Category Archives: Methods

Coding Library Cognitive Maps

After Donna Lanclos’s recent post on using my library cognitive mapping method, a few people asked me to briefly write up my approach to coding the drawings.

I developed the cognitive mapping exercise based on the sketch maps protocol used by Kevin Lynch in The Image of the Citywhich was introduced to me by an urban planner I met during my fieldwork on the Polish-German border.  Incidentally, I thought I was the first person to apply this method to libraries, until I ran across Mark Horan’s 1999 article, “What Students See: Sketch Maps as Tools for Assessing Knowledge of Libraries,” which used the same urban planning source materials to develop a very similar approach.

The examples I discuss below are drawn from the ERIAL Project, and the exact instructions I gave students were as follows:

“You will be given 6 minutes to draw from memory a map of the [NAME] Library. Every two minutes you will be asked to change the color of your pen in the following order: 1. Blue, 2. Green, 3. Red. After the six minutes is complete, please label the features on your map. Please try to be as complete as possible, and don’t worry about the quality of the drawing!”

This method assumes that the things people most associate with their “mental map” of the  library will appear as elements in the drawing, and that the most important things (or strongest associations) will appear earlier.  Therefore, by changing the pen colors, this approach creates both a spatial dimension and a temporal dimension.

The mapping activity was conducted away from library building itself both to obtain a diverse cross-section of students (e.g. students who do not regularly use the library) and to obtain a picture of how students conceptualize the library’s spaces that was not influenced by any direct visual references.

We used this protocol at four of the five ERIAL project libraries, but for simplicity, I’ll just use examples from one library.  The floor plan of this particular library looks like this:



Students were allowed an open interpretation of the instructions, which resulted in the wide range of approaches.  For example:




Coding these images basically involves counting the elements drawn in order to construct two indexes:  a identification index, which is the number of times that an element is drawn divided by the total number of individuals participating (i.e. the percentage of the time the element occurs), and representativeness index, which is the number of times an element is drawn divided by the number of times that category of element is drawn (e.g. the number of times a study room on the first floor is drawn divided by the number of times all study rooms are drawn) (See Colette Cauvin’s “Cognitive and cartographic representations : towards a comprehensive approach” for additional discussion).  I also constructed a temporal index for each element by coding the three colors in order (1 = Blue, 2 = Green, 3 = Red) and calculating the mean value for each element (you could do more complicated things by combining the indexes if you are mathematically inclined, however, I’ve found that these three get at most  questions).   You can set up a spreadsheet in excel to do this coding, or utilize the visual coding built into a QDA software package.   This process can be time consuming as every element must be coded.   You also need to decide which categories you will use (e.g.  “chairs,”  “computers,” “rooms”, etc.).  The presence or absence of all elements need to be coded for for every drawing, so if you find a new element in a later drawing, you need to go back and code for it in all the previous drawings (this is akin to coding against a closed codebook).

This is all fairly straightforward, except that there can be a lot of ambiguity in the drawings and you will have to decide rules for when something “counts” (this is why having students label things helps).

For example, in the drawing below, the computer stations (circled in orange) are clearly labeled, so these might be coded as element = “first floor information commons computers,” category = “computers,” time=1.


In contrast, the following drawing has unlabeled squares and rectangles (circled in orange) where there are tables and periodicals shelving.  In this case, the coder must decide what the element represents.  Since the squares are in the correct position, we coded these as tables, and since the rectangles are the correct shape and in the correct position we coded these as periodical shelves.  This can obviously become complicated, and you will need to decide what rules work for your particular context.


Some high index elements that we identified were reference & circulation desks, computer workstations, and study rooms, while low identification elements were librarian offices, journals, and new books areas.  Importantly, much high traffic library real estate was taken up by low-identification elements.  In this way, the blank spaces of the drawings can also be especially informative.  For example, in the following drawing almost the entire left side of the library is blank space.
This area is where current periodicals are shelved.
While I think this method is extremely informative for researchers, I would also recommend using caution in interpreting the results.  The assumptions about the association between an individual’s conception of  the library and the drawn representation can be questioned, and there are also a variety of potential sampling bias issues in the way sketch map methods are usually collected (e.g. problems stemming from convenience samples).   I therefore recommend utilizing this approach in conjunction with additional interviewing methods that can corroborate and add context to the findings.


Creative Commons License
Biblio|Ethno|Historio|Graphy by Andrew Asher & Donna Lanclos is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported License.