This chapter enables the user to begin using the Manager program and to create a simple survey in a few hours. It is a 'how to' chapter that progresses from some preliminary design tips and considerations, to the creation of blank instruments and modules, and then to the actual entry of question records into the Manager. The chapter ends with some special features of the OSU CHRR CAPI programs that include using loops, symbols, and rosters. The list of topics below is a quick reference to the elements discussed in this chapter.
|
Introduction (this page) Design Considerations Stylistic Conventions
Access the Manager Design Function |
Begin Sample List of Questions Begin Entering Sample Questions
|
In the Manager program eleven distinct question types are used to define the question records of an instrument. The question record is the principal driving force of an instrument's logic and flow. This chapter focuses on the creation of question records (as opposed to help, pick list, roster, alternate language, etc.) and illustrates each question type by examples. Auxiliary records (those linked to the question record) are depicted when applicable.
In general there are two broad categories of question types:
No Experience
If your assignment is to design a CAPI survey and you have no experience, follow the topics in this chapter sequentially. The layout takes the user step by step through the authoring process. First general design considerations are presented, then, paper questions are transformed into CAPI records, question type by question type. Each CAPI record is accompanied by the display of what the interviewer would see during the interview, wherever applicable. The steps assemble a simple, but somewhat contrived, survey instrument. This instrument can then be transformed into a field questionnaire by exporting it from the Manager to the Survey program.
Some Experience
If you already have some experience in survey design, you may wish to use this chapter as a "how to" topical reference to grasp concepts and see how it is done in the Manager and Survey programs. If you need detailed information about question types, record types, fields in records, functions, and specific syntax refer to the "Reference" section in Chapter 8.
If your assignment is to setup and administrate the Manager program, you will want to refer to other appropriate sections of this manual such as "Manager Installation," "Manager Configuration" and " Manager Startup."
To begin, first determine whether you must start from a paper copy of an instrument or if you have an existing DOS version.
For the purposes of this manual, we must distinguish between questionnaire design and survey design: questionnaire design comprises the design of the questions used in an instrument; survey design encompasses the design of an instrument and/or multiple instruments into the field environment. We further define questionnaire design as the phase of the survey project that takes place up to and, to some extent, including the use of the Manager program to input the questions in computer form. This means that questionnaire design begins with the inception of an instrument or a survey project and ends with an approved set of questions that will be exported to the Survey program.
Who Is the Designer’s Audience?
The most important person in executing an interview is the interviewer. The interviewer is someone who is very friendly and social; they can extract the details of your life before you are aware of what is happening. They are self-starters and resourceful. It is these attributes that had made the NLSY79 completion rates the envy of the industry. The success of a project is, more than anything else, a direct reflection of the interviewer and the field managers who supervise them. If these people are supported in the design process and their morale kept high in field support, the survey will have success. If they become befuddled, perplexed, and are demoralized, watch out.
Interviewers may not typically be long on technical skills. The best lesson interviewers taught CHRR in 1989 was that the key to a successful CAPI interview is to employ a relatively small number of question types. When interviewers were asked to work with full-screen questions that had lots of blanks to fill out and somewhat obscure keyboard commands, error rates and training time increased and interviewer morale decreased.
Based upon careful observation of interviewer training, OSU CAPI design strategy stresses what may be called "reduced instruction set interviewing." OSU CAPI builds up the instrument from a relatively few question types. These question types are individually simple to administer, but can be combined to form an extraordinarily long and complex interview. The interviewer encounters relatively few of the available question types which makes training easier.
There are, of course, skip instructions and other mechanical operations that are complex and take place without interviewer intervention. The designer has additional question types available to create and carry out the sophistication of a complex instrument. As a designer there are two important objectives in designing a good instrument.
For the most part a draft questionnaire for the Manager program should look very much like a conventional paper and pencil questionnaire. If the design process begins with a list of questions, the list should be augmented to include not only the question text, but also the "coding frame" features discussed below. These features include not only the question number and text but also name, type of response (datum) to be collected, its format and length, the next question’s name, and skip instructions that are contingent upon the answer. If a pick list is used, its name and the items in the list should be included with the question. This pattern will account for the vast majority of questions. When there is a "check item" (i.e., something that the computer must verify) that will conduct a skip based upon responses to several questions, design these to look like what is usually seen in a paper and pencil questionnaire. A sample list of questions follows, see "Sample List of Questions."
If you are designing a self-administered instrument, it must be very simple and clear. More care should be taken in making the self-administered instrument simple, because preparation, practice, and instruction of the respondent cannot be carried out in the field environment.
Listed below are some stylistic conventions that CHRR has found useful in the design process. When designing and creating questions consider using the following conventions. Some will facilitate transition to inputting the records into the Manager program, see the information under "Coding Frames ," and some will simplify interviewer training, see "Interviewer Instructions" , and "Hand Cards."
A "coding frame" is a design template used to structure the coarse, initial questions. The coding frame is a preliminary design tool that exhibits the question name, the question text, the range of possible answers, and the skip logic prescribed by the answer given. If a pick list is used, its name and the items in the list should be included with the question. The coding frame may also include the format and length of the response, the limits that bound it, and so on. It gathers these elements together and resembles, not surprisingly, the question record as it appears in the Manager’s standard question design template. This method facilitates transition from the primary question to the Manager question record.
The scheme used in this chapter first presents a sample list of initial questions that a designer might receive from a design team or client. See "Sample List of Questions." The questions in the sample list cover all of the possible question types used by the Manager and Survey programs. In the subsequent section entitled "Entering Manager-Survey Questions" each sample question is transformed into a coding and correlated with the corresponding Manager records and Survey screens. A discussion of the correlations accompanies the illustrations.
All Manager records displayed are created by using the Design Questionnaire function (Design tab, Questionnaire button) of the Manager program.
Every question in the survey instrument needs to have a unique name that identifies it. (Question names may bear a resemblance to question numbers but question numbers order the questions sequentially and are specified in the RecPos field.
Most questions in the survey instrument will gather a response datum. When designing and creating the questions, the time must be taken to consider the type of response(s) that will be collected.
Sometimes allowances must be made for responses that don't fit the coding frame. Use the "Other (SPECIFY)" category to allow this.
Whenever specifying a length it should always be set to a minimum of '2'. The reason for this is because the Survey program always allows the respondent to refuse to respond or to record a 'don't know' response. The integer values for these two responses contain a minus sign and the integer itself, which requires at least two characters in the answer file:
Give answer categories in a list number values in case you want to use these variables later on. For example, ten questions further along you may wish to insert an instruction (machine check item) that verifies whether the response to question number 4 was equal to ‘1’ or greater than ’6’ and, consequently, conduct a skip that depends upon which condition applies.
Use UPPER CASE letters to convey Instructions to the interviewer that are NOT read to the respondent. Use this same convention when you do not want the interviewer to read response choices to the respondent. For example, the following phrases would NOT be read to the respondent:
If you want the interviewer to read the choices, type them in lower case (capitalize the first word of line). Consistent use of UPPER CASE can become a handy training tool.
This section presents one OSU CAPI solution to the use of long "pick lists." When using a long pick list remember a "Show Card" or "Hand Card" may be used that gives the possible answer options for the respondent to read. The responses may then be collected by using Text Entry or Numeric Entry question types. See the section entitled "Hand Card."
Next Steps
Start with the "Sample List of Questions" or "Entering Manager-Survey Questions."
To begin creating a new instrument, see Creating a New Instrument, Chapter 7 - Quick Start.
To begin creating records in the new instrument, see Creating New Records, Chapter 7 - Quick Start.
Below is a sample list of questions that designers might create or receive from the question design team. This list will be used to briefly illustrate the steps in transition from a coarse list of questions to a Survey instrument. The section following the list will display each question in its "coding frame", the Manager record, and the Survey screen that appears to the interviewer.
DEMO Survey Questions
In the following section each question in this list is first transformed into a "coding frame," The coding frames contain an elaboration of the question number and text. Each elaboration contains a label that corresponds to a field name in the Manager record and a ‘value’ to be placed in that field.
Next Step
Go to "Entering Manager-Survey Questions" or view "Sample Question in Coding Frames."