|







| |
 Our
bureau service has a set of comprehensive standard checks ... Our comprehensive bureau
service includes the following checks, which we undertake as standard on all
of the surveys we process (where appropriate) - ensuring that your results are
always as accurate as possible! QUESTIONNAIRE
DESIGN
 |
If we set up your questionnaire
(creating the questions definitions file .QDF for you) the content is always
independently checked. The "checker" pays particular
attention to question text, numbering, spelling and routing. |
 |
Once the questionnaire file has
been checked, the data entry screens themselves are examined to ensure that
all entries and filters (skips) are correct and any on-screen instructions
are easy to follow. |
PAPER
QUESTIONNAIRES (PREPARATION)
 |
Before data is entered from paper
questionnaires, they are checked, sorted and a unique serial number is
allocated to each. |
 |
Any printed questionnaire routing is also
checked at this stage. |
CODING
 |
If required, the open ended
questions (verbatim texts) included on your questionnaires are coded.
We are happy to work with supplied code lists, those from previous waves, or
to generate new lists, as required. Whenever appropriate, new code lists are
created using the 5 bar gate
(positive/neutral/negative) method. |
 |
Code lists are always available
for client approval. |
 |
Whenever time scales allow, the QPSMR
questionnaire file is then updated before data entry, so that the coded
response texts will be displayed at the screen when responses are selected. |
 |
10% of the coding undertaken is
quality checked, as standard. |
DATA ENTRY AND
VERIFICATION
 |
Data from your questionnaires is
then entered using the QPSMR Input program, following filters (skips) and
using "hold" screens. Only valid responses are available for
selection, ensuring your data is as "clean" as possible. |
 |
We randomly "spot
check" paper questionnaires against the data entered and 10% of all of
the work returned from each member of our data entry team is independently
verified. |
CHECKING YOUR INFORMATION
 |
When data has been entered a
topline report is produced. This shows summary counts of all of your
questionnaire information, including the answers which have been allocated
to reject categories. This report is then thoroughly
interrogated to ensure all entries based on the total sample add to the
correct number of respondents, and all filtered entries add to the
appropriate filtered sample. |
 | Any unused coded responses are
located (by referring back to the paper questionnaires and code lists) and
the information is updated accordingly. Also, if data has been entered
for an open ended response, which has not had text allocated to it, this is
fully investigated and modified as required. |
 |
All blank responses are checked
against the completed paper questionnaires to ensure that the respondent did
in fact leave the question empty. If the question was blank because of
a punching error, the response is updated and the entire questionnaire is
verified. |
 |
Frequency counts are produced for
all integer and weight numeric entries. This ensures that the
responses entered are appropriate for the question being asked. For
example, if a time is collected as a 24 hour clock, we would check that all
answers are acceptable - a time of 25.13 would obviously be incorrect!
Any unacceptable answers would be located and corrected, and once again the
entire questionnaire would be verified. |
 |
If appropriate, cross tables are
also generated to further check the information collected. For
example, we might run "Day by Date", or "Promotion by
Store" tables to ensure that the data has been collected and captured
correctly. |
 |
If automatic range checks were
not set for all numeric entries when the questionnaire was defined,
these entries are validated to your specification (for example, an
"amount spent" question, for which only values between £5 and
£12 are acceptable responses). |
ANALYSIS
 |
When producing tables, the
results for all column and row entries are checked to ensure that they add
to the correct filtered or unfiltered total. |
 | Printed score values (often
applied to rating scale responses in order to generate mean scores and other
statistics) are always checked to ensure the value is appropriate, and
correct, for each response. |
 |
The definitions for complex
variables (combinations of many existing entries) are always independently
checked to ensure that they include all of the required respondents. |
 |
If table arithmetic is used, or
"banked" tables are created, these are always thoroughly checked
to ensure calculations are accurate and that appropriate columns are
accumulated (overlaid). |
 |
The tables file itself is always
checked to ensure that all requested tables are present. |
EXPORTED DATA AND TABLES
 |
If you require QPSMR data or tables
in a format suitable for use with another program, whenever possible the
exported information is loaded into the appropriate program to ensure that
the transfer process, for you or your client, is swift and trouble free. |
If you would like us to help with any part of
your data processing, simply request a quotation.
|