10/12 Sampling Methods

Population vs Sample

  • If sample accurately represents the population, the results can be generalized to the population

  • Sampling frame - a list of the people/items that make up a population from which a sample is taken

  • Sampling error - different between results from sample and results from population; larger samples reduce error

Ways of sampling:

  • Probability sampling - random selection of participants; each member of the population has an equal chance of being selected

    • Simple Random Sampling - everyone has an equal chance of being selected. Complete list and RNG

    • Systematic - every Kth individual. Calculate sampling interval (k) = pop/sample

    • Stratified - break population into strata and sample within each group

    • Cluster - break population into clusters based on natural groupings and randomly select which clusters to completely sample

  • Nonprobability sampling

    • Convenience

    • Purposive - attempt to draw a representative group based on a researcher’s best judgement

    • Snowball - initial respondents obtained through nonprobability, but others are referred into the study

    • Quota - participants selected non-randomly from different strata or groups


Probability sampling is always better if you want generalizable results.

Nonprobability is better for:

  • exploratory research

  • cost-effectiveness

  • time constraints

  • pilot studies


10/14 Hypothesis Testing

What makes a good hypothesis?

  • Testable

  • Compatible with current knowledge

  • Logically consistent

  • Simple

Null hypothesis - default assumption that there’s no relationship between the variables

Alternative hypothesis - relation between the variables

  • We test the null, not the alternative. If the null is rejected, the alternative explains the relationship

Type I error - false positive; rejecting the null when it’s actually true and no relationship exists.

Type II error - false negative; failing to reject the null when a relationship exists.

P-value - the probability of getting observed results if the assumption is true

Significance level - how much type I error is acceptable

Results are statistically significant if the p-value is smaller than the chosen risk level (ex. .01, .05) and the null is rejected.