Scribd
Upload
10 views3 pages

MODULE IN STATISTICS - Why Your Type of Data Matters

Nominal, ordinal, interval, and ratio data types have different properties that determine what statistical analyses can be used. [1] Nominal data uses categories without mathematical relationships, allowing only non-parametric tests. [2] Ordinal data ranks categories but distances between ranks are unknown, also limiting analyses to non-parametric tests. [3] Interval data has equal distances between points, enabling parametric tests and calculations of mean and standard deviation.

Uploaded by

MARCOS GUADALOPE
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
10 views3 pages

MODULE IN STATISTICS - Why Your Type of Data Matters

Nominal, ordinal, interval, and ratio data types have different properties that determine what statistical analyses can be used. [1] Nominal data uses categories without mathematical relationships, allowing only non-parametric tests. [2] Ordinal data ranks categories but distances between ranks are unknown, also limiting analyses to non-parametric tests. [3] Interval data has equal distances between points, enabling parametric tests and calculations of mean and standard deviation.

Uploaded by

MARCOS GUADALOPE
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Why Your Type of Data Matters

Types of measurement for variables include nominal, ordinal, interval and ratio
levels.

Table 1. Each Type of Data, Corresponding Examples, and the Type of Statistics
You Can Use

Types of
Data/Scale of
Nominal Ordinal Interval Ratio
measurement

Examples Gender High school Standardized Balance in


(categories of class rankings test scores your savings
male and (e.g. SATs) account
female)

Marital status The order that Temperature Exam scores


(categories of participants measured in (percentage
married or not) finished a task Fahrenheit or correct on a
(I. e. first, Celcius test)
second, third)

Political party Bra cup sizes Ratings using Reaction time


(categories of a scale of 1 to
Democrat, 7 (or any other
republican or number range)
Independent, to measure
etc.) opinions

Hair color T-shirt sizes IQ scores


(categories of (SM, MED, L,
blonde, XL)
brunette,
redhead,
other)

Months of the Personality


Year Measures

Type of Mode, Median and Mean and Mean and


descriptive frequencies, range standard standard
statistic to use and deviation deviation
percentages

Type of non- Chi square


parametric test
to use

1
Type of means Mann- t-tests and t-tests and
test to use Whitney-U or ANOVA ANOVA
Wilcoxon

Type of Spearman’s Pearson’s r Pearson’s r


correlation to rho and multiple and multiple
use regression regression

Nominal data nominal measures present numerals that cannot be


used for meaning mathematical calculations.(Examples: Area code,
Student ID numbers, numbers on the Jersey of basketball players, Social
Security Numbers). Would you derive any valuable information from
calculating the mean of the Area Codes of participants, or Social Security
Numbers? Such calculations would be meaningless, and you can likely
see why. The same constraints limits the statistical analyses you can use
to test your hypothesis if you collected nominal data.

Based on Table 1. you will see that nominal data only allow you to run
non-parametric tests. For example, you can run a chi-square analysis with
nominal data, which compares observed frequencies to frequencies that
would be expected under the null hypothesis. In other words is the
observed number of items in each category different from a theoretically
expected number of observations in the categories?

Ordinal data. You might be interested in the rank order rating of the
variable you are measuring, or the sequence of events that took place, in
which case you would ordinal data. As opposed to nominal data, with
ordinal data you do know that one value is greater than another. Ordinal
variables have categories, just like nominal data, but the categories have
meaningful order. In fact, sometimes ordinal data are called ranked
because the categories can be put in order or ranked.

When would you use this type of data? Suppose you are interested in
determining if there is a relationship between birth order and high school

2
rank. Both of this variables would generate ordinal data. With ordinal data,
there are no equal differences between values. When considering birh
order, the first born could be two years older than the second born, with
the third sibling born six years after that. The same is true for high school
rank. These ranks do not indicate that the students ranked as 1 and 2 are
equally different in their GPA than student 2 and 3.

When conducting statistical analysis with ordinal data, like nominal


data, you are limited to non-parametric statistics. Like nominal data,
calculations like addition and subtraction with ordinal data are not
meaningful. That said, you can use your ordinal data and compare
medians and test the relationship between variables.

Interval data. Parametric statistics require addition and only interval


and ratio measures allow for addition. With interval variables, the levels
have meaningful order and the space between each number on the scale
is equal. Think about a thermometer and the markings on it that represent
10- degree intervals. Because this is an interval scale, not only do you
know that 90 degrees is hotter than 80, you know that each interval on the
thermometer differs by 10 degrees. That is why interval variables are
named interval to represent these equal increments across the variable.

You also generate this kind of data when using a LIkert-type scale. As
you see in the table when you collect interval data, you can run parametric
statistics to test your hypotheses. With this type of data you can calculate
mean and standard deviation, run independent t-tests or any type of
analysis of variance (ANOVA).

Ratio data. For ratio data, the value of 0 equals none. A simple way
to decide if a variable represents ratio data is to see if you can double the
value, (e.g. a person who is six feet tall is twice as tall as one who is three
feet tall.

You might also like