Terms of good sample
In general, a good sample is which can represent as much as possible the characteristics of the population. In the language of measurement, meaning that the sample must be valid, which can measure something that should be measured. A valid sample is determined by two considerations.
First: The accuracy or precision, the degree of absence of "bias" (error) in the sample. In other words, the less the error rate is in the sample, the more accurate the sample. Benchmarks for "bias" or error is population.
Cooper and Emory (1995) states that "there is no systematic variance"
that means is there is no diversity of measurement due to the influence
of known or unknown, which led to scores tend to lead to a certain
point.
Second: Precision. The second criterion is to have a good sample estimate of the level of precision. Precision refers to the question of how close we are to estimate population characteristics. There has never been a sample that can fully represent the characteristics of the population. Therefore,
in each sampling always attached keasalahan-error, which is known as
"sampling error" Precision is measured by standard deviation (standard
error). The
smaller the difference between the standard deviation obtained from the
sample (S) with a standard deviation of the population (s), the higher
the level of precision. Though
not always, the level of precision may be increased by adding the
number of samples, the fault may be reduced if the number of samples
plus (Kerlinger, 1973).
