go to the School of Psychology home page Go to the UNE home page
Chapter 7 - Analysing the Data Part IV - Analysis of Variance Chapter 1 - Behavioural Science and research Chapter 2 - Research Design Chapter 3 - Collecting the Data Chapter 4 - Analysing the Data Part I - Descriptive Statistics Chapter 5 - Analysing the Data Part II - Inferential Statistics Chapter 6 - Analysing the Data Part III - Common Statistical Tests Frequency distributions Central tendancy Variability The normal distribution Transformations Standard scores - Z scores Correlation and regression Linear regression Readings and links

 

Chapter 4: Analysing the Data
Part II : Descriptive Statistics

 

Dichotomisation

As discussed in Chapter 1, variables can be classified in many ways. One way is continuous or discrete. A continuous variable (e.g., length or IQ) takes on many values, it is not restricted to just a few values such as gender (takes two values) or days of the week (takes on seven values). A variable that takes on only two values is a dichotomous variable. Male/female, yes/no, agree/disagree, true/false, present/absent, less than/more than, lowest half/highest half, experimental group/control group, are all examples of dichotomous variables.

A continuous variable is said to contain more information about a construct because it measures it more accurately or more sensitively. Asking a person if they Agree or Disagree to a question does not give as much information about that personŐs level of agreement as does a seven point (Likert) scale

1

2

3

4

5

6

7

Strongly Disagree

Moderately Disagree

Somewhat Disagree

Ambivalent

Somewhat Agree

Moderately Agree

Strongly Agree

So, in general, you should use continuous variables wherever possible. However, there are times when only dichotomous measures are possible. There are also times when dichotomous variables are useful in their own right. This is mainly when you are only interested in broad comparisons. To compare two or three or four broad groupings can sometimes lead to a clearer understanding of relationships in data than considering continuous data.

It is possible to convert continuous measurements to smaller numbers of categories by recoding the variable. We could recode height into below average or above average (call them 0 and 1, or 1 and 2). We could convert age into three categories of young, middle, and old. We could recode the above Likert scale to Disagree (all responses of 1, 2, and 3) and Agree (all responses of 5, 6, and 7) and ignore a response of 4. Whenever you do this you are losing information about the construct being measured. Note it is not possible to convert from a dichotomous variable to a higher number of categories. If you only have Agree/Disagree data, you cannot recode into a seven-point scale. This is because you do not have the information needed to regain that level of sensitivity.

 

 

 

© Copyright 2000 University of New England, Armidale, NSW, 2351. All rights reserved

UNE homepage Maintained by Dr Ian Price
Email: iprice@turing.une.edu.au