Qualitative Research-iii

Welcome to Part-III of my observations about Qualitative Research. As promised, we’ll take a quick tour of the different sampling strategies used in research. Since I am limited by goal of brevity and the blogosphere’s attention span, I am fully aware that my list is not exhaustive. If you are hungry for greater detail please do refer to a classic such as Leslie Kish’s, “Survey Sampling”, priced at $103 and potentially hazardous at 664 pages (close to 2 pounds, it’ll hurt if you knock yourself on the head with its pristine and crisp pages).

Quantitative Approaches

Simple Random Sampling

This is the most basic method everybody and anybody interested in surveying have come across.

1. Generate a numbered list of potential candidates
   1. Be sure not to introduce any bias. In other words be very careful not to systematically exclude participant group.
   2. The Chicago Tribune overlooked this simple fact and looked on a pile of embarrassment.
2. Write a random number generator, or use one of many available online.
3. Select one participant per random number…

…and voila! Hit the target participants with your survey.

Systematic Random Sampling

Systematic random sampling is often used in studies of processing and manufacturing units. People who work in Quality quite appreciate this method.

1. Decide on a random number “n” via the random number generator
2. Make a list of participants or take note of a flow of observations
3. Select every n^th person or observation
   1. Be sure that there is no pattern to the flow of observations
   2. For example, in a household survey if every 5^th house in the block is a doctor’s household, then avoid this method of selection

Stratified Random Sampling

As the name suggests first split the population into meaningful groups or “strata”. You could divide the households according to the educational level of the primary breadwinner of the families- those with or less than a high school certificate, graduation, and post grad degrees.

1. Make sure your stratification makes sense and is relevant to the research. Dividing the population in a household income survey by their degree is driven by the notion (possibly false) that higher one’s degree, greater their potential income
2. Perform either a Simple or Systematic random sampling of the members of these groups

Multistage Cluster Sampling

1. Get a list of clusters, example, branches of your organization
2. Randomly sample clusters from that list
3. Then randomly sample persons from the samples under consideration
   1. This method is expensive
   2. You should be able to justify your clusters

Qualitative Approaches

The goal of qualitative studies is to figure out reasons and sources of heterogeneity. They are often constrained by very tight budget restrictions. So the surveying has got to be time and cost effective.

Convenience Sampling

This sampling is based convenience. Pick observations that are readily available. Remember what we did in grade school when we had to collect money on behalf of the school for charities? We first asked our parents, the neighbors, our friends’ parents, and acquaintances etc.

Snowball Sampling

1. Target persons related to your focus of study
2. Ask and use their reference to fill your sample

Quota Sampling

1. Figure out how the population looks like in terms of specific qualities
2. Create quotas based on these qualities
3. Select people for each quota

Theoretical Sampling

This method can be best understood through the diagram below.

To end this series, let us look at an important choke point of carrying out surveys- namely, poor response rates. So how does one go about fixing this? It is not possible to force people to participate in your study so you should entice them instead with a personal invitation or certificate of participation, and offer an incentive to participate- free soda coupon etc.

That’s all friends. Thank you very much for taking an interest in this blog series.