| As you recall from the
Data Collection section of The Power of
Proof, quantitative collection techniques use numerical
data to make calculations and draw conclusions in terms
of percentages, proportions, and other values. These
data are more definitive than qualitative data, thus
often easier to organize and analyze.
Here are a few simple steps to begin your analysis
of quantitative data.
- Describe the responding persons, organizations,
or communities using frequencies for each demographic
or other descriptive item of information.
-
Be sure to give the total number
(n=) for each descriptive item. This is important,
especially if some people didn’t answer
some questions so that the numbers differ from
question to question.
-
Consider reporting the range
for each descriptive item (e.g., the youngest
participant was “18 years old” and
the oldest was “67 years old”)
-
You may want to use percentages
as well as counts.
-
If appropriate, describe the
participants and the comparison group separately.
- For questions whose answers are reported as rating
scales or rankings (e.g., strongly disagree
= 1, strongly agree = 5) consider computing
a mean, or average, for each question.
Example: Suppose 40 people
responded to an Attitude item. Let’s
say that one person responded “1” (strongly
disagree), three people responded “2”
(disagree), eight people responded “3”
(neither agree nor disagree), 17 people
responded “4” (agree), and
11 people responded “5” (strongly
agree). The mean (average) score would be (1x1)+(3x2)+(8x3)+(17x4)+(11x5)
divided by 40. This is equal to 154/40, or 3.85.
Thus, on average, people were closest to the response
“agree”.
- If a series of questions go together (e.g., a set
of attitude questions), consider making a scale
out of these items and computing one score for the
entire scale.
Example: Suppose we have
ten attitude questions, each with a possible responses
of “1” (strongly disagree),
“2” (disagree), “3”
(neither agree nor disagree), “4”
(agree), and “5” (strongly
agree). If Representative Smith’s responses
to the ten questions were: agree (4), agree
(4), strongly agree (5), agree
(4), disagree (2), agree (4),
agree (4), disagree (2), strongly
agree (5), neither agree nor disagree
(3), then her total score would be:4+4+5+4+2+4+4+2+5+3
= 37. This could also be expressed as an average
response (37/10 questions = 3.7).
- For outputs or outcomes expressed in categories
(e.g., attended/did not attend; smoke/don’t
smoke), consider calculating proportions,
rates, or ratios.
- Proportion: A proportion
is a part of the whole.
Example: Suppose
18 legislators of the 25 who were visited by
volunteers from your program voted for the smoke-free
environment legislation. The proportion would
be 18/25 = .72. Expressed as a percentage, this
would be .72x100=72%
-
Rate: A rate
is a special type of proportion. It has a specific
time period associated with it, and it is expressed
in standard units (e.g., per 100, per 1,000, per
100,000).
Example: Suppose
you polled 150 legislators at the end of September,
and 102 of them intended to support the smoke-free
environment legislation. Suppose you polled
them again at the end of October, and 138 of
them intended to support the legislation.
Incidence: The incidence rate is new
instances of the outcome of interest in
a specified period of time divided by the
population of interest, expressed per standard
unit of population. In this example, the incidence
of new support per 100 legislators
during the month of October is (138-102)/150
x 100, or 24 per 100.
Point Prevalence. The point prevalence
rate is all instances of the outcome
of interest at a specified point in time
divided by the population of interest, expressed
per standard unit of population. In this example,
the prevalence of support at the end of September
is 102/150 x 100, or 68 per 100. The prevalence
of support at the end of October is 138/150
x 100, or 92 per 100.
-
Ratio: A ratio
is a mathematical way to compare two numbers.
If we want to compare a to b,
we calculate the ratio by dividing a (the number
before the word “to”), by
b (the number after the word “to").
Example:
Continuing the example above, at the end of
October, the ratio of legislators who supported
the legislation (138) to those who did not (150-138
= 12) is 138/12, or 11.5, usually reported 11.5
to 1. This means that, at the end of October,
for every legislator who did not support the
legislation there were 11.5 who did.
- If your evaluation questions ask about the change
in an output or outcome, subtract the value
at the beginning of the program from the value at
the end of the program. For example, if you
want to know how much the Attitude score increased,
subtract the Attitude score at pretest from the Attitude
score at posttest.
Back to Analyze the Data section
---------------
Bibliography
Source: Frechtling, J., Sharp, L., &
Westat, eds. (1997). Analyzing qualitative data. In:
User-friendly handbook for mixed method evaluations.
Arlington, VA: National Science Foundation, Directorate
for Education and Human Resource.
Source: Health Canada. (1996). Analysing
and interpreting data. In: Guide to project evaluation:
A participatory approach. Ottawa: Minister of Health
& Welfare Canada.
Source: McNamara, C. (1999). Analyzing,
interpreting and reporting basic research results.
Retrieved July 21, 2004 from The Management Assistance
Program for Nonprofits web site:
http://www.mapnp.org/library/research/analyze.htm |
