To summarize, altmetrics should build on existing statistical and social science methods for developing reliable measures. The draft white paper from the NISO altmetrics project suggests many interesting potential action items, but does not yet incorporate, suggest or reference a framework for systematic definition or evaluation of metrics.
NISO offered a recent opportunity to comment on the draft recommendation on their ‘Altmetrics Standards Project’. MIT is a non-voting NISO member, and I am the current ‘representative’ to NISO. The following is my commentary, on the draft recommendation. You may also be interested in reading the other commentaries on this draft.
Response to request for public comments on on ‘NISO Altmetrics Standards Project White Paper ’
Scholarly metrics should be broadly understood as measurement constructs applied to the domain of scholarly/research (broadly, any form of rigorous enquiry), outputs, actors, impacts (i.e. broader consequences), and the relationships among them. Most traditional formal scholarly metrics, such as the H-Index, Journal impact Factor, and citation count, are relatively simple summary statistics applied to the attributes of a corpus of bibliographic citations extracted from a selection of peer-reviewed journals. The Altmetrics movement aims to develop more sophisticated measures, based on a broader set of attributes, and covering a deeper corpus of outputs.
As the Draft aptly notes, in general our current scholarly metrics, and the decision systems around them are far from rigorous: “Unfortunately, the scientific rigor applied to using these numbers for evaluation is often far below the rigor scholars use in their own scholarship.” 
The Draft takes a step towards a more rigorous understanding of alt metrics. It’s primary contribution is to suggest a set of potential action items to increase clarity and understanding.
However, the Draft does not yet identify either the key elements of a rigorous (or systematic) foundation for defining scholarly metrics, their properties, and quality. Nor does the Draft identify key research in evaluation and measurement that provide a potential foundation. The aim of these comments is to start to fill this structural.
Informally speaking, good scholarly metrics are fit for use in a scholarly incentive system. More formally, most scholarly metrics are parts of larger evaluation and incentive systems, where the metric is used to support descriptive and predictive/causal inference, in support of some decision.
Defining metrics formally in this way also helps to clarify what characteristics of metrics are important for determining their quality and usefulness.
– Characteristics supporting any inference. Classical test theory is well developed in this area.  Useful metric supports some form of inference, and reliable inference requires reliablilty. Informally, good metrics should yield the similar results across repeated measurements of the same purported phenomenon.
– Characteristics supporting descriptive inference. Since an objective of most incentive systems is descriptive, good measures must have appropriate measurement validity.  In informal terms, all measures should be internally consistent; and the metric should be related to the concept being measured.
– Characteristics supporting prediction or intervention. Since objective of most incentive systems is both descriptive and predictive/causal inference, good measures must aid accurate and unbiased inference.  In informal terms, the metric should demonstrably be able to increase the accuracy of predicting something relevant to scholarly evaluation.
– Characteristics supporting decisions. Decision theory is well developed in this area : The usefulness of metrics is dependent on the cost of computing the metric, and the value of the information that the metric produces. The value of the information depends on the expected value of the optimal decisions that would be produced with and without that information. In informal terms, good metrics provide information that helps one avoid costly mistakes, and good metrics cost less than the expected of the mistakes one avoids by using them.
– Characteristics supporting evaluation systems. This is a more complex area, but the field of game theory and mechanism design are most relevant. Measures that are used in a strategic context must be resistant to manipulation — either (a) requiring extensive resources to manipulate, (b) requiring extensive coordination across independent actors to manipulate, or by (c) inventing truthful revelation. Trust engineering is another relevant area — characteristics such as transparency, monitoring, and punishment of bad behavior, among other systems factors, may have substantial effects. 
The above characteristics comprise a large part of the scientific basis for assessing the quality and usefulness of scholarly metrics. They are necessarily abstract, but closely related to the categories of action items already in the report. In particular to Definitions; Research Evaluation; Data Quality; and Grouping. Specifically, we recommend adding the following action items respectively:
– [Definitions] Develop specific definitions of altmetrics that are consistent with best practice in the social-science field on the development of measures
– [Research evaluation] – Promote evaluation of the construct and predictive validity of individual scholarly metrics, compared to the best available evaluations of scholarly impact.
– [Data Quality and Gaming] – Promote the evaluation and documentation of the reliability of measures, their predictive validity, cost of computing, potential value of information, and susceptibility to manipulation based on the resources available, incentives, or collaboration among parties.
 NISO Altmetrics Standards Project White Paper, Draft 4, June 6 2014; page 8
 See chapter 5-7 in Raykov, Tenko, and George A. Marcoulides. Introduction to psychometric theory. Taylor & Francis, 2010.
 See chapter 6 in Raykov, Tenko, and George A. Marcoulides. Introduction to psychometric theory. Taylor & Francis, 2010.
 See chapter 7 in Raykov, Tenko, and George A. Marcoulides. Introduction to psychometric theory. Taylor & Francis, 2010.
 See Morgan, Stephen L., and Christopher Winship. Counterfactuals and causal inference: Methods and principles for social research. Cambridge University Press, 2007.
 See Pratt, John Winsor, Howard Raiffa, and Robert Schlaifer. Introduction to statistical decision theory. MIT press, 1995.
 See ch 7. in Fudenberg, Drew, and Jean Tirole. “Game theory, 1991.” Cambridge, Massachusetts (1991).
 Schneier, Bruce. Liars and outliers: enabling the trust that society needs to thrive. John Wiley & Sons, 2012.
New Discovery Tools for Digital Humanities and Spatial Data (Summary of the July, Brown Bag Talk by Lex Berman)
My colleague, (Merrick) Lex Berman, who is Web Service Manager & GIS Specialist, at the Center for Geographic Analysis at Harvard presented this as part of the Program on Information Science Brown Bag Series. Lex is an expert in applications related to digital humanities, GIS, and Chinese history — and has developed many interesting tools in this area.
In his talk, Lex notes how the library catalog has evolved from the description of items in physical collections into a wide-reaching net of services and tools for managing both physical collections and networked resources: The line between descriptive metadata and actual content is becoming blurred. Librarians and catalogers are now in the position of being not only docents of collections, but innovators in digital research, and this opens up a number of opportunities for retooling library discovery tools. His presentation will presented survey of methods and projects that have extended traditional catalogs of libraries and museums into online collections of digital objects in the field of humanities — focusing on projects that use historical place names and geographic identifiers for linked open data will be discussed.
A number of themes ran through Lex’s presentation: One theme is the unbinding of information — how collections are split into pieces that can be repurposed, but which also need to be linked to their context to remain understandable. Another theme is that knowledge is no longer bounded, footnotes and references are no longer stopping points, from the point of view of the user, all collections are unbounded, and the line between references to information and the information itself has become increasingly blurred. A third theme was the pervasiveness of information about place and space — all human activity takes place within a specific context of time and space, and implicit references to places exist in many places in the library catalog such as in the titles, and descriptions of works. A fourth them is that user expectations are changing – they expect instant, machine -readable information, geospatial information, mapping, and facetting as a matter of course.
Lex suggested a number of entry points for Libraries to investigate and pilot spatial discovery:
- Build connections to existing catalogs, which already have implicit reference to space and place
- Expose information through simple API’s and formats, like GEORSS
- Use and contribute to open services like gazetteers
Tracking scholarly outputs has always been a part of the academic enterprise. However the dramatic increase in publication and collaboration over the last three decades is driving new, more scaleable approaches. A central challenge to understanding the rapidly growing scholarly universe is the problem of collecting complete and unambiguous data on who (among researchers, scholars, students and other members of the enterprise) has contributed in what ways to what outputs (e.g., articles, data, software, patents) with the support of which institutions (e.g. as funders, host institutions, publishers). In short, a full understanding of research requires those involved in the research enterprise to use public, reliable identifiers.
In June, I had the pleasure of speaking on a panel at the “Twelfth Annual ARIES EMUG Users Group Meeting” that aimed to provide an overview of the major new trends in the area of scholarly identifiers.
The presentation embedded below provides an overview of ORCID researcher identifiers; their role in integrating systems for managing, evaluating, and tracking scholarly outputs; and the broader integration of researcher identifiers with publication, funder, and institutional identifiers.
Most of the credit for the presentation itself is due to ORCID Executive Director Laure Haak who developed the majority of the presentation materials — those which describe ORCID and developments around it. And there are indeed many ORCID-related developments to relate.
My additions attempt to sum up the larger context, in which ORCID, and researcher identifiers play a key role.
It has been widely remarked that the sheer number of publications and researchers has grown dramatically over the last three decades. And it is not simply the numbers that are changing. Authors are changing — increasingly students, “citizen-scientists”, software developers, data curators and others author or make substantial intellectual contributions to, scholarly works. Authorship is changing — science, and the creations of scientific outputs involves wider collaborations, and a wider potential variety of research roles. Scholarly works are changing — recognized outputs of scholarship not only include traditional research articles and books, but also datasets, nano-publications, software, videos, and dynamic “digital scholarship”. And evaluation is changing to reflect the increasing volume, granularity, and richness of measures available, and the increasing sophistication of statistical and computational methods for network and textual analysis.
The tools, methods, and infrastructure for tracking, evaluating, attributing, understanding patterns of scholarship are under pressure to adapt to these changes. ORCID is part of this — it is a key tool for adapting to changes in the scale and nature of scholarly production. It’s a community-based system for researcher identification, based on standardized definitions, open source, an open API, and open data.
ORCID provides a mechanism for robustly identification of researchers – it aims to solve the problem of understanding the “who” in research. Increasingly, ORCID is also integrating with solutions to address the “which”, and “what”.
Effective sustained long-term integration of multiple domains requires work at multiple levels:
- At the abstract level, integration involves the coordination of vocabularies, schemas, taxonomies or ontologies that link or cross domain boundaries.
- At the systems level, integration requires accessible API’s that provide hooks to access domain specific identifiers, linkages, or content.
- At the user level, integration requires human-computer-interface design must expose and domain-specific information, and leverage this to increase ease-of-use and data integrity, and support and document needs to be available.
- At the organizational level, integration requires engagement with the evolution of standards and implementation, and organizations driving these, in other domains. Especially in this rapidly changing ecosystem, one must frequently monitor integration points to anticipate or mitigate incompatible changes.
ORCID is making rapid progress in integrating with systems that address the “which” of research. ORCID id’s are now integrated into manuscript management systems and publisher’s workflows and CrossRef DOI indexing with the result that these id’s are now increasingly part of the core metadata associated with publications.
ORCID now uses standard Ringold identifiers to identify institutions such as employers. (Ringold identifiers are in the process of being mapped to ISNI institutional identifiers as well — which will further integrate ORCID and ISNI.) These institutional identifiers are seamlessly integrated into the ORCID UI which help users of ORCID auto-complete institutional names, and increases data integrity. These ID’s are part of of the ORCID schema and exposed through the open API . And ORCID engages with Ringold on an institutional level so that institutional identifiers can be added on the request of ORCID members.
Similarly, ORCID now uses FundRef identifiers to identify funding agencies and awards. These too are integrated at points in the UI, schema, and API. Search and link wizards can push FundRef identifier into the ORCID registry along with other information about each award.
Full integration of data across the next-generation of the scholarly ecosystem will involve more of the “what” of research. This includes associating publication, institutional, and individual identifiers with a wider variety of scholarly outputs, including data sets and software; and developing standardized information about the types of relationships among outputs, institutions, and people — particularly the many different types and degrees of contribution that members of collaborations make to research and to its products.
ORCID has been taking steps in this direction, including a DataCite – search and link wizard for datasets, working to expand work-types supported in the ORCID schemas, and working with the community to develop and enhance existing schemas and workflows; and working with CASRAI to develop an approach to embedding researcher identifiers into peer review. This is just the tip of the iceberg, however, and the scholarly ecosystem has considerable ways to go before it will reflect the many emerging forms of scholarly outputs and roles that contributors take in relation to these.