Reporting Meaningful Results: One Statistical Agency’s Explorations

• Activity Number: 82 - Contributed Poster Presentations: Government Statistics Section
• Type: Contributed
• Date/Time: Monday, August 3, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Government Statistics Section
• Author(s): Marilyn Seastrom*
• Companies: NCES/IES/Dept. of ED
• Keywords: Reporting standards; utility; quality; meaningful results; p values

Abstract

NCES has struggled with this issue over the last 35 years. From calls in the mid-1980’s for products to have both utility and quality guided by a written set of Statistical Standards, to specific parameters placed on the reporting of statistical findings in NCES Statistical Standards published in 1987, 1992, 2003, and 2012; NCES has struggled with the topic of how best to use the field of statistics to support the description and analysis of NCES data. Over the years, different approached have been proposed, discussed, and tried--to no avail. The recent discussions within the profession over whether or not we should continue our devotion to significance tests with a p value of .05 has opened up a new platform for NCES to use in identifying a way forward. There are hints of movement in this direction on the part of NCES as early as the 2003 NCES Statistical Standards. More recently NCES held a seminar on the role of a .05 p-value and convened two NISS panels to review NCES practices and provide ideas for a path forward. Our early efforts in this direction and the recent panel results will be discussed in this poster presentation.