Orditerval/Ordiratio Variables in IR Research

I know it’s kind of freaky name. In the second class of the first week, we talked about three types of variable: nominal, ordinal, and interval/ratio. Some variables are flexible, can fall into any category depending on context. Also one category will help the others to define each other. Especially ordinal variable, we basically cannot measure the distance between categories. However, a lot of people appreciate the exact number shown in the result as it makes the report look more scientifically. Therefore, they often threat ordinal variables as interval/ratio variables.

For example, in relevance judgment area, most studies are intended to find out how people make judgment by let subjects define the level of relevance of document retrieved. One technique I found interesting is continuum. The scale of the continuum is ranged from X- not relevant at all to Y-absolutely relevant. Then researchers observe the degree of relevance by measuring the distance of the continuum and reported as interval/ratio numbers.

In another context, there was a similar study observing the relevance degree by letting the subjects squeezing the measure object as how relevant is the document retrieved. Yes! The focus will be spotted on the reliability of this method. Anyway, as the result of the study, they present the result as the interval/ratio number which may be the “average” degree of relevance. Doesn’t it sound weird for you?


Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / เปลี่ยนแปลง )

Twitter picture

You are commenting using your Twitter account. Log Out / เปลี่ยนแปลง )

Facebook photo

You are commenting using your Facebook account. Log Out / เปลี่ยนแปลง )

Google+ photo

You are commenting using your Google+ account. Log Out / เปลี่ยนแปลง )

Connecting to %s