In the context of estimating parameters for a uniquely distributed set of independent and identically distributed random variables, I am examining the following probability density function $ f(x|\theta) = \frac{3\theta^3}{x^4} \mathbf{1}_{[\theta,\infty)}(x) $ with $ \theta > 0 $.

(a) For the estimation, I am tasked to derive the maximum likelihood estimator (MLE) for $ \theta $ based on a sample $ X_1, \ldots, X_n $. I have started by setting up the likelihood function and taking the derivative with respect to $ \theta $, but I always get that $\hat{\theta} = 0$ because:

$$\frac{\partial l(\theta \mid X)}{\partial \theta} = \frac{3n}{\theta}$$

Is this correct?

(b) After deriving the MLE, I need to determine its distribution and whether the estimator is unbiased for $ \theta $. I am aware that the unbiasedness of an estimator is linked to its expected value equalling the parameter. However, I am struggling to find the distribution of the MLE, how do I do this? Is there any theorem for the distribution of the MLE or there is a hidden "hint" in the question?

  • 2
    $\begingroup$ You probably meant the log-likelihood $\endgroup$
    – Firebug
    Dec 1, 2023 at 12:54
  • $\begingroup$ Please provide the expression you derived for the (log-)likelihood $\endgroup$
    – Firebug
    Dec 1, 2023 at 13:12
  • 1
    $\begingroup$ of course! The log-likelihood that i've found is: $$l (\theta \mid X) = n \log{3} + 3n \log{\theta} - 4\log {\sum_{i = 1}^{n} X_i}$$ But i fell like im missing something, in this approach i completely ignored $\mathbf{1}_{[\theta,\infty)}(x)$ that is just a long way to say that the density is defined at the points where $x \ge \theta$ $\endgroup$
    – Occhima
    Dec 1, 2023 at 13:18
  • 3
    $\begingroup$ You shouldn't ignore that term $\endgroup$
    – Firebug
    Dec 1, 2023 at 13:43
  • $\begingroup$ Best to edit the new information into the question as comments are not permanent. $\endgroup$
    – mdewey
    Dec 1, 2023 at 15:53

1 Answer 1


As said in the comments, you cannot ignore the indicator functions.

Note that, for $\theta \in \mathbb R_{> 0}, n \in \mathbb N_{\geq 1}$, we have $$ \begin{align} \prod_{i=1}^n \left[\frac{3\theta^3}{x_i^4} \mathbf 1_{\left[\theta, \infty\right)}\left(x_i\right)\right] &\overset{\theta}{\propto} \theta^{3n} \prod_{i=1}^n \mathbf 1_{\left[\theta, \infty\right)}\left(x_i\right) \\ &= \begin{cases} \theta^{3n}, &x_i\geq \theta \;\forall i \in \{1,\ldots, n\}\\ 0, &\text{otw.} \end{cases} \\ &= \begin{cases} \theta^{3n}, &\theta \leq \min\left(\left\{x_i: i \in \left\{1, \ldots, n\right\}\right\}\right)\\ 0, &\text{otw.} \end{cases} \end{align} $$ for the likelihood function of $\theta$.

Also, $\theta \mapsto \theta^{3n}$ is strictly increasing on $(0, \infty)$ for $n \in \mathbb N_{\geq 1}$.

Can you take it from here?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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