First 389 words of the document:
Descriptive and inferential statistics
Which statistic to use?
Measures of central tendency; informs us about central (middle) values for a set of data.
They are averages ways of calculating a typical value for a set of data.
An average can be calculated in different ways:
MEAN: adding up all the scores and dividing by how many scores there are.
It makes use of all values of all the data but can be unrepresentative of the data as a whole if
there are extreme values not appropriate for nominal data.
MEDIAN: is the middle value in an ordered list.
It is not affected by extreme scores but is not as sensitive as the mean because not all values
are reflected in the median not appropriate for nominal data.
MODE: the value that is most common in a data set.
It is the only method appropriate when the data are in categories (such as number of people
who like pink) i.e. nominal data, but can be used for all kinds of data. It is not a useful way of
describing data when there are several modes.
Measures of dispersion; informs us about the spread of data.
RANGE: calculated by finding the difference between their highest and lowest score in a data set.
This is easy to calculate but may be affected by extreme values.
STANDARD DEVIATION: expresses the spread of the data around the mean. This is more precise
measure because all the values of the data are taken into account. However some characteristics of
the data are not expressed such as the influence of extreme values.
Graphs; Graphics provide a means of eyeballing your data and seeing the results at a glance.
BAR CHART: The height of the bar represents frequency. Suitable for words and numbers i.e. all levels
SCATTERGRAM: Suitable for correlation data, a dot or a cross is shown for each pair of values. If the
dots form a pattern going from bottom line to top right, this indicated a positive correlation,
whereas top left to bottom right suggests negative correlation. If there is no detectable pattern
there is a zero correlation.
Other pages in this set
Here's a taster:
Inferential tests require time and patience but they are the only way to determine whether the
results of a study are significant i.e. a real effect has been demonstrated as opposed to a chance
pattern that looks meaningful.
When deciding which test is appropriate in any situation, you can ask yourself the three questions:
Justifying your choice:
SPEAMANS'S RHO: A test of correlation is needed, as the hypothesis predicted a correlation. The
data involved rating made by participants that are ordinal data.…read more