Disproportionate Stratified Sampling in Case Studies
In proportionate stratified sampling, the size of each stratum is proportionate to its representation within the general population. Conversely, in disproportionate stratified sampling the sizes of different groups may vary and not represent the percentage of the any particular group within the larger population. Sample points are awarded to each stratum and must be assigned correctly to assure accuracy in the study.
Reliance on the Researcher
Disproportionate stratified sampling places a great amount of reliance on the skills of the researcher for accurate results. If errors are made in the allotment of points to any group, the results may be skewed toward a particular stratum. In most cases proportionate stratified sampling offers a higher degree of accuracy with fewer margins for error, but disproportionate stratified sampling may be appropriate for certain types of studies.
When Is Disproportionate Sampling Used?
This type of sampling is most appropriate where one or more of the subgroups is very small in comparison to other groups, or where the target of the study is a specific and oversampling of a group may provide more accurate results. For instance, a study on the effectiveness of a Medicaid program might purposely over sample the group with the lowest annual income since they are more likely to have contact with the Medicaid program.
Example of Disproportionate Stratified Sampling
A city has a population of 100,000 residents with 5000 residents earning more than $50,000 per year and 15,000 residents earning less than $50,000 per year. The researcher chooses a sample of 100 people from each group. The sampling fraction for the higher income group is 100/5000 or .02% while the fraction for the lower income group is 100/15000 or .006%. The points assigned to each group are the inverse of the fraction or 50 points for the higher income group and 150 points for the lower income group.
Separating Results of Disproportionate Stratified Sampling
In a disproportionate analysis of opinions between a large group and a small group, for instance a company with 300 male employees and 50 female employees, assigning points to each group can effectively remove the opinions of the smaller group. It may be better to consider the two groups as separate populations for the purposes of the analysis rather than as subgroups. When analyzing specific subjects like employment opportunities for women, a disproportionate stratified sampling of this population may not provide the most accurate results.
Advantages of Disproportionate Stratified Sampling
Disproportionate stratified sampling provides the greatest advantage in the ability to study the responses of subgroups. It is useful where some subgroups are small and a proportionate sample might include only one or two individuals of a particular subgroup. In this case, the views of the individuals interviewed may not be representative of the subgroup and a disproportionate sampling might provide more accurate responses. The results can be equalized by assigning points to each group.
Disproportionate stratified sampling may provide greater accuracy than other sampling methods when one or two of the subgroups are extremely small or if a particular subject of the study is more relevant to one subgroup than another. It also provides the opportunity to analyze the results from each subgroup and how it relates to the general population being sampled.