12.01.2015 Views

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

26<br />

Choosing a statistical test<br />

There are very many statistical tests available to<br />

the researcher. Which test one employs depends<br />

on several factors, for example:<br />

<br />

<br />

<br />

<br />

<br />

<br />

the purpose of the analysis (e.g. to describe<br />

or explore data, to test a hypothesis, to seek<br />

correlations, to identify the effects of one or<br />

more independent variables on a dependent<br />

variable, to identify differences between two or<br />

more groups; to lo<strong>ok</strong> for underlying groupings<br />

of data, to report effect sizes)<br />

the kinds of data with which one is working<br />

(parametric and non-parametric)<br />

the scales of data being used (nominal, ordinal,<br />

interval, ratio)<br />

the number of groups in the sample<br />

the assumptions in the tests<br />

whether the samples are independent of each<br />

other or related to each other.<br />

Researchers wishing to use statistics will need to<br />

ask questions such as:<br />

<br />

<br />

<br />

<br />

<br />

What statistics do I need to answer my research<br />

questions<br />

Are the data parametric or non-parametric<br />

How many groups are there (e.g. two, three or<br />

more)<br />

Are the groups related or independent<br />

What kind of test do I need (e.g. a difference<br />

test, a correlation, factor analysis, regression)<br />

We have addressed several of these points in<br />

the preceding chapters; those not addressed in<br />

previous chapters are addressed here. In this<br />

chapter we draw together the threads of the<br />

discussion of statistical analysis and address what,<br />

for many researchers, can be a nightmare: deciding<br />

which statistical tests to use. In the interests<br />

of clarity we have decided to use tables and<br />

graphic means of presenting the issues in this<br />

chapter (see http://www.routledge.com/textbo<strong>ok</strong>s/<br />

9780415368780 – Chapter 26, file 26.1.doc).<br />

How many samples<br />

In addition to the scale of data being used<br />

(nominal, ordinal, interval, ratio), the kind of<br />

statistic that one calculates depends in part<br />

on, first, whether the samples are related to,<br />

or independent of, each other, and second, the<br />

number of samples in the test. With regard to<br />

the first point, as we have seen in previous<br />

chapters, different statistics are sometimes used<br />

when groups are related to each other and when<br />

they are independent of each other. Groups will<br />

be independent when they have no relationship<br />

to each other, e.g. in conducting a test to see if<br />

there is any difference between the voting of males<br />

and females on a particular item, say mathematics<br />

performance. The tests that one could use here are,<br />

for example: the chi-square test (for nominal data),<br />

the Mann-Whitney U test and Kruskal-Wallis<br />

(for ordinal data), and the t-test and analysis of<br />

variance (ANOVA) for interval and ratio data.<br />

However, there are times when the groups<br />

might be related. For example, we may wish to<br />

measure the performance of the same group at<br />

two points in time – before and after a particular<br />

intervention – or we may wish to measure the<br />

voting of the same group on two different factors,<br />

say preference for mathematics and preference for<br />

music. Here it is not different groups that are being<br />

involved, but the same group on two occasions and<br />

the same two on two variables respectively. In this<br />

case different statistics would have to be used, for<br />

example the Wilcoxon test, the Friedman test, the<br />

t-test for paired samples, and the sign test. Let us

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!