Skip to main content icon/video/no-internet
icon backReturn to Entries

Data Transformation

Broadly speaking, data transformation refers to the conversion of the value of a given data point, using some kind of consistent mathematical transformation. There are an almost limitless number of ways in which one can transform data, depending on the needs of the research project or problems at hand. The current entry discusses some of the data transformations more commonly seen in communication research, the instances in which they would be used, and their practical utility to the communication scholar. These include transformation into standard scores, inverse scoring, dichotomizing, and log transformations. Where applicable, some of the shortcomings and tradeoffs associated with these transformations are also addressed.

Transformation Into Standard Scores

Perhaps the most common form of data transformation is the conversion to standard score or z score. Stand scores use the standard deviation to represent the position of a data point in an overall distribution, expressed as:

z = Raw score − Mean score Standard deviation

The z score represents the number of units of standard deviation that a given data point falls from the mean of the distribution. Thought differently, the computation of z scores transforms the original distribution of scores in an array of data into a normalized distribution with a mean of 0 and a standard deviation of 1. This particular transformation allows for the comparisons of data points that may have been measured using different scalars. One can make sense of data distributions gathered using different measurement systems, as standardizing the distributions allows one to evaluate the position of individual data points in the distribution.

Inverse Scoring

Another common practice in data transformation is the reversing of scores on a particular data point. There are a number of reasons for such transformations, but most common is its applicability to multiple item scales, such as semantic differential or Likert scales. These scales typically attempt to tap into abstract or complex constructs by asking numerous questions that are logically related and should be somehow correlated mathematically.

It is common practice to ask questions that require reverse coding. As communication scholars ask multiple items on questionnaires, it is good practice to ask items that should logically correlate negatively with the construct at large. This allows the researcher to check for problems such as participant fatigue or researcher effects, as a respondent giving consistent answers regarding his or her attitude or position on a given construct should provide answers that correlate consistently with the true score in question. Although there are a number of mathematical tests that can be performed to evaluate the effectiveness of scale in tapping the construct (such as checks of internal consistency and reliability), the first step in the process is to reverse-code items that require it, such that all items are scored in the same way, and that an increase in value for either the scale as a whole or an individual item represents some kind of increase in the value of the construct being measured.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Read next

More like this

Sage Recommends

We found other relevant content for you on other Sage platforms.

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution