Research And References


Want to learn more about how natural language processing works, or why feedback is important, or just how we figured out that we can get all this information from a couple hundred words of feedback? Well, we used a bunch of studies and research in charting the ideas and technology behind this platform, so now you can nerd out with us.

Feedback For Good
How Feedback Helps Everyone

Why We Don’t Come: Patient Perceptions on No-Shows
Lacy, N.L., et al
A qualitative study of clinic patients who missed an appointment yields transferrable, linguistic-based information about the impact of emotion and perception of respect on the client’s willingness to follow through with a planned activity, or if they will return to the facility again should the need arise. While not a nonprofit, this helps demonstrate the effect of emotion and experience on if they will return for continued services. Feedback (and psychometric analysis) gives this information to providers so they can plan better and create environments that encourage the person to follow through or return.

Read the Study

Local Consumer Review Survey 2020
An analysis of how a potential participant or consumer will lean towards businesses which have at least one relatable review online, as well as what they hope for if there is to be continued engagement (even if the review was less than positive). People act the same way when seeking referrals for services, preferring organizations which have a review about what they might expect from someone with whom they feel they can relate.

Read the Analysis

Voice of the Customer 2021
MacDonald, S.
A corporate understanding of how proper, comprehensive collection and analysis of user feedback can significantly multiply returns. Returns might be repeat visitors, referrals, funding, and other KPIs. This blog post contains additional source information and studies as reference.

Read the Blog

Providing the Providers
How Nonprofits and Service Providers Can Use Quantitative Data

Eusocial Theory of Suicide Risk: Clinical Presentations and Commentary
Kirk, K., and Branson, Y.
A clinical look at the effects service providers, counseling, and community have on the Veteran population. This paper notes the importance of finding social and health opportunities locally, and with those in the same population. This commentary also makes note of how Pathfinder Labs can help both Veterans and providers through the feedback process.

Read the Commentary

Nonprofits and Evaluation: Empirical evidence from the field
Carman, J., and Fredericks, K.
This study, published in 2008, defines metrics and data nonprofits can collect and, with help, analyze for optimizing their impact and performance in the community. The authors acknowledge most data is used for marketing, fundraising, and internal assessment and offer there are ways to fully capitalize on information.

Read the Article

Accounting and Economic Biases in Donations to NGO's: is Scaling the New Overhead Myth in Charity Giving?
Byrd, J. and Cote, J.;
A 2017 analytic look at the common method for measuring nonprofit effectivenes: assessing the administrative expenses versus funding spent on programming along with the ability to scale programs. The researchers examine how these elements, when used as the defining metric, can significantly hamper the ability for programs to be effective regardless of size or scale.

Read the Analysis

Playing the Numbers Game
Garvin, S. et al;
A 2016 examination of the complexities and various factors that are overlooked when the operations-to-programming ratio is overused in measuring nonprofit performance. The study also posits better, more reliable metrics for understanding the impact programs have on the population they serve.

Read the Study

Military and Veterans in the Workplace
Feedback for Corporate Employee Support

Workplace Perceptions of Veterans and Nonveterans in the Workplace
Teclaw et al;
A research study in 2016 with milvets employed with the Department of Veterans Affairs found that Military and Veterans in the workplace are often less satisfied with their civilian employment when it came to most areas of interpersonal engagement. Programming for socializing, support, and community-building may have a significant impact on retention. This community prefers to communicate with and trust within its population group, indicating programs that provide more opportunities for this type of interaction at work would help increase employee satisfaction.

Read the Research

Workplace Social Support in Job Satisfaction Among Veterans
Harris et al;
A 2017 scientific study discovered Veterans who lack resources which meet their social, mental health, and general wellness needs are more likely to miss work and have deteriorating work functionality. The introduction of specific programs and the inclusion of anonymous, external resources and services for the targeted community may assist in increased productivity and overall quality of employees.

Read the Study

Rand: More Research on Veteran Employment
Batka and Curry Hall
A 2016 Rand Corporation study examined how Military members and Veterans hired as employees offer substantial benefits to companies, but are not often assessed as a group. Better understanding of how they adjust and contribute - through targeted surveys and analysis - could likely lead to better recruitment and hiring decisions.

Read the Article

Technology Validation
NLP as a Reliable Tool

NOTE: This is where it gets really deep in the data forest, so for this section we added a little snippet about applying the research.

Applying a Computer-Assisted Tool for Semantic Analysis of Writing: Uses for STEM and ELL
Smith-Keiling, B., and Hyun, H.
This study creates a controlled group assessment for creating a corpus that can identify development and cognitive process in students. By developing the specific group of words for the different student segments, the researchers found they could identify specific patterns unique to the students.

How We See It: Pathfinder Labs has built a corpus for the Veteran population with the aggregated feedback of its members. Given the evidence in this study we can - as the corpus continues to grow - extrapolate to the Military, Veteran, Spouse, and more subsets of the population to get more and more accurate probability assessments.

Read the Study

Personality in 100,000 Words: A large-scale analysis of personality and word use among bloggers
Yarkoni, T.
A quantitative analysis of how written language in opinion and other unedited content correlates to results using word count and syntax processors, and how this process can reasonably be used to measure personality attributes.

How We See It: Simple validation that opinion content can create personality assessments! We just need more and more feedback for more and more words...

Read the Analysis

Ethical considerations of the GermEval20 Task 1. IQ assessment with natural language processing: Forbidden research or gain of knowledge?
Johannssen, D., et al
This paper examines inherent bias in IQ and other traditional tests such as self-reported personality scoring, regardless of validation. It notes that AI and NLP can assist in minimizing or eliminating such bias to measure traits, but results should be used ethically.

How We See It: Boy do we hate statistical and survey bias in reporting, and boy do we hate the misuse of personality assessments. We used this and similar studies to build our statistical models around reducing bias, and to continue enhancing privacy and security to avoid misuse of individual or population information.

Read the Comparison

What do we know when we LIWC a person? Text analysis as an assessment tool for traits, personal concerns and life stories.
Pennebaker, J. et al
This meta-analysis compares several methods of text analysis for verification, validation, and system. It also examines syntax and connecting words, and how they are linked to understanding emotion and personality.

How We See It: This is a good explainer for understanding how text analysis leads to personality traits, and why we need AI and refining algorithms like the ones we developed to make it all work.

Read the Meta-Analysis

Psychological Aspects of Natural Language Use: Our Words, Our Selves
Chung, C.K., and Pennebaker, J.
This document - part of a larger handbook on personality and metrics - defines LIWC and defines its use in several aspects of personality assessments, including identifying areas for future study.

How We See It: Dr. Pennebaker is one of the grand poobahs (technical term) for linguistic analysis, and this article plus the handbook it came from is a basis for how we came up with what Pathfinder Labs could do.

Read the Chapter

The Psychological Meaning of Words: LIWC and Computerized Text Analysis Methods
Tausczyk, Y., and Pennebaker, J.
This paper details the validation of the Linguistic Inquiry and Word Count (LIWC) system upon which much of the Pathfinder Labs analysis is based.

How We See It: A validation study is just that, giving us evidence to use as a platform we can build upon. By offering statistical evidence that the basics work, it provides sound footing for us to create the rest of the Pathfinder Labs algorithms, machine learning processes, and intellectual property.

Read the Validation

Natural Language Analysis and the Psychology of Verbal Behavior: The Past, Present, and Future States of the Field
Boyd, R., and Schwartz, A.
A general definition paper on the history and future of natural language processing techniques to understand and define personality traits, including cautions and pitfalls.

How We See It: Some of the considerations we wanted to keep in mind as we built the technology, like the limitations of probability and the issues with things like shorter text responses.

Learn More About NLP