coding - Survey and Behavior Survey and Behavior Coding...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Survey and Behavior Survey and Behavior Coding Coding Coding • Coding is the process of assigning numerical values to pre­specified categories • Codes may be assigned automatically via computer or manually by a person using a coding manual (codebook) Coding Questionnaire Data Coding Questionnaire Data For Every Questionnaire Variable – Each variable (var) should have a name – Typically var names should begin with a letter – Keep it short (3­12 characters) – Var names should be meaningful, and – The question # is generally placed at the end of var. name Scales Scales • Define the positive end of the scale – If possible, assign the larger numbers to the positive end and smaller numbers to the negative end – If there is a “neutral” category, it should be assigned the middle value – Be consistent with the direction of your scales despite their attribute Closed­ended Questions Closed­ended Questions • Numerical codes are typically assigned and specified during the development of the questionnaire item questionnaire • Numerical codes may be visibly displayed on the • • • • Establish standard codes: 7, 97, 997, 9997 Not applicable/Non­responsive 8, 98, 998, 9998 Don’t know 9, 99, 999, 9999 Refused EXAMPLES EXAMPLES • Should all closed­ended questionnaire items be coded? Example 1: Example 2: Answer1………….…...1 Yes……………. …….1 Answer2……………….2 No…………………...2 Answer3…………......3 Don’t know………..8 Other (SPECIFY)…..4 Refused…………….9 __________________ Don’t know…..........8 Codebook Codebook • In the event that you choose not to show response codes directly on the questionnaire, a codebook that identifies each code for each questionnaire item is used for coding during the backend Regardless of your survey method, a codebook is a useful tool. A codebook is a technical manual outlining the questionnaire and any associated programming. The codebook also identifies what the various numbers and letters (codes) mean, and any special instructions on how to use the data properly. • Codebook Contents Codebook Contents 1. A basic description of the study: who/what/how 2. Sampling information: description of the population and sample 3. Number of expected observations, type of interview, canned responses to FAQs and any required study specific information, etc. 4. Details about the data: columns in which specific variables can be found, whether they are character or numeric, and if numeric, what format. 5. Text of the questions and responses: some even have how many people responded a particular way. • Sort responses to open­ended questions into meaningful categories • Typically, responses will be divided into both superordinate and subordinate categories based on commonality. Open­ended Questions Open­ended Questions • Assign a numerical value to categories • Example: Your data reveals a class of items that is related to issues concerning the economy, within this category of items you notice that thus far respondents have included at least 5 variations of this theme, for each respondent, assigned codes will vary because of response, therefore, for one respondent the following codes are applicable for this item: – Superordinate Category: Economic (Code #10) – Subordinate Category Codes: Cost of food (Code #11), Cost of Gas (12), and Banking (Code # 15) codes would be entered into a statistical database: 10, 11, 12, and 15 • For this particular questionnaire item and for this respondentthe following • Codebook: A coding manual used to assign numerical values to each survey response Coding Definitions Coding Definitions • Acceptable Code: A code assigned by an automated coder that is deemed acceptable by an error control algorithm • Expert Coding: Coding by human experts considered the “GOLD” standard • Error Control: An algorithm for controlling the estimated error rate produced by an automated coder • Error Rate: The ratio of the number of correctly coded cases divided by the number of acceptable codes • Feedback Loop: Mechanism for improving an automated coder based on the results of coding test data • Production Rate: The ratio of the number of cases with acceptable codes assigned divided by the total number of cases Different Types of Coding Different Types of Coding • Closed­ended Questions: Pre­specified numerical categories • Open­ended Responses: Project specific manual • Industry and Occupation: Census Bureau coding index • Medical: International Classification of Diseases, 9th revision • Geographic: Federal Information Processing Standards Index or the US General Services Geographic Index • Military: Department of Defense Occupational Conversion Manuel • Dietary: US Department of Agriculture • Automated and Behavioral Automated Coding Automated Coding • Codes are assigned using a computer without human interaction • It is very rare that a computer will assign 100% of the codes to all cases, so a coding manual is usually produced to supplement computer coding doing it manually depends on the relative cost associated with the activity • Typically, the decision to use automated coding over • The process of automated coding involves: carefully constructed category classifications, training, test, and validation data; software development; and error and quality control methods Behavioral Coding Behavioral Coding • Focuses on the interaction between the interviewer and the respondent • Relies on overt cues during questionnaire administration; Interviews typically tape recorded through observation • It is a passive activity, no probing, completed solely • Behavior codes are designed to identify and examine situations that exemplify interactions that indicate different types of cognitive issues in interviewing Behavioral codes assess: Behavioral codes assess: 1) interviewer question­asking behavior (ex.: skip instructions, administration) 2) immediate response behavior of respondents (requests for clarifications, refusals to answer questions) 3) respondent interruptions, and 4) the number of times and length of the number of pauses respondents’ use while answering questions ...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online