Please enable JavaScript.
Coggle requires JavaScript to display documents.
Producing data (Samples and Surveys (Non-sampling Error (Wording bias:…
Producing data
Samples and Surveys
Sampling Error
Sampling error is the result of testing a small area, instead of an entire population causing a sample bias.
Convience sample
Consists of data found in easy to access locations, giving a smaller sample space than if more accurately tested.
If one was trying to give a poll to see who people were voting for in their area, going to a one party rally may be an example of connivence sample because the average person in this example would not reflect the average person.
-
Undercoverage: An area not entirely covered, not getting a full sample space.
Hard to reach places, or just plain laziness can cause a lack of data, causing one to have unevenly distributed data, and a lack of response.
-
Bias
A bias is inaccurately sourced data, due to a conflict or interference
Good sampling techniques
SRS
A SRS is a random sample chosen from a larger population that is intended to find an even distribution of results.
The advantage to SRS is having a completely random sample, which drastically saves on survey costs for similar results. The disadvantage to this method is the lack of a 100% accurate representation of data.
Stratified SRS
A Stratified SRS divides the population into multiple sections, and picks random samples from each smaller group. This helps to prevent uneven distribution of data.
The advantage of Stratified SRS is having an even distribution of data, but the downside is that it is not a completely random sample.
Cluster Sampling
Cluster Sampling is when an area that reflects the general population of the sample space is selected for surveying.
Cluster sampling is a good tool for testing areas with similar distribution of data such as behavioral data. This does not work very well for data types based on income by area, because smaller samples such as this are not necessarily an accurate representation of the big picture.
Observational Study
An observational study is based on information already available information, based on peoples natural tendencies. This information may not be entirely accurate, as people lie, and may change their habits in some attempt to sway the test results.
If someone were to stand outside of a hotel to monitor how many people walk in and out to gather data about how many people were staying there, it would be an example of an observational study because no cooperation is needed and the information is readily available.
Exprtiment
In an experiment, the person administering the tests has complete control, giving specific instructions to the test subject, but the downside is that this may not create entirely organic results that would occur naturally in the real world, but this could also inflict possible harm on its subjects, as opposed to using information already available.
An experiment takes place in an entirely controlled setting, allowing for there to be no biases or errors in recorded data. Aside from this method not being 'organic', it is the most accurate way to test data.
Lurking Variable
A lurking variable is a variable that can be an underlying cause, that actually drives both data points observed, as opposed to one of the observed points causing the other.
If for a science project someone were to grow plants, testing their ability to grow using different substances as nutrients, there is a possibility that some plants may die regardless of the substance given.
Placebo effect
The placebo effect is a beneficial reaction to an administered form of treatment, to a patient who is told it might help them.
Good experiment design
To have a good experiment, it is important to have An even distribution of subjects, and for the questions to be airtight. It is also important for there to be no room for bias, and the subjects must be accurately and efficiently tested.
Blinding or Masking
Masking
When one withholds information for the sake of preventing bias, commonly used alongside a placebo.
Double blind: When neither the subject nor the tester has access to information about who gets the placebo, to give the tester more free reign to communicate how they want with the subject.
-
Holistic survey
For this survey, I will survey what kind of cellular device each student uses.
Each student will be assigned a number, and 100 will be randomly picked. From there, they will be asked what device they use