Questions

Week Two Listen To Me First

Speakers: Narrator, SME

Narrator: Welcome to the Week Two podcast for CJA/335. This week, we’ll discuss survey design and describing data.

What are some ways to describe data?

SME: Usually when you hear a description, you think about adjectives and vocabulary to visualize a noun. In statistics, description is more about values and significant calculations regarding a data set. Basically, data is described by descriptive statistics. Descriptive statistics include mean, median, standard deviation, range, minimum, maximum, and quartiles. As you become more familiar with statistics and analyzing data, these descriptive statistics help you visualize that data. Here’s an example of descriptive statistics in your life: if you remember grade school, test results were provided back to you with descriptive statistics. Not only do you get your score, you are provided the average (mean) score, the median (middle value score), and you are told in what quartile your score resides, such as the top 25 percent.

Narrator: Can you discuss bias and how it can arise in survey design?

SME: Bias often has a negative connotation, but understanding that the presence of bias in any aspect of life is almost inevitable is key to this discussion. Bias in survey design can occur by the method of selection and participation. This matters because of who has been excluded from a survey (non-respondent bias) and those who were included (reporting bias). As a survey researcher or designer, it is almost impossible to eliminate bias, but you can limit the effect of bias on data by focusing on two things: who the target population is and what you are trying to determine. The target population must be represented among those who are surveyed. The questions in the survey must be focused on what you are trying to determine, such as a correlation between dinner choice and dietary preference. In addition to controlling the sampling, bias can be eliminated by ensuring that questions in the survey will be interpreted in the same way by all respondents. This is really where skill is important in survey design.

Narrator: So, why is it important to understand data distribution?

SME: Understanding data distribution builds on our earlier discussion of knowing measures of central tendency—mean, mode, and median. Data distribution gives you an idea of how your sample or population is affected. This feeds into bias, but more importantly, it allows the analyst or researcher see what aspect of the data must be further analyzed and explored, as when you find that your data is skewed. There are three common data distributions: normal, left-skewed, and right-skewed. A normal distribution is what you often see; it's the bell-shaped curve, where the majority of the data groups and peaks in the middle. When you understand mean, median, and mode, you can readily identify those points in a distribution, figuring out if the peak is either the mean or mode.

Narrator: What key point should students focus on this week?

SME: The concepts in Week Two build on the concepts of Week One. Students should review what was covered last week prior to delving into this week’s concepts. Don't be discouraged: statistics only get easier with greater familiarity.