Transcript Steps in a proper study

Survey methods 1
Methods of study





History
Content analysis
Experimental design
Focus group
Survey
2
Survey

Is defined as a "means for gathering
information about the characteristics,
actions, or opinions of a large group of
people." (AIO)

Surveys can also be used to assess needs,
evaluate demand, and examine impact.
3
Steps in a proper study


A problem statement or a goal
Set a measurable objective for your survey
(See http://www.quorum-pr.com/Risorse/probjectives.pdf)




Define the population
Define the sampling choice
Determine the method for administering
Develop an opening statement: the question
of anonymity versus confidentiality.
4
Steps in a proper study …






Structuring the survey
Institutional approval
Pre-test
Corrections
Code sheet and software considerations
Training the interviewers
5
Steps in a proper study …





Conducting the interviews
Verification as necessary
Data entry and debugging
Initial data presentation
Interpreting the data, in context
6
Types of data are related to stats

Nominal: categorical, only by type.

Ordinal: Rank order.

Interval: Rating scale of 0 to 100, arbitrary
divisions.

Ratio: Percentages and true zero.
7
Benchmark or?

Benchmark: a moment in time

Longitudinal: at intervals over time
8
Population

The group that you are going to study.

The population should be tightly defined.

Think “population frame.”
9
Sample

Non probability sample – a respondent is
chosen at a place of convenience.

Probability sample – a respondent has an
equal opportunity of being in the survey.
This sample is drawn typically from a list.
10
Convenience sample

If you wanted to do a study of what
shoppers thought about Global Village, you
would choose folks shopping at Global
Village.
11
Probability sample

Everyone has an equal chance of being
chosen as a respondent.

Folks are chosen as an Nth sample from a
complete list.

Folks could be chosen proportionately from
a list, say Freshman, Sophomores, Juniors,
and Seniors.
12
Nth sample

400 respondents from a population of 4,000

Divide 400 into 4,000. The answer is 10.

Pick a number at random between 1 and 10.
See http://www.programurl.com/software/random-number-generator.htm.

Say that number is 7. The first respondent is
7, the second is 17, the third is 27 and so on.
13
Sample size

Sample size depends upon time, money, and
desired error rate.

The error rate is 1/square root of the sample
size.
14
Sample error rate

A sample of 400 produces an error rate of

1/square root of 400 is 20.

1 divided by 10 is .05

The error rate is + or – 5 percent.
15
Sample error rate

A sample of 1,600 would produce an error rate of +
or – 2.5%.

The square root of 1,600 is 40.

The error rate is 1/40.

The error rate is + or – 2.5%.
Confused see, http://www.robertniles.com/stats/mean.shtml.
16
Population sizes and sample size

Population
+/- 3%

200
300
400
500
100
150
200
250



If you want the table for this, see Frey, L.R., Botan, C., & Kreps, G.L. (1999).
Investigating Communication. You need this list for populations to about 1,000.
17
Cooperation rate

This is the percentage of those who agree to
participate.

The cooperation rate varies from 75% to 25%
depending upon the method used, survey
fatigue, and how much money that you have.

This means you might have to choose more.
18
Type of question format

Tunnel: similarly organized questions

Funnel: begins with broad open questions
followed by narrower closed questions

Inverted funnel: begins with narrow, closed
questions and builds to broader, open
questions.
19
Arranging the format

Treat all questions as a whole.

Group questions logically.

Place easy to answer questions at the end
because of respondent fatigue.
20
Arranging the format

Place sensitive questions in the middle, after
the respondent is comfortable and before
she is tired.

Six or more similar items cause fatigue. Vary
what you ask.
21
Wording

Use complete sentences.

Write questions in conversation tones.

Underline, italicize, or bold important items.
22
Wording

Ask about a limited time period, specify it.
Yesterday, week, month, but don’t make it a
memory test.

Use other, but sparingly.

Be careful with letting someone answer
don’t know.
23
Wording…

Use standard demographic questions.

Avoid all inclusive terms such as always or
never.
24
Wording …

If you use opened-ended questions ask the
positive ones first, the negative ones second.
(What do you like best, what do you like
least.)

Don’t embarrass. Ask “What is the highest
school level you completed,” rather than
“Did you graduate from College?”
25
Good survey questions are

Straightforward

Ask about only one issue (if it has an and
break it up)

Don’t load questions. “Don’t ask, Don’t you
think that…”
26
Demographic block





Age
Educational level
Gender
Occupation
Place of residence
27
Educational categories







Does not have any formal education
Some elementary education
Finished elementary education
Finished secondary, did not attend
university or college
Graduated from college
Some post-graduate education
Post-graduate education
28
Categorical questions


Should have equal groupings
What is your age?
Less than 10
 10-19
 20-29
 30-39
 40-49


Or see, http://factfinder.census.gov/servlet/QTTable?_bm=y&-geo_id=01000US&qr_name=DEC_2000_SF1_U_QTP12&-ds_name=DEC_2000_SF1_U.
29
Cultural concerns

Symbolic words are abstract. They do not
mean the same time to all people

Abstract: love, honor, be safe, dignity

Concrete: table, chair, love, house?
30
Symbol



I love you.
I love broccoli.
I love this game.
31
Semantic rules

I will call you soon.

Value
Potency
Activity




good, bad
strong, weak
fast, slow
Low-context culture
High-context culture
good, strong, fast
good, weak, slow
32
Likert-scales

Gauge the degree to which there is
agreement or disagreement

5, 7, 9, 11 point scales, depending upon
sensitivity and number of respondents
33
Likert scale sample

Circle one answer for each question

I think research is exciting

SA
A
N or DNA
D
SD
But consider “forced choice.”
34
Or, rephrase the question

0
50
100
150

Respondents report a number to rate
something.
35
Thurstone scales (rocket science)





Place a check mark in the blank next to each
statement with which you agree.
__ This class is more challenging than the
other course I am taking.
__This class teaches me valuable
information.
__This class is what I expected.
__This class is interesting.
36
Semantic differential scales

Rate the concept, “the president” according
to the way you feel about it by placing an X
on each of the seven point scales to indicate
your evaluation.
37
The scale
The president
Reliable _:__: __:__:__:__:__Unreliable*
Uninformed _:__: __:__:__:__:__Informed
Unqualified _:__: __:__:__:__:__Qualified
Dishonest _:__: __:__:__:__:__Honest
Intelligent _:__: __:__:__:__:__Unintelligent*
* Indicates reverse scoring
38
Guttmann scale
Consider a person who is Chinese, and circle
the number next to each of the situations
listed below in which you would accept a
person from this group.
39
The scale
1.
2.
3.
4.
5.
6.
As a visitor to my country
To citizenship in my country
To employment in my occupation
To my street as my neighbor
To my club as a personal friend
To close kinship by marriage
Bogardus Social Distance Scale
40
If you are stuck, look at

Rubin, R.B., Palmgreen, P. & Sypher, H.E.
(2004). Communication Research Measures:
A Sourcebook. Mahwah, New Jersey:
Lawrence Erlbaum Associations.
41
Presenting the data

Asking the questions so that you get what
you want.

Statistics for People who (Think They) Hate
Statistics
42