Behavioral Experiment Packages Rated

Behavioral Experiment Software Survey Results

Here's a one-paragraph summary of the survey results; details below: E-Prime is the most popular package of those surveyed, but the majority of folks are using either E-Prime, DMDX, or some flavor of PsyScope. E-Prime, PsyScope, SuperLab, and ERTS are all rated as easy to build experiments with, and about equally so. DMDX and NESU are seen as slightly harder. Presentation and MatLab are notably the hardest of the commonly used packages.

A Nice Font for Working Papers

The attractive font authors like Andrew Gelman use for working papers like this one (pdf) is American Typewriter, available in Office for Macintosh. But not in Office for Windows.

Krosnick's Forthcoming Survey Design Handbook

There's a nice preview of Jon Krosnick's forthcoming survey design handbook at the link below. Krosnick, whom I was fortunate enough to study with at the Summer Institute in Political Psychology, has done the research to give what I think will be definitive answers on design conundrums such as providing (or not) a "don't know" option. I've often been frustrated by the lack of a one-stop how-to on rigorous survey design, so I can hardly wait for his book.

Harvard University Program on Survey Research at the Institute for Quantitative Social Science

The conference centered around the research Krosnick has done into a century's worth of survey methodology studies for his forthcoming book, The Handbook of Questionnaire Design: Insights from Social and Cognitive Psychology ( Oxford ).

Surveys have been an investigative staple since the earliest years of the social sciences. But surprisingly, though most research-methods textbooks include informal discussions of questionnaire design, they tend to treat it as an intuitive art rather than a scientific skill governed by formal rules for optimizing data quality.

...

Another surprising result involves "don't know" responses. Many researchers have presumed for decades that it is wise to offer a "don't know" option to respondents, because many people genuinely lack the information necessary to answer some survey questions. But Krosnick found instead that offering a "don't know" option mostly lures people who have real opinions to decline to answer, as a way to minimize the effort they devote to the process. By omitting the "don't know" option, researchers can measure the real opinions held by more people.