1. Quantative studies
   1. Every element has known non-zero equal probability of being sampled.
   2. Involves random selection at some point
2. 'Qualitative studies'
   1. NOT statistical representativeness but rep of emergent themes
3. STRATIFICATION
   1. most effective when
      1. Variability within strata are minimized
      2. Variability between strata are maximized
      3. The variables upon which the population is stratified are strongly correlated with the desired dependent variable.
   2. Advantages
      1. dif. sampling tech for dif. subpopulations.
      2. Focuses on important subpopulations and ignores irrelevant ones.
      3. Permits greater balancing of statistical power of tests of differences between strata by sampling equal numbers from strata varying widely in size.
      4. Improves the accuracy/efficiency of estimation.
   3. Disadvantages
      1. Requires selection of relevant stratification variables which can be difficult.
      2. Not useful when there are no homogeneous subgroups.
      3. Can be expensive to implement.
4. PROBABILISTIC
   1. SIMPLE RANDOM
      1. Random (e.g. computer gen) selection from defined eligible population
   2. SYSTEMATIC
      1. Pick the 1st one randomly --> select every nth subject thereafter
   3. STRATIFIED RANDOM
      1. Divide the eligible pop into desired strata (e.g. age) -->apply random sampling to EACH stratum --> each stratum is thus represented proportionately
   4. MULTISTAGE RANDOM
      1. Ramdom sampling in >=2 stages in selection. i.e. random sampling in subset of sample as well e.g. school-->pupils
   5. CLUSTER
      1. Randomly select clusters (e.g. centres) of samples and choose ALL subjects in that cluster i.e. don't randomize the subset
5. NON-PROBABILISTIC
   1. PURPOSEFUL
      1. identify a population of interest -->develop a systematic way of selecting cases that is not based on advanced knowledge of how the outcomes would appear.
   2. THEORETICAL
      1. sampling is guided by theoretical concepts /categories generated thr' literature search
   3. QUOTA
      1. divide pop in mutually exclusive segments --> select subjects NON-RANDOMLY as per specified proportion (i.e. like stratified sample but non-random)
   4. SNOWBALLING
      1. one subject suggests other subjects (good design for study in drug users)
   5. CONVENIENCE / OPPURTUNISTIC
      1. sample that is easy to find and close at hand is picked