By Shalin Hai-Jew, Kansas State University 

In the common everyday use, a “crosswalk” is a defined way for people to get from one side of a street to another.  In an analytical context, a “crosswalk” connects two systems around similarities or overlaps.   

A “crosswalk analysis,” defined analogically, is a way of connecting various similar or disparate objects to each other:  constructs like (1) metadata schemas, standards, competencies, and frameworks, and elements like (2) data.  Crosswalk analyses enable the connecting of prior-unrelated concepts and informational contents to enhance comparability and contrast-ability.  They enable the harmonizing of ideas and data.  A crosswalk enables comparability not only between types (such as metadata schema to metadata schema, framework to framework) but across different types (such as professional standards to professional competencies, metadata schemas to data, frameworks to professional standards).  (Figure 1)  

 

Figure 1.  A Crosswalk Analysis (with bridging or connective functions) 

The actual work is harder than it may seem. After all, those who create metadata schemas, professional standards structures, professional competencies systems, frameworks, and data…have different conceptual (and mental) models.  The various constructs and data may have different origination times.  The respective organizations and individuals who contribute ideas have differing contexts and local cultures. Their intentions for creating the various constructs and datasets often vary. Even if language is highly precise and specific, people may imbue words and phrases with different understandings, given the polysemic nature of language.  Professional standards are often written in “multi-barreled” ways, so an accurate crosswalk may have to consider every mentioned element…on both sides (on both constructs being compared).  Crosswalks should not only show that there is a possible relational bridging but the intensity of that connection and the related dimensionality of that connection.  (How strong is the relationship, and on what elements is there similarity vs. non-similarity?)  

During the making of a crosswalk, it is important not to introduce error—such as through mischaracterizing information, losing information, skewing information, and so on.  It is important to be thorough.  If there is no buildable-crosswalk between elements, then that should be noted.   In general, two systems are connected at the most common and most granular level (applied standards, applied competencies, data columns), not at the coarser levels of granularity like categories.  The crosswalk analysis should first deal with what is actually organically present in the respective compared systems (constructs or data), even if further analyses may consider what could be there.  

For a crosswalk to effectively bridge between constructs and data, the crosswalk has to identify (pseudo-)equivalencies and connectivity in accurate ways.  It would make sense for crosswalks to be validated/invalidated by subject matter experts/content experts or other evaluators.  

Beyond accuracy, crosswalk analyses have to inform on “gaps” between the compared constructs (or data).  They have to inform human decision making and design (data, program, and other types), in practical and effective ways.  They have to be useful for follow-on analyses and informed actions.  

For the insiders to the target field, they may ask the following:  

The systems analyzed in a crosswalk analysis also have to be kept up-to-date as each system changes.  Depending on how quickly the information expires, there may be implications for how to set up and maintain crosswalk analyses.  (Crosswalk analyses also expire fairly quickly, depending on the systems on which these are predicated.)  

Crosswalk analysis are thought to be “lateral” (or one-way) mappings (Caplan, 2003, as cited in “Schema_crosswalk,” Sept. 28, 2018).  This suggests that if the crosswalk is to be two-ways, two separate analyses have to be done (Figure 2).  The latter analysis of Competencies Y to Standards A not only should shed new light on the target Competencies Y (and Standards A) but also increase general intimacy with and understandings of the two elements.  

Figure 2.  Lateral Nature (One Directionality) of Some Crosswalk Analyses (requiring separate mappings and no assumption of reversibility) 

Another level of abstraction and complexity from the original crosswalk may involve “transitivity,” or the relationships between Metadata Schema A and C in Figure 3.  It would seem that the main challenge has to do with maintaining coherence while engaging in these respective analyses.  

 

Figure 3.  Crosswalk Transitivity 

There are accepted professional standards for (metadata) schema crosswalks, to enhance the uses of structured type data.  These define how similar data may be connected as similar and equivalent sorts of data for datasets to be “union” queried.  However, in terms of other areas, standards have not yet been defined for how to build effective crosswalks.  “Schema crosswalks” enable the integration of various data structures for data discoverability and querying.  Some schema crosswalks are overlaid over content management systems (CMSes). 

A simple structure for a crosswalk analysis involves having a main seeding construct and its sub-elements in some related data columns, a set of “crosswalk” columns to the right of that, and then beyond that, the construct and / or data being compared.  There are additional columns that provide information about the respective data records for both System 1 and System 2.   Each row shows some aspect of the first construct being compared to the latter one, via the crosswalk, which is in between.  This structure is amenable for setup in a data table, a matrix, and / or a spreadsheet.  Once proof-of-concept has been established, crosswalks may be built into databases.  

System 1 (Focal System)	Crosswalk	System 2  (Local)

Table 1.  A Basic Structure for a Crosswalk Analysis (Purposefully Empty) 

As part of a workplace project, I conducted multiple related crosswalk analyses, based on national standards and local learning competencies (and vice versa).  Experientially, creating the respective crosswalks requires keeping both systems being compared as intact as possible.  It also requires multiple read-throughs of each system on their own to understand their internal structures before working towards bridging.  While crosswalk analyses are supposed to be one-directional, it is very hard not to see the crosswalks as multi-way…until one actually switches up the systems for a follow-on crosswalk analysis and sees how that reordering changes the insights.  The identification of the element that fits into the crosswalk is generally pretty obvious—the most granular aspects with the most overlap.  (A crosswalk analysis is a Venn diagram albeit not in visual format, and the overlap is where the crosswalk occurs.)  In the analysis, it is important to be as clear-headed as possible about what the granular elements mean on both sides but still be liberal in terms of mapping (and not to expect one-for-one mapping).  An analyst has to be comfortable with the gappiness between the linked elements.  

Figure 4.  A Venn Diagram Analogy to the Crosswalk Analysis

The work itself, in my experience, is painstaking and meticulous.  It is also political, in terms of what the suggestions may be for the development of courses and degree programs and learner experiences (but these considerations are also outside the purview of an external analyst).  

Taking one set of standards and setting it up as a crosswalk analysis template, with different local learning objective / outcomes / competencies (interrelated) is fairly straightforward.  Moving this to a simple database and maintaining coherence is something else and requires much more design thinking (of interfaces, of forms, of data design).  

“Schema crosswalk.”  (2018, Sept. 28).  Wikipedia. Retrieved Dec. 1, 2018, from https://en.wikipedia.org/wiki/Schema_crosswalk.  

Shalin Hai-Jew works as an instructional designer at Kansas State University.  Her email is shalin@k-state.edu.