Experiments are designed to isolate and measure characteristics of one specific variable. Because of the necessity to isolate one characteristic -- such as measuring a mouse's spatial memory while under the influence of a memory-altering drug -- experimental results have to be generalized with caution. One way to augment the credibility of an experiment's results is to perform it with a large sample size to make the results more representative of an entire population.

## Increased External Validity

Experiment and survey results are more credible with a larger sample size.

If one wanted to determine what proportion of Americans drink soda on a daily basis, and get the correct answer, every American's soda consumption habits would need to be gauged. This would require colossal amounts of time and money. The other extreme would be to ask a very small number of Americans about their soda consumption. But these results could not be generalized with confidence to the entire population of America. The way to get the best approximation of the true proportion who drink soda daily would be to ask as large a sample of Americans as possible. This increases the findings' external validity, or likelihood that the results are truly indicative of a trend in the population.

## Accuracy

A larger sample size reduces the margin of error.

Margin of error is the range of sampling error in a survey's results. For example, according to the Universal Accreditation Board, if you have a population of 100,000 people and survey 384 of them about their soda drinking habits, there will be a 5 percent margin of error under standard confidence levels. If our experiment finds that 40 percent of the 384 polled drink soda daily, the margin of error means the real number could be anywhere between 35 and 45 percent. If we increased our sample size to 2,400 people, the margin of error decreases to 2 percent. Now our result of 40 percent is much more accurate, because the true range lies within 38 and 42 percent. Greater sample size leads to greater accuracy.

## Segmentation

With a large sample size, you could segment your measurements to determine how well young mice perform vs. old mice.

Measuring a large sample size allows one to divide the sample into segments, or smaller portions of the sample that share characteristics. Segmentation yields more and deeper information about a population. If one had tested the maze-running ability of a large sample of mice influenced by a memory-altering drug, male and female segments of the sample could analyzed. Or one could break down the total numbers and see if results were affected by the age of the mice. With a large sample, conclusions can be confidently drawn about subgroups of a sample, whereas this is not possible with a small sample.

## Capture of Diversity and Outliers

A large sample will capture the full diversity of a population better than a small one.

When a sample size is large, it is more likely to be representative of the full diversity of the population from whence it came. Likewise, the larger the sample, the greater the odds that outliers will be captured. Although outliers sometimes complicate the statistical analysis of data, observing and accounting for them gives a more realistic picture of the characteristics of the population.