Data
Surveys
Clearly define the problem
Decide what statistics are needed
type of questions needed
what type of scale
4-100 point
Feelings/Perceptions
As big a scale as possible
truer average for a 10 point scale then a 4 point scale
Sample Survey
representative part of a whole group
inferential statistics
unbiased sampling techniques
get as large a population as possible
smallest group hsould have is 30 in a survey
disadvantages
risk
insufficient sample size
lack of representatives
means error
Contents of a Survey
Cover Letter
Purpose - why to complete
How it will be important to the person
IRB
Who is sponsoring the survey
Confidentiality of results
Instructions
What to do if you don't know an answer
don't skip question
Questions (survey)
Follow-up with non-responders
Data Analysis
Interval Data
Ordinal data
descriptive - report frequencies (or %
Satifisfaction data
Nominal Data
descriptive - report frequencies (or %
age groups, male/female (no way to find out mean of group
Ratio Analysis is the 4th Indexes???
response rate
Subtopic
sent out divided by # sent back
random Sample
contains least amount of sampling error
it is an unbiased sample
2 conditions
each person in population has a chance to be chosen
the selection of any member must have no influence on any other selection
Stratified Random Sampling
combination of independent samples
homogeneous groups within heterogeneous population
superior to a simple random
divided into smaller homo groups before sample
less variation
Non-random Sample
every nth person is chosen
nonrandom
produce larger sampling errors
Systematic
primarily for ease and speed of identifying participants
choose the kth member in the pop.
where K is equal to the population size divided by the required sample size
once started-selection is nonrandom
Judgement/Puposive Sampling
ask expert to define members
Types of Questions
background questions
demographics
Closed
lists possible answers
determine feelings/opinions
Open-ended
respondents write the answer out in their own words
divise a method for categorizing
infinite number of possible answers
construct categories for each possible answer in advance
Comment Boxes
qualitative
gets at the why?
Humanize the data
Intensity
7 point scale
give a neutral
I don't know
I have no judgement
Tips
Start with general and then go more indepth
ask easier quetsions first