Describing, Clarifying and Presenting Data

2. Characteristic, Variable and Measurement

2.2. Ensuring measurements are valid, reliable and unbiased

When you are thinking about measurement as it relates to variables, you also need to think about other issues. For instance, you should ask yourself:

• How accurate or unbiased are my measurements?
• How reliable are my measurements? (An average of repeated measurements is usually more reliable than a single measurement.)

Consider again the characteristic ‘intelligence’. The table below outlines how your measurements could be valid, reliable and unbiased?

Ensuring the language that you use does not affect your measurements

The language or terms you use can prejudice the response to questions. Consider, for instance, how responses to a survey might change if the word ‘problem’ is substituted for ‘crisis’, or if ‘termination’ or ‘murder of the foetus’ is substituted for ‘abortion’. Your choice of language can bias a response by suggesting to the respondent which answer they think is correct. It is, therefore, important that the language and terms you use are neutral.

Let’s look at an survey in which the variable used to define a characteristic (in this case what is ‘in’) was criticised as being both biased and invalid.

 SCENARIO

The company Levi Strauss conducted a survey across university campuses in the US[3]. The students surveyed were provided with a list of clothing items and were asked to select what was 'in' that year. The list consisted of:

• Levi's 501 jeans
• T-shirts with graphics
• 1960s-inspired clothing
• Lycra/spandex clothing
• Overalls
• Patriotic-themed clothing
• Decorated denim
• Printed pull-on beach pants
• Long-sleeved hooded T-shirts
• Neon-coloured clothing

Levi-Strauss used the results of the survey in a marketing campaign.

In each question below a statement is made about the survey. Decide whether the statement suggests aspects of the survey were

a) invalid, or
b) biased.