Why is random sampling a better choice when making an inference?

Random sampling is a completely non biased and purely random and reliable way of choosing samples, which represent the whole population. Everyone has an equal chance of getting selected here.

Random selection does not divert the survey in one direction that is why it is a better choice when making an inference about something.

For example:

1. Lets say we are randomly selecting students from each class in a school to get to know their favorite subject.

This is random as anyone can be selected and anyone can answer.

2. But if we say that we choose the students from an art class to know their favorite subject, then it will be a biased choice as the students will tend to say arts is their favorite subject. So. a valid inference cannot be made.

Because it is more reliable, with inference you can be biased toward one side. Random is random and has no bias.