There is something which is incorrect in terms of the terminology. Everyone does it.
Please pardon me for going geek in this post.
There is a tendency to use the term 'sample size' when it really means 'number of observations'.
E.g., The Detroit Tigers have played 3 games this season so the number of observations = 3.
Sample size would be take all 162 games and draw 3 games at random to 'represent' those 162 games.
If these first 3 games 'represent the 162 games' then one is saying that the Tigers will win 162 games and not give up an earned run.
These (first) 3 games don't represent a 'sample' of the season but the first 3 games of the start of the season.
Or a Tiger batter might have had 15 of say 500 at bats.
Typically sample refers to a defined population, e.g. all freshmen, enrolled on a given date, at College X, and a sample is some 'subset' of that population. The rest everyone knows: "random" for which statisticians have many different methods to draw a sample.
The reason for a sample, obviously, is that it is much easier to draw a small sample randomly (e.g., 1,000 registered voters in Ohio) then to interview all registered Ohio voters.
Sorry for getting off topic.