Method of collecting data in statistics?

Answer 1

Your question is very general. I will give you some suggestions and perhaps you can rephrase your question to a specific problem. I believe the question can be rephrased to how a statistician may approach obtaining valid data for the purposes of interpretation.

Generally, data is collected with the purpose of making inferences to a larger population which can not be surveyed. So, in statistics, the key to collecting data is that it is representative of the larger population that you are interested in.

The statistician has choices to make in a planned observational or experimental study. The simple random selection may be appropriate in many cases, for example, in a quality control situation, where a sample of parts from a larger batch of parts are selected and tested.

More complex sampling schemes are possible, still with the intent that the data can provide a significant, meaningful understanding of the population. The means to reduce biases in these surveys is very important.

Data can be complicated, and may not tell the full story. For instance, let's say that one road has a high number of accidents. Is it a problem of the road condition, the drivers that use that road, poor signs, too many exits, etc. In this example, statistics and other information can help point to the most important factors.

It should be noted that surveys are not the only way of collecting data. In education, data may be in the form of tests scores, GPA, etc. In media research, content analysis is frequently used to count and/or categorize randomly sampled media content (for example, comparing the volume or tone of war coverage in newspapers to television). The list of alternatives to survey research is extensive, but in all cases, the principles of random sampling and statistical assumptions still apply.