Gendered Evaluations

Hi, my name is Alex Richter (she/her). I am a recent graduate from Reed College and post bac researcher with Anna Ritz. This is my second summer doing research with Anna Ritz. The project of this summer dealt with letters of recommendation and the role of gendered words/phrases.

The Problem

There are certain words and ways of describing people that appear in letters of recommendation that are gendered. These issues are implicit and occur regardless of the race, gender, or ethnicity of the writer. This issues leads to a disparity of women in specific fields due to the importance of letters of recommendation in graduate school acceptances and hiring.

A very easy example is that women are more likely to be praised in letters of recommendation for their communication or being nice. Men on the other hand are likely to be praised on their ability to do a task or a skill that is relevant to the position. It should also be noted that most of the research in this topic deals with a gender binary of just men and women. Our work tries to avoid the binary by focusing on pronouns, but it should be noted that in discussing this topic phrases such as “Male vs Female” or “Man vs Woman” are common place.

Why does it matter?

Letters of recommendation are important as they allow a better insight into a persons’ character. It is an important factor in hiring a candidate or accepting a candidate into any program. It shows more than just the accomplishments but speaks to who the individual is and how they are perceived by someone who knows them.

This work matters especially at Reed because of the emphasis we put on letters of recommendation. Notoriously, Reed does not give out grades unless requested and while transcripts with letter grades exist the letter grade concept is de-emphasized here. Reed focuses on quality and understanding as opposed to emphasizing grades. As stated in the letter that accompanies all transcripts, “Since its founding, Reed College has adhered to a distinctive educational philosophy, a critical component of which is the evaluation and grading of student academic work.”

Screenshot of PDF included with Reed transcripts to highlight grading policy

The Pre-Existing Work

In 2018, there was a project done by a graduate student named Mollie Marr and other collaborators. The repository for this project can be found here. This work entailed writing detectors that could be run in order to see if specific words were present in letters of recommendation. There were many detectors that were written by multiple developers. Each detector had been extensively researched and was backed by various publications.

The preexisting detectors that were implemented or suggested:
Female Detector / Male Detector: Identifies gendered female or male terms respectively.
Gendered Words Detector: Identifies whether the gender of the candidate is being brought to attention unnecessarily.
Personal Life Detector: Letters of recommendation for women are more likely to mention facts about their personal life.
Publication Detector: Letters of recommendation for women are less likely to mention publications and projects.
Effort (vs Accomplishment) Detector: This one is from a very impactful paper in this topic by Trix and Psenka. Women are more likely to be described using effort words while men are more likely to be described as accomplished or having an innate ability to do the task required.
Superlative Detector: Letters for women are less likely to contain superlatives. If they do, they usually describe women in the context of emotional terms.
Conditional Superlative Detector: This deals with superlatives that are “hedged” by restricting the population to only women.

The summer work for this project is dealing primarily with implementing more detectors and making this into an easily accessible resource for people. Future work would deal with creating a google docs or Microsoft word plug in or website. A major issue in this work has been dealing with the lack of a cohesive data set of letters of recommendation. This is due to the difficulty in anonymizing letters of recommendation.

Below is a list of citations that was included in the genderbias repository for any desired future inquiry into this problem.

Publications, Projects, and Research
Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.

Superlatives
Dutt, K., Pfaff, D. L., Bernstein, A. F., Dillard, J. S., & Block, C. J. (2016). Gender differences in recommendation letters for postdoctoral fellowships in geoscience. Nature Geoscience, 9(11), 805.


Schmader, T., Whitehead, J., & Wysocki, V. H. (2007). A linguistic comparison of letters of recommendation for male and female chemistry and biochemistry job applicants. Sex Roles, 57(7-8), 509-514.

Nouns
Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.

Minimal Assurance
Isaac, C., Chertoff, J., Lee, B., & Carnes, M. (2011). Do students’ and authors’ genders affect evaluations? A linguistic analysis of medical student performance evaluations. Academic Medicine, 86(1), 59.


Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.

Effort
Deaux, K. & Emswiller, T., “Explanations of successful performance on sex-linked tasks: What is skill for the male is luck for the female,” Journal of Personality and Social Psychology 29(1974): 80-85


Isaac, C., Chertoff, J., Lee, B., & Carnes, M. (2011). Do students’ and authors’ genders affect evaluations? A linguistic analysis of medical student performance evaluations. Academic Medicine, 86(1), 59.


Steinpreis, R., Anders, K.A., & Ritzke, D., “The impact of gender on the review of the curricula vitae of job applicants and tenure candidates: A national empirical study,” Sex Roles 41(1999): 509-528

Personal Life
Isaac, C., Chertoff, J., Lee, B., & Carnes, M. (2011). Do students’ and authors’ genders affect evaluations? A linguistic analysis of medical student performance evaluations. Academic Medicine, 86(1), 59.


Madera, J. M., Hebl, M. R., & Martin, R. C. (2009). Gender and letters of recommendation for academia: Agentic and communal differences. Journal of Applied Psychology, 94(6), 1591.


Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.

Gender Stereotypes
Axelson RD, Solow CM, Ferguson KJ, Cohen MB. Assessing implicit gender bias in Medical Student Performance Evaluations. Eval Health Prof. 2010 Sep;33(3):365-85.


Eagly, A.H.; Karau, S.J., “Role congruity theory of prejudice toward female leaders,” Psychological Review 109, no. 3 (July 2002): 573-597.; Ridgeway, 2002.


Foschi M. Double standards for competence: theory and research. Ann Rev Soc. 2000;26:21–42.


Gaucher, D., Friesen, J., & Kay, A. C. (2011, March 7). Evidence That Gendered Wording in Job Advertisements Exists and Sustains Gender Inequality. Journal of Personality and Social Psychology.


Hirshfield LE. ‘‘She’s not good with crying’’: the effect of gender expectations on graduate students’ assessments of their principal investigators. Gender Educ. 2014;26(6):601–617.


Madera, J. M., Hebl, M. R., & Martin, R. C. (2009). Gender and letters of recommendation for academia: Agentic and communal differences. Journal of Applied Psychology, 94(6), 1591.


Ross DA, Boatright D, Nunez-Smith M, Jordan A, Chekroud A, Moore EZ (2017) Differences in words used to describe racial and gender groups in Medical Student Performance Evaluations. PLoS ONE 12(8): e0181659.


Sprague J, Massoni K. Student evaluations and gendered expectations: what we can’t count can hurt us. Sex Roles. 2005;53(11):779–793.


Steinpreis RE, Anders KA, Ritzke D. The impact of gender on the review of the curricula vitae of job applicants and tenure candidates: a national empirical study. Sex Roles. 1999;41(7):509–528.


Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.


Wenneras C, Wold A. Nepotism and sexism in peer review. Nature. 1997;387(6631):341–343.

Raise Doubt
Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.


Madera, J. M., Hebl, M. R., Dial, H., Martin, R., & Valian, V. (2019). Raising doubt in letters of recommendation for academia: gender differences and their impact. Journal of Business and Psychology, 34(3), 287-303.

Shorter Length of Letters
Dutt, K., Pfaff, D. L., Bernstein, A. F., Dillard, J. S., & Block, C. J. (2016). Gender differences in recommendation letters for postdoctoral fellowships in geoscience. Nature Geoscience, 9(11), 805.


Trix, F. & Psenka, C., “Exploring the color of glass: Letters of recommendation for female and male medical faculty,” Discourse & Society 14(2003): 191-220.

Hypergraph Algorithms

Hey everyone! I am Alex Richter. I am a rising junior in math and computer science working with Anna Ritz this summer on hypergraph algorithms. The algorithms that I implemented this summer were from a paper co-authored by Nick Franzese, Anna Ritz, and Adam Groce [1]. The implementation of the algorithms already existed in Python; however, I learned Java and implemented them in Java to be incorporated into a plugin written in Java. 

We are working on incorporating the algorithm into the application Cytoscape via a plugin called ReactomeFIViz. In doing so, we hope that scientists will begin to examine hypergraph topology when checking to see if two molecules are connected within a cell. In viewing a different topology as opposed to simply directed or bipartite graphs, different pathways can be discovered as well as some nodes in a graph topology might be connected within a cell but not in a hypergraph topology. The potential for hypergraph connectivity in better understanding connectivity within a cell is a field that deserves to be further explored. 

The algorithm that I implemented relaxes the definition of B-connectivity and determines if two nodes are connected in a hypergraph topology. Here are some definitions from the paper: 

  • Directed hypergraphs represent reactions with many-to-many relationships, where each hyperedge e = (Te, He) has a set of entities in the tail Te and a set of entities in the head He (here, the tail denotes the hyperedge inputs and the head denotes the hyperedge outputs). [1]
  • B-connectivity: requires all the nodes in the tail of a hyperedge to be visited before it can be traversed. This definition has a natural biological meaning in reaction networks: B-connectivity requires that all reactants of a reaction must be present in order for any product of that reaction is reachable. Unlike the compound graph rules, B-connectivity describes the strictest version of connectivity, and it is the only rule used to traverse hyperedges. [1]
    • Given a directed hypergraph and a source set SV, a node uV is B-connected to S if either (a) uS or (b) there exists a hyperedge e = (Te, He) where uHe and each element in Te is B-connected to S. We use to denote the set of nodes that are B-connected to S in . [1]

To be B-connected implies that in order to traverse a hyperedge all of the nodes in the tail set (input set) must be present. B-connectivity can be a little bit difficult to understand at first glance. It is a harsh restriction on hypergraph connectivity. Two nodes are B-connected if one of two things is true: either both nodes are in the source set so obviously they are B-connected or the node we are seeing if is B-connected is located in the head (output) set of a hyperedge that where each element in the tail (input) set is connected to original node. 

b_visit Algorithm 

The algorithm that I implemented is two parts. The first algorithm is called b_visit and takes in a hypergraph and a set of source nodes, s. The b_visit algorithm will then return three sets: the set of traversed hyperedges from s, the set of B-connected nodes to s, and the set of restricted hyperedges. Restricted hyperedges are edges that can be reached but not traversed because it is missing an input in the tail to make the hyperedge traversable. For example, given a source set of A, B for this hypergraph, the set of restrictive hyperedges would be: 

Tail Set Head Set 
G, DL
D, H M
I, N O
Set of Restrictive Hyperedges

The set of traversable hyperedges would be: 

Tail Set Head Set 
AC
A, BD
BE
BF
E, FI
Set of Traversable Hyperedges

The set of B-Connected nodes to A, B would then be: C, D, E, F, I. 

Image taken from paper [1].

b_relaxation Algorithm 

The b_visit algorithm helps when trying to find out which nodes are B-connected to the source set, or the set of nodes the algorithm is called on. This is helpful to find those immediate nodes; however, since the requirements for B-connectivity are so strict, this does not include all of the nodes that can be reached from the source set in the hypergraph. Through calling b_visit multiple times and replacing the requirements for B-connectivity, more nodes can be viewed as connected in the hypergraph. 

The second algorithm from the paper is called b_relaxation which makes calls to the b_visit algorithm. The input for b_relaxation is once again the hypergraph and a set of source nodes, s, on which the algorithm will be called. Through multiple calls to b_visit, b_relaxation relaxes the strict conditions of B-connectivity. The algorithm runs on the entire hypergraph to determine which hypernodes are B-connected to the source set. The return value of the algorithm is a dictionary of k-distances where the key is the node and the value is the distance. K-distance is easiest understood as the number of times the algorithm was run until it could reach the node in question. So if it was in the source set for instance, the k-distance for the node would be 0. In relaxing the harsh conditions of B-connectivity, the algorithm is essentially traversing restrictive hyperedges to find all of the nodes that are connected in the hypergraph. 

Using the same toy example, at the end of b_relaxation, it would return a dictionary of distances. 

KeyValue
A0 (source set) 
B0 (source set) 
C0
D0
E0
F0
GINFINITY
HINFINITY
I0
JINFINITY
KINFINITY
L1
M1
NINFINITY
O1
P2
Q2
R3
Dictionary of distances for toy hypergraph after running b_relaxation algorithm
Image take from paper [1].

Anna and I are working with our collaborators in incorporating this into ReactomeFIViz before adding it to Cytoscape. The next steps include cleaning up the code, making sure it works properly in the plugin, and considering implementing further hypergraph algorithms. 

References: 

[1] Franzese N, Groce A, Murali TM, Ritz A (2019) Hypergraph-based connectivity measures for signaling pathway topologies. PLOS Computational Biology 15(10): e1007384. https://doi.org/10.1371/journal.pcbi.1007384