# scale issues

#### Dr.Appalayya

##### New Member
Please look at the following descriptive labels for a scale of importance.

Extremely critical (7)

Highly important (6)

Very Important (5)

Important (4)

Moderately Important (3)

Of Little Importance (2)

Unimportant (1)

This scale is used for about 40 variables and a sample of 150 respondents is chosen to administer the instrument with above scale.

What do you call the scale given above?-nominal, ordinal or ratio?
For what kinds of analysis is above scale used for: factor analysis, multiple regression, correlation co-efficient, mean, and standard deviation.

#### jayashree_ramana

##### New Member
This is an ordinal scale, because there is a ranking of the labels with respect to each other.

The measure of central tendency for such scales are median or mode, mean, standard deviation cannot be defined for such scales.

#### Dr.Appalayya

##### New Member
This is an ordinal scale, because there is a ranking of the labels with respect to each other.

The measure of central tendency for such scales are median or mode, mean, standard deviation cannot be defined for such scales.
What kind of analytical tools can be employed on such scales? Thanks, Mrs. Jayashree.

#### terzi

##### TS Contributor
What kind of analytical tools can be employed on such scales?
Many non-parametrics tests can be performed with ranked data. You can use non-parametric correlation measures (Such as Spearman correlation or Kendall's), tests for testing difference, etc. There are also regression methods available for rank data, although are different models from those found in "Normal" Linear Rgression

Now, certain authors use ordinal response to certain analysis that involve continuous variables. Ofir and Simonson (2001) developed a Principal Component Analysis on 18 questions with a 5-option ordinal scale.

So, some ordinal measures, such as Likert Scales, can be used as continuous but this requires a deeper knowledge of the statistical methods you wish to apply.

#### Dr.Appalayya

##### New Member
Please read my original post. Is it wrong if I treat the numbers as scores or quantities, and that way, I could employ the tools like mean, standard deviation, Pearson's Correlation and Factor Analysis?