Suppose that there are two (2) candidates (i. E., jones and johns) in the upcoming presidential election. Sara notes that she has discussed the presidential election candidates with 21 friends, and 14 said that they are voting for candidate jones. Sara is therefore convinced that candidate jones will win the election because jones gets more than 50% of votes.

The statement made by Sara: "candidate jones will win the election because jones gets more than 50% of votes" cannot be considered true, when analyzing how she arrived to this claim.

According to statistical principles, when conducting inference, the results of a sample can only be generalized to the whole population if randomization exists in the data collection mechanism. Random samples are representative of the underlying population.

If the aim of Sara is to forecast the results of the elections, her population of interest is formed by all US citizens with the right to vote. Hence, a random sample should be collected out of the total members of this population. Her group of friends is not a random sample, and hence, not a representative sample and the results obtained cannot be generalized to forecast the behaviour of all US voters.

Also, a larger number of observations should be needed, once the sample has been randomly selected.