BL ECO - Spada - A. Indagine statistica - 1a parte

Summary

The lecture introduces statistical surveys, the information they produce, and the main ways statistical data and surveys are classified.

What a Statistical Survey Studies

A statistical survey is a structured process for collecting data to produce statistical information: concise descriptions of a phenomenon across a group or population.

Statistics is especially useful when a phenomenon varies among the units being studied. For example, if everyone had the same age, calculating the population’s average age would provide little information. Variability is what makes a phenomenon statistically interesting.

• Population: The collective group or phenomenon under study.
• Statistical unit: The individual item being observed or measured. It might be a person, a product, or another appropriate unit.
• Character (or statistical variable): A feature observed on a statistical unit, such as age, weight, occupation, or education level.
• Modality: The value or category a character takes for a particular unit. For age, examples include 15, 39, or 51 years.

Collecting Data

The data-collection method depends on the information needed. Examples include:

• Questionnaires, such as those used for opinion polls.
• Measuring instruments, such as a scale for measuring product weight.
• Counting, such as recording how many people enter a shop.
• Existing administrative or business records, such as company accounts.

A survey can observe multiple characters for each statistical unit—for example, a person’s age, weight, and education.

Types of Statistical Characters

Qualitative Characters

Qualitative characters describe non-numerical qualities or categories. Their modalities are called attributes or categories. Examples include education level, occupation, or the presence or absence of a feature.

• Nominal (unordered) characters: Categories have no meaningful order, such as religion, birthplace, or occupation.
• Ordinal characters: Categories have a meaningful order, such as satisfaction ratings from “not at all satisfied” to “very satisfied,” or competition rankings.
• Cyclical ordered characters: Categories have an order that repeats in a cycle, such as the months of the year.

Quantitative Characters

Quantitative characters take numerical values and are also called statistical variables.

• Interval-scale variables: Differences are meaningful, but there is no absolute zero, so ratios are not meaningful. For example, with temperature in degrees Celsius, 40°C is not “twice as hot” as 20°C, though the difference is 20 degrees.
• Ratio-scale variables: There is an absolute zero, so both differences and ratios are meaningful. For example, an income of €1,000 is twice an income of €500.
• Discrete variables: Take countable values, typically whole numbers, such as the number of children.
• Continuous variables: Can be measured to increasing degrees of precision, such as weight or height. The recorded precision depends on the measuring instrument.
• Transferable variables: Can shift from one statistical unit to another, such as income or employees moving to another employer.
• Non-transferable variables: Are tied to the unit, such as a person’s age or weight.

Total and Sample Surveys

Surveys can be classified by how much of the population is observed:

• Total survey (census): Observes every unit in the population.
• Sample survey: Observes only a subset of the population, called a sample, and uses those observations to draw conclusions about the wider population.

Total Surveys

Advantages

• Provide extensive information and detail.
• Can support later surveys, including sample surveys.
• Do not have sampling error because the entire population is observed.

Disadvantages

• Require more time, money, and organizational effort.
• May produce less precise data in practice when the survey is very large, because it is harder to examine every unit carefully.

Sample Surveys

Advantages

• Reduce costs, time, and organizational workload.
• Allow more focused, detailed, or careful observations of the units included.

Disadvantages

• Have sampling error, since only part of the population is observed.
• Cannot extend results to the entire population with certainty; statistical inference generalizes them probabilistically. For example, a population estimate may be stated with a 95% confidence level rather than as a certainty.

When a Sample Survey Is Necessary

A sample survey may be necessary, rather than merely preferable, in these situations:

• The population is effectively unknown or unbounded, such as fish in a river or an ongoing production process.
• Observing a unit destroys or consumes it. Examples include testing how long a light bulb lasts, testing pasta’s cooking resistance, or taking a blood sample. Testing every unit would be impractical or impossible.

Speakers and Sources

• One lecturer, identified in the video title as A. Spada.
• The lecture also refers to teaching slides, tables, and illustrative examples. No other speakers are featured.