Printable View
AnyOne has any input on how to design a database which is based on input from
survery/questionaire. It is essentially data on arthritis patients who will
fill these surveys at least twice a year. Data will be used to do statistical
analysis using Stata.
This is my second post, no help on the first one. We are a non-profit organization
trying to make our ends meet and this redeishing of our database would greatly
help us.
Anybody's experience is helpful to me in this regard.
Hassan,
How many questins are on your questionaire? What other demographic information
is being collected. Are the questions singel choice from several possible
answers? This should be very straightforward, but there are always possible
quirks.
Arthur Wood
"Hassan Mir" <hassan@arthritis-research.org> wrote:
>
>AnyOne has any input on how to design a database which is based on input
from
>survery/questionaire. It is essentially data on arthritis patients who will
>fill these surveys at least twice a year. Data will be used to do statistical
>analysis using Stata.
>
>This is my second post, no help on the first one. We are a non-profit organization
>trying to make our ends meet and this redeishing of our database would greatly
>help us.
>
>
>Anybody's experience is helpful to me in this regard.
>
What is your time frame? What database are you using? What technical expertise
do you require? Are you a techie who just needs some help with the database?
I can help if the time frame is reasonable. Send me an email.
"Hassan Mir" <hassan@arthritis-research.org> wrote:
>
>AnyOne has any input on how to design a database which is based on input
from
>survery/questionaire. It is essentially data on arthritis patients who will
>fill these surveys at least twice a year. Data will be used to do statistical
>analysis using Stata.
>
>This is my second post, no help on the first one. We are a non-profit organization
>trying to make our ends meet and this redeishing of our database would greatly
>help us.
>
>
>Anybody's experience is helpful to me in this regard.
>
Since you will have several questionaires over time and they may vary in
content consider -
Questionaire(id) was released on Date
Questionaire(id) has Title
Question(id) was placed on Questionaire in Position
Question has Question Text
Answer Set is associated with Question
Answer Set includes Response in Position
Response has Response Meaning
Patient answered Question on Questionaire with Response.
Patient replied on Date
Title 1/1/2000
Date Taken _____ 1 2 3 4 5
1 Are you feeling OK today?
2. How well are you taking your Medication?
Title is the questionaire title
1/1/2000 is the release date
"Are you feeling OK today" is question 1.
Remember question 1 is merely an ID for the question.
Question just happens to be in position 1 on this
questionaire. It could be question lsoition 4 on some
other questionaire. Questions are a set of questions.
they are then associated to a questionaire.
Taking (1 2 3 4 5 ) this is the answer set for the question.
This same answer set can be used for may questions.
1 = Very Good
2 = Good
3 = Average
4 = not soo good
5 = bad
This is meaning of each response.
There is a set of answer sets. Examples:
(1 2 3 4 5) (y n) (1,2,3), etc.
Design the system such that:
There is list is questions that can be assigned to any
questionaire.
Q question has exactly one answer set. If the wording is the
same and the answer set is different then it is a differet
question.
each response in an answer set has a meaning.
Each question is in some position on the questionaire.
Each response is in some position (1,2,3,4,5).
The partient response:
Patient responded to Questionaire on Date
Key (PAtient, Questionaire, Date-time) in case they can take
it twice on the same day. Key is all rows.
Assoicate to that
Patient Questionaire, Question, Date-Time, Response.
From there is is a matter of stats.
Pat
"Hassan Mir" <hassan@arthritis-research.org> wrote:
>
>AnyOne has any input on how to design a database which is based on input
from
>survery/questionaire. It is essentially data on arthritis patients who will
>fill these surveys at least twice a year. Data will be used to do statistical
>analysis using Stata.
>
>This is my second post, no help on the first one. We are a non-profit organization
>trying to make our ends meet and this redeishing of our database would greatly
>help us.
>
Is the data from the survey numeric; i.e., does the patient rank "Pain Intensity"
from 1-5? Or is this a short answer survey? Numeric data can be scanned
or keyed in under headings (e.g. "Pain", "Range of Motion", etc.)for stat
analysis. Short answer surveys are difficult to quantify; and so don't extrapolate
well into stats.
>
>Anybody's experience is helpful to me in this regard.
>
This is just a loose set of ideas - I strongly recommend working with an experienced
data architect or database designer - but if you can't find the time or $
to do so, the following ideas might help:
Table 1: Questionnaire info. Columns would include questionnaire release
date, number of questionnaires sent, number of responses received, number
of questions in the questionnaire, and other information that is normalized
at the questionnaire level. The key would be a unique sequence number, and
you could implement secondary indices that prevent duplication by release
date, or some other information.
Table 2: Question info. Each question would have a sequence number in alignment
with the number of the question as issued in the questionnaire. You would
also have a column called question type, which would be a foreign key from
a supporting table of question types (text, indicator (y/n, m/f, etc...),
range (1-5), etc...). This allows you the flexibility to handle any sort
of question.
Table 3: Question text. If the question required a text response, you store
the text here, keyed off the question info keys.
Table 4: Question indicator. This table contains the selected value, and
the other allowed values.
Table 5: Question range. This table contains the selected value, and the
other allowed values.
Table 6: Question type. This table contains the allowable types.
This is just a five-minute overview. This structure can and should be significantly
refined, but at least it gives you a place to start thinking about how to
tackle this.
I'm going for an MBA specializing in IT management, and had to do a questionnaire
that required this sort of analysis, and with the exception of questionnaire
versions, this is how I tackled it. I do not think that this is the only
or the best solution, it's just my 5-minute solution.
Good luck.
Sharon
