Services

​

 

 

What we do

 

CodingExperts turns your unstructured qualitative survey data into insightful, actionable management information.  Partnering with suppliers who have decades of research experience, we can offer turnkey research from inception through reporting.  In addition to coding, we are proud to offer the following services:

 

• ​Questionnaire design

​

• Column coding paper questionnaires, data entry and verbatim capture

​

• Sampling; access to over 30 million worldwide panelists

​

• Online survey platform and programming including advanced research techniques such as conjoints, sophisticated quota logic,        segmentation algorithms, along with extensive rotation and randomization options.  We have customizable templates to personalize the look and feel of your survey, and respondent-friendly presentations with interactive questions using card sorts, sliders, button grids, concept clickers, virtual shelves and more

​

• Online, real-time reporting with customized reports and the ability to download your survey data 24/7

​

• Crosstabs in IBM SPSS Quantum with deliverables in Word and/or Excel

​

• Advanced statistical testing

​

• Pivot tables

​

• Charts and graphs in Word, Excel and PowerPoint

 

Experience

 

• We provide coding services in many market segments including consumer research, healthcare, pharmaceutical/biotech, naming/branding, customer/employee satisfaction, events/professional organizations, education, high tech etc. 

​​

• CodingExperts has completed over 2,000 medical studies in the past 11 years (5,000+ total studies), so we are well-versed in the field.  As a third-party resource, we have done a substantial amount of work for GSK, Pfizer, Merck, Abbott, Roche and Boehringer Ingelheim among other drug manufacturers.

 

• Each coder has been trained in the nuances of medical coding and all senior coders have a minimum of five years of experience within the pharmaceutical industry.  They have also studied training materials and passed standard manufacturer pharmacological tests.

 

• We offer multi-language capabilities which include in-language coding as well as translations. We offer full European language compatibility, as well as Chinese, Japanese, Korean and more.

 

 

Coding Platform

 

• We use Language Logic's Ascribe™ coding software for our projects. The system provides tools to help code more accurately and gives us the ability to change and adapt the codebooks and coding throughout the life of the study.

 

• Ascribe offers several AI/automated coding options; however, unless requested, all coding is done manually by human coders. 

 

• Appropriate situations for AI coding include projects where a simple overview or topline reporting is all that is necessary, large data sets that need to be processed within hours and as a cost-saving measure for studies where the budget does not allow for manual coding.

 

 

Processes

 

Codebuilding

 

• Open ended coding conveys the voice of the customer in a way that is not possible with close ended questions.  We create hierarchical codeframes structured in logical fashion to provide a “360° view” of the respondents’ thoughts and feelings.

 

• A minimum of 25% of the data is sampled in order to build preliminary codeframes.  Exact percentages are based on quotas, available data and delivery requirements. 

 

• Initial codes are developed based on an understanding of the study objectives provided through client information, the questionnaire, stimuli used, and most importantly, the actual verbatim responses. 

 

 Coding

 

• The coder(s) are briefed on each new study and supplied with all relevant client and study information and documents, including related close-ended questions for context where applicable.

 

• Excluding extenuating circumstances such as timing, size/scope of study and in-language/translation considerations, standard procedure is for a single coder to code a study in its entirety.  This aids in familiarity with all materials, instructions, codebooks, consistency in wording of new codes, application of codes and a reduction in subjective bias.  

 

• The coder adds and modifies codeframes while coding, so while initial codes are set on a percentage of the data, the final codes are built on 100% of the verbatim responses. This allows us to provide the most complete and comprehensive picture possible.

 

• In general, we code the data and then submit codes for approval.  If specifically requested or in the case of unusual/complex studies, we may obtain preliminary codebook approval before coding commences or early in the coding process.  This is to ensure the codes properly address the client’s goals. We then fine-tune the code text, as necessary to most accurately reflect the verbatim responses as well as to break out less frequently mentioned themes and nuances.   

 

• If we have received preliminary code approval, any added and/or modified codes are marked for ease of reference and are approved again by the client, if necessary.

 

• Any preliminary data previously delivered that may be affected by codebook changes will be updated, rechecked and resent to the client unless otherwise instructed.

​

​

 

Quality Control

 

Codebook Quality

 

• All codebooks are quality checked for the following:

 

• Frequencies in each code are checked to make sure there are a minimum number or percentage of responses based on the requirements of the client, all the way down to single mention codes, if requested.

 

• Codes with high frequencies are checked to be certain they are not too broad in nature as well as to verify they have not been overused

 

• Nets and codes are checked to make sure each is distinct and unambiguous without any redundancy

 

• Net and code text is checked for clarity and understandability

 

 Coding Quality

 

• Industry standard allows for a 10% margin of error in open-ended coding; however, our internal acceptable standards are 5% or less.

 

• While coding, the coder is responsible for communicating any data issues including truncated responses, duplicated responses, a high percentage of gibberish and any other anomalies they find.

 

• Each coder is responsible for checking his/her work upon the completion of each question. This process includes reviewing all miscellaneous codes to determine if additional codes need to be created.  We work towards a goal of having a maximum of 3% in the miscellaneous codes, although there are occasions when the percentage is unavoidably higher.

 

• Once the coder has completed their quality check, the data is reviewed by a supervisor.  A minimum of 10% is checked to confirm that the work was done accurately and thoroughly.  For example, if a response has three distinct thoughts, there should be three codes applied.  On the rare occasion when there are issues, we will drill down and review as much data as is necessary to ensure the quality of the coding meets our standards.

 

• On large, complex or time-sensitive studies, the quality will be checked after an initial data set has been coded.  Ascribe allows us to mark verbatims as quality reviewed, so when we resume quality checking upon coding completion, we can identify responses that have not been previously reviewed.

 

• ‘Don’t know’, ‘Nothing/none’ and ‘N/A’ codes are checked to verify that they are appropriately coded and are exclusive codes.

 

• We have the capability to do double-blind coding when necessary.  That entails two coders working on the same verbatims individually.  The coding is then compared and where there is a discrepancy, the supervisor determines the most appropriate code(s).  The broader the question and more detailed the codes, the more inconsistencies can occur, many or most of which will be legitimate interpretation differences, so this method is only recommended when absolutely necessary.

 

Client Deliverables

 

• We can provide codebooks and coded data in a variety of formats along with specialty outputs such as SPSS and Dimension specs.

 

• Frequencies and statistical testing for individual open end questions as well as across question iterations and shared codebooks without the need for elaborate tabulations.

 

• Specialty requests: we will do our best to accommodate any special requirements you may have.