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.

328 QUESTIONNAIRES<br />

(Circle one number)<br />

1 2 3 4 5 6<br />

Strongly − − − − − − Strongly<br />

agree<br />

disagree<br />

Let us say that one member of staff circled 1,<br />

eight staff circled 2, twelve staff circled 3, nine<br />

staff circled 4, two staff circled 5, and seven<br />

staff circled 6. There being no mid-point on this<br />

continuum, the researcher could infer that those<br />

respondents who circled 1, 2 or 3 were in some<br />

measure of agreement, while those respondents<br />

who circled 4, 5 or 6 were in some measure of<br />

disagreement. That would be very useful for, say,<br />

a headteacher, in publicly displaying agreement,<br />

there being twenty-one staff (1 + 8 + 12) agreeing<br />

with the statement and eighteen (9 + 2 + 7)<br />

displaying a measure of disagreement. However,<br />

one could point out that the measure of ‘strongly<br />

disagree’ attracted seven staff – a very strong<br />

feeling – which was not true for the ‘strongly agree’<br />

category, which attracted only one member of staff.<br />

The extremity of the voting has been lost in a crude<br />

aggregation.<br />

Further, if the researcher were to aggregate the<br />

scoring around the two mid-point categories (3<br />

and 4) there would be twenty-one members of staff<br />

represented, leaving nine (1 + 8) from categories<br />

1and2andnine(2+ 7) from categories 5 and<br />

6; adding together categories 1, 2, 5 and 6, a<br />

total of eighteen is reached, which is less than the<br />

twenty-one total of the two categories 3 and 4. It<br />

seems on this scenario that it is far from clear that<br />

there was agreement with the statement from the<br />

staff; indeed taking the high incidence of ‘strongly<br />

disagree’, it could be argued that those staff who<br />

were perhaps ambivalent (categories 3 and 4),<br />

coupled with those who registered a ‘strongly<br />

disagree’, indicate not agreement but disagreement<br />

with the statement.<br />

The interpretation of data has to be handled<br />

very carefully; ordering data to suit a researcher’s<br />

own purposes might be very alluring but quite<br />

illegitimate. The golden rule here is that crude data<br />

can yield only crude interpretation; subtle statistics<br />

require subtle data. The interpretation of data must<br />

not distort the data unfairly. Rating scale questions<br />

are treated as ordinal data (see Part Five), using<br />

modal scores and non-parametric data analysis,<br />

though one can find very many examples where<br />

this rule has been violated, and non-parametric<br />

data have been treated as parametric data. This is<br />

unacceptable.<br />

It has been suggested that the attraction of<br />

rating scales is that they provide more opportunity<br />

than dichotomous questions for rendering data<br />

more sensitive and responsive to respondents. This<br />

makes rating scales particularly useful for tapping<br />

attitudes, perceptions and opinions. The need for a<br />

pilot study to devise and refine categories, making<br />

them exhaustive and discrete, has been suggested<br />

as a necessary part of this type of data collection.<br />

Questionnaires that are going to yield numerical<br />

or word-based data can be analysed using computer<br />

programmes (for example SPSS or Ethnograph<br />

respectively). If the researcher intends to process<br />

the data using a computer package it is essential<br />

that the layout and coding system of the<br />

questionnaire are appropriate for that particular<br />

computer package. Instructions for layout in order<br />

to facilitate data entry are contained in manuals<br />

that accompany such packages.<br />

Rating scales are more sensitive instruments<br />

than dichotomous scales. Nevertheless, they are<br />

limited in their usefulness to researchers by their<br />

fixity of response caused by the need to select from<br />

agivenchoice.Aquestionnairemightbetailored<br />

even more to respondents by including open-ended<br />

questions to which they can reply in their own<br />

terms and own opinions. We consider these later.<br />

Constant sum questions<br />

In this type of question respondents are asked to<br />

distribute a given number of marks (points) between<br />

a range of items (see http://www.routledge.<br />

com/textbo<strong>ok</strong>s/9780415368780 – Chapter 15, file<br />

15.8. ppt). For example:<br />

Please distribute a total of 10 points among the<br />

sentences that you think most closely describe your<br />

behaviour. You may distribute these freely: they may<br />

be spread out, or awarded to only a few statements,<br />

or all allocated to a single sentence if you wish.

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

Saved successfully!

Ooh no, something went wrong!