Types of Data
(based on Hinkle, Wiersma, & Jurs, 2003 refs)
Collected data are the results of the measurement of factors. For example, a student's knowledge of biology might be measured by a test or a written laboratory report. A grade on a test or lab represents measurement of knowledge. If a teacher examines the types of questions asked on a test, different levels of understanding are bound to be determined by the types of questions asked. Perhaps some questions are factual in nature, only requiring students to recall information. Some might be conceptual, which would utilize more higher-order thinking skills. Yet others might be analytical in nature, which, too, would be more higher-order computational skills. In any event, understanding is assessed, and assigned a numerical value which translates to a grade that depicts the measurement of mastery of information.
Not all measurement is the same. Some measures are more accurate than others. Saying a UConn basketball player tall is different from saying that her height is six foot five inches (or 1.96 meters, if I am being a responsible, metric-oriented scientist). There is a level of accuracy associated with the quantified measurement that is not present in the qualitative description of tall.
It is reasonable to say that some measurements are more amenable to accuracy than others. We can much more easily measure the basketball player's wingspan that we can measure an affective trait, like anxiety before the big game against Tennessee.
When choosing a statistical method to evaluate data, it is important to consider the accuracy of the type of measurement used. Scales of measurement are hierarchically categorized based on their level of accuracy. From least accurate to most accurate, the scales are: i.) nominal, ii.) ordinal, iii.) interval, and iv.) ratio.
contributed by Frank LaBanca, EdD
Nominal Scale
The least accurate measurement scale is termed nominal. This is sometimes referred to as categorical data. As the name implies, the measurements are classified by categories based on some defined characteristics. Generally, the number of objects in each category is counted for a total. Gender and ethnic background would be examples of nominal data that might be used in an educational setting.
Example:
Using gender as a nominal data source, the two categories (cases, or levels) are male and female. A tally of males and females can be counted to determine how many objects (in this case, individuals) fit into each of the two nominal cases.
Nominal data has the following properties:
- Data categories are mutually exclusive. An object can belong to one and only one category.
- There is no logical order (or reason for a logical order) for categories.
contributed by Frank LaBanca, EdD
Ordinal Scale
One of the key features to nominal data is that there is no logical order for categories. However, in an ordinal scale, categories still exist, but there is a logical organization and ordering to the categories. Ordinal scale is sometimes referred to as rank data. Scores can be ranked from highest to lowest, and then categorized within that order framework. The letter grading system (A, B, C, D, F) is an example of ordinal scale data.
Examples:
1. We know that someone who gets a grade of A has a higher grade than a person with a grade of B. However, we cannot infer that the distance between students with grades of A and B respectively are equal from students with grades of B and C.
2. A cooperating teacher has had four student teachers over the years and is asked to rank them from best to worse. He assigns them values:
- Jim = 1
- Susie = 2
- Roberta = 3
- Carl = 4
We can't say for certain that a Jim (1) compared to a Susie (2) is equal distance from Susie (2) to Roberta (3). So although we often assign a numerical value to each, we must be extremely cautious when considering differences. The process of essay writing will be much easier with MarvelousEssays.Com as there are a lot of highly professional and talented writers who are always eager to help you out with any sort of academic assignments regardless of the complexity levels. I do know what I�m talking about! 1-2 may not equal 2-3 on the ordinal scale
Ordindal data has the following properties:
- Data categories are mutually exclusive
- Data categories have a logical order
- Data categories are scaled or ranked according to the amount of a particular characteristic present
contributed by Frank LaBanca, EdD
Interval
Interval level data has all of the properties of nominal and ordinal with the addition of intervals between categories being equal. Sometimes the interval scale is referred to as the equal unit scale.
Examples:
1. We might ask someone if they agree or disagree with a statement. If the scale is 4-point, for example
- Strongly agree
- Agree
- Disagree
- Strongly disagree
We are assuming that the distance from strongly agree to agree is the same as agree is to disagree. This means that we can interpret differences in the distance along the scale. If we contrast this to an ordinal scale, we can only talk about differences in order, not differences in the degree of order. In this case, we must be very careful to ensure that our distances along the scale make logical sense. Sometimes we would term this as "equally appearing intervals."
2. Dates are also interval data. A 1-week treatment from September 1 to September 7 is half of a 2-week treatment from September 1 to September 14.
3. Although not applicable to Educational Research, temperature is also an interval data scale. Temperature is an important model to consider because there is a zero on a temperature scale, but notice that zero is NOT the absence of the trait or the start of the scale. Zero is still a temperature - it is just another value along the scale's continuum.
Interval data has the following properties:
- Data categories are mutually exclusive
- Data categories have a logical order
- Data categories are scaled
- There are equal distances between characteristics and they are represented by equal distances in the numbers assigned to the categories.
- Zero is just a point along the scale.
contributed by Frank LaBanca, EdD
Ratio
The highest level in the measurement scale hierarch is the ratio scale. Ratio-level data is generally considered the most precise method of measurment. Ratio data is similar to interval data with the added feature of having a true zero point. The true zero represents the absence of the characteristic that is being measured. Unfortunately, ratio data is not often available in social science/educational research.
Examples:
1. Physical science data is often available as ratio data. For example: mass, length, or energy.
2. In social research some ratio data examples: age, years of teacher experience, score on a 100-point test.
Ratio data has the following properties:
- Data categories are mutually exclusive
- Data categories have a logical order
- Data categories are scaled
- There are equal distances between characteristics and they are represented by equal distances in the numbers assigned to the categories.
- Zero is a point on the scale which represents the absence of a characteristic
contributed by Frank LaBanca, EdD
Summary
The four levels of measurement are as follows:
| Data Type | Explanation |
|---|---|
| Nominal | Categories without order |
| Ordinal | Ordered categories |
| Interval | Ordered categories with equal units between categories |
| Ratio | Ordered categories with equal units between categories and contains a true zero point |
contributed by Frank LaBanca, EdD
Don't forget that the acronym for the levels of data is NOIR, or black, in French. Very helpful hint from Dr. Delcourt.
contributed by Susan Guertin
