1

KFP Project

Anna Bird, Ashley Owlett, Erica Cucinella, Sicilia Randazzo, Ira Hogg

College of Social Work, University of South Carolina

SOWK 352-001: Intro to Field Education

Dr. Owens

April 24, 2023

Introduction

Anxiety disorders are the most common mental disorders affecting individuals in the

United States (Brown, 2016). The American Psychological Association defines anxiety as "an

emotion characterized by apprehension and somatic symptoms of tension in which an individual

anticipates impending danger, catastrophe, or misfortune (American Psychological Association,

2023)." Anxiety can lead the body to meet the assumed threat, with muscles becoming tense,

breathing faster, and the heart beating more rapidly. College students dealing with anxiety is an

ongoing struggle continuing to worsen. According to the American College Health Association,

in 2021, 63% of college students in the United States felt overwhelming anxiety (LeBlanc and

Marques, 2019). Behavioral Medicine clinicians at Boston University conducted a survey. They

found that 21.9% of students claimed that in 2016 anxiety affected their academic performance,

including receiving a lower grade on an exam or project, an incomplete, or dropping a course.

This number increased by 18.2% from a 2008 survey (Brown, 2016). The most common causes

of anxiety among college students are financial obligations, social interactions, adjusting to the

college atmosphere, and academic workload, which can affect academic success, self-esteem,

physical health, and the ability to sleep (Naceanceno et al., 2021).

Across the United States, female college students experience stress and anxiety daily.

Specific phobias, social anxiety disorder, and generalized anxiety disorder are the most common

anxiety orders affecting women in the United States (ADAA, 2023). Further research claims

anxiety disorders occur twice as often in women than men (Willett, 2011). Focusing on female

college students allows us a more intimate lens as we look into their situations and learn more

about their mental and emotional struggles in school. To improve female college students' mental

and emotional well-being, researchers must compare the effectiveness of the two interventions
 3

for reducing anxiety. With exhaustive methods, researchers may have a better chance of finding

help and offering guidance for these young women based on their experiences.

Anxiety is typically treated by the on-campus health center or student wellness center on

college campuses. This type of intervention is less effective because college campuses generally

need help to keep up with the demand for health services (Field, 2022). Although colleges have

used one-on-one counseling for years with moderate success, it takes weeks to access counseling

services on university campuses, so it is not accessible. According to the latest numbers from the

Center for Collegiate Mental Health, caseloads among United States universities have an average

of 90 clients per counselor per year (Field, 2022). In 2019, 90% of counseling center directors

reported an increase in students seeking services (Abrams, 2019). Because of this increased

number of students seeking services, health centers often have to set session limits and space out

appointments. Further, due to the coronavirus pandemic, subsequent budget cuts and financial

downturns nationwide have led student mental health issues to go unaddressed as there are

increased levels of need from students (Abrams, 2019).

If college institutions implement diverse intervention options to address female student

anxiety disorders, then the frequency and severity of this issue would decrease for those utilizing

those interventions. Understanding the efficacy of fully automated therapy would allow female

college students 24/7 access to therapy services. Therapy would be readily available and

convenient if an individual struggles with anxiety and cannot make an in-person appointment.

Our study focuses on if traditional-based counseling in colleges or Cognitive Behavior Therapy

(CBT) via a fully automated conversational agent is more effective in lowering rates of anxiety

among female college students (Fitzpatrick et al., 2017). An automated conversational agent
 4

delivers cognitive behavior therapy through brief, daily conversations and mood tracking within

an instant messenger app (Fitzpatrick et al., 2017).

Understanding the efficacy of this intervention will impact the target population in that

CBT will be more easily accessible and lead to lower anxiety levels among female college

students. The research question we aim to answer is when narrowing the scope to traditional

counseling methods versus CBT via a fully automated conversational agent (Woebot), which

technique will prove most successful among female college students? Our hypothesis is that if

college institutions implement diverse intervention options to address female student anxiety

disorders, the frequency and severity of this issue would decrease for those utilizing those

interventions.

Literature Review

Traditional Cognitive Based Therapy

The Center for Collegiate Mental Health at Penn State University conducted a study

where 1,454 therapists saw 9,895 students across 108 university counseling centers. They were

assessed by routine psychological treatment (CCAPS) for symptoms of depression, generalized

anxiety, social anxiety, eating concerns, hostility, and alcohol use (CCMH, 2019). Throughout

the study, the university counseling centers measured the change during and after treatment and

the effectiveness of counseling services provided by university counseling centers (CCMH,

2019).

Results of this study indicate students who entered university counseling centers with

elevated levels of concern (depression, generalized anxiety, social anxiety, eating concerns,

hostility, and alcohol use) improved over time with the implementation of counseling services

(CCMH, 2019). Clinicians working at university counseling centers were very effective at
 5

treating students, and results showed that after treatment, students do not return for distress upon

receiving services (CCMH, 2019). For example, 52.6% of students reliably improved, including

24.1% of students with generalized anxiety (CCHM, 2019).

In a study by Farabaugh et al. (2018), thirty students, ages 18-24, with a clinical

diagnosis of mild to severe depression were recruited from several local universities in the

Boston area to participate in a one-group pretest post-test intervention to determine if a 12-week

CBT intervention reduced depressive symptoms. These outcomes were measured at baseline and

post-test using the Beck Depression Inventory, which includes a 21-item measure with questions

about core symptoms of depression. 76.6% of the participants were female. The findings from

this study included significantly decreased depressive severity, suicidality, anxiety, frequency of

negative cognitions, and dysfunctional attitudes once students completed therapy (Farabaugh et

al., 2018).

To dive deeper into the effects of cognitive behavioral therapy on young adult women,

we will also look at the impact on pregnant women as they fit into our chosen demographic. In

an article titled "Cognitive behavioral therapy and pharmacotherapy for anxiety: Treatment

preferences and credibility among pregnant and non-pregnant women," Joanna Arch examines

the most effective therapy for lessening anxiety symptoms (Arch, 2014). Recruitment for this

experiment was fulfilled by posting links on pregnancy and women-related websites, CraigsList,

and a site called Mechanical Turk (Arch, 2014). These links connected women to the study

consent and survey, where they indicated that they met the inclusion criteria: female, currently

pregnant, English-speaking, and at least 18 years old (Arch, 2014). For the non-pregnancy

portion of the study, women had to meet the same criteria but also indicate they were not

currently pregnant and were between the ages of 18 and 44, to match the age range of the
 6

pregnant sample (Arch, 2014). Of the participants recruited, 377 pregnant women and 399 non-

pregnant women completed the study.

Treatment credibility and willingness were measured to determine the participants' desire

for cognitive behavioral therapy when battling anxiety. The study consisted of one-page

descriptions of multiple treatments for anxiety, including CBT and pharmacotherapy. The goal

was to present a thorough treatment rationale and describe the risks and benefits of each

treatment (Arch, 2014). The participants then filled out a questionnaire rating their willingness to

participate in each form of therapy. Women, pregnant and non-pregnant, both preferred anxiety-

related treatment that included psychotherapy. In response to treatment descriptions, both groups

rated CBT more favorably than pharmacotherapy on treatment willingness, credibility, and

concern, with this preference significantly greater among pregnant than non-pregnant women

(Arch, 2014). Pregnancy status was unrelated to CBT ratings.

In concluding the study, pregnant and non-pregnant women chose CBT over all other

treatment options. In the future, in this particular research, the status of being pregnant did not

pose a significant difference in the desire for CBT as their chosen form of therapy. The

effectiveness of CBT outweighed other treatment plans. Helping the participants focus more on

restructuring their cognitions to improve anxious behavior proved the most effective choice.

CBT via a conversational agent

In a research study, Attridge, Morfitt, Roseborough, and Jones examined the impact of

the Self-Directed Internet-Based Cognitive-Behavioral Therapy Model and college students with

anxiety and other illnesses. The study was conducted among 951 college students from four

colleges and universities in the Midwest region of the United States (Attridge et al., 2020). Over

six months, students in the study were measured through a Self-Directed Internet-Based
 7

Cognitive Behavioral Therapy Model for clinical symptoms and put into different groups based

on their condition: depression, stress, anxiety, social anxiety, and insomnia (Attridge et al.,

2020). Various factors were examined, such as age, gender, comprehensive preliminary

assessment, duration, live coach support, and peer support (Attridge et al., 2020).

The results from the study show that students who used the Self-Directed Internet-Based

Cognitive Behavioral Therapy Model improved with more than one session in areas such as

support, satisfaction, academic outcomes, and successful integration into university counseling

(Attridge et al., 2020). Significantly lower rates of stress and anxiety levels were found in more

male college students than female college students (Attridge et al., 2020). Overall, the Self-

Directed Cognitive Behavioral Therapy Model is promising as a supplement to traditional in-

person counseling services at colleges and universities (Attridge et al., 2020).

Furthermore, another study, conducted by Gabrielli et al. (2021), aimed to determine if a

psychoeducational chatbot successfully supported healthy coping with stress and anxiety among

university students. One group of 71 university students was enrolled voluntarily and asked to

use the Atena psychoeducational chatbot 2 times per week for 4 weeks. Over half (67.6%) of the

students in the study were female, and all students were first-year college students, with an

average age of 20.6 years old (Gabrielli et al., 2021). Evidence-based approaches and

intervention strategies informed the conversations between Atena and the user. The goal of the

conversations was to support the participant in self-reflecting on personal thoughts and emotions

experienced in stressful settings. One limitation of this study is that it reports data from the

preliminary evaluation of a proof-of-concept chatbot intervention targeting a homogeneous

population of university students without a control group. Also, the study was conducted during

the second wave of the coronavirus pandemic, which increased rigid restrictions on social
 8

behavior and educational practices that might have strongly impacted our participants' mental

well-being and reduced the intervention's positive effect.

Results indicated that chatbots effectively deployed to university students positively

affect their mental state. Results also showed Atena was effective because there was a significant

improvement in the ability of participants to describe and accept their emotions and practice

mindfulness and self-reflection based on conversations with the chatbot (Gabrielli et al., 2021).

Online stress management interventions were shown to be more effective for students with

higher levels of stress, anxiety, and depression than traditional in-person counseling (Gabrielli et

al., 2021).

These findings were also supported in a case study conducted by Fitzpatrick, Darcy, and

Vierhile (2017), which examined the efficiency of web-based Cognitive Behavior Therapy as a

fully automated conversational agent (Woebot) in treating college students experiencing anxiety

symptoms and depression. The study examined seventy individuals aged 18-28 recruited from an

online university platform. Of the research pool, 67% of the participants were female (Fitzpatrick

et al., 2 017). Using the General Anxiety Disorder-7 online screening tool, the data indicated a

significant reduction in anxiety levels from participants utilizing the automated conversational

agent. The results yielded greater emotional awareness among college students using CBT in

brief, daily conversations and mood tracking (Fitzpatrick et al., 2017). Participants expressed that

the required daily check-ins and the empathic manner of the automated bot encouraged personal

accountability and emotional insight. With over two-thirds of the participants female, the study

denotes CBT via Woebot to effectively reduce anxiety levels among female college students.

This study was not without limitations. Participating individuals expressed concerns with

the Woebot's inability to understand specific responses from the members or produce replies to
 9

unexpected answers. Members also noted a repetitive nature to some encounters preventing the

conversations from feeling natural. Subsequently, multiple participants reported technical issues

and glitches that hindered their interactions. Considering this study was conducted at one

university with a small research group, the results may not be representative of all college

students experiencing anxiety symptoms.

These studies suggest that cognitive-based therapy conducted by an online agent can be

effective when used to help female college students combat anxiety. Among the population

studied, CBT via a conversational agent provided greater emotional awareness and improved

overall mental health. If college institutions implement diverse intervention options to address

female student anxiety disorders, the frequency and severity of this issue would decrease for

those utilizing those interventions.

Methods

Research Design

This research aims to determine if cognitive-based therapy via a conversational agent is

more effective than traditional CBT for lowering the rates of anxiety among female college

students. We will use a quasi-experimental design and utilize a pre, mid, and post-survey to

determine the effectiveness of each intervention. This design is the best option for our research

because it demonstrates causality between an intervention and an outcome (Harris et al., 2006).

Throughout the study, it allows for measurements of non-randomly selected control groups of

our demographic. By considering the case study's financial aspect, this design proves the most

cost-efficient to facilitate our study. We will collect qualitative data by using surveys to assess

the participant's progress and compare the results from the group utilizing CBT via a

conversational agent and traditional CBT.

While this design is most effective for our research, it has strengths and weaknesses. The

strengths of this approach are that it is inexpensive and more feasible to conduct, and many

flexible design options can be utilized depending on the content of the study. The quasi-

experimental design can also administer significant evidence without randomization (Office for

Health Improvement and Disparities, 2021). Weaknesses are the need for a control group,

making it difficult to determine if the intervention method caused the results. It also does not

consider pre-existing variables outside the experiment that could have influenced the

participant's responses. Threats to internal validity include maturation and attrition. Maturation

refers to an individual's growth and development, whether physical, emotional, or mental

(Saleem, 2019). Factors that play a role in our study include tiredness, loss of patience, and

boredom, particularly among college students, which could impact the validity of the research.

These factors could also lead to attrition or participants dropping out of the study or not

participating (Saleem, 2019).

Sampling Design

The ideal sample size for this study is 30 participants. This relatively small sample size is

most effective as accurate measurements can be made, and errors and biases can be easily

controlled and identified. 15 participants will receive CBT via a conversational agent, and the

remaining 15 will receive traditional cognitive-based therapy. The sampling method that will be

used is non-probability quota sampling. The sample of participants will be obtained voluntarily

from students at the University of South Carolina. The University of South Carolina's counseling

center will post a QR code around campus with a quick questionnaire about the participant's

demographics and choose 30 students who fit the criteria. Non-probability sampling is necessary
 11

because the study requires a specific demographic for this study: college-aged females with

anxiety symptoms.

Non-probability sampling, specifically quota sampling, is the best choice for our

research. Quota sampling focuses on a clear and specific population, and our particular group is

female college students with anxiety (Statistics How To, 2022). The strengths of non-probability

sampling are that, in general, it is fast, easy, cost and time-effective, and accessible for our

college demographic (Statistics How To, 2022). Finding respondents is easy and simplistic,

which increases the likelihood of our population participating. The weaknesses of non-

probability quota sampling are that it is subjective by nature because of the conductor's bias,

beliefs, and prejudices (Java T Point, n.d.). This aspect is critical in determining who will

participate in a study. To add, specifically, quota sampling needs multiple researchers. No two

people will have the same amount of competence; therefore, the conclusions drawn from the

study may not be consistent (Java T Point, n.d.).

Data Collection/Measurement

The survey we administer will assess the changes in the level of anxiety experienced by

each participant. This information will help us to determine which intervention is more effective

in lowering rates of anxiety among our target population. The 84-day research will reveal the

significance of traditional in-person CBT or counseling via an artificially intelligent chatbot.

Some clients may prefer a human presence facilitating the counseling sessions, while others may

opt for alone time with an automated app designed to act as the client's conversational agent.

When measuring anxiety levels during the pre, mid, and post-therapeutic intervention, the

dependent variable is the level of anxiety captured, and the independent variable is the two

comparable interventions being studied. A behavioral and cognition assessment survey will act
 12

as our measurement tool. The survey will ask questions about anxiety symptoms, with the

allowed answers being chosen on a numerical scale. The scale will display the lowest interval,

"1" as strongly disagree, "2" as disagree, "3" as unsure, "4" as agree, and the highest interval, "5"

as strongly agree. The answers will be tallied and totaled to equate a score we will refer to an

answer key. The answer key will determine the client's level of anxiety.

We will administer this survey 3 times: pre, mid, and post-intervention, and participants

will be separated into 2 groups. Half of the participants will participate in traditional in-person

CBT for 12 weeks, and the other 15 will participate in CBT via an online conversational agent

application for 12 weeks. The first survey will be taken before engaging in these interventions to

gauge the anxiety level of the participants before partaking in the study. Then, there will be a 6-

week check-in where the same survey is administered. This check-in will show signs of

resilience or resistance from the participants. Also, it will allow us to assist anyone facing

difficulties using the non-traditional CBT method. Lastly, a post-intervention survey will

determine which intervention was more successful in lowering anxiety rates. It will assess

participants' opinions on technical ease with using the conversational app and time management

in traditional in-person sessions. This particular data will identify the better counseling solution,

depending on many variables that could hinder clients from receiving treatment for anxiety.
 13

References

Abrams, Z. (n.d.). A crunch at college counseling centers. Monitor on Psychology. Retrieved

February 21, 2023, from https://www.apa.org/monitor/2020/09/crunch-college-counseling

Anxiety - American Psychological Association. (n.d.). Retrieved February 6, 2023, from

https://www.apa.org/topics/anxiety/

Arch, J. J. (2014). Cognitive behavioral therapy and pharmacotherapy for anxiety: Treatment

preferences and credibility among pregnant and non-pregnant women. Behaviour Research

and Therapy, 52, 53–60. https://doi.org/10.1016/j.brat.2013.11.003

Attridge, M. D., Morfitt, R. C., Roseborough, D. J., & Jones, E. R. (2020, July 23). Internet-

based cognitive-behavioral therapy for college students with anxiety, depression, social

anxiety, or insomnia: Four single-group longitudinal studies of archival commercial data

and replication of employee user study. JMIR Formative Research.

https://formative.jmir.org/2020/7/e17712/

Bishop, J. B., Bauer, K. W., & Becker, E. T. (1998). APA PsycNet. 502 Bad Gateway.

https://psycnet.apa.org/record/1998-00588-007

Brown, J. (2016, October 2). Anxiety: The most common mental health diagnosis in college

students: Bu Today. Boston University. Retrieved February 6, 2023, from

https://www.bu.edu/articles/2016/college-students-anxiety-and-depression/CCMH. (2019,

June 19). Effectiveness of college counseling centers. Home.

https://ccmh.psu.edu/index.php?

option=com_dailyplanetblog&view=entry&year=2022&month=05&day=31&id=25:effecti

veness-of-college-counseling-centers
 14

Facts & Statistics: Anxiety and Depression. Facts & Statistics | Anxiety and

Depression. (n.d.). Retrieved February 21, 2023, from https://adaa.org/understanding-

anxiety/facts-statistics

Farabaugh, A., Nyer, M.B., Holt, D.J. et al. CBT Delivered in a Specialized Depression Clinic

for College Students with Depressive Symptoms. J Rat-Emo Cognitive-Behav Ther 37, 52–

61 (2019). https://doi.org/10.1007/s10942-018-0300-z

Field, K. (2022, May 11). Inside a health center struggling with the College Mental Health

Crisis. The Hechinger Report. Retrieved February 21, 2023, from

https://hechingerreport.org/inside-a-college-counseling-center-struggling-with-the-student-

mental-health-crisis/

Fitzpatrick, Darcy, A., & Vierhile, M. (2017). Delivering Cognitive Behavior Therapy to

Young Adults With Symptoms of Depression and Anxiety Using a Fully Automated

Conversational Agent (Woebot): A Randomized Controlled Trial. JMIR Mental Health,

4(2), e19–e19. https://mental.jmir.org/2017/2/e19/

Gabrielli, Silvia, et al. “Engagement and Effectiveness of a Healthy-Coping Intervention via

Chatbot for University Students during the COVID-19 Pandemic: Mixed Methods Proof-

of-Concept Study.” JMIR MHealth and UHealth, JMIR Publications Inc., Toronto, Canada,

28 May 2021, https://mhealth.jmir.org/2021/5/e27965.

Harris, A., McGregor, J. C., Perencevich, E. N., Furuno, J. P., Zhu, J., De, P., & Finkelstein, J.

(2006). The Use and Interpretation of Quasi-Experimental Studies in Medical Informatics.

Journal of the American Medical Informatics Association, 13(1), 16–23.

https://doi.org/10.1197/jamia.m1749
 15

Java T Point. (n.d.). Advantages and disadvantages of non-probability sampling - Javatpoint.

www.javatpoint.com. https://www.javatpoint.com/advantages-and-disadvantages-of-non-

probability-sampling

LeBlanc, N., & Marques, L. (2019, August 27). Anxiety in college: What we know and how

to Cope. Harvard Health. Retrieved February 6, 2023, from

https://www.health.harvard.edu/blog/anxiety-in-college-what-we-know-and-how-to-cope-

2019052816729

Lewin, T. (2011, January 27). Record level of stress found in college freshmen. The New

York Times. Retrieved April 5, 2023, from

https://www.nytimes.com/2011/01/27/education/27colleges.html?_r=1

Mcalpine, K. (2021, February 17). Depression, anxiety, loneliness are peaking in college

students. Boston University. Retrieved April 5, 2023, from

https://www.bu.edu/articles/2021/depression-anxiety-loneliness-are-peaking-in-college-

students/

Naceanceno, K. (2021). A comparison of anxiety levels among college students - harding

university. Retrieved February 22, 2023, from

https://scholarworks.harding.edu/cgi/viewcontent.cgi?article=1013&context=jger

O’Connell Kent, Jackson, A., Robinson, M., Rashleigh, C., & Timulak, L. (2021). Emotion‐

focused therapy for symptoms of generalised anxiety in a student population: An

exploratory study. Counselling and Psychotherapy Research, 21(2), 260–268.

https://doi.org/10.1002/capr.12346

Quasi-experimental study: comparative studies. (2021, September 8). GOV.UK.

https://www.gov.uk/guidance/quasi-experimental-study-comparative-studies
 16

Saleem, A. (2021, June 4). Threats To Internal Validity – 9 Biggest Threats To Research

Validity In 2020. The Life Virtue. https://www.thelifevirtue.com/threats-to-internal-

validity/

Statistics How To. (2022, October 29). Non-probability sampling: Definition, types.

https://www.statisticshowto.com/probability-and-statistics/sampling-in-statistics/non-

probability-sampling/

Willett, C. (2011, March 14). Female students more likely to experience anxiety: What can

you do?Strong Women Strong Girls. https://swsg.org/female-students-more-likely-to-

experi ence-anxiety-what-can-you-do/