Statistics for Modern Life: 06 Common Research Techniques

Robert Ramstetter By Robert Ramstetter, 16th Jun 2015 | Follow this author | RSS Feed | Short URL http://nut.bz/282m368d/
Posted in Wikinut>Guides>Science>Numbers And Maths

In order to have proper statistics on a subject, the method of data collection is most important.

Common Research Methods in Statistics:

It is time to take a break from the mathematical portion of Statistics and remember that data collection and the methods of doing so are just as important. When compiling data for Statistics, there are different methods that may be employed. Some may be better suited for certain types of surveys, while others are more appropriate under different circumstances. As you will see, some methods do not produce reliable data at all. Unfortunately, many survey results are the product of unreliable data collection methods. Deciding the type of data collection is paramount to achieving a reliable survey.
One of the terms with which you will need to familiarize yourself is the term “population”. For statistical purposes, population is used to define those whom you are compiling your data. If, for instance, you wanted to compile average BMI (Body Mass Index) for vegetarians living in England under the age of 50, then your population would be all of the vegetarians in England under the age of 50. It would not represent the whole population of England. However, it would not be possible to record the BMI of all vegetarians under 50 living in England, so a reliable representative sample must be taken. In order to conduct your research, you would need to take a sample survey of people under the age of 50 living in England who are also vegetarians. The method in which you gather your data can vary in accuracy and have a tremendous impact on your result.

Randomized Experiments vs Observational Studies

A randomized experiment is one in which a portion of a group is manipulated and compared to subjects in the same group who were not manipulated. An example of this would be a pharmaceutical company that is conducting a survey to determine the effectiveness of a new drug. Let’s say they conduct the survey on two hundred people. Out of that group, half were given the drug and the other half were given a placebo.
There are two variables in studies such as these. The first is the explanatory variable. This is the feature that is being manipulated on the subjects of the study. In this case, it would be whether they received the drug or received the placebo. The second variable is called the outcome variable, or it can also be called the response variable. In the drug study, the outcome, or response variable would be drug effect vs. no drug effect.
An observational study, on the other hand, is where there is no forced manipulation. The manipulation occurs naturally and only the results are recorded. There are many times that an observational study is the only possible method. Suppose, for instance, you wanted to conduct a study on how many people continued to use heroin after becoming addicted. There is no way a study could be conducted by giving heroin to a group of people and allowing them to become addicted. The only logical way would be to study those who already had an addiction. This is an observational study, where you research without manipulating the outcome. You simply observe.

Random Sampling

In order to do a truly representative sample survey of a particular population, you must ensure that everybody has an equal chance of participating. In other words, if you wanted to survey the political affiliations of a certain city, you would not get a true representation if you went to an upscale mall and asked people who were shopping at a jewelry store. On the other hand, you would not get a true representation either if you polled people watching a little league ball team. In the first example, you would be more likely to sample more affluent members of the community. In the second example, you would be more likely to sample those who had small children. Even if you called people at random at 6:00 in the evening from random numbers in the phone book, you would only get samples of those people who had phones and were home at 6:00 in the evening. Anyone who did not have a phone, or had evening jobs or other commitments would be left out of the survey. Therefore, it is most important that you design your survey around the most representative population sample possible.
Some of the worst methods of collecting data is one of the most commonly observed ones. One example would be asking readers, television viewers, or web site browsers to respond to a question. This may be whether or not you think a particular candidate’s actions make them unworthy of your trust. You probably see this type of survey all of the time. The results will then be posted, saying that 94% of the people feel that the particular candidate is not trustworthy. In truth, it is not the candidate’s trustworthiness, but the survey itself that is not trustworthy.
There are numerous issues with these types of surveys that make them not even worth the paper on which they are printed (or the electrons on which they are stored). The first is the audience itself. How was this survey conducted? Many times it will be conducted during a news story about the candidate’s issue. Anybody who is watching already has an opinion or curiosity about the candidate, or they would not be watching. Also, the story that is being covered is directing their opinions toward the fact that the candidate is not trustworthy. Finally, only those who have a strong opinion will bother to respond. After examining the facts, you can correctly conclude that the results of the survey are complete rubbish and meant only to advance the agenda of the group who is conducting the survey.
Taking a proper sample is the backbone of your statistical analysis. Without it, the best mathematical formulas are of little use.

This is part 6 of the series "Statistics for Modern Life".
Statistics for Modern Life: 01 An Introduction
Statistics for Modern Life: 02 Beginning Terminology
Statistics for Modern Life: 03 Ratio Scales
Statistics for Modern Life: 04 Learning to Sum
Statistics for Modern Life: 05 Mean, Median, & Mode

Tags

Math, Mathematics, Observational Studies, Randomized Experiments, Sample, Statistics, Surveys

Meet the author

author avatar Robert Ramstetter
Robert Ramstetter is a world traveler and writer of short stories, full length novels, and a vast array of technical articles.

Share this page

moderator Mark Gordon Brown moderated this page.
If you have any complaints about this content, please let us know

Comments

author avatar Mark Gordon Brown
17th Jun 2015 (#)

It is good to have a proper understanding of statistics.

Reply to this comment

author avatar Robert Ramstetter
17th Jun 2015 (#)

I couldn't agree more.

Reply to this comment

author avatar Carol Roach
22nd Jun 2015 (#)

in psychology a survey is asking people to respond by questions asked. The drug survey you mentioned is an experiment, or clinical study.

Those surveys for candidates are useful, not garbage. First of all you need people who want to participate and know the candidates regardless of what the latest bad review. People have been found to be very loyal and do not change their political view on a whim.

Also these are the people you want, what good is asking someone who does not know one candidate from another and couldn't care less. That information is totally useless.

Reply to this comment

author avatar Robert Ramstetter
23rd Jun 2015 (#)

I totally agree with you if the data is collected for a useful purpose, such as the examples that you mentioned. However, I am referring to the surveys that you see all of the time in articles that are used as a self-promoting tool for the articles themselves. I appreciate and value your input and I would like to know how you feel about this comment.

Reply to this comment

Add a comment
Username
Can't login?
Password