RESEARCH METHOD COHEN ok

506 QUANTITATIVE DATA ANALYSIS
structure of a questionnaire survey very carefully
in order to assist data entry for computer reading
and analysis; an inappropriate layout may obstruct
data entry and subsequent analysis by computer.
The planning of data analysis will need to consider:
What needs to be done with the data when
they have been collected – how will they be
processed and analysed
How will the results of the analysis be verified,
cross-checked and validated
Decisions will need to be taken with regard to
the statistical tests that will be used in data
analysis as this will affect the layout of research
items (for example in a questionnaire), and the
computer packages that are available for processing
quantitative and qualitative data, e.g. SPSS and
NUD.IST respectively.
Reliability
We need to know how reliable is our instrument
for data collection. Reliability in quantitative
analysis takes two main forms, both of which
are measures of internal consistency: the split-half
technique and the alpha coefficient. Both calculate
acoefficientofreliabilitythatcanliebetween0
and 1. The split-half reliability has been discussed
in an earlier chapter. The formula given is:
r =
2r
1 + r
where r = the actual correlation between the
halves of the instrument (this requires the
instrument to be able to be divided into two
matched halves in terms of content and difficulty).
So, for example, if the correlation coefficient
between the two halves is 0.85 then the formula
would be worked out thus:
r = 2(0.85)
1 + 0.85 = 1.70
1.85 = 0.919
Hence the split-half reliability coefficient is 0.919,
which is very high. SPSS automatically calculates
split-half reliability at the click of a button.
An alternative calculation of reliability as
internal consistency can be found in Cronbach’s
alpha, frequently referred to simply as the alpha
coefficient of reliability. The Cronbach alpha
provides a coefficient of inter-item correlations,
that is, the correlation of each item with the sum of
all the other items. This is a measure of the internal
consistency among the items (not, for example, the
people). It is the average correlation among all the
items in question, and is used for multi-item scales.
SPSS calculates Cronbach’s alpha at the click of a
button; the formula for alpha is:
nr ii
alpha =
1 + (n − 1)r ii
where n = the number of items in the test or
survey (e.g. questionnaire) and r ii = the average
of all the inter-item correlations. Let us imagine
that the number of items in the survey is ten, and
that the average correlation is 0.738. The alpha
correlation can be calculated thus:
alpha =
nr ii
10(.738)
=
1 + (n − 1)r ii 1 + (10 − 1).738
= 7.38
7.64 = 0.97
This yields an alpha coefficient of 0.97, which
is very high. The alpha coefficients are set out in
Table 1 of the Appendices of Statistical Tables. For
the split-half coefficient and the alpha coefficient
the following guidelines can be used:
>0.90 very highly reliable
0.80–0.90 highly reliable
0.70–0.79 reliable
0.60–0.69 marginally/minimally reliable