12.01.2015 Views

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

342 QUESTIONNAIRES<br />

The above outline describes a particular kind<br />

of pilot: one that does not focus on data, but on<br />

matters of coverage and format, gaining feedback<br />

from a limited number of respondents and experts<br />

on the items set out above.<br />

There is a second type of pilot. This is one<br />

which starts with a long list of items and, through<br />

statistical analysis and feedback, reduces those<br />

items (Kgaile and Morrison 2006). For example,<br />

a researcher may generate an initial list of,<br />

for example, 120 items to be included in a<br />

questionnaire, and wish to know which items<br />

to excise. A pilot is conducted on a sizeable<br />

and representative number of respondents (e.g.<br />

50–100) and this generates real data – numerical<br />

responses. These data can be analysed for the<br />

following factors:<br />

<br />

<br />

<br />

<br />

Reliability: those items with low reliability<br />

(Cronbach’s alpha for internal consistency: see<br />

Part Five) can be removed.<br />

Collinearity: if items correlate very strongly<br />

with others then a decision can be taken<br />

to remove one or more of them, provided,<br />

of course, that this does not result in the<br />

loss of important areas of the research (i.e.<br />

human judgement would have to prevail over<br />

statistical analysis).<br />

Multiple regression: those items with low betas<br />

(see Part Five) can be removed, provided,<br />

of course, that this does not result in the<br />

loss of important areas of the research (i.e.<br />

human judgement would have to prevail over<br />

statistical analysis).<br />

Factor analysis: to identify clusters of key<br />

variables and to identify redundant items (see<br />

Part Five).<br />

As a result of such analysis, the items for<br />

removal can be identified, and this can result<br />

in a questionnaire of manageable proportions. It is<br />

important to have a good-sized and representative<br />

sample here in order to generate reliable data<br />

for statistical analysis; too few respondents to this<br />

type of pilot and this may result in important items<br />

being excluded from the final questionnaire.<br />

Practical considerations in questionnaire<br />

design<br />

Taking the issues discussed so far in questionnaire<br />

design, a range of practical implications for<br />

designing a questionnaire can be highlighted:<br />

Operationalize the purposes of the questionnaire<br />

carefully.<br />

Be prepared to have a pre-pilot to generate<br />

items for a pilot questionnaire, and then be<br />

ready to modify the pilot questionnaire for the<br />

final version.<br />

If the pilot includes many items, and the<br />

intention is to reduce the number of items<br />

through statistical analysis or feedback, then<br />

be prepared to have a second round of piloting,<br />

after the first pilot has been modified.<br />

Decide on the most appropriate type of<br />

question – dichotomous, multiple choice, rank<br />

orderings, rating scales, constant sum, ratio,<br />

closed, open.<br />

Ensure that every issue has been explored<br />

exhaustively and comprehensively; decide on<br />

the content and explore it in depth and<br />

breadth.<br />

Use several items to measure a specific<br />

attribute, concept or issue.<br />

Ensure that the data acquired will answer the<br />

research questions.<br />

Ask more closed than open questions for ease<br />

of analysis (particularly in a large sample).<br />

Balance comprehensiveness and exhaustive<br />

coverage of issues with the demotivating factor<br />

of having respondents complete several pages<br />

of a questionnaire.<br />

Ask only one thing at a time in a question. Use<br />

single sentences per item wherever possible.<br />

Keep response categories simple.<br />

Avoid jargon.<br />

Keep statements in the present tense wherever<br />

possible.<br />

Strive to be unambiguous and clear in the<br />

wording.<br />

Be simple, clear and brief wherever possible.<br />

Clarify the kinds of responses required in open<br />

questions.

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

Saved successfully!

Ooh no, something went wrong!