JOHN N. WILLIAMS, affiliation not provided to SSRN
Email: johnwilliams@smu.edu.sg
ERIC W. K. TSANG, University of Texas at Dallas - School of Management
Email: ewktsang@utdallas.edu

In their article, Generalizing Generalizability in Information Systems Research published in Information Systems Research, Lee and Baskerville (2003) make the useful attempt to clarify generalization and distinguish four types of generalization. We show that this discussion is flawed in various ways. In particular, Lee and Baskerville do not appreciate the disastrous consequences - even for their own account - of their central claim that Hume's problem of induction (originally published in 1748) is insoluble. Moreover, their characterization of generalization is confused in some respects. For example, what they call generalization is not generalization in the sense used by researchers in science, their definition of induction is demonstrably too narrow, and their characterization of generalization contradicts their own definition of it. We propose remedies of all these flaws. Central among these remedies is our argument that we should adopt the view, at least as a working hypothesis, that Hume's problem of induction has a solution.

MARK W. BELL, Indiana University Bloomington - Department of Telecommunications
Email: typewritermark@gmail.com
EDWARD CASTRONOVA, Indiana University Bloomington - Department of Telecommunications, CESifo (Center for Economic Studies and Ifo Institute for Economic Research)
Email: castro@indiana.edu
GERT G. WAGNER, Berlin University of Technology, German Institute for Economic Research (DIW Berlin), Institute for the Study of Labor (IZA)
Email: G.Wagner@ww.tu-berlin.de

Changes in communication technology have allowed for the expansion of data collection modes in survey research. The proliferation of the computer has allowed the creation of web and computer assisted auto-interview data collection modes. Virtual worlds are a new application of computer technology that once again expands the data collection modes by VASI (Virtual Assisted Self Interviewing). The Virtual Data Collection Interface (VDCI) developed at Indiana University in collaboration with the German Socio-Economic Panel Study (SOEP) allows survey researchers access to the population of virtual worlds in fully immersive Heads-up Display (HUD)-based survey instruments. This expansion needs careful consideration for its applicability to the researcher's question but offers a high level of data integrity and expanded survey availability and automation. Current open questions of the VASI method are an optimal sampling frame and sampling procedures within e. g. a virtual world like Second Life (SL). Further multi-modal studies are proposed to aid in evaluating the VDCI and placing it in context of other data collection modes.

DAVID A. BRAY, Emory University - Department of Decision & Information Analysis
Email: dbray@bus.emory.edu
ROBERT LAUBACHER, Massachusetts Institute of Technology (MIT) - Center for Coordination Science (CCS)
Email: rjl@mit.edu
THOMAS W. MALONE, Massachusetts Institute of Technology (MIT) - Sloan School of Management
Email: malone@mit.edu

This document discusses a few versatile, tool-like crowd performance algorithms necessary for humans to employ internet technologies to make better decisions collectively. This document also readily acknowledges that true collective intelligence approaches are foremost about organizational culture change and encouraging shared group norms (i.e., values) of sharing, openness, transparency, and collaboration. Humans will need to buy-in to the approaches and technology tools for any collective intelligence approach to work. The scope of this document has two sections:

The first section dives into some of the known theory and empirical evidence supporting a case that collective intelligence does, in fact, lead to better decision outcomes - specifically that collective intelligence employing recent advances in internet technologies can provide greater advantages than traditional organizational methods of making decisions.

The second section takes these underpinnings and then extends them to consider a tool and associated crowd performance algorithms that could be coupled into a packaged generic system that serves as a staging ground for an organization to launch collective intelligence efforts. This discussion considers how different features would be aligned with desired human and organizational factors, as well as what outcomes could be expected from such a generic system approach.

DAVID A. BRAY, Emory University - Department of Decision & Information Analysis
Email: dbray@bus.emory.edu
JERRY MECHLING, affiliation not provided to SSRN
Email: jerry_mechling@harvard.edu
BENN KONSYNSKI, Emory University - Goizueta Business School
Email: Benn_Konsynski@bus.emory.edu
HOLLI SEMETKO, Emory University
Email: holli.semetko@emory.edu

Within the next few pages, we outline how the U.S. President might employ novel arrangements of human and technology elements to produce improved government efforts involving the collective insights of all government works to: (1) better serve the public, (2) protect our country, and (3) generate innovative solutions that both address present day problems and produce a better future for us all. We seek to discuss insights into collective intelligence in large organizations - which, for the most part, has not been discussed with regard to the U.S. government - and translate the themes of these books to improve the Executive Branch.