Understanding Social Science Research by Black Thomas R.;

Understanding Social Science Research by Black Thomas R.;

Author:Black, Thomas R.;
Language: eng
Format: epub
ISBN: 334609
Publisher: SAGE Publications
Published: 2001-09-17T16:00:00+00:00

FIGURE 6.3 Example of a negatively skewed distribution for a measured trait

Most psychologists still argue that intelligence is normally distributed and as a consequence, psychometricians ensure that IQ tests produce normal distributions of results. It is processes employed during test construction that make it highly probable the instrument will generate scores forming such a distribution for representative samples of the population. This involves selecting questions that provide the optimal amount of spread in scores.

Alternatively, tests designed by teachers and examination boards may well be criterion referenced (actual grades are determined by comparing scores against specific predetermined criteria) rather than norm referenced (designed to produce a normal distribution with grades based on how examinees perform relative to each other). The design of such tests makes no assumptions about the shape of the distribution of scores and, therefore, the choice of the questions does not force the shape one way or another. There has been a tendency for criterion-referenced tests to produce negatively skewed distributions (with long tails at the low end) such as the one shown in Figure 6.3. Since the objectives and criteria for success for such tests tend to be well defined and well understood by the examinees, they tend to be better prepared for them, and consequently scores tend to bunch towards the high end.

While other shapes will appear, these three basic categories of distributions will suffice for the following discussion on choice of statistics (numbers) that would best describe a group characteristic. Some other shapes will be discussed later. Frequently in social science research, it is assumed that the distribution of scores on questionnaires or other instruments is normally distributed, without showing any evidence. This is based upon similar arguments for assuming intelligence is normally distributed and often presented without corroborating evidence. As a reader, you may not have recourse to the raw data and must trust that the researcher has checked it, but rest assured not all traits or characteristics are normally distributed. For example, income across most societies has a long tail containing a relatively few high-income people and would produce a graph with a shape like that of Figure 6.2.


Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.