Introduction to Descriptive and Inferential Statistics
Data Collection and Descriptive Statistics
Organizing data is the first step in statistical analysis, often involving the use of frequency tables to summarize observations.
Frequency Table Example: Favorite Colors
- Green: observations
- Purple: observation
- Blue: observations
- Pink: observations
- Teal: observations
- Yellow: observations
- Red: observation
- Total Sample Size ():
Descriptive Statistics Defined: This branch of statistics involves the collection, organization, and summarization of observed data.
Summaries can be expressed as fractions or percentages.
- Example: In a group of people, if like blue, the summary is .
- Calculating the percentage: , which can be rounded to .
- Example: For the color teal, the summary would be .
Inferential Statistics and Generalization
- The Goal of Inferential Statistics: The purpose is to take a limited set of data (a sample) and use it to draw conclusions or generalize about a much larger group (a population).
- The Problem of Certainty: It is impossible to know a population value with absolute certainty without asking every single member of that population.
- For example, we cannot definitively state that of all students at a university like blue based solely on a sample of students. Another class or sample group might yield different results (e.g., green or yellow might be more popular elsewhere).
- The "X" Variable: In inferential statistics, we represent the true population value as . Since we often cannot ask everyone (e.g., asking every student in the state of Georgia their favorite color), the goal is to get as close to the value of as possible using sample data.
- The Challenge of Large Populations: Some populations are physically or logistically impossible to measure entirely.
- Example: Whale Lifespans. To find the exact average lifespan of every whale on the planet, one would have to observe every single whale until death. If a researcher only has data for whales, they must draw conclusions for the entire species based on that extremely limited data set.
Confidence Intervals and Levels of Confidence
- Because we can never know the exact population parameter for certain, statisticians use intervals to estimate where the true value lies.
- Confidence Levels: We can state with a certain degree of confidence (e.g., or ) that the real population value is located between two specific numbers.
- The true population parameter is expected to fall within an interval defined by a lower bound and an upper bound. The mathematical process for narrowing down these values is typically covered in advanced stages of a statistics course.
Core Vocabulary in Statistics
- Population: The entire group that is being studied. In any statistical study, there is only one defined population.
- Sample: A subset or a smaller group selected from the population. While there is only one population, there are potentially hundreds of thousands of different possible samples.
- A subset is a smaller group taken out of a larger set (e.g., selecting only even numbers from the set of numbers from to ).
- Individual: A member of the population. While often referring to people, an "individual" in statistics can be more abstract, such as a specific color in a population of favorite colors.
- Statistic: A numerical summary of a sample.
- Example: If of the students in a specific class like pink, that is a statistic because it describes that specific sample.
- Parameter: A numerical summary of an entire population. These values are often difficult or impossible to obtain in real-world scenarios.
- Abstract Populations: In statistics, a population is not limited to people in a geographic area. It can refer to groups of attributes or measurements, such as:
- The lifespans of all whales (focusing on the ages rather than the whales themselves).
- The gas mileages of all automobiles (focusing on the fuel efficiency figures).
- The favorite colors of an entire student body.
Practice Examples: Parameter vs. Statistic
- Case 1: US Governors. Following the national midterm election, of the governors of the U.S. were women.
- Classification: Parameter.
- Reasoning: This value describes the entire population of U.S. governors (all of them), not just a sample.
- Case 2: Calculus Midterm. The average score for a class of students taking a calculus midterm was .
- Classification: Parameter.
- Reasoning: The problem describes the entire group in question (the class of ) and does not specify a larger population from which these students were sampled. If the class is the entirety of the group being studied, the summary is a parameter.
- Case 3: National Substance Abuse Survey. In a national survey on substance abuse, of graders reported using illicit drugs within the past month.
- Classification: Statistic.
- Reasoning: The keyword "survey" indicates that a sample was taken. The describes the sample of individuals who responded to the survey.
- Population Identification: The population in this case would be all graders in the United States.
- Case 4: Bullying Poll. A poll of middle school students in the U.S. found that of them had been bullied.
- Classification: Statistic.
- Reasoning: A "poll" functions like a survey, representing a sample of the larger group.
- Population Identification: Every middle schooler in the United States.
Questions & Discussion
Question: Why can't we say of all university students like blue based on the class data?
Answer: The reasons include a limited sample size and the fact that the entire campus was not surveyed. Additionally, there is inherent variability; different classes might have different favorite colors like green or yellow. We only have a limited set of data to work with.
Question: What would the population be in the national survey on substance abuse?
Discussion: The population isn't just people in the nation or those who use substances; it is specifically the entire group the sample is meant to represent. In this case, it would be all graders in the United States. The respondents to the survey constitute the sample.