I'm having a bit of trouble understanding exactly why there is a "1" in the general simple Binary Choice Model where $Y_i$ can take a value of either $0$ or $1$. We also assume that the conditional distribution of $\epsilon$ given $X$ is a standard normal distribution.

For example, in my notes, our econometrics professor wrote the following:

$$E[Y_i|X_i=t]=P[B_0+B_1X_i\geq\epsilon_i|X_i=t]$$ and the next step was: $$E[Y_i|X_i=t]=E[1(B_0+B_1X_i\geq\epsilon_i|X_i=t]$$

My question is about this $1$ that keeps popping up. What is the significance of this? Also How can we go from the line above to the line below?

  • 9
    $\begingroup$ Although your expressions are syntactically incorrect (probably just some typographical errors in the markup), it's likely the "$1$" refers to the indicator function. $\endgroup$ – whuber Apr 14 '15 at 23:11
  • $\begingroup$ I see. But can you please explain how to go from one line to the other? Thank you. $\endgroup$ – nicefella Apr 15 '15 at 20:30
  • 1
    $\begingroup$ By definition, the expectation is a sum of values times probabilities. When the values are only $1$ and $0$, multiplication by $0$ introduces nothing while multiplication by $1$ leaves just the probabilities. $\endgroup$ – whuber Apr 15 '15 at 20:33

As @whuber commented, the most plausible guess is that your $1$ refers to the indicator function.

You go from one line to the other by noting that the indicator function can only take two values: $1$ if the event it indicates is true and $0$ when it is false, so (again, just following @whuber's comment)

$E[I_{\{B_0+B_1X_i \geq\epsilon_i\}}|X_i=t] = $, by expectation definition for a discrete variable,

$\mathbb{P}(I_{\{B_0+B_1X_i \geq\epsilon_i\}}=1|X_i=t).1 + \mathbb{P}(I_{\{B_0+B_1X_i \geq\epsilon_i\}}=0|X_i=t).0 = $

$\mathbb{P}(I_{\{B_0+B_1X_i \geq\epsilon_i\}}=1|X_i=t)$

And since the events $\{I_{\{B_0+B_1X_i \geq\epsilon_i\}}=1\}$ and $\{B_0+B_1X_i \geq\epsilon_i\}$ are the same event (the indicator equals one if and only if the event on the right occurs), you have that:

$\mathbb{P}(I_{\{B_0+B_1X_i \geq\epsilon_i\}}=1|X_i=t) =$ $\mathbb{P}(B_0+B_1X_i \geq\epsilon_i|X_i=t)$

And the equality you were looking for.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.