably unrealistic. However, we do not expect this bias to critically harm our analysis.

Explanation / Answer

Benchmark and Control Variables

Control Variables

Control variables in regression analysis are the ones in a model that when kept at constant value give the value for change in the dependent variable for unit change in the variable not being kept at a constant value.

In the given model, Xi represents the control variables sex, age, marital status, day of the week, education, race, children less than 5 years of age, year dummies, and, state dummies.

In this model, if we need to know the impact of unit change in one variable; the other variables are considered constant. The coefficient beta values represent the change in the dependent variable with unit change in a particular variable, for instance, state, when all other variables are considered to be having constant values.

Benchmark

In the model for sleep hours, Married is the benchmark for marital status, weekdays (Monday through Thursday) is the benchmark for days of the week, no degree is the benchmark for education and white is the benchmark for race.

These values as the benchmark for marital status, weekdays, education, and, race represent that the sleep duration value derived will be that for weekdays for a white married man with no degree. These values represent the base group. Changing the values for these variables will bring about a change in the value of the dependent variable for a change in the benchmark value for that particular variable under consideration.

Related Questions

Navigate

• Browse (All)
• Browse H
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.