Welcome to my Thinking World…

Abstract:

In some surveys, respondents are confronted with selecting a response from an extremely long list of response options. For example, in the National Survey of College Graduates (NSCG) mail survey, respondents are asked to select the job code that best describes their work from a long list of job codes (listed on a different page) that are categorized and arranged alphabetically to facilitate easy selection. While various aspects of the findings from this survey have been published, for instance, the effect of previously sent token incentives on subsequent contact rates (see Dillman (2007), page 241), little is known about the measurement error aspect of using extremely long response lists. Read more

Abstract:

The crisis in coverage of the landline telephone frame has reached a point where researchers are testing alternatives to (or at a minimum, complements to) traditional landline random digit dialing (RDD). Ideally, a new sampling methodology would address not only the coverage challenges posed by cell phone only households and unlisted households in zero-listed banks, but also provide a basis for developing survey designs that could address the growing nonresponse issue which also plagues traditional RDD approaches. One such approach proven to be promising and potentially cost effective involves use of address-based samples using large comprehensive databases, such as the U.S. Postal Service Delivery Sequence File (DSF). Read more

Abstract:

In Web surveys, progress indicators inform respondents of their progress, using a variety of design approaches. A graphic-based method usually involves the image of a progress bar, while a text-based method involves using simple text (often percent completed).  Sometimes a combination of graphic- and text-based designs is used (see Couper, Traugott, & Lamias, 2001; Heerwegh & Loosveldt, 2006). Progress indicators can also be displayed on every screen or intermittently, or at the respondents’ request (see Conrad et al., 2005).  The main argument for using a progress indicator is that it motivates respondents to complete the survey by making the end of the survey appear visibly nearer with every question answered. Read more

Abstract:

Decreasing survey response rates are a growing concern in survey research, principally because survey estimates may be biased by selective nonresponse (Kypri, Stephenson, & Langley, 2004). One of the methods of assessing nonresponse bias is to compare those who respond late to a survey with those who respond early, in terms of the topic of interest. In this paper, we draw upon data obtained from multiple panel surveys conducted by the Gallup Panel in order to examine whether early, intermediate, and late respondents differ significantly – either in terms of demographics or in terms of the answers that they provide to survey questions. Read more

Abstract:

In mail surveys, researchers often face the challenge of too little space in the questionnaire for the number of questions they wish to administer. Oftentimes certain structuring methods are used in order to maximize the data gathering efficiency of a given set of questions. One such method involves using a grid format in which two or more questions with the same response options are grouped together in a ‘grid’ of rows and columns. In this way, the response options are not repeated for every question in the grid, which minimizes not only the use of space in the questionnaire, but also respondent’s burden of comprehending new scale for each question in the grid. Read more