Guidelines for       

Collaboration with the BRC      


Statisticians in the BRC work closely with investigators to achieve research and career development goals. Our services include assistance in the following areas: study design, sample size determination, data analysis, manuscript preparation, grant applications and IRB proposal preparation. This one-page summary will clarify for you how to prepare for your consultation.

I. For Data Analysis Projects:    

 
    1. What is your hypothesis?   

  • What is the primary, specfic question you want answered?   
  • What are the secondary questions that you want answered?   
  • Are you interested in subset of data? (e.g., do analysis on female patients only)   
  • What association or correlation among measurements, are you interested in?  (e.g., Does patient's age affect blood pressure?  Do older patients tend to have higher blood pressure?)   

 
    2.  How will your outcomes and independent variables be measured? 

  • What are the data set and variables involved in the analysis you need?   
  • Please explain what the variables in your dataset refer to.  [e.g., look at the assoication between: yyy (column H) and zzz (column C)]  
  • If you want to group/split your dataset based on a specific variable, please specify your grouping criteria.  (e.g., make three groups in relation to patient ages: 0 - 8 years, 8 - 12 years, 12 - 20 years)  

 

II. For Grant and IRB Proposals:   

     1. How do you want your sample size determination done and what is your hypothesis? 
            • What is the primary, specific question you want answered? 
            • What are the secondary questions that you want answered?  

      2. What is the study design (e.g. hypothesis, sample size determination, etc.)?  

      3. How will you collect the information? (e.g., using a questionnaire in your research)  

      4. Do you want to address other outcomes?  

 

III.  Data Format      

Please prepare by identifying the variables in the data via coding and names.  If you entered your data in Exel, several basic rules are   necessary so that your data may be read into a statistical package easily.  Please be aware of the  following: 

  1. Include unique variable names in the first row of your spreadsheet.  
      
  2. Do not color code or have blank rowes within your spreadsheet.  If information is important enough to "offset" be sure a variable is included that would indicate what you would have color coded. 
            
  3. Data need unique (numeric) study ids (not identifiable, per HIPAA privacy rules).  Row numbers in Excel do not count.  Use one row of data for each unique, independent study subject. 
            
  4. Do not mix "string" (text) and "numeric" data types. 
            
  5. Numeric coding is preferable (e.g. 0=no, 1=yes) with a data codebook, or a data dictionary either in a separate sheet or document.  
           
  6.  If your dataset has scale variables, please explain what the scale means.  (e.g., pain score 0-10, while 0 means no pain and 10 means the most severe pain)