is the data nominal tallied, or ordinal (ranked)?

Is the data nominal tallied or ordinal (ranked)?

As you now know, a non-parametric-relationship type test will be used for at least one of the following reasons:

As you now know, a non-parametric-relationship type test will be used for at least one of the following reasons:1. The distribution is skewed

As you now know, a non-parametric-relationship type test will be used for at least one of the following reasons:1. The distribution is skewedOR

As you now know, a non-parametric-relationship type test will be used for at least one of the following reasons:1. The distribution is skewed2. The data is ordinal (ranked)

As you now know, a non-parametric-relationship type test will be used for at least one of the following reasons:1. The distribution is skewed2. The data is ordinal (ranked)OR

As you now know, a non-parametric-relationship type test will be used for at least one of the following reasons:1. The distribution is skewed2. The data is ordinal (ranked)3. The data is nominal tallied

The purpose of this presentation is to help you determine if your problem has:

The purpose of this presentation is to help you determine if your problem has:

At least one variable that is Ordinal (Rank Ordered)

The purpose of this presentation is to help you determine if your problem has:

or

At least one variable that is Ordinal (Rank Ordered)

The purpose of this presentation is to help you determine if your problem has:

Both variables that are NOMINAL TALLIED

At least one variable that is Ordinal (Rank Ordered)

Let’s begin with:

Let’s begin with:

At least one Ordinal (Rank

Ordered) Variable

There are two ways to express ordinal data:

As rank-ordered data:

As rank-ordered data:

1st, 2nd, 3rd, 4th, 5th . . .

Or as percentiles:

Or as percentiles:

1%, 10%, 50% or 99%

Or as percentiles:

1%, 10%, 50% or 99%

A percentile means the percent of observations, scores or

persons below a point (e.g., 10%le means 10% of all observations, scores or

persons fall below this point)

Here is an example of a question of independence with one rank-ordered variable:

Is the Nielsen rating rankings for TV shows independent of shows that lean conservatively or liberally?

Is the Nielsen rating rankings for TV shows independent of shows that lean conservatively or liberally?

SHOW LEANS1 = Conservative2 = Liberal

NIELSEN RATINGS

RANKINGS2 4th

2 2nd

1 1st

1 3rd

Is the Nielsen rating rankings for TV shows independent of shows that lean conservatively or liberally?

SHOW LEANS1 = Conservative2 = Liberal

NIELSEN RATINGS

RANKINGS2 4th

2 2nd

1 1st

1 3rd

This is ordinal or rank

ordered data.

So we would select:

Both variables that are NOMINAL TALLIED

At least one variables that is Ordinal (Rank Ordered)

Nominal Tallied data is simply the amount of certain levels within a category.

For example:

Gender is a category with two levels:

Gender is a category with two levels:- Male

Gender is a category with two levels:- Male - Female

Gender is a category with two levels:- Male - Female

“Tallied” simply means the number in each of the levels of the category.

For example:

For example:- Male - 46- Female - 54

For example:- Male - 46- Female - 54

These are called the

tallies.

Now let’s add age as a variable.

Age can be a category with as many levels as desired:

Age can be a category with as many levels as desired:1 – Infant/toddler (0-2 years)2 – Children (3-12 years)3 – Teenagers (13-19 years)4 – Young Adults (20-39 years)5 – Middle Age (40 – 64 years)6 – Seniors (65 and older)

Age can be a category with as many levels as desired:1 – Infant/toddler (0-2 years) - 1002 – Children (3-12 years) - 803 – Teenagers (13-19 years) - 2004 – Young Adults (20-39 years) - 2885 – Middle Age (40 – 64 years) - 2016 – Seniors (65 and older) - 86

Age can be a category with as many levels as desired:1 – Infant/toddler (0-2 years) - 1002 – Children (3-12 years) - 803 – Teenagers (13-19 years) - 2004 – Young Adults (20-39 years) - 2885 – Middle Age (40 – 64 years) - 2016 – Seniors (65 and older) - 86

These are called nominal

TALLIES.

Let’s imagine that you want to know if age is independent of gender.

Let’s imagine that you want to know if age is independent of gender. Your expectation is that the number of persons that are at a particular gender level and age level are the same.

So, if you selected a sample of 600, you would expect that the “nominal tallied” data set would look like this:

So, if you selected a sample of 600, you would expect that the “nominal tallied” data set would look like this:

0-2 3-12 12-19 20-39 40-64 65+MaleFemale

So, if you selected a sample of 600, you would expect that the “nominal tallied” data set would look like this:

0-2 3-12 12-19 20-39 40-64 65+Male 50 50 50 50 50 50

Female 50 50 50 50 50 50

But let’s say you collect a sample of 600 that is actually arrayed as follows:

But let’s say you collect a sample of 600 that is actually arrayed as follows:

0-2 3-12 12-19 20-39 40-64 65+

Male 100 25 80 100 80 60

Female 25 25 20 25 20 40

But let’s say you collect a sample of 600 that is actually arrayed as follows:

0-2 3-12 12-19 20-39 40-64 65+Male 100 25 80 100 80 60

Female 25 25 20 25 20 40

Because the data is not arrayed as expected then gender and age may not be independent of one another.

But let’s say you collect a sample of 600 that is actually arrayed as follows:

0-2 3-12 12-19 20-39 40-64 65+Male 100 25 80 100 80 60

Female 25 25 20 25 20 40

Independence can be tested for statistical significance.

So we would select:

At least one variables that is

Ordinal (Rank Ordered)

Both variables that are NOMINAL TALLIED

What type of variables does your problem have?

What type of variables does your problem have?

Both variables that are NOMINAL TALLIED

At least one variable that is ORDINAL (Rank Ordered)