THE ROLE OF MINDFULNESS, PERCEIVED DISCRIMINATION, AND - RELATED DISTRESS IN PREDICTING HEALTH BEHAVIORS AND GLYCEMIC CONTROL

Leah M. Bogusch

A Dissertation

Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

August 2020

Committee:

William H. O'Brien, Advisor

David Tobar Graduate Faculty Representative

Abby Braden

Howard Casey Cromwell ii

ABSTRACT

William H. O’Brien, Advisor

For persons with diabetes, adherence to treatment recommendations, such as medication

adherence and following diet and exercise guidelines, is often difficult and subject to multiple

influences, including psychological well-being and social stressors. Such influences include self- reported microaggressions, mindfulness, depressive symptoms, self-care behaviors, and glycemic control. A model of relationships between these variables was proposed for testing through structural equation modeling. 337 Participants over the age of 18 years with diagnoses of diabetes were recruited from Amazon’s Mechanical Turk website to take a survey of measures assessing these variables. Tests of the hypothesized model indicated poor fit, and the model was respecified to remove diet and exercise behaviors, which resulted in satisfactory fit. Between- groups differences were assessed to investigate potential differences in the model between participants. Findings were generally consistent with hypotheses that better psychological well- being and less frequent microaggressions would be associated with improved self-care behaviors,

including diet, exercise, and taking medication regularly. Some differences were noted in the

magnitude of relationships between participants from the United States or , but, the models

were generally similar between groups. Limitations included possible misrepresentation of

participant character, lack of health literacy, and use of cross-sectional data. This study informs

future research on interventions for improvements in treatment adherence for people with

diabetes, including interventions for improving mindfulness skills and interventions for

decreasing impact of microaggressions.

iii

Especially to my grandmother, Evelyn Jahr Greene, whose unshakeable faith in the power of

education touched everyone she met.

To my mother, Brenda, who always models tenacity, a deeply rooted sense of justice, and

outspokenness. Oyáhitike iná. To my father, John, who taught his daughter to love science and

math and that every problem has a solution. To my brother, Christopher, who saved my sanity countless times with his ability to find the humor in any situation. Nína iĥámayaye s’a misúŋka.

To my partner, Ryan Buckman, who made as many sacrifices as any graduate student while I completed my degree and encouraged me many times to continue writing, even when I wanted to

give up. iv

ACKNOWLEDGMENTS

Many thanks to my dear friends, Serena Wong and Mary Moeller, whose assistance with statistics, proofing, and preparing this manuscript was invaluable. I am grateful to my mentors

Drs. Deborah Altschul, Deidre Begay, and Lindsay Smart for helping me to prepare for my defense, for offering many resources, and especially for all their wise, creative advice and encouragement. Thanks to my committee members, Drs. Abby Braden, Casey Cromwell, and

David Tobar, for their helpful feedback and support. And thanks to my advisor, Dr. William

O’Brien. I will always be grateful for everything he has taught me about being a values-driven researcher, clinician, and mentor.

v

TABLE OF CONTENTS

Page

INTRODUCTION ...... 1

Perceived Discrimination and Well-Being ...... 2

Relationships to Depressive Symptoms and Mental Well-Being ...... 3

Relationships to Medication Adherence ...... 5

Relationships to Diet and Exercise ...... 6

Relationships to Glycemic Control ...... 8

Depressive Symptoms and Diabetes-Related Distress ...... 9

Medication Adherence ...... 10

Diet and Exercise ...... 12

Glycemic Control ...... 14

Diabetes-Related Distress ...... 15

Medication Adherence ...... 16

Diet and Exercise ...... 17

Glycemic Control ...... 18

Mindfulness and Well-Being ...... 20

Depressive Symptoms and Diabetes-Related Distress ...... 20

Diet and Exercise ...... 21

Glycemic Control ...... 23

Summary and Hypotheses...... 24

METHODS ...... 26

Participants ...... 26 vi

Measures ...... 26

Demographics ...... 26

Microaggressions ...... 26

Mindfulness...... 28

Depressive Symptoms ...... 29

Diabetes-Related Distress ...... 30

Health Behaviors ...... 30

Medication Adherence ...... 31

Glycemic Control ...... 32

Procedure ...... 32

Analysis Plan: Structural Equation Modeling...... 33

Rationale ...... 33

Statistical Programs ...... 33

Power ...... 33

Statistical Assumptions ...... 34

Descriptives...... 35

Specification ...... 36

Identification ...... 36

Model Estimation ...... 38

Respecification ...... 40

RESULTS ...... 41

Participant Characteristics ...... 41

Education and Socioeconomic Status ...... 43 vii

Diabetes Outcomes ...... 44

Statistical Assumptions ...... 46

Exploratory Between-Groups Analyses of Main Study Variables ...... 46

Mindfulness...... 47

Microaggressions ...... 47

Depressive Symptoms ...... 48

Diabetes-Related Distress ...... 48

Health Behaviors ...... 48

Medication Adherence ...... 48

Glycemic Control ...... 48

Time Since Diagnosis ...... 49

Time Since Last Physician’s Appointment ...... 49

Type of Medication ...... 49

Main Analyses ...... 49

CFA ...... 49

SEM ...... 50

Respecification ...... 51

Relationships Between Main Study Variables ...... 52

Hypothesis 1...... 52

Hypothesis 2...... 52

Hypothesis 3...... 53

Hypothesis 4...... 54

Hypothesis 5...... 54 viii

Exploratory Multigroup Analysis ...... 55

Mindfulness...... 56

Microaggressions ...... 57

Depressive Symptoms ...... 57

Diabetes-Related Distress ...... 57

DISCUSSION ...... 58

Limitations ...... 63

Implications and Future Directions ...... 68

Conclusions ...... 70

REFERENCES ...... 71

APPENDIX A. TABLES ...... 92

APPENDIX B. FIGURES ...... 113

APPENDIX C. RECRUITMENT SCRIPT ...... 120

APPENDIX D. INFORMED CONSENT...... 121

APPENDIX E. STUDY SCREENER...... 123

APPENDIX F. RACIAL AND ETHNIC MICROAGGRESSIONS SCALE ...... 124

APPENDIX G. CENTERS FOR EPIDEMIOLOGICAL STUDIES- DEPRESSION ...... 126

APPENDIX H. PROBLEM AREAS IN DIABETES ...... 127

APPENDIX I. FIVE FACET MINDFULNESS QUESTIONNAIRE ...... 128

APPENDIX J. HEALTH BEHAVIORS ...... 130

APPENDIX K. MORISKY MEDICATION ADHERENCE SCALE ...... 131

APPENDIX L. IRB APPROVAL LETTER ...... 132 1

INTRODUCTION

Type 1 and are chronic health conditions resulting in difficulty managing blood sugar levels, or glycemic control, that affect approximately 30.3 million people in the

United States. is developed in childhood and is caused by an autoimmune condition that prevents the body from making its own insulin. In contrast, type 2 diabetes manifests later on in adulthood and is typically attributed to the development of insulin resistance (Centers for Control and Prevention [CDC], 2017). It is estimated that Type 2 diabetes makes up approximately 90-95% of all diabetes cases and that it is one of the leading causes of excess morbidity and premature death (CDC, 2020).

Glycemic control in diabetes is generally treated through medication management, adherence to a diabetic diet, and regular exercise (CDC, 2020). In spite of these recommendations, many people struggle with maintaining such a lifestyle (Garcia-Perez et al.,

2013). Because treatment for type 2 diabetes has many substantial lifestyle changes that must be maintained over time, nonadherence is a significant problem. For example, in a review of a prescription database it was estimated that about 69% of people diagnosed with diabetes were adherent as defined by filling their prescriptions at least 80% of a calendar year (Kirkman et al.,

2015). Similarly, an analysis of another e-prescription database revealed that 22.9% of prescriptions written for diabetes were never filled (Fischer et al., 2010). Bailey and Kodack

(2011) argued that poor adherence to diabetes medication regimens is associated with

complexity, cost, poor understanding of risks and benefits of adherence, and depression. Since

behavioral adherence is a crucial part of maintaining health and managing glycemic levels for

people with diabetes, it is important to research reasons for nonadherence to recommended

regimens. 2

Moreover, health disparities have been noted for decades in the prevalence of diabetes, and being a person of color has been associated with worse glycemic control in the United

States. For example, prevalence rates are currently much higher among American Indian/Alaska

Natives (15.1%), non-Hispanic Blacks (12.7%), and Hispanics (12.1%), in comparison to non-

Hispanic Whites (7.4%) and Asians (8.0%) (CDC, 2020). Perceived discrimination has been linked with poorer health behaviors in general (Pascoe & Richman, 2009). In their meta-analysis of studies examining relationships between discrimination and health, Pascoe and Richman

(2009) reported a small effect size of -0.18 for the relationship between perceived discrimination and health behaviors, such as, sleep quality, diet, exercise, and medication adherence.

The following sections review the literature on potential predictors, including perceived discrimination, as well as mediators and moderators of glycemic control. These include perceived socio-environmental influences, such as perceived discrimination. Relationships between psychological factors, such as diabetes-related distress, depressive symptoms, and mindfulness are examined. Finally, behavioral predictors of glycemic control are discussed, especially self-reported medication adherence, dietary habits, and frequency of exercise.

Perceived Discrimination and Well-Being

One variable that could account for health disparities and poorer treament adherence is perceived discrimination. In comparison to lifetime prevalence rates of perceived discrimination, prevalence rates of perceived discrimination experienced in provider settings are generally low, ranging from 4%- 14% (Lyles et al., 2011; Peek et al., 2011; Piette et al., 2006;

Ryan et al., 2008). However, Gonzales and colleagues (2013) found that 67% of American

Indian women with type 2 diabetes reported experiencing some perceived discrimination from healthcare providers. 3

A longitudinal study by Fuller-Rowell et al. (2018) was conducted to collect data on socioeconomic disadvantage, self-rated health, and perceived discrimination at baseline, nine years, and eighteen years among a sample of adults throughtout the United States. They found a significant negative relationship between socioeconomic disadvantage and self-rated health, and that this relationship was mediated by perceived discrimination. To investigate whether these relationships could be explained by individual factors, the authors included neuroticism and negative affect in the model as predictors, but the model held. Similarly, in D’Anna and colleagues’ (2010) study of California residents, the relationship between perceived discrimination and physical limitations was consistent across all socioeconomic levels.

Perceived discrimination, particularly in healthcare settings, may be related to health outcomes for people with diabetes. These perceived social interactions may also have a relationship to individual psychological factors related to diabetes outcomes. Perceived microaggressions are a form of discrimination in which bias or stereotypes toward people of color or other underrepresented groups are expressed in subtle or unconscious ways (Nadal,

2011). Perceived microaggressions may be experienced on a frequent basis by people of color, and may contribute to poorer health through worse psychological health.

Relationships to Depressive Symptoms and Mental Well-Being

Research on relationships between perceived discrimination or microaggressions, diabetes-related distress, and depression is limited. LeBron and colleagues (2014) conducted a self-report survey of African Americans and Latinos with type 2 diabetes to investigate associations between these variables. They found that, among Latinos, perceived discrimination was associated with greater diabetes-related distress and more depressive symptoms, but there were no such relationships among the African American participants. 4

Other research has been conducted that relates perceived discrimination,

microaggressions, and mental health more generally. According to a literature review by

Nadimpalli and Hutchinson (2012), several studies have demonstrated significant negative relationships between perceived discrimination and mental health outcomes for Asian

Americans. In a longitudinal study of Marine recruits, Foynes and coauthors (2015) collected information on baseline physical health, mental health, social support, self-esteem, and self- reported race-based discrimination. After controlling for other baseline variables, self-reported race-based discrimination significantly predicted worse physical health and physical functioning

11 years later. Other cross-sectional self-report studies have demonstrated a positive relationship

between self-reported discrimination and poorer mental well-being in general (Hagiwara et al.,

2015; Himmelstein et al., 2015; Penner et al., 2009; Sanders-Philips et al., 2014). Nadimpalli

and colleagues (2016) conducted a study of Sikh Indian Americans living in New York to

determine relationships between perceived discrimination and mental and physical health. They

found a significant relationship between perceived discrimination and worse self-reported

mental health. In contrast, a cross-sectional study among Latinos with a preference for using

English, perceived microaggressions were not associated with likelihood of reporting emotional symptoms (Anderson & Finch, 2017).

Lee and colleagues (2018) conducted a cross-sectional study examining relationships

between cortisol, anxiety symptoms, depressive symptoms, and self-reported discrimination

among emerging African American adults. They found that self-reported discrimination was

associated with increased anxiety and depressive symptoms, but only anxiety symptoms were

associated with cortisol levels. 5

Generally speaking, perceived discrimination and microaggressions tend to be associated

with worse mental well-being, including self-reported depressive symptoms, anxiety, and

general mental health, though there are some documented differences between samples. Gaps in

the literature suggest that additional research is needed to assess relationships between microaggressions and mental well-being among people with diabetes.

Relationships to Medication Adherence

Perceived discrimination and microaggressions may be associated to worse self-care, including medication adherence. In a self-report study conducted among low-income hypertensive Black patients seen at a community-based primary care setting, greater perceived discrimination was associated with worse medication adherence at baseline (Forsyth et al.,

2014). In contrast, Peek and colleagues (2011) found that, among a sample of community- dwelling adults with diabetes, self-reported perceived healthcare discrimination was not associated with self-monitoring of blood glucose.

Perceived discrimination can exert an adverse impact on other treatment outcomes among persons with diabetes. According to at least one study, perceived discrimination from healthcare providers seemed to be related to worse diabetes outcomes. For example, a study conducted among African Americans with hypertension evaluated relationships between perceived racism, trust in the physician, and medication adherence (Cuffee et al., 2013).

Researchers found that trust in the physician was a mediator of the relationship between perceived discrimination and medication adherence, such that lower discrimination was associated with greater trust, which was in turn associated with better medication adherence.

Among a self-report survey and medical review study of American Indian women with type 2 diabetes receiving through tribal services, healthcare discrimination was associated 6

with decreased likelihood of being current on preventive tests, including clinical breast

examinations and Pap tests, after controlling for demographic covariates (Gonzales et al., 2013).

Among a sample of community members in Chicago, chronic perceived discrimination was

associated with reduced likelihood of having tests for diabetes among Puerto Ricans

(Benjamins, 2012). Another study conducted among Black/African American participants found

that baseline perceived discrimination was negatively associated with general health at pre-test,

4 weeks, and 16 weeks and with worse adherence to provider recommendations at 4 and 16

weeks (Penner et al., 2009). Moreover, there is a documented relationship between perceived

discrimination and health behaviors, such as substance use, smoking, sleep, and exercise

(Pascoe & Richman, 2009).

Current research indicates that perceived discrimination, particularly in healthcare

settings, is associated with poorer medication adherence behaviors, general health, and

engagement in healthcare. These relationships among people with diabetes remain understudied,

including relationships between general perceived discrimination and medication adherence.

Relationships to Diet and Exercise

The relationship between perceived discrimination, microaggressions, and health

outcomes for people with diabetes may be explained through worse diet and exercise. One

explanation for difficulty with maintaining a healthy diet may be that food is used as a coping

mechanism, particularly comfort foods, which are often foods that are high in saturated fat or sugar. Nadimpalli and colleagues (2017) conducted a survey among a sample of South Asian adults living in the United States to examine the types of foods associated with perceived

discrimination. In their sample, perceived discrimination was related to increased intake of

sweets but was not associated with intake of fruits and vegetables. The authors proposed that 7

eating such comfort foods could reduce the physiological stress associated with perceived

discrimination. In their study, conducted among low-income hypertensive Black patients,

Forsyth and colleagues (2014) found that greater perceived discrimination was related to worse

diet at baseline, but perceived discrimination at baseline predicted improvements in healthy

eating behaviors 12 months later. In one particularly relevant study, Sittner and colleagues’

(2018) conducted a study of American Indians with type 2 diabetes in the northern Midwest to

investigate relationships between everyday microaggressions, diabetes-related distress,

depressive symptoms, and health behaviors, including diet and exercise. Authors used structural

equation modeling to examine these relationships. They found that microaggressions

significantly predicted increased diabetes-related distress and depressive symptoms. However,

only diabetes-related distress was associated in turn with worse health behaviors; depressive

symptoms were not significantly associated with health behaviors.

Researchers examining relationships between perceived discrimination and physical

activity have reported mixed results. For example, a study by Forsyth and colleagues (2014)

found that, among low-income hypertensive Black patients, self-reported perceived

discrimination was not associated with exercise behaviors. Similarly, Edwards and Cunningham

(2013) analyzed a subsample of data collected among a sample of Texan residents. Like Forsyth

and colleagues (2014), they found that, among people of color, reports of racism were not

associated with physical activity. However, they found that racism moderated the relationship

between opportunities to exercise and physical activity such that when racism was high, there was a stronger relationship between opportunities to exercise and physical activity. In contrast, a cross-sectional survey of residents of California was conducted to assess relationships between racism and health behaviors. Researchers found that greater self-reported racism was associated 8

with increased likelihood of walking, though the authors note that a confounding variable may

be socioeconomic status, especially if those of lower socioeconomic status use walking as

transportation (Shariff-Marco et al., 2010).

Relationships to Glycemic Control

Relationships between perceived discrimination, microaggressions, and glycemic control

among individuals with diabetes are also understudied. A cross-sectional study by Moody-Ayers and coauthors (2005) investigated relationships between perceived discrimination and glycemic control among African American older adults, but no relationship between these variables was demonstrated. A study conducted by Gonzales and colleagues (2014) among American Indian women with type 2 diabetes examined relationships between perceived discrimination from healthcare providers and glycemic control. They found that perceived discrimination was associated with increased likelihood of A1c greater than 7.0%, a cutoff indicating worse glycemic control. Another study by Wagner et al. (2015) was conducted among Black and

White participants with type 2 diabetes. These participants responded to self-report measures of diabetes-related distress, depressive symptoms, and perceived discrimination. Their continuous blood glucose (CBG) was measured over 24 hours, and perceived discrimination was related to higher CBG, and this relationship was mediated by diabetes-related distress. Depressive symptoms were not signifiantly related to CBG.

To summarize, the current literature examining relationships between perceived discrimination, microaggressions, and medication adherence, diet, and exercise is limited. There seems to be converging support that perceived discrimination is associated with worse health outcomes, especially for people of color with diabetes. Such a relationship may play a large part in explaining the existence of health disparities among people with diabetes. Sufficient evidence 9

of relationships between discrimination and worse treatment adherence exists to support

additional research to clarify these behaviors and investigate potential mediators and moderators

should these relationships become well-documented in the literature.

Depressive Symptoms and Diabetes-Related Distress

In following sections, research on two additional predictors that relate to worse diabetes

outcomes will be reviewed. Specifically, these sections will examine the relationship between

two constructs, diabetes-related distress and depressive symptoms, and their relationships to

diabetes outcomes. Diabetes-related distress and depressive symptoms are related and somewhat overlapping constructs in that they represent psychological distress; however, longitudinal research demonstrates they predict diabetes-related outcomes in unique ways.

In a cross-sectional study of adolescents with type 1 diabetes, Baucom and colleagues

(2015) found that baseline depressive symptoms were associated with worse diabetes-related distress. Multiple cross-sectional studies examining depressive symptoms and diabetes-related distress as predictors of diabetes outcomes have found that depressive symptoms and diabetes- related distress are, in general, moderately correlated, but these estimates can range from small to very large (r = .15 - .87) (Aikens, 2012; Fisher et al., 2010; Gonzalez et al., 2008; Gonzalez et al., 2015; Lee et al., 2014; Reddy et al., 2013; Ting et al., 2011; Tsujii et al., 2012; van Bastelaar et al., 2010). In at least one study, there was no association between diabetes-related distress and depressive symptoms (Park et al., 2015). It is important to note that diabetes-related distress does not appear to differ based on type of diabetes diagnosis (van Bastelaar et al., 2010).

Studies directly comparing relationships between depressive symptoms and diabetes- related distress and diabetes outcomes report mixed findings. According to Schmitt and colleagues (2015), depressive symptoms were associated with increased diabetes-related distress, 10 which was in turn associated with worse diabetes outcomes among a group of people with either type 1 or type 2 diabetes. Additionally, Aikens (2012) found in their longitudinal self-report study of individuals with type 2 diabetes that there was no interaction between level of depressive symptoms and diabetes-related distress and diabetes outcomes (A1c, diet, exercise, and SMBG). Higher DRD scores predict worse diet, exercise, and medication adherence through worse depressive symptoms (Gonzalez et al., 2008). According to Reddy et al. (2013), people with diabetes who also had a past diagnosis of a depressive disorder had a higher DRD score than those without a history of depressive disorder, even after controlling for current depressive symptoms. After reviewing these studies, there is no clear pattern of findings, and it is unclear whether diabetes-related distress is a mediator for depressive symptoms or vice versa, or if these two constructs are differentially related to health behaviors.

Medication Adherence

In a similar fashion, longitudinal studies of the relationship between depressive symptoms and medication adherence do not clearly highlight a relationship between these variables, as results are not consistent. A longitudinal study revealed that depressive symptoms were related to better self-reported self-monitoring of blood glucose at baseline and 6-month follow-up, but baseline depressive symptoms were unrelated to medication adherence at baseline and 6 months (Aikens, 2012). To investigate relationships between depressive symptoms and self-reported medication adherence, a longitudinal study conducted by Katon and colleagues

(2009) collected information on depressive symptoms at baseline and prescriptions filled for diabetes medication over time among a sample of adults with type 2 diabetes. They found that, among people with poorer disease control, depressive symptoms were associated with worse adherence to medication. 11

Cross-sectional research of relationships between depressive symptoms and medication adherence has been more extensive and consistent than longitudinal research. For example, in

Baucom and colleagues’ (2015) study of adolescents with type 1 diabetes, participants’ depressive scores moderated the relationship between diabetes-related distress and medication adherence such that the relationship between diabetes-related distress and worse medication adherence was strengthened among participants with greater depressive symptoms. Of interest,

Kilbourne et al. (2005) investigated relationships between depressive symptoms and various types of reports of medication adherence, including patient self-report and medical provider’s report, among participants with type 2 diabetes recruited from Veteran’s Affairs clinics. Results indicated that depressed patients were more likely to self-report worse medication adherence, though provider report of medication was not related to depressive symptoms. Several other cross-sectional self-report studies of adults with diabetes have replicated the relationship between depressive symptoms and worse medication adherence and self-care (Chew et al., 2015;

Devarajooh & Chinna, 2017; Egede & Osborn, 2010; Gentil et al., 2017). Among participants with type 2 diabetes of low socio-economic status, depressive symptoms have been associated with worse medication adherence in a cross-sectional self-report study, but only among those in the lower three quartiles (Osborn et al., 2014). Gonzalez and colleagues’ (2008) study examined relationships between depressive symptoms, diabetes-related distress, and self-care behaviors in a cross-sectional sample of adults with type 2 diabetes. They found that diabetes-related distress was a significant predictor of medication adherence and self-monitoring of blood glucose, but that these relationships were reduced to non-significance and were better explained by depressive symptoms once they were added to the model. Sakraida & Weber (2016) examined relationships between specific depressive symptoms and self-care behaviors in a cross-sectional self-report 12 study of adults with type 2 diabetes and chronic kidney disease. They found that increased report of irritability and worthlessness were negatively associated with self-reported frequency of monitoring blood glucose.

Diet and Exercise

Studies of depressive symptoms and diet among people with diabetes generally tend to support the assertion that depressive symptoms are negatively correlated with healthy eating habits among this population. Sumlin and colleagues’ (2014) systematic review of studies investigated relationships between depressive symptoms and dietary and exercise habits among people with type 1 and type 2 diabetes. They uncovered 21 studies between the years 2000 and

2012 that documented relationships between depressive symptoms and dietary adherence. Of these, 20 studies described a negative relationship, with effect sizes ranging from -0.21 to -0.53 derived from the studies that provided numerical data. Similarly, 15 studies investigated relationships between depressive symptoms and exercise. Of these, 13 studies described a negative relationship between depressive symptoms and physical activity, with effect sizes ranging from -0.17 to -0.46 (Sumlin et al., 2014).

Longitudinal studies tend to support the direction of an inverse relationship between depressive symptoms and diet and exercise habits. A longitudinal study conducted by Katon et al. (2010) examined relationships between depressive symptoms and diet and exercise among adults with type 1 and 2 diabetes. At baseline, participants with no depressive symptoms had more days of healthy eating and exercise than other participants. After five years, those with no depressive symptoms had more days of healthy eating and exercise than those with worse depressive symptoms, indicating no changes in health behaviors over time for these groups. 13

Cross-sectional research tends to follow the trend of an inverse relationship between depressive symptoms and diet as well. Dipnall et al. (2015) compared dietary habits of participants with and without type 2 diabetes. They found that healthy dietary patterns (e.g. eating vegetables, fruits, and whole grains) were associated with lower depressive symptoms for all participants. However, they found that eating sweets was associated with worse depressive symptoms among participants with diabetes. Gonzalez and colleagues (2008) found that depressive symptoms were related to worse carbohydrate spacing and reduced intake of fruits and vegetables in a sample of adults with type 2 diabetes. Naiker et al.’s (2017) cross-sectional study of adults in Norway with type 2 diabetes showed a positive relationship between depressive symptoms and self-reported intake of saturated fats. Other studies have uncovered a relationship between depressive symptoms and worse general diet, including those of lower socio-economic status (Osborn et al., 2014). Sakraida & Weber (2016) examined relationships between specific depressive symptoms and self-care behaviors in a cross-sectional self-report study of adults with type 2 diabetes and CKD. They found that increased report of agitation and self-criticism were negatively associated with eating fruits and vegetables, while changes in appetite were associated with worse carbohydrate spacing.

Studies of relationships between depressive symptoms and exercise generally report an inverse relationship, though some exceptions exist. A longitudinal study by Aikens (2012) demonstrated that depressive symptoms were related to more frequent self-reported exercise at baseline and 6-month follow-up. Loprinzi and coauthors (2013) conducted a study in which participants with diabetes completed self-report measures of depressive symptoms, and then wore an accelerometer for a week. Depressive symptoms were associated with less moderate- and high-intensity exercise, but were unrelated to light exercise. Similarly, Gonzalez and 14 colleagues (2008) found that depressive symptoms were associated with worse exercise in a cross-sectional sample of adults with type 2 diabetes. A cross-sectional study conducted by

Naiker et al. (2017) indicated that depressive symptoms was associated with physical inactivity among participants with type 2 diabetes. In Sakraida & Weber’s (2016) cross-sectional self- report study of specific depressive symptoms and self-reported exercise among adults with type

2 diabetes and chronic kidney disease found that changes in energy were associated with decreased self-reported exercise.

Glycemic Control

The relationship between depressive symptoms and glycemic control has not been demonstrated unambiguously in the current literature, despite several longitudinal and cross- sectional studies evaluating this relationship. Aikens (2012) found in a longitudinal study that depressive symptoms were unrelated to glycemic control at baseline and at 6 month follow-up.

Similarly, another longitudinal study of participants with type 2 diabetes was conducted by

Fisher and colleagues (2010). Results demonstrated no significant relationship between baseline depressive symptoms and A1c 6 months later. In contrast, A1c was significantly higher in a longitudinal study of military veterans with type 2 diabetes with diagnoses of depression versus those with no such diagnoses according to a review of health records at baseline, and this relationship remained stable over the course of three years (Richardson et al., 2008).

Cross-sectional research of the relationship between depressive symptoms and glycemic control is limited, but generally supports the relationship between depressive symptoms and glycemic control. One cross-sectional study of adults with diabetes reported a marginally significant relationship (p = .06) between depressive symptoms and glycemic control after controlling for race (Lee et al., 2009). Other cross-sectional studies have documented expected 15 relationships between depressive symptoms and worse A1c in various populations with type 1 and type 2 diabetes (Lustman et al., 2005; Naiker et al., 2017). In contrast, a study conducted among adolescents with type 1 diabetes living in Malaysia investigated relationships between depressive symptoms and glycemic control, but this relationship was reduced to non-significance after quality of life was included in the model (Tan et al., 2005).

In conclusion, studies conducted to evaluate the relationship between depressive symptoms and diabetes outcomes, such as medication adherence, diet, exercise, and glycemic control, have generally supported the position that depressive symptoms tend to be associated with poorer outcomes among people with type 2 diabetes. However, the relationships are not always consistent, especially in longitudinal research, indicating that depressive symptoms may not be a true predictor of diabetes outcomes or that additional mediator or moderator variables may be at play. One such mediator variable may be diabetes-related distress.

Diabetes-Related Distress

One key psychological factor related to adherence among people with diabetes is diabetes-related distress. Diabetes-related distress is conceptualized as a cognitive and emotional response to stressors associated with diabetes, such as concerns about blood glucose levels, difficulties with maintaining a healthy diet and an exercise regimen, and fear of adverse health outcomes (Reddy et al., 2013). Diabetes-related distress has been linked to multiple health behaviors and glycemic control, especially poorer blood glucose management as measured by

A1c, reduced medication adherence, a less healthy diet, and reduced time spent exercising

(Polonsky, et al, 1995; Reddy et al., 2013; Snoek et al., 2000; Weinger & Jacobson, 2001;

Zagarins et al., 2012). 16

Medication Adherence

The relationship between diabetes-related distress and glycemic control may be explained

by medication adherence. Bogusch and O’Brien’s (2016) unpublished meta-analysis of

correlational studies of diabetes-related distress and medication adherence reported a small effect

size of this relationship (Zr = -.11) based on 10 studies.

In a longitudinal study of patients with type 2 diabetes, Aikens (2012) found that

diabetes-related distress at baseline was associated with worse medication adherence at 6-months

follow-up. In a cross-sectional study of patients with type 2 diabetes, diabetes-related distress

was associated with medication non-adherence (Gonzalez et al., 2008). In another cross-sectional

sample of participants with type 2 diabetes meeting criteria for Major Depressive Disorder,

Gonzalez et al., (2015) found that diabetes-related distress was associated with worse glycemic

control and that this relationship was mediated by worse medication adherence. In contrast,

Chew et al. (2015) did not find a relationship between diabetes-related distress and medication

adherence after controlling for demographic variables in a cross-sectional survey of adults with

type 2 diabetes seeking services at public health clinics in Malaysia. Other cross-sectional studies

of adults with type 2 diabetes have reported negative associations between diabetes-related

distress and medication adherence or self-care (Manan et al., 2014). However, at least one cross- sectional self-report study of outpatients with type 2 diabetes in Italy reported a positive association between diabetes-related distress and increased frequency of self-monitoring of blood glucose (Pintuadi et al., 2015), and three other cross-sectional self-report studies of patients with type 1 or type 2 diabetes reported no association between diabetes-related distress and frequency of self-monitoring of blood glucose (Polonsky et al., 2005; Potter et al., 2015; Ting et al., 2011). 17

Diet and Exercise

Studies examining relationships between diabetes-related distress and diet are generally

cross-sectional and are inconsistent in terms of their findings. For instance, Aikens (2012) found that self-reported diabetes-related distress was not associated with diet at 6-months follow-up. In a cross-sectional self-report study of patients with type 2 diabetes, diabetes-related distress was associated with worse general diet, increases in consumption of high-fat foods, and decreases in carbohydrate spacing, but in consumption of fruits and vegetables (Gonzalez et al., 2008). Zhou and colleagues (2017) found that greater diabetes-related distress was associated with decreased likelihood of development of a meal plan among adults with type 2 diabetes in . Park and coauthors (2018) conducted a cross-sectional self-report study of adults with type 2 diabetes to investigate relationships between diabetes-related distress and diet. Results indicated that increases in emotion-oriented coping mediated the relationship between diabetes-related distress and greater emotional eating, restrained eating, and eating behaviors cued by external factors

(e.g. snacks kept within reach). Multiple other cross-sectional self-report surveys of patients with

type 2 diabetes recruited through endocrinology clinics, and primary care facilities have

produced results indicating that increased diabetes-related distress was associated with worse

dietary behaviors (Fisher et al., 2012; Huang et al. 2010; Pintaudi et al., 2015; Polonsky et al.,

2005; Potter et al., 2015) though at least one cross-sectional self-report study has not found a

relationship between diabetes-related distress and meal planning (Ting et al., 2011).

Results of studies examining relationships between diabetes-related distress and exercise

have been limited and mixed. Aikens (2012) found that self-reported diabetes-related distress

was not associated with physical activity at 6-months follow-up. Cross-sectional self-report

studies of patients with type 1 and type 2 diabetes indicated that diabetes-related distress was 18

associated with less frequent physical activity (Gonzalez et al., 2008; Huang et al., 2020;

Pintaudi et al., 2015; Polonsky et al., 2005; Potter et al., 2015; Zhou et al., 2017), though at least

one cross-sectional self-report study has reported no association between diabetes-related distress

and exercise (Ting et al., 2011).

Glycemic Control

A meta-analysis conducted by Bogusch and O’Brien (2016) investigated effect sizes of

all correlational studies published before 2016 (k = 35) of diabetes-related distress and glycemic control. Results indicated a small effect size of diabetes-related distress on glycemic control (Zr

= .18), such that increased diabetes-related distress was associated with worse glycemic control.

In longitudinal studies, diabetes-related distress consistently predicts glycemic control. For

instance, when investigating the relationship between diabetes-related distress and glycemic

control among participants with type 2 diabetes, Aikens (2012) found that diabetes-related

distress at baseline was associated with worse A1c at 6-months follow-up. Similarly, Fisher and

colleagues (2010) found that diabetes-related distress at baseline was associated with worse A1c at 9- and 18-months follow-up in a mostly non-White sample of participants diagnosed with type

2 diabetes. In yet another longitudinal study among a group of Japanese patients with well-

controlled (mean A1c < 7.0%) diabetes, diabetes-related distress was associated with increased

glycemic control at 6- and 12-months follow-up and that this relationship was explained by

decreases in self-efficacy and adherence (Nakahara et al., 2006).

Cross-sectional studies of relationships between diabetes-related distress and glycemic

control tend to follow the pattern of longitudinal research. In particular, diabetes-related distress

has been investigated among populations with type 1 and type 2 diabetes as a mediator of the

relationship between depressive symptoms and glycemic control such that depressive symptoms 19 are associated with greater diabetes-related distress, which are in turn associated with worse glycemic control (Schmitt et al., 2015; van Bastelaar et al., 2010). One cross-sectional study using self-report and review of medical records investigated mediators of the relationship between diabetes-related distress and A1c among a group of Danish participants with type 2 diabetes (Rogvi et al., 2012). Diabetes-related distress explained a significant amount of variance, and explained significantly more variance in A1c in the non-insulin-treated subsample.

Among individuals with low socio-economic status and type 2 diabetes, diabetes-related distress has been identified as a mediator for the relationship between food insecurity and poor glycemic control (Seligman et al., 2012). Various cross-sectional studies of people diagnosed with type 1 and type 2 diabetes have demonstrated a relationship between diabetes-related distress and worse

A1c, including samples of outpatient populations, veterans, Korean employees, individuals taking insulin alone, patients receiving outpatient care at diabetes centers in Greece and Italy, and patients living in urban areas (Franks et al., 2012; Gonzalez et al., 2015; Han & Cotter, 2015;

Huang et al., 2010; Lee et al., 2014; Nichols et al., 2000; Papathanasiou et al., 2014; Park et al.,

2015; Pintaudi et al., 2015; Polonsky et al., 2005; Potter et al., 2015; Reddy et al., 2013; Ting et al., 2011; Tol et al., 2012; Tsujii et al., 2012; Zhou et al., 2017). One cross-sectional self-report study conducted among outpatients with type 1 or 2 diabetes has reported no association between diabetes-related distress and A1c (Polonsky et al., 2015). It is important to note that, because of the cross-sectional nature of these studies, causality and directionality for these studies cannot be inferred.

In sum, the existing literature, including a meta-analysis, longitudinal studies, and other correlational research, supports a small effect size to describe the relationship between diabetes- related distress and glycemic control. The literature relating diabetes-related distress to diabetes 20

treatment adherence is less clear but tends to indicate a relationship between diabetes-related distress and poorer diabetes outcomes. Additional research is needed to clarify such relationships and mechanisms through which diabetes-related distress is related to glycemic control.

Mindfulness and Well-Being

A correlate and possible protective factor for improved health is mindfulness.

Mindfulness has been investigated as a correlate of better health behaviors among multiple populations, especially those with chronic health problems. The concept of mindfulness arises from traditional Buddhist teaching and mindfulness meditation. It has been described as an

attentive quality of consciousness of the present moment, including physical sensations, thoughts, and emotions, in a curious and non-judgmental way (Grabovac et al., 2011). Poorer treatment adherence and metabolic control may be at least partially explained by avoidance.

When choosing to engage in treatment-adherent behavior, such as checking blood sugar, unpleasant thoughts and feelings related to having diabetes may arise. Similarly, dietary restrictions can evoke negative thoughts and emotions such as feelings of deprivation, hunger,

and unfairness. By avoiding these behaviors, the unpleasant thoughts and feelings associated

with having diabetes are also avoided. However, the avoidance and subsequent nonperformance

of important treatment-adherent behavior can lead to problems with managing blood glucose.

People who are more mindful may be able to recognize these unpleasant feelings without acting

on the urge to avoid adherent behaviors like checking blood glucose.

Depressive Symptoms and Diabetes-Related Distress

One way that mindfulness may exert beneficial effects on health might be through lower

psychological distress, such as levels of depressive symptoms. Indeed, researchers have reported

that higher levels of mindfulness have generally been related to less depressive symptoms 21

(Bowlin & Baer, 2012; Masuda & Tully, 2012). Among participants seeking treatment for either

depression or anxiety disorders, mindfulness was associated with lower depressive symptoms

prior to treatment (Desrosiers et al., 2013). Another study found a similar relationship between

mindfulness and depressive symptoms in a community sample, with greater mindfulness

associated with a more positive state of mind (Branstrom et al., 2011). In each of these studies, it

is posited that mindfulness is associated with an acceptance of negative thoughts without

responding emotionally to negative thoughts, which in turn may have been associated with lower

depressive symptomatology.

Research on mindfulness and diabetes-related distress has typically been done through

intervention studies of mindfulness-based interventions for people with diabetes. Bogusch and

O’Brien (2019) reported a medium effect size (d = .41) of mindfulness interventions for reducing

diabetes-related distress from pre-test to post-test in their meta-analysis of treatment effects of

mindfulness interventions. Research on trait mindfulness among people with diabetes is

understudied, but trait mindfulness was associated with lower diabetes-related distress in

Brown’s (2014) cross-sectional survey of adults with diabetes.

Diet and Exercise

Research investigating treatment adherence, particularly diet and exercise, contributing to

glycemic control have been limited but promising. In one cross-sectional self-report study of

adults with type 2 diabetes, mindfulness was associated with better adherence to diet, including

eating more fruits and vegetables and eating fewer carbohydrates (Fanning et al., 2018). In a

cross-sectional survey of Dutch adults with type 1 and type 2 diabetes, researchers found that mindfulness was a significant predictor of less emotional eating, less eating in response to external cues, and increases in restrained eating (Tak et al, 2015). 22

Additional studies of relationships between mindfulness and dietary habits have been

conducted among other populations but have resulted in variable findings. For instance,

Clevenger and colleagues (2018) found a negative relationship between mindfulness and junk

food intake among a group of children but found no relationship between mindfulness and intake

of fruits and vegetables. In contrast, Gilbert and Waltz (2010) found a relationship between

mindfulness and lower fat intake and greater fruit and vegetable intake in a sample of college

students. In other cross-sectional college samples, greater mindfulness has been associated with

better diet in general (Lentz & Brown, 2018; Murphy et al., 2012; Slonim et al., 2015), and less

self-reported binge eating (Roberts & Danoff-Burg, 2010). However, findings from longitudinal

research do not demonstrate that mindfulness predicts eating habits over time in college samples

(Murphy et al., 2012). Additionally, reports of lower mindfulness scores have been associated

with worse eating habits among undergraduate college students and MTurk workers, including

self-reported uncontrolled eating and greater calorie consumption (Jordan et al., 2014). Taken

together, this collection of studies may indicate that mindfulness may behave differently among

different populations and that, at this time, the relationship between mindfulness and dietary

habits is still not well understood.

The relationship between mindfulness and exercise has been investigated, although an

extremely limited number of cross-sectional studies have evaluated the relationship between

mindfulness and exercise among people with diabetes. Fanning and colleagues (2018) found no

association between mindfulness and self-reported exercise habits in a sample of adults with type

2 diabetes.

As it is with dietary habits, the relationship of mindfulness with exercise is unclear among other populations. Ulmer and colleagues (2010) surveyed a group of adults who utilized 23

their local YMCAs to assess relationships between mindfulness and exercise. They found that

people who are regular exercisers and people who report success in reaching exercise goals tend

to report greater mindfulness scores. In a cross-sectional survey of Dutch adults with type 1 and

type 2 diabetes, a group of researchers found that self-reported state mindfulness while

exercising was more strongly associated with physical activity than self-reported trait

mindfulness (Tsafou et al., 2017). Clevenger and colleagues (2018) found no relationship

between mindfulness and duration of physical activity among a sample of elementary-aged

school children. Cross-sectional research conducted among college students has demonstrated

relationships between mindfulness and increased physical activity, greater enjoyment of physical

activity, and lower perceived activity restriction (Gilbert and Waltz, 2010; Lentz & Brown, 2018;

Roberts & Danoff-Burg, 2010); though longitudinal research does not support the association

between baseline mindfulness and frequency of exercise at follow-up (Murphy et al., 2012).

Awareness while eating and lower emotional eating were both associated with increased duration

of physical activity among college students, though overall mindful eating ratings were not associated with physical activity (Moor et al., 2013).

Glycemic Control

Investigations of the relationship between mindfulness to glycemic control have been extremely limited and the results of these studies are contradictory. In one study, participants with type 2 diabetes were recruited through a prospective cohort sample that was created in the

New England Area as part of a larger, cross-sectional study (Loucks et al., 2016). Mindfulness was associated with increased likelihood of having plasma glucose levels that fell into the normal range. In another cross-sectional self-report study of adults with type 2 diabetes, researchers

found no association between mindfulness and self-monitoring of blood glucose after controlling 24

for demographic variables (Fanning et al., 2018). Similarly, Caluyong and colleagues (2015) found no relationship between mindfulness facets self-monitoring of blood glucose among a sample of adults with type 2 diabetes.

In conclusion, there is a paucity of research investigating relationships between

mindfulness, psychological well-being, and diabetes-related treatment adherence among

populations with diabetes. However, current findings are promising. In addition, research

conducted among other populations, while preliminary, appears to support relationships between

mindfulness and treatment adherence relevant to populations with diabetes.

Summary and Hypotheses

Perceived discrimination, diabetes-related distress, depressive symptoms, and mindfulness may all relate to diabetes-relevant behaviors especially medication adherence, dietary habits, and exercise. Understanding relationships between these variables may be a starting point for development of interventions to improve well-being and adherence to treatment behaviors among people with diabetes. In turn, improvements in well-being and treatment adherence may be associated with better glycemic control. The present study investigated the

proposed relationships among these variables using SEM.

1. Greater perceived microaggressions will be associated with increased self-reported

depressive symptoms.

2. Higher levels of mindfulness will be associated with lower self-reported depressive

symptoms.

3. Microaggressions will be associated with increased diabetes-related distress through

increased depressive symptoms. Mindfulness will be associated with decreased

diabetes-related distress through decreased depressive symptoms. 25

4. Diabetes-related distress will be associated with lower levels of reported treatment

adherence, namely, worse diet, exercise, and medication adherence.

5. Diabetes-related distress will be associated with worse glycemic control, which will

be explained by worse adherence to diabetes treatment.

The relationships described above are expected and outlined in Figure B1. 26

METHODS Participants

All procedures were approved by the Bowling Green State University Institutional

Review Board. Subjects were 337 registered workers on Amazon's Mechanical Turk (MTurk) website and were recruited through this website. They were required to be 18 years or older in age and endorse a medical diagnosis of type 1 or type 2 diabetes in a screener (see Appendix C).

Participants were awarded $1 to their Amazon Worker account after completing the measures.

Measures

Demographics

General demographic information was collected, including the following: age, gender, marital status, location, race, ethnicity, education level, and yearly household income. In addition, demographics relevant to diabetes were collected including time since diagnosis of diabetes. latest self-reported A1c, and time since meeting with a medical professional regarding diabetes care. Participant demographics for the entire sample (n = 337) are reported in Tables

A1-4. Descriptive data for the following measures is reported with multivariate outliers removed in Table A5 (n = 334). Independent samples t-tests were conducted with country (i.e. India,

USA) and study variables (e.g. mindfulness, perceived microaggressions) to examine between- group differences and to determine the need for exploratory multigroup analysis (see Tables A6-

7). See Appendices E-K for measures described in this section.

Microaggressions

Microaggressions were assessed using the Racial and Ethnic Microaggressions Scale

(REMS; Nadal, 2011). The REMS is a 45-item questionnaire in which participants rated the frequency of experiencing common microaggressions on the following scale: 0 “I did not 27

experience this event in the past six months”, 1 “I experienced this event 1 time in the past six

months,” 2 “I experienced this event 2 times in the past six months,” 3 “I experienced this event

3 times in the past six months,” 4 “I experienced this event 4 times in the past six months,” or 5

“I experienced this event 5 times in the past six months.’” Higher scores indicate greater

frequency of encountering microaggressions. It contains six subscales: Assumptions of

Inferiority, Second-Class Citizenship and Assumptions of Criminality, Microinvalidations,

Exoticization/Assumptions of Similarity, Environmental Microaggressions, and Workplace and

School Microaggressions. The REMS was developed and validated in a sample of African

American, Latino/a, Asian, and multiracial participants. The REMS was demonstrated to have good convergent validity as it has been positively correlated with other measures of self-reported racism in varying samples (Mekawi & Todd, 2018; Nadal, 2011) and negatively correlated with multiple measures of quality of life, such as social functioning and emotional well-being (Nadal et al., 2017).

In this study, the mean total REMS score was 2.42 (SD = 1.24, Range 0-4.51, α = .985).

For the subscale Assumptions of Inferiority, the mean score was 2.35 (SD = 1.35, Range 0-4.75,

α = .947). For the subscale Second-Class Citizenship and Assumptions of Criminality, the mean

score was 2.32 (SD = 1.38, Range 0-4.86, α = .943). For the subscale Microinvalidations, the

mean score was 2.36 (SD = 1.30, Range 0-4.67, α = .939). For the subscale

Exoticization/Assumptions of Similarity, the mean score was 2.42 (SD = 1.37, Range 0-4.78, α =

.948). For the subscale Environmental Microaggressions, the mean score was 2.70 (SD = 1.24,

Range 0-5.00, α = .908). The mean score for the Workplace and School Microaggressions scale

was 2.31 (SD = 1.37, Range 0-4.80, α = .917). These means are higher than reported in previous

research (see Mekawi & Todd, 2018; Nadal et al., 2011). 28

Mindfulness

Mindfulness was assessed using the Five-Facet Mindfulness Questionnaire (FFMQ; Baer et

al., 2006). The FFMQ is a 39-item questionnaire in which participants rated the extent to which

various statements describe them in general on a scale ranging from 1 (“never or rarely true”) to 5

(“very often or always true.”). The items on this scale measured the tendency to be mindful in

everyday life using items that reflected everyday experiences. Negative items were reverse-scored,

and then all items were summed and averaged to create the total scale score for mindfulness with

higher scores indicating higher levels of mindfulness. Among other samples, the FFMQ has been

found to have convergent validity in that it was positively correlated with other measures of

mindfulness (Baer et al., 2006) and divergent validity as it was negatively correlated with measures

of thought suppression and dissociation (de Bruin et al., 2012).

The FFMQ also includes five subscales that assess five facets of mindfulness: act with

awareness (“I find it difficult to stay focused on what’s happening in the present moment.”), describe

(“I’m good at finding words to describe my feelings.”), non-judge (“I tell myself I shouldn’t be

feeling the way I’m feeling.”), observe ( “I pay attention to physical experiences, such as the wind in

my hair or sun on my face.”), and nonreact (“I watch my feelings without getting carried away by

them.”). The nonreact subscale contained seven items; all other subscales contain eight items each.

In the present study, the mean FFMQ score was 3.07 (SD = 0.29; range = 2.19- 3.95; α =

.717). The mean score for the act with awareness subscale was 2.80 (SD = 0.78; range = 1.25-5.00; α

= .833), for the describe subscale was 3.15 (SD = 0.50; range = 1.44-4.81; α = .480), for the non-

judge subscale was 2.75 (SD = 0.76; range = 1.00-5.00; α = .852), for the non-react subscale was

3.27 (SD = 0.70; range = 1.00-5.00; α = .786), and for the observe subscale was 3.42 (SD = 0.70; range = 1.25-5.00; α = .793). 29

Depressive Symptoms

Depressive symptoms were assessed using the Center for Epidemiologic Studies-

Depression Scale (CES-D; Radloff, 1977). The CES-D was designed to measure depressive

symptoms among the general population and in chronically depressed populations. Participants indicated how often they experienced 20 different symptoms over the past week on a scale ranging from 0 (“rarely or none of the time (less than 1 day)”) to 3 (“most or all of the time (5-7 days).”). Positive items were reverse-scored, and items were summed to create a total scale score. Scores could range from 0 to 60 with higher scores indicating higher levels of depressive symptomatology. A score of 16 or higher on the CES-D has been identified as a cut-off for being

at risk for clinical depression. Internal reliability CES-D scores have been moderately to highly

correlated with scores derived from the Hamilton Clinician’s Rating scale and the Raskin

Depression Rating scale (Hamilton, 1960; Raskin, Schulterbrandt, Reatig, & Me-Keon, 1969),

which are both measures of depression. Scores collected from the CES-D and Bradburn Positive

Affect scale (Bradburn, 1969) were negatively correlated, suggesting that the CES-D has

discriminant validity as well (Radloff, 1977).

In the present study, the mean CES-D score was 29.57 (SD = 9.91; range = .00– 58.00; α

= .868), and 89.9% of participants met the cut-off score for being at risk for clinical depression

(i.e., 16 or higher). Depressive symptoms reported here were higher on average than reported in

previous research among people with diabetes (Fisher et al., 2007; Wagner & Abbot, 2007). The

rate of participants exceeding the cut-off for at risk of clinical depression is much higher than

reported in previous research among people with diabetes (about 22% reported by Fisher et al.,

2007). 30

Diabetes-Related Distress

Diabetes-related distress was measured with the Problem Areas in Diabetes scale (PAID;

Reddy et al., 2013). The PAID is a 20-item scale in which participants indicated their level of distress. Possible scores range from 0 (“Not a problem”) to 4 (“Serious problem”). Items are summed and multiplied by 1.25 to produce scores ranging from 0-100. The PAID demonstrated convergent validity, as it has been moderately and positively correlated with the PhQ-9, a measure of depressive symptoms, and with poorer glycemic control (Reddy et al., 2013).

Discriminant validity has also been observed as it is negatively correlated with self-reported physical and mental health (Reddy et al., 2013). In the present study, the mean PAID score was

51.98 (SD = 21.03; range = 0- 96.25; α = .949), which is higher than reported in previous studies

(Reddy et al., 2013; Snoek et al., 2000), though Polonsky and colleagues (1995) reported a mean

PAID score of 54.5 among a sample of individuals with poorly controlled diabetes.

Health Behaviors

Dietary and exercise behaviors were assessed with a scale used in similar studies of health behaviors (Fekete et al., 2009; Fekete et al., 2006). The ten dietary items on the scale were created using guidelines suggested in the Dietary Guidelines for Americans 2010 report (U.S.

Department of Agriculture and U.S. Department of Health and Human Services, 2010). The exercise items on the scale were created using guidelines suggested by the 2008 Physical

Activity Guidelines for Americans report (U.S. Department of Health and Human Services,

2008). Participants responded to statements asking them to indicate the extent to which they engaged in specific dietary and exercise behaviors during the previous month on a scale ranging from 0 (“Never”) to 3 (“All of the Time (Every Day)”). 31

As the reliability score for this measure was low (α = .443), the item-scale correlations were examined to identify items that did not contribute to the reliability of the scale. The final scale included items 1, 3, 4, 6, 8, 11, 12, and 13. Mean for this scale was 1.74 (SD = 0.50; range

= 0.22-2.88, α = .732). For the exercise subscale, the mean was 1.73 (SD = 0.51; range = 0.00-

2.88; α = .503). For the diet subscale, the mean was 1.60 (SD = 0.34 ; range = 0.67-2.56; α =

.586).

Individual items were included as part of the confirmatory factor analysis described in the analysis plan section (see page 31) and were included as potential items onto which the

“Treatment Adherence” latent variable factor would load.

Medication Adherence

Medication adherence was measured by use of an adapted version of the Morisky

Medication Adherence Scale (MMAS; Morisky et al., 2008). The MMAS is an 8-item measure that assesses behaviors related to taking medication regularly. The instructions for this measure were altered to the following: “Please answer each question based on your personal experience with your diabetes medication.” Items 1-7 are “yes” or “no” questions. The response to item 8

(“How often do you have difficulty remembering to take your diabetes medication) is Likert- style, with possible responses ranging from 0: Never/Rarely to 4: All of the time. To calculate a total score, negatively-worded items were reverse scored. “Yes” responses to items 1-7 were each 1 point, and responses to item 8 were as follows: a response of 1 resulted in an additional

.25 point, a response of 2 resulted in an additional .50 point, a response of 3 resulted in a .75 point, and a response of 4 resulted in an additional 1 point. Greater scores indicated better medication adherence (Korb-Savoldelli et al., 2012). In other samples, the MMAS has been found to demonstrate convergent validity, with relationships being observed with measures of 32

beliefs about medication, treatment satisfaction (Plakas et al., 2016; Reynolds et al., 2012). For

this study, the average score on the MMAS was 3.98 (SD = 1.81; range = 0.00-7.75, α = .547),

which is indicative of poor medication adherence, and lower than reported in previous research

(MMAS < 6; Morisky et al., 1986; Morisky et al., 2008). Item-scale correlations indicated that dropping any items would not lead to substantially improved reliability.

Individual MMAS items were included as part of the confirmatory factor analysis

described in the analysis plan section (see page 31) and were included as potential items onto

which the “Treatment Adherence” latent variable factor would load.

Glycemic Control

Glycemic control was measured by participants’ response to the question “Please indicate

your latest A1c.”

Procedure

The project was approved by the Bowling Green State University Institutional Review

Board. Participants were recruited on the Amazon Mechanical Turk website. They clicked the

survey link and endorsed that they agreed to participate after reading the informed consent (see

Appendix D). Participants were asked to check any current health conditions for screening

purposes before they were qualified for the study in order to increase likelihood of recruiting

participants who were truly diagnosed with diabetes. Participants who selected diagnoses for

"diabetes" were notified that they were eligible to participate and were directed to continue with

the survey. Participants who did not check either of these boxes were informed that they did not

qualify for the survey, were thanked for their willingness to participate, and were directed to exit

the survey. Upon completion of the survey, participants were awarded $1.00 to their Mechanical 33

Turk account. Informed consent, recruitment script, survey measures, and IRB approval are

included in Appendices C-L.

Analysis Plan: Structural Equation Modeling

Rationale

Structural equation modeling was selected as the statistical method to evaluate the hypotheses. Specifically, a confirmatory factor analysis and partially latent regression analysis were conducted to evaluate the model. Structural equation modeling using these methods has a few benefits in comparison to traditional regressions. Structural equation models include latent variables, which are variables that are not directly measured. These variables can combine multiple measures into a single construct; for example, items assessing exercise and diet can be combined into a latent variable of “Treatment Adherence” (Kline, 2011). The structure of these latent variables are informed by theory and evaluated statistically with confirmatory factor analysis. In addition, SEM using partially latent regression analysis allows for testing multiple directional relationships at once. This means that a single test can be performed to test larger theoretical models.

Statistical Programs

SPSS version 24 was used to evaluate descriptive statistics and to conduct t-tests. Amos

Graphics version 26 was used to conduct structural equation modeling analyses.

Power

For structural equation modeling, there is no clear rule for sample size. Kline (2011) recommends a ratio of observations to parameters of 20:1, while others recommend smaller ratios of 10:1 or even 5:1 (Bentler & Chou, 1987; Schreiber et al., 2006). Other researchers recommend a general rule of a sample size of 300 (Comrey & Lee, 2013; Tabachnick & Fidell, 34

2013). As guidelines are unclear, we sought a sample size of at least 300, and used Daniel

Soper’s (2020) sample size calculator for structural equation models to calculate the minimum sample size to detect the a medium effect size. For the hypothesized structural regression model

(Figure B3), a minimum sample size of 230 was calculated.

Statistical Assumptions

Assumptions of using confirmatory factor analysis and structural equation modeling were examined and addressed, including missing data, univariate and multivariate outliers, univariate and multivariate normality, and extreme collinearity. Any violations of assumptions were addressed prior to conducting the analyses.

First missing data were addressed. The percentage of missing data were calculated by variable. If less than 5% of the data for a specific variable was missing, a single-imputation method of mean substitution was used to replace these data. If greater than 10% of data for a single participant was missing, that participant’s data were removed.

To assess for univariate outliers, Z-scores of each variable were calculated to identify univariate outliers as z-scores with values greater than 3. These cases were replaced with scores that were three standard deviations above or below the mean for their respective scale (Field,

2011). Univariate normality was then evaluated. Variables used in the present analyses were determined to be normally distributed through a 1) a visual inspection of histograms and 2) skewness and kurtosis absolute values of less than +/- 3 and +/- 10, respectively, as structural equation modeling is a robust statistical method (Brown, 2006). Linearity and homoscedasticity were established through examination of scatterplots of the residuals for each endogenous variable (Field, 2011). 35

Assumptions of multivariate normality were examined, including assessment for multivariate outliers, which were identified through Mahalanobis distances. Based on the final sample size of 337, a Mahalonobis distance of greater than 22 was set to identify influential cases per Barnett and Lewis (1978). Multivariate normality was assessed with Mardia’s index

(Kline, 2011). A Mardia’s coefficient of less than 3 is considered an indicator of multivariate normality (Mardia, 1970; Mardia, 1985).

Procedures for identification and addressing extreme collinearity were followed as per

Kline (2011). A squared multiple correlation (R2) was calculated between each variable and other variables. Any observations in which R2 > .90 were interpreted as extreme collinearity. To address extreme collinearity, redundant variables were combined into a composite.

Use of structural equation modeling analysis includes the assumptions of positive definiteness (Kline, 2011). This requires that a data matrix is nonsingular, or that the inverse of the data matrix is able to be derived. In addition, all eigenvalues in the data matrix must be positive, and determinants of the data matrix must be greater than zero. Finally, none of the correlations that are part of the data matrix may be out-of-bounds. In other words, the maximum absolute value of a covariance of two variables may not exceed the square root of the product of the variances of each variable. Any such value greater than this limit is said to be out-of-bounds and renders the data matrix impossible to derive.

Descriptives

Means, standard deviations, and range were calculated for all relevant variables and are reported in Table A5. Correlations between measures and items included in the structural equation model were calculated. Alpha coefficients for all measures were calculated. 36

Specification

The first step of structural equation modeling is model specification, or the creation of a theoretical model (see Figure B1) of all latent and measured variables (Kline, 2011). Latent variables are those that are unable to be measured directly and are therefore measured indirectly.

For example, in the proposed model, Treatment Adherence is a latent variable that was measured through self-reported medication adherence, exercise, and diet. Measured variables are defined as variables that were directly measured, such as total mindfulness score. Relationships between these variables were represented in the specification model.

Structural equation models can be recursive or non-recursive. This model hypothesized is a recursive path model because 1) causality is hypothesized to flow in one direction and 2) disturbances (i.e. residual variance of outcome variables) are theoretically uncorrelated.

The final model examined was a partially latent structural regression model, as both measured and latent variables are included in the model.

Identification

The second step in SEM was model identification or the determination that it was mathematically possible for the model to be analyzed. The final model was a partially latent structural regression model. For such a model to be identified, it must be identified after being respecified as a Confirmatory Factor Analysis (CFA), or measurement, model and as a path model. Given that both the measurement and path models (see Figures B1 and B2) are recursive, these are both assumed to be identified. Per Kline (2011), to be identified, the model must meet two other criteria: 1) degrees of freedom must be greater than or equal to zero and 2) every latent variable, including residuals, is assigned a scale. To determine the degrees of freedom, the following preliminary calculations are made: 37

df = p – q

Where df is the degrees of freedom, p is the number of observations (i.e. the number of entries in the proposed covariance matrix in the lower diagonal), and q is the number of free parameters (e.g. variances, covariances, and direct effects). The equation for observations is as follows:

p = v(v +1)/ 2

Where v is the number of observed variables represented in the model.

For this analysis, 1 latent variable was proposed with 16 items. Mindfulness, perceived microaggressions, depressive symptoms, diabetes-related distress, and A1c were also included as observed variables in the proposed model; therefore q = 44 and v = 21. It follows that

df = [v(v +1)/2] – q

df = [21(21 +1)/2] – 44

df = 231 – 44

df = 187

As the df was greater than 0, the proposed model was identified.

To address the second criterion of identification, every latent variable was assigned a scale. This was accomplished through unit loading identification, or to set selected unstandardized path coefficients to a fixed constant of 1.0 (Kline, 2011). This was done for each endogenous disturbance as well as for one factor loading for each latent variable onto a single indicator, which acted as a reference variable, as the remaining indicators were scaled in reference to it. 38

Model Estimation

To evaluate the hypothesized model, a two-step process was done as per Kline (2011).

First, the structural regression model (e.g. Figure B1) was respecified to a measurement model

(e.g. Figure B2) using latent variables to confirm factor structure with confirmatory factor

analysis. Then the path model (e.g. Figure B1) was estimated and respecified as needed.

To evaluate the measurement and path models, a maximum likelihood method was

conducted. This is the most commonly used estimation method in the literature as it generates

statistics that maximize the likelihood that observed covariances are drawn from a multivariate

normal population (Kline, 2011). To evaluate the measurement model, a maximum likelihood

(ML) confirmatory factor analysis (CFA) was conducted to 1) assess for fit of the measurement

model and 2) identify items to which the latent variables loaded poorly. Multiple measures of fit

were used to evaluate both the measurement and path models. Chi-square statistics were used as

an indicator of difference in fit between the hypothesized model and the actual model, with p-

values greater than .05 indicating a good fit (Kline, 2011). However, a drawback of chi-square tests are that they are sensitive to large sample sizes such that greater sample sizes often produce p-values less than .05 (Brown, 2006; Diamantopoulos & Sigauw, 2000).

As a result, other indices of fit were examined, including the Comparative Fit Index

(CFI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean

Square Residual (SRMR). The CFI is a “goodness-of-fit” measure, that measures the improvement in fit of the tested model in comparison to the independence model, or a null

hypothesis model in which population covariances between variables are assumed to be 0 (Kline,

2011). Greater values indicate better fit, with a value of 1.0 indicating perfect fit. Typical

standards for good fit for the CFI are values greater than .95 (Hu & Bentler, 1999). The RMSEA 39

is a parsimony-adjusted measure that follows a non-central chi-square distribution, with smaller

values indicating a better fit. It generally decreases with increases in degrees of freedom, or increased model parsimony. Typical standards for good fit for the RMSEA are less than .08, with values less than .05 preferred (Browne & Cudeck, 1993). The SRMR is calculated using covariance residuals, which would be zero in a model with perfect fit. As a result, lower values indicate better model fit, with values less than or equal to .05 considered adequate fit (Kline,

2011).

In addition to examination of fit indices, the correlations residuals matrix was also examined to identify poorly specified items as needed (Kline, 2011). Following examination of indices of fit, individual items were removed from the measurement model if factor loadings were less than .40, and correlations between factors were examined for discriminant validity, with correlations less than .80 as an indicator that factors measured distinct constructs (Kline,

2011).

To examine the path model, a Maximum Likelihood (ML) estimation method was used to estimate relationships between factors. To evaluate fit of the overall model, the model chi- square, RMSEA, CFI, and SRMR were examined. Standards for these indices were the same as in the confirmatory factor analysis. Standardized and unstandardized regression weights were calculated as part of this analysis to estimate the strength of relationships between factors.

Methods as described by Hayes (2017) were used to evaluate hypotheses related to mediation. Namely, unstandardized indirect effects were calculated, and a 90% Confidence interval was calculated by bootstrapping. P-values associated with an unstandardized indirect effect are also applicable to the respective standardized indirect effect (i.e. the product of two standardized regression weights). 40

Respecification

If the analyzed path model was a poor fit for the data, the model was respecified using

theoretical rationale(s) and analyzed as an alternative model to investigate whether alternative models were a better fit for the data.

41

RESULTS

Participant Characteristics

A total number of 1988 Mechanical Turk (MTurk) workers took the screener. Of these

potential participants, 537 met inclusion criteria of a diagnosis of diabetes. Of these, 99 participants failed one or both of the attention checks and were excluded from the data, leaving

438 participants. Another 101 participants indicated that they had received a diagnosis of

Alzheimer’s disorder in addition to diabetes from a medical professional (Table A2 lists frequency of other comorbid conditions). These participants were excluded as it was not possible to determine their level of cognitive functioning, and many of the respondents that endorsed a diagnosis of Alzheimer’s also reported multiple chronic illnesses, which may be an indicator of

over-reporting in an attempt to qualify for the study. This resulted in a sample size of 337

participants. (Appendices A and B include tables and figures of data presented in the remainder

of this section.)

Table A1 presents the participant characteristics of the final sample of participants

summarized. Of the 334 participants for whom location was recorded, 65.6% (n = 219) were

from India, 31.7% (n = 106) were from the United States, 0.9% were from Canada (n = 3), 0.6%

(n = 2) were from Russia, 0.3% (n = 1) were from Kenya, 0.3% (n = 1) were from Venezuela,

0.3% (n = 1) were from Japan, and 0.3% (n = 1) were from China. Since a substantial number of participants were from India and the United States, demographic data for each of these subsets are reported in addition to aggregate data.

Of participants who reported age (n = 316), participants were on average 32.57 (SD =

9.37, Range = 20-70) years of age. Of participants who indicated gender (n = 337), male participants made up 61.6% of the sample, and females made up 37.4% of the sample. Of the 42

337 participants who indicated race, 62.7% were Asian (n = 210), followed by 20.8% White (n =

70), followed by 7.7% Black/African American (n = 26). A total of 3.3% of participants were

Native American (n = 11) and 3.0% reported that they were multiracial (n = 10). Of the remaining participants, 0.3% (n = 1) reported Native Hawaiian/Pacific Islander as their race and

1.2% of the sample (n = 4) reported their race as “other.” Of the 337 participants who reported ethnicity, the majority of the sample were non-Hispanic (72.7%; n = 88).

Of participants from India (n = 219) who reported age (n = 204), participants were on average 30.15 (SD = 6.93, Range = 22-67) years of age. Of participants who indicated gender (n

= 219), male participants made up 64.4% of the sample, and females made up 35.6% of the sample. Of the 219 participants who indicated race, 90.0% were Asian (n = 197), followed by

4.6% Native American (n = 10), followed by 2.7% White (n = 6). A total of 2.3% of participants were multiracial (n = 5) and 0.5% reported that they were Black/African American (n = 1). Of the 219 participants who reported ethnicity, the majority of the sample were non-Hispanic

(70.8%; n = 155).

Of participants from the United States (n = 106) who reported age (n = 103), participants were on average 36.97 (SD = 11.92, Range = 20-70) years of age. Of participants who indicated gender (n = 105), male participants made up 55.7% of the sample, and females made up 43.4% of the sample. Of the 104 participants who indicated race, 57.5% were White (n = 61), followed by 22.6% Black (n = 24), followed by 8.5% Asian (n = 9). A total of 4.7% of participants were multiracial (n = 5) and 2.8% reported that they were some other race (n = 3). Of the remaining participants, 0.9% (n = 1) reported Native American as their race and 0.9% of the sample (n = 1) reported their race as Native Hawaiian/Pacific Islander. Of the 105 participants who reported ethnicity, the majority of the sample were non-Hispanic (77.4%; n = 82). 43

Education and Socioeconomic Status

Participants were generally well-educated, with most reporting attending at least some college. Of the 337 participants who reported education level, 55.2% (n = 186) reported receiving a 4 year college degree, followed by 16.0% (n = 54) of participants who reported earning a Master’s degree, 12.4% (n = 42) who attended some college, 9.8% (n = 33) who earned a 2 year degree, 2.7% (n = 9) who earned a high school diploma or equivalent, 2.4% (n = 8) who earned a professional degree (e.g. MD, DDS, JV), and 0.3% (n = 1) who earned a doctoral degree.

In the India sample (n = 219), all participants reported education level. About 63.0% (n =

138) reported receiving a 4 year college degree, followed by 16.0% (n = 35) of participants who reported earning a Master’s degree, 10.0% (n = 22) who attended some college, 8.2% (n = 18) who earned a 2 year degree, 2.3% (n = 5) who earned a professional degree, and 0.5% (n = 1) earned a high school diploma or equivalent. Income is not reported for the sample from India, as the items asked participants to report their income in dollars.

In the US sample (n = 106), 105 participants reported education level. About 39.6% (n =

42) reported receiving a 4 year college degree, followed by 18.9% (n = 20) of participants who reported attending some college, 16.0% (n = 17) who earned a Master’s degree, 13.2% (n = 14) who earned a 2 year degree, 7.5% (n = 8) who earned a high school diploma or equivalent, 2.8%

(n = 3) who earned a professional degree (e.g. MD, DDS, JV), and 0.9% (n = 1) who earned a doctoral degree. Of the 105 participants who indicated household income, 3.8% (n =4) indicated income of less than $20,000. 37.7% (n = 40) indicated income of $20,000-$49,999, 50.9% (n =

54) indicated income of $50,000-$99,999, 5.7% (n = 6) reported income of $100,000-$149,999, and 0.9% (n = 1) reported a household income of greater the $150,000. 44

Diabetes Outcomes

Information related to diabetes outcomes are listed in Table A3. Of the 337 participants

who indicated diabetes type, 50.4% (n = 170) indicated that they had been diagnosed with Type

1 diabetes, 38.9% (n = 131) indicated that they were diagnosed with type 2 diabetes, and 9.2% (n

= 31) indicated “I’m not sure.”

In regard to glycemic control, only 203 participants had responded to the item with a

numeric response, with a mean of 33.41 (SD = 185.75). After outliers and medically unlikely scores (A1c > 20 or < 4) were removed, a total of 150 responses were retained with a mean of

6.34 (SD = 1.08). This is lower than means of A1c control reported in other investigations of people with diabetes (e.g. M = 7.75, SD = 1.63; Range: 4.60-14.30; see Reddy et al., 2013).

Regarding medication for diabetes, 47.8% (n = 161) of participants reported that they only take oral medication to treat their diabetes, 32.0% reported that they take insulin only,

12.8% (n = 43) reported that they take both oral medication and insulin, and 7.4% (n = 25) reported that they do not take medication to treat their diabetes. On average, participants received a diagnosis of diabetes about 42.52 months ago (SD = 66.30; n = 147). Participants reported seeing a medical professional an average of 13.78 weeks prior to participating in this survey (SD

= 25.11; n = 228).

In the India sample, 219 participants indicated diabetes type. 56.2% (n = 123) indicated that they had been diagnosed with Type 1 diabetes, 30.1% (n = 66) indicated that they were diagnosed with type 2 diabetes, and 13.7% (n = 30) indicated “I’m not sure.” They reported a mean A1c of 6.18 (SD = 1.06, n = 84). Regarding medication for diabetes, 47.9% (n = 105) of participants reported that they only take oral medication to treat their diabetes, 34.2% (n = 75) reported that they take insulin only, 14.6% (n = 32) reported that they take both oral medication 45

and insulin, and 3.2% (n = 7) reported that they do not take medication to treat their diabetes. On

average, participants received a diagnosis of diabetes about 20.52 months ago (SD = 66.30; n =

89). Participants reported seeing a medical professional an average of 13.58 weeks prior to

participating in this survey (SD = 23.57; n = 138).

In the USA sample, 104 indicated diabetes type. In this group, 57.5% (n = 61) indicated that they had been diagnosed with Type 2 diabetes, 39.6% (n = 42) indicated that they were diagnosed with type 1 diabetes, and 0.9% (n = 1) indicated “I’m not sure.” They reported a mean

A1c of 6.57 (SD = 1.10, n = 61). Regarding medication for diabetes, 46.2% (n = 49) of participants reported that they only take oral medication to treat their diabetes, 27.4% (n = 29) reported that they take insulin only, 10.4% (n = 11) reported that they take both oral medication and insulin, and 16.0% (n = 17) reported that they do not take medication to treat their diabetes.

On average, participants received a diagnosis of diabetes about 80.59 months ago (SD = 95.35; n

= 54). Participants reported seeing a medical professional an average of 14.29 weeks prior to participating in this survey (SD = 28.46; n = 83).

Medication adherence was assessed with a modified version of the Morisky Medication

Adherence Scale and an item to assess frequency of self-monitoring of blood glucose. For this study, the average score on the MMAS was 3.98 (SD = 1.81). Morisky et al. (2008) suggest that highly adherent patients have scores of 8, medium adherent patients have scores of 6 to <8, and low adherent patients have scores of <6. Thus, the sample was reporting poor average medication adherence. Possible responses to the item “How often do you measure your blood glucose?” ranged from 0: Never to 4: More than once per day. Participants reported a mean of 2.72 (SD =

0.98; range = 1-5), indicating frequency of monitoring blood glucose between 3-4 times per week and daily. 46

Statistical Assumptions

Following examination of univariate and multivariate normality, three multivariate

outliers were uncovered by examination of Mahalanobis distances of > 22. As a result, the

sample size for the following analyses was n = 334. Data otherwise met criteria for univariate

and multivariate normality. Examination of squared multiple correlations did not yield values

suggesting extreme collinearity. The data matrix met the criteria of positive definiteness. The

inverse of the data matrix was able to be derived, and all eigenvalues were positive. None of the

correlations included in the matrix were out of bounds.

Exploratory Between-Groups Analyses of Main Study Variables

Two large subsets in the sample were identified: 219 participants from India, and 106

participants from the United States. In addition, type 2 diabetes is the most common type of

diabetes in both countries, but type 1 diabetes is more prevalent in India than in other countries

(CDC, 2020; Unnikrishnan et al., 2016). Prevalence rates of diabetes in India varies widely by

region, with reports of prevalence rates ranging from less than 1% to 38.0% (Unnikrishnan et al.,

2016). A chi-square test was conducted to identify differences in diabetes type by country to

investigate whether this differences was present in this sample.

There was a significant association between country and diabetes type (type 1 or type 2),

χ2(1) = 16.02, p < .001. Participants from India were more likely to report Type 1 diabetes

whereas participants from the USA were more likely to report Type 2 diabetes.

Following this analysis, an exploratory factorial Multivariate Analysis of Variance

(MANOVA) was conducted to identify possible significant differences in study variables between groups and whether there were interaction effects of country and diabetes type. This test 47

was chosen to reduce likelihood of Type 1 error by conducting a single statistical test (Field,

2011). Any significant between-group differences were used to guide subsequent, main analyses.

Country (India and United States) and diabetes type (type 1 or type 2) were entered as

categorical predictors of A1c, time since diagnosis, time since last doctor’s visit, and total scores

of the following measures: mindfulness, microaggressions, depressive symptoms, diabetes-

related distress, medication adherence, and health behaviors. Using Pillai’s trace statistic, there

was a significant main effect of country on study variables, V = 0.378, F(9, 64) = 4.318, p < .001.

There was no significant main effect of diabetes type, V = 0.196, F(9, 64) = 1.733, p = .100, or interaction effect V = 0.078, F(9, 64) = 0.598, p = .794.

Separate univariate ANOVAs were conducted on the outcome variables as post hoc analysis.

Mindfulness

There was no significant main effect of country on mindfulness, F(1, 72) = 0.052, p =

.822.

Microaggressions

There was a significant main effect of country on perceived microaggressions, F(1, 72) =

27.437, p < .001, such that participants from India reported more frequent experience of microaggressions (M = 2.57, SD = 0.97) than participants from the United States (M = 1.09, SD

= 1.24). 48

Depressive Symptoms

There was a significant main effect of country on depressive symptoms, F(1, 72) = 5.42, p = .023 such that participants from India reported more symptoms (M = 30.37, SD = 6.55) than participants from the United States (M = 24.66, SD = 13.23).

Diabetes-Related Distress

There was a significant main effect of country on diabetes-related distress, F(1, 72) =

15.611, p < .001 such that participants from India reported more diabetes-related distress (M =

59.04, SD = 13.46) than participants from the United States (M = 38.87, SD = 25.30).

Health Behaviors

There was a significant main effect of country on depressive symptoms, F(1, 72) = 7.550, p = .008 such that participants from India reported more frequent health behaviors (M = 1.91, SD

= 0.49) than participants from the United States (M = 1.52, SD = 0.56).

Medication Adherence

There was a significant main effect of country on medication adherence, F(1, 72) =

10.033, p = .002 such that participants from India reported greater medication adherence (M =

4.11, SD = 1.53) than participants from the United States (M = 2.58, SD = 2.14).

Glycemic Control

There was no significant main effect of country on glycemic control, F(1, 72) = 0.383, p

= .538. 49

Time Since Diagnosis

There was a significant main effect of country on time since diagnosis, F(1, 72) = 10.973, p = .001 such that participants from India tended to report a more recent diagnosis (M = 26.31,

SD = 24.00), than those from the USA (M = 62.80, SD = 62.56).

Time Since Last Physician’s Appointment

There was no significant main effect of country on time since last visit to a physician,

F(1, 72) = 0.017, p = .895.

Type of Medication

A chi-square test was conducted to identify differences in medication by country. There was a significant association between country and type of medication, χ2(3) = 18.02, p < .001.

Based on the odds ratio, the odds of someone from India taking oral medication alone to manage

diabetes was 2.14 times higher than those living in the United States. The odds of someone from

India taking insulin alone to manage diabetes was 2.59. The odds of someone from India taking

both insulin and oral medications to manage diabetes was 2.90. The odds of someone from India

not taking medications to manage diabetes as 0.41.

Main Analyses

CFA

The factor structure of the latent variables was evaluated with a Maximum Likelihood

confirmatory factor analysis (CFA). Input data for the CFA and path models 1 and 2 are

summarized in Table A8. Per Soper’s (2020) sample size calculator, a sample size of 218 was

required for this analysis. However, a sample size of 150 was used, as this was the number of

participants who had responded to the item assessing A1c, and the amount of missing data for

this item was greater than the 5% set for mean substitution. 50

The estimation process converged onto a solution; however, the initial model fit was

poor. The model chi-square was significant, χ2(104) = 401.50, p < .001, indicating that the

proposed model was significantly different from the data. Additional fit indices indicated that the

hypothesized model was a poor fit to the data, RMSEA = 0.093 (90% CI = .083-.102, pclose <

.001); CFI = .544, SRMR = 0.096. Model fit statistics for all models are presented in Table A10.

All of the medication adherence items loaded poorly onto the latent variable of Treatment

Adherence, so the model was respecified into two separate latent variables: Health Behaviors and

Medication Adherence. Following this respecification, items with standardized factor loadings less than .40 were removed, including item 4 from the health behavior measure and items 1, 3, 5,

7, and 8 from the Morisky Medication Adherence measure.

The final model chi-square was not significant, χ2(19) = 28.91, p = .064, indicating that

the model was not significantly different from the data. Additional fit indices were examined.

Indices of fit indicated that the hypothesized model fit adequately to the data, RMSEA = 0.059

(90% CI = .000-.100, pclose = .332); CFI = .943, SRMR = .040. Factor loadings for each item are

listed in Table A11. Measurement error variance for each item is included in Table A10. Factor

variances and covariances are found in Table A12. The measurement model respecified in this

CFA is presented in Figure B4, and the updated structural regression model that was tested in

following analyses is presented in Figure B5.

SEM

Structural equation modeling was used to test hypotheses about direction of relationships between variables. Significant data were missing (55%) for A1c, as a result, the sample size for

SEM analysis was reduced to the 150 valid responses to test the hypotheses. However, this 51

resulted in a significant loss of power, as a sample size of 212 was required for this analysis

(Soper, 2020).

The congruence between model and data were less than adequate (see Table A10 for

model fit indices). The model chi-square was significant, χ2(60)=132.33, p < .001, indicating that

the model was significantly different from the data. However, as the chi-square test is sensitive

to large sample sizes, additional fit indices were examined. Indices of fit indicated that the

hypothesized model were a poor fit to the data, RMSEA = 0.085 (90% CI = .064-.106, pclose <

.004); CFI = .888, SRMR = .097.

Respecification

The latent variable of Health Behaviors was removed in model respecification (see Figure

B6). In the current literature, the relationship between diabetes-related distress and health behaviors, particularly exercise and diet, is not consistent, and when such relationships are present, they are often small (Aikens, 2012; Fisher et al., 2012; Gonzalez et al., 2008; Huang et al. 2010; Pintaudi et al., 2015; Polonsky et al., 2005; Potter et al., 2015; Ting et al., 2011). As a result, Health Behaviors was removed to investigate fit of the model in its absence. This improved power somewhat, as a sample size of 177 was needed to detect a medium effect size

(Soper, 2020).

The congruence between model and data were adequate. Indices of fit indicated that the model fit the data well. The model chi-square was significant, χ2(16) = 18.53 , p = .294,

indicating that the proposed model was not significantly different from the data. Indices of fit

indicated that the hypothesized model fit well to the data, RMSEA = 0.033 (90% CI = .000-.086, pclose = .645); CFI = .993; SRMR = .042. 52

Relationships Between Main Study Variables

P-values associated with unstandardized regression coefficients were used to assess the significance of a relationship. Standardized regression coefficients were used to interpret the magnitude of relationships between factors, with coefficients of .10 interpreted as small, coefficients of .30 interpreted as medium, and coefficients around .50 as large (Kline, 1998). The results presented in this section are summarized in Tables A14-15.

Hypothesis 1

Greater perceived microaggressions will be associated with increased self-reported depressive symptoms.

The unstandardized path coefficient between perceived microaggressions and depressive symptoms was positive and significant (b = 3.74, SE = 0.42, p < .001). The standardized path coefficient of 0.48 indicated a large relationship between these variables. As a result, the hypothesis that microaggressions were associated with increased self-reported depressive symptoms was supported. The data used to evaluate this hypothesis are also summarized in

Table A14.

Hypothesis 2

Higher levels of mindfulness will be associated with lower self-reported depressive symptoms.

The unstandardized path coefficient between mindfulness and depressive symptoms was negative and significant (b = -17.520, SE = 1.941, p < .001). The standardized path coefficient of

-0.49 indicated a large relationship between these variables. As a result, the hypothesis that 53 mindfulness would be associated with decreased self-reported depressive symptoms was supported. The data used to evaluate this hypothesis are also summarized in Table A14.

Hypothesis 3

Microaggressions will be associated with increased diabetes-related distress through increased depressive symptoms. Mindfulness will be associated with decreased diabetes-related distress through decreased depressive symptoms.

To evaluate whether depressive symptoms mediated the relationship between predictor variables and diabetes-related distress, the standardized regression coefficients between the predictor and criterion variables were multiplied to obtain a standardized indirect effect. The significance of this indirect effect was tested through bootstrapping procedures. Unstandardized indirect effects were computed for each of the 10,000 bootstrapped samples, and a 90% confidence interval was computed.

The standardized indirect effect of microaggressions on diabetes-related distress through depressive symptoms was 0.25, which corresponded to a small, positive relationship between these variables. The bootstrapped unstandardized indirect effect was significant (b = 4.28; SE =

.86; CIbootstrap = 3.09 – 6.03; p = .004). Therefore, the hypothesis that microaggressions would be related to diabetes-related distress through depressive symptoms was supported.

The standardized indirect effect of mindfulness on diabetes-related distress through depressive symptoms was -0.26, indicating a small, inverse relationship between these variables.

The bootstrapped unstandardized indirect effect was significant (b = -20.04; SE = 4.22; CIbootstrap

= -29.27 - -14.91; p = .003). Therefore, the hypothesis that mindfulness would be related to 54 diabetes-related distress through depressive symptoms was supported. The data used to evaluate this hypothesis are also summarized in Table A15.

Hypothesis 4

Increases in diabetes-related distress will be associated with lower levels of Treatment

Adherence, namely, worse diet, exercise, and medication adherence.

The unstandardized regression coefficient between diabetes-related distress and medication adherence was 0.002 (SE = .006, p = .219), indicating a nonsignificant relationship between these variables. Therefore, the hypothesis that higher levels of diabetes-related distress would be associated with poorer medication adherence was not supported (see Table A14).

The unstandardized regression coefficient between diabetes-related distress and diet and exercise was 0.01 (SE = .002, p < .001), indicating a significant relationship between these variables. The standardized regressions coefficient indicated a medium, positive relationship (β =

0.31). Therefore, the hypothesis that higher levels of diabetes-related distress would be associated with poorer diet and exercise behaviors was not supported. Instead, higher levels of distress were associated with higher levels of adherence.

Hypothesis 5

Increases in diabetes-related distress will be associated with worse glycemic control, which will be explained by worse Treatment Adherence.

The standardized effect of diabetes-related distress on glycemic control through medication adherence was -.12, indicating a small, inverse relationship. The bootstrapped unstandardized indirect effect was not significant (b = -.006; SE = .006; CIbootstrap = -.016, .002; 55

p = .230). Therefore, the hypothesis that diabetes-related distress would be related to glycemic

control through medication adherence was not supported.

The latent variable measuring diet and exercise was not retained in the respecified model;

therefore, a separate regression model was conducted to determine whether health behaviors were a mediator of the relationship between diabetes-related distress and glycemic control.

A mediation analysis was conducted using Andrew Hayes’ Process macro for SPSS with

diabetes-related distress entered as a predictor variable, glycemic control as the criterion variable, and health behaviors entered as the mediator variable. The standardized effect of diabetes-related distress on glycemic control through health behaviors was -.03, indicating a small, inverse relationship. The bootstrapped unstandardized indirect effect was not significant

(b = -.001; SE = .001; CIbootstrap = -.004, .001; p = .293). Therefore, the hypothesis that diabetes-

related distress was related to glycemic control through medication adherence was not supported.

The data used to evaluate this hypothesis are also summarized in Table A15.

Exploratory Multigroup Analysis

Two nested models were compared to determine whether the hypothesized model changed as a function of country and with the exclusion of A1c. This was done for a few reasons.

The first was that excluding A1c allowed the sample size of participants to be expanded to 301, as there was significant missing data for this item. This allowed for greater power for each model. Secondly, there were several differences in study variables between countries; therefore, it was of interest to investigate whether there were significant differences in relationships between study variables as a function of country. Other countries could not be included in this

multi-group analysis due to smaller sample sizes. The data presented in this section is 56

summarized in Tables A16-18, and the structural regression model was specified as in Figure B7

(model 3).

Within this sample, the congruence between model and data were adequate. Indices of fit indicated that the model fit the data well and values of fit were similar to fit of Model 2, which

was tested in the main analyses. The model chi-square was significant, χ2(16) = 18.530, p = .294,

indicating that the proposed model was not significantly different from the data. Indices of fit

indicated that the hypothesized model fit well to the data, RMSEA = 0.033 (90% CI = .000-.086, pclose = .645); CFI = .993; SRMR = .038. Power was greatly improved, as a sample size of 170 was calculated to detect a medium effect size (Soper, 2020).

The nested model comparison indicated that the two total models were significantly different (CMIN(9) = 19.808, p = .011). The paths between each major study variable were examined for between-group differences.

Mindfulness

There was a significant difference between groups for the relationship between mindfulness and depressive symptoms (CMIN(1) = 8.752, p = .032) and for the relationship between mindfulness and diabetes-related distress (CMIN(1) = 4.613, p = .032).

An examination of standardized coefficients between groups indicated a weaker relationship between mindfulness and depressive symptoms for the sample of participants from

India than for participants from the United States (βIndia = -.214; βUSA = -.577). In addition, there

was a weaker relationship between mindfulness and diabetes-related distress for the sample of participants from India than for participants from the United States (βIndia = -.032; βUSA = -.181) 57

Microaggressions

There was no significant difference between groups in the relationship between microaggressions and depressive symptoms (CMIN(1) = .063, p = .802) or microaggressions and diabetes-related distress (CMIN(1) = 2.064, p = .151).

Depressive Symptoms

There was a significant difference between groups for the relationship between depressive symptoms and diabetes-related distress (CMIN(1) = 4.378, p = .036).

An examination of standardized coefficients between groups indicated a weaker relationship between depressive symptoms and diabetes-related distress for the sample of participants from India than for participants from the United States (βIndia = .433; βUSA = .725).

Diabetes-Related Distress

There was no significant difference between groups for the relationship between diabetes-related distress and medication adherence (CMIN(1) = 1.933, p = .164).

58

DISCUSSION

This was a study of perceived microaggressions, psychological well-being, treatment adherence, and glycemic control among adult Mechanical Turk workers with diabetes. The purpose of this study was to test a model of relationships between these variables and to determine the strength of relationships between these variables. This study also unintentionally included an international sample of participants, including two major subsets of participants, one from the United States and the other from India. This resulted in collection of interesting data with unexpected findings as well as the creation of conceptual and statistical challenges that are discussed throughout the remainder of the chapter.

To review hypothesized relationships, the first hypothesized relationship was that greater perceived microaggressions would be associated with increased self-reported depressive symptoms, which was supported by the data. The magnitude and nature of this relationship was large and positive, indicating that as frequency of perceived microaggressions increased, self- reported depressive symptoms also increased. Participants from India tended to report greater frequency of perceived microaggressions than those from the United States, though there were no group-based differences between the relationships between microaggressions and other variables. This is perhaps contrary to expectations, since microaggressions likely exist in a different cultural context between the two countries. For instance, region of origin and the social have been associated with perceived discrimination (Agarwal & Priyanka,

2017; Sohi & Singh, 2016; Verma & Acharya, 2018). However, colorism, or a preference for lighter skin and European features, is experienced in both the United States and India and is highly associated with perceived discrimination (Hunter, 2007; Misra, 2015). As a result, colorism may account for lack of significant differences between groups. While Asians are the 59 numerical majority in India and retain significant political power, there exists in India a history of imperialism by the British Empire that has contributed to present-day, pervasive colorism and language affiliation of Asian Indian elite to English (Chand, 2011; Mishra, 2015). Future research conducted internationally should include measures of perceived discrimination based on colorism to test this hypothesis.

The second hypothesized relationship was that higher levels of mindfulness would be associated with lower self-reported depressive symptoms. Mindfulness may represent a protective factor for health-related psychological variables. Several researchers have documented inverse relationships between mindfulness and ill-being, especially depressive symptoms, but also anxiety, burnout, and general health (Di Benedetto & Swadling, 2014; Bogusch et al., 2016;

Bowlin & Baer, 2012; Roberts & Danoff-Burg, 2010; Masuda & Tully, 2012). In addition, interventions to teach mindfulness skills, such as Mindfulness-based Cognitive Therapy, have been associated with reductions in depressive symptoms (Goldberg et al., 2019).

The third hypothesized relationships were that depressive symptoms would mediate relationships between microaggressions and mindfulness and diabetes-related distress. These relationships were supported by the data. These findings follow the trend of research that demonstrates relationships of perceived discrimination to psychological distress in general populations (Foynes et al., 2015; Hagiwara et al., 2015; Himmelstein et al., 2015; LeBron et al.,

2014; Nadimpalli & Hutchinson, 2012; Penner et al., 2009; Sanders-Philips et al., 2014).

However, the relationship between microaggressions and diabetes-related distress represents a connection between perceived social interactions and distress related to health among people with diabetes in particular. These findings may serve as a foundation for creation of interventions for improving distress, and possibly health behaviors, among people with diabetes. 60

In regard to relationships between mindfulness and diabetes-related distress, this

relationship was supported and adds to the current, limited literature of trait mindfulness.

Alternatively, mindfulness-based interventions have been associated with decreases in diabetes-

related distress (Bogusch & O’Brien, 2019). Notably, the relationship between mindfulness and

depressive symptoms and diabetes-related distress was weaker among participants from India in

comparison to participants from the United States, though there was no between-group

difference in mindfulness itself. Bishop and colleagues (2004) note that all people have some

level of mindfulness, though the level of mindfulness can change with practice. There may be

cultural variables at play that influence the strength of the relationship between mindfulness and

other constructs; for instance, in Western countries, like the United States, seeking happiness is a

common societal value (Schmidt, 2011). Future research investigating correlates of mindfulness

in international samples should include such measures.

The relationship between depressive symptoms and diabetes-related distress falls in line

with extensive previous research indicating a positive relationship between diabetes-related distress and depressive symptoms (Aikens, 2012; Fisher et al., 2010; Gonzalez et al., 2008;

Gonzalez, et al., 2015; Lee et al., 2014; Reddy et al., 2013; Ting et al., 2011; Tsujii et al., 2012; van Bastelaar et al., 2010). The sample deviates from previous research in its extremely high rate of self-reported depressive symptoms. Nearly 90% of participants met the CES-D cutoff for

depressive symptoms, and the mean CES-D score was also above the clinical cutoff. This

prevalence rate is much higher than rates of depressive disorders in other MTurk samples (Ophir

et al., 2020) It is important to remember that the CES-D alone is not sufficient to diagnose

depressive disorders and that clinical judgment is needed to determine whether a depressive

disorder is truly present. One possible explanation is that the presence of such high depressive 61 symptoms may be indicative of significant distress in individuals with diabetes. Prevalence of depressive symptoms is much higher in chronic illness populations than in the general population

(Kang et al., 2015). In addition, samples drawn from MTurk have reported poorer general health in comparison to data collected in more traditional methods (Walters et al., 2018). Moreover, about half of participants (50.7%) reported at least one additional comorbid health condition, including hypertension (17.2%), depression (15.7%), high cholesterol (15.4%), and chronic kidney disease (15.1%), which may lead to even greater increases in depressive symptoms.

Managing multiple chronic illnesses likely impacts quality of life and may lead to depressive symptoms. Other researchers have reported a so-called dose-response relationship between comorbid health conditions and depressive symptoms, with increases in diagnoses associated with increases in self-reported depressive symptoms (Gunn et al, 2012).

Increases in diabetes-related distress were hypothesized to be associated with lower frequency of self-reported Treatment Adherence, including exercise and diet, and medication adherence. Contrary to expectations, there was a small and positive relationship between diabetes-related distress and health behaviors of diet and exercise. This follow trends of previous research, which has reported small, though generally inverse, relationships between diabetes- related distress and better diet, exercise, and medication adherence (Aikens, 2012; Bogusch &

O’Brien, 2016; Fisher et al., 2012; Gonzalez et al., 2008; Huang et al. 2010; Park et al., 2018;

Pintaudi et al., 2015; Polonsky et al., 2005; Potter et al., 2015). In their discussion of health behaviors, Park and Iacocca (2014) write that health behaviors occur in complex social contexts, and are therefore subject to the influence of other variables, and may themselves be used as coping strategies for psychological distress. In this particular sample, most participants reported a comorbid health condition, which may have motivated some participants to participate in 62

increased health behaviors. For some participants, improving diet and exercise may have served

as strategies for stress reduction and also as problem-solving strategies to address poor glycemic control. For other participants, eating comfort foods and avoiding exercise may be activities used as avoidance strategies to cope with psychological distress associated with diabetes. Existence of

these two groups in the data may have effectively resulted in findings of no relationships

between health behaviors and other study variables. Regardless, lack of a relationship was an

atypical finding in the context of the present literature, so future research should be conducted to

replicate these findings and to assess whether these relationships differ between groups with and

without multiple chronic health conditions and whether participants tend to use avoidance or

approach coping strategies.

Finally, we hypothesized that increased frequency of diet, exercise, and medication

adherence would be associated with improved glycemic control, represented by self-reported

A1c. Unexpectedly, these relationships were weak and not supported by the data. Participants

generally reported poor medication adherence, and over half endorsed engaging in diet and

exercise included in the analysis “some of the time” or “often.” It is important to note that

engagement in frequent health behaviors does not necessarily preclude engaging in unhealthy

behaviors. Indeed, a similar proportion of participants reported frequent fast food consumption

and eating refined grains, but such items were not included in the structural regression model. An

alternative, and more likely, explanation for lack of a relationship to A1c for these variables is

that most participants had poor recall of their diet, exercise, medication adherence, and A1c,

which resulted in unexpected relationships. Harwell and colleagues (2002) found that only about

24% of participants could recall a value for their latest A1c and that their responses were often

inaccurate. As a result, the present study’s data of glycemic control are likely inaccurate and 63

more of a measure of health literacy, as indicated by the number of participants who responded

with medically unlikely A1c values and those who did not respond at all. Future research

including measures of A1c or other tests of glycemic control would likely obtain more accurate

results by direct data collection or review of medical records.

The presence of a latent factor of Treatment Adherence was generally not supported by

the data as hypothesized and did not fit well to the structural regression model. Items measuring

medication adherence, diet, and exercise are all recommended as important treatment factors

(CDC, 2020), these items did not load well onto a single latent factor, and confirmatory factor

analysis supported two latent factors in the measurement model. In addition, the structural

regression model was a poorer fit to the data with the inclusion of health behaviors of diet and

exercise. In total, this suggests that measures of diet and exercise are two separate construct that

are independently related to other psychological variables.

Limitations

This study has several limitations, including limitations related to the sample of

participants collected. The survey was not set up to be limited by region, and as a result, the

present sample included a significant percentage of participants from India. This is not

inconsistent with previous international MTurk samples; in fact, as many as 16% of MTurk

workers after from India, and it is the country with the second-highest number of MTurk workers

after the United States, which make up 75% of these workers (Difallah et al., 2018). This means

that care must be taken in interpreting and generalizing results to their respective populations.

However, this limitation is also in some ways a strength, as the different samples permit

comparisons of psychosocial and behavioral variables among people with diabetes from different countries. 64

One drawback of the cross-cultural sample was that there are differences in prevalence

rates of the type of diabetes between countries, which may limit generalizations of study

findings. Type 1 diabetes is a significant problem in India, as India has the second-largest

population of children with type 1 diabetes in the world (International Diabetes Federation [IDF],

2019). Type 2 diabetes is prevalent in the United States, with about 90-95% of individuals with

diabetes diagnosed as having type 2 diabetes (CDC, 2020). While the exploratory MANOVA revealed that there was no main effect of diabetes type on study variables, other researchers have reported effects of diabetes type on psychological well-being (Schmitt et al., 2015; Wardian et al., 2018). Diabetes type may influence psychological distress in its etiology. Specifically, Type

2 diabetes is reversible to some extent with changes in diet and exercise, while Type 1 diabetes is not curable (American Diabetes Association, 2015). This may lead participants with Type 1 diabetes to view their disease as uncontrollable, while people with Type 2 diabetes may perceive their disease as within their power to control. Another difference between countries included

significant differences in time since diagnosis, with participants from India reporting a significantly shorter time since diagnosis. This may have contributed to increased distress among participants from India, who had less time to adapt to their chronic illness diagnosis than people from the United States.

Other studies of MTurk workers have found that they report poorer general health and higher depressive symptoms than in samples collected by telephone or paper-and-pencil (Ophir et al., 2020; Walters et al., 2018). Indeed, many participants reported a comorbid health condition, with 50.7% of participants reporting at least one chronic health condition in addition to a diagnosis of diabetes. Participants with more than one chronic illness may experience additional distress related to managing multiple health conditions, which may have increased 65 reports of psychological distress, especially depressive symptoms. This makes it difficult to determine the extent to which such distress is associated strictly with living with diabetes, since individuals with multiple chronic health conditions were not ruled out. In addition, individuals with comorbid chronic illnesses may have difficulty with maintaining treatment adherence, such as physical activity. Consequently, the results of this study are likely only generalizable to individuals with poorly controlled diabetes that have multiple chronic health conditions.

Another limitation is that many of the individuals who screened in for reporting a diagnosis of diabetes were screened out for reporting a diagnosis of Alzheimer’s dementia (n =

129. A large percentage (71.3%) of these individuals reported that they were diagnosed with all the chronic illnesses listed in the screener. While it is possible that individual participants may have been diagnosed with multiple chronic illnesses, it is more likely that those who endorsed several comorbid medical conditions were over-reporting in an attempt to qualify for the study

(Wessling et al., 2017). In future research involving online collection of data on platforms such as MTurk, the recommendations of Wessling and colleagues (2017) may be helpful: to use a two-stage survey which asks participants to identify personal characteristics in the initial stage without providing the incentive for participation.

In a similar vein, there are some concerns when collecting research online whether participants are responding honestly. Previous researchers report concerns about participants responding blindly, without considering their responses, as evidenced by little variation in responses (Wessling et al., 2017). The mindfulness measure from the India sample had a standard deviation of 0.14, which may indicate such responding. However, all standard deviations for individual respondents from India were greater than zero (range: 0.40-1.76) for this measure, indicating that some variation occurred on an individual level. Some of these more 66 restricted variations in scores may have contributed to lack of findings of relationships between mindfulness and diabetes-related distress. Also, it is important to consider that lack of variation in responding does not necessarily indicate inattention, but may reflect participant’s true scores.

Another possible explanation for lack of expected relationships may be psychometric problems with the health behaviors and medication adherence measures. Reliabilities for the health behaviors and medication adherence measures were low, and several items from each measure were dropped following the initial confirmatory factor analysis. This indicated that all of the items on each of these measures did not seem to correlate well to one another, and that they did not all load onto the same construct. This puts into question the reliability of these measures for this sample as well as their validity in measuring the constructs of interest, namely health behaviors and medication adherence. Consequently, these measures were not adequate measures of participant health behaviors and medication adherence, which may account for absence of relationships between diabetes-related distress, health behaviors, and medication adherence. The measures used here were also self-report of health behaviors, which can be subject to social desirability bias in online surveys (Brenner & DeLamater, 2014; Widmar, et al.,

2016). Alternative measures of health behaviors, such as photographing meals or activity trackers, may reduce the effect of such bias and provide more objective measure of diet and exercise (Prochaska et al., 2001).

Another concern of this study is that the data collected were cross-sectional, meaning that directionality of the relationships between variables is unable to be determined for certain. In the model presented in this manuscript, outcome variables, such as glycemic control, may in fact precede other predictor variables, such as diabetes-related distress. For example, high glycemic levels may contribute to feelings of increased distress and concern about living with diabetes as 67 well as feelings of low mood associated with depressive symptoms. In a similar fashion, greater diabetes-related distress or depressive symptoms may contribute to perceptions of poorer treatment by others and increase perceptions of microaggressions. Increased psychological distress could also contribute to greater avoidance behaviors, which would decrease mindfulness.

To further complicate this discussion, relationships delineated in the present study may exist in reality, but feedback loops may exist such that outcome variables may influence predictor variables over time. The drawback of using cross-sectional data is that it is impossible to make temporal or causal claims without collection of longitudinal data. However, use of cross- sectional data in structural equation modeling is a common practice (Kline, 2011). This practice is to be used with the support of the existing literature in order to generate a rationale for directionality of relationships. Future longitudinal research should be conducted to support claims of directionality of relationships by collecting data on all of these constructs at multiple time points in order to to evaluate which models are the best fit to the data.

There was some difficulty with assessment of glycemic control that resulted in significant missing data. Unusual and missing values were prevalent for this particular item. This likely is a marker of poor health literacy, as A1c is an indicator for glycemic control for persons with diabetes. As a result, it is difficult to know for certain whether the collected values are accurate, but previous research suggests that self-report is a poor form of collecting this type of medical data (Harwell et al., 2002). Future research including glycemic control should include other, objective data collection in addition to self-report.

Because of these issues with A1c, a main study variable, there were some concerns related to power of the statistical tests. This was especially true in regard to models 1 and 2, which were more complex models that were, ironically, intended to assess the relationship of 68 psychosocial and behavioral variables to glycemic control. The exploratory analysis of the larger sample appeared similar in strength and direction of study variables in which the smaller sample was used. Future research should be done to replicate these results in order to determine with greater confidence relationships of study variables to glycemic control.

Implications and Future Directions

The sample collected here reflects an international sample from the United States and

India who report high levels of psychological distress and multiple chronic health conditions.

Here, we consider implications of our findings for populations included in this sample and suggestions for future research and intervention.

Microaggressions and mindfulness were both related to depressive symptoms and diabetes-related distress. While causal claims cannot be made, these relationships inform development of future interventions for individuals with diabetes, perhaps both in the United

States as well as in India, as the relationships between these variables across samples were significant, though sometimes different in magnitude. Microaggressions may be a point of intervention for decreasing psychological distress, and mindfulness may be a useful skill for improving psychological distress. Future research should include longitudinal and experimental designs to assess directionality and causality of the relationships between these variables in order to inform development of interventions.

Other potential points of intervention may be interventions to improve health literacy. In the present study, a significant proportion of participants could not recall their A1c or provided values that were extreme. This falls into line with previous research documenting poor recall and understanding of measures of glycemic control (Harwell et al., 2002). Additional studies should 69 be conducted to develop interventions to improve understanding of A1c, including what A1c means and how to use this information to inform self-care.

There were significantly more microaggressions reported in the sample of participants from India than in the sample from the United States, though the relationships between microaggressions and other variables were similar. These findings perhaps lead to more questions than conclusions about cross-cultural comparisons of health-related variables. On one hand, the REMS was developed in the cultural context of the United States, and may not capture well the picture of microaggressions in the cultural context of India. The structure of racial, ethnic, and social classes and the relationships between these structures, including power dynamics, are likely very different than in the United States. On the other hand, the impact of global imperialism, and the influence of the rule of the British Empire, are still felt in the present day, and are characterized by an alignment to Whiteness (Chand, 2011; Mishra, 2015). It is unclear whether the data collected here contain the entire picture of perceptions of what microaggressions in India, though the measure used in this study indicated that the experiences of participants from the India sample were associated with greater psychological distress.

Regardless, the construct validity of use of the REMS in an Indian sample is outside of the scope of this study. Future research may find that there are cross-cultural differences in perceptions of microaggressions and their relationships to psychological distress and health behaviors.

One unexpected finding was the high rate of depressive symptoms. This most likely indicates high levels of distress and a great need for individuals in the sample for intervention to cope with living with chronic illnesses. Indeed, this may indicate a need for health care professionals to routinely screen for depressive symptoms among patients with diabetes and multiple chronic health conditions. In addition, this supports the increased push for integrated 70

health care as more and more psychologists are placed in primary health care settings to facilitate

greater access to mental health care (Johnson & Marrero, 2016). However, it is important to note

that the extremely high rates of depressive symptoms in the present sample do not follow current

trends in the existing literature; therefore, additional research should be conducted among similar populations to document rates of depressive symptoms.

Conclusions

This study investigated relationships between perceptions of social interactions,

psychological measures of well-being, self-care behaviors, and glycemic control among persons

with diabetes. Structural equation modeling was used to evaluate these relationships in an

international sample of MTurk workers in which participants were generally from the United

States or India. Tests of the hypothesized models were generally consistent with hypotheses that

better psychological well-being and less frequent microaggressions would be associated with

improved self-care behaviors, including diet, exercise, and taking medication regularly. Some

differences were noted in the magnitude of relationships between participants from the United

States or India, but, the models were generally similar between groups. This study informs future

research on interventions for improvements in treatment adherence for people with diabetes,

including interventions that may include mindfulness skills and interventions for decreasing

impact of microaggressions. 71

REFERENCES

Agarwal, S., & Priyanka, M. (2017). Perceived caste discrimination, self-esteem and self-

determination among college students belonging to scheduled castes. Journal of

Psychosocial Research, 12(1), 147.

American Diabetes Association. (2015). Standards of medical care in diabetes—2015 abridged

for primary care providers. Clinical Diabetes: A Publication of the American Diabetes

Association, 33(2), 97.

Anderson, K. F. & Finch, J. K. (2017). The role of microaggressions, stress, and acculturation in

understanding Latino health outcomes in the USA. Race and Social Problems, 9(3), 218-

233.

Aikens, J. E. (2012). Prospective associations between emotional distress and poor outcomes in

type 2 diabetes. Diabetes Care, 35(12), 2472-2478.

Baer, R. A., Smith, G. T., Hopkins, J., Krietemeyer, J., & Toney, L. (2006). Using self-report

assessment methods to explore facets of mindfulness. Assessment, 13(1), 27-45.

Bailey, C. J., & Kodack, M. (2011). Patient adherence to medication requirements for therapy of

type 2 diabetes. International Journal of Clinical Practice, 65(3), 314-322.

Barnett, V. & Lewis, T. (1978). Outliers in statistical data. Wiley.

Baucom, K. J. W., Queen, T. L., Wiebe, D. J., Turner, S. L., Wolfe, K. L., Godbey, E. I., … &

Berg, C. A. (2015). Depressive symptoms, daily stress, and adherence in late adolescents

with type 1 diabetes. Health Psychology, 34(5), 522-530.

Benjamins, M. R. (2012). Race/ethnic discrimination and preventive service utilitzation in a

sample of whites, Blacks, Mexicans, and Puerto Ricans. Medical Care, 50(10), 870-876. 72

Bentler, P. M., & Chou, C. P. (1987). Practical issues in structural modeling. Sociological

Methods & Research, 16(1), 78-117.

Bishop, S. R., Lau, M., Shapiro, S., Carlson, L., Anderson, N. D., Carmody, J., ... & Devins, G.

(2004). Mindfulness: A proposed operational definition. Clinical Psychology: Science

and Practice, 11(3), 230-241.

Bogusch, L. M., Fekete, E. M., & Skinta, M. D. (2016). Anxiety and depressive symptoms as

mediators of trait mindfulness and sleep quality in emerging adults. Mindfulness, 7(4),

962-970.

Bogusch, L. M., & O'Brien, W. H. (2019). The effects of mindfulness-based interventions on

diabetes-related distress, quality of life, and metabolic control among persons with

diabetes: A meta-analytic review. Behavioral , 45(1), 19-29.

Bogusch, L. M. & O’Brien, W. H. (2016). Mindfulness and Acceptance for Type 2 Diabetes: A

Meta-Analysis. Unpublished manuscript.

Bowlin, S. L., & Baer, R. A. (2012). Relationships between mindfulness, self-control, and

psychological functioning. Personality and Individual Differences, 52(3), 411-415.

Bradburn, N. M. (1969). The structure of psychological well-being. Aldine.

Bränström, R., Duncan, L. G., & Moskowitz, J. T. (2011). The association between dispositional

mindfulness, psychological well‐being, and perceived health in a Swedish population‐

based sample. British Journal of Health Psychology, 16(2), 300-316.

Brenner, P. S., & DeLamater, J. D. (2014). Social desirability bias in self-reports of physical

activity: Is an exercise identity the culprit? Social Indicators Research, 117(2), 489-504.

Brown, T. A. (2006). Confirmatory factor analysis for applied researchers. The Guilford Press. 73

Brown, J. B. (2014). The relationships between mindfulness, diabetes-related distress, selected

demographic variables, and self-management in adults with type 2 diabetes (Publication

No. 9781303972935) [Doctoral dissertation, University of North Carolina at

Greensboro]. ProQuest Dissertations and Theses Global.

Browne M. W. & Cudeck R. (1993). Alternative ways of assessing model fit. In Bollen K, Long

J. (Eds.), Testing structural equation models (pp. 136-162). Sage Publications.

Caluyong, M. B., Zambrana, A. F., Romanow, H. C., Nathan, H. J., Nahas, R., & Poulin, P. A.

(2015). The relationship between mindfulness, depression, diabetes self-care, and health-

related quality of life in patients with type 2 diabetes. Mindfulness, 6(6), 1313-1321.

Carlson, A., & Frazão, E. (2012, May 1). Are healthy foods really more expensive? It depends

on how you measure the price. United States Department of Agriculture.

https://ageconsearch.umn.edu/record/142357/files/eib96_1_.pdf

Centers for Disease Control and Prevention (2020). National diabetes statistics report, 2020.

Centers for Disease Control and Prevention.

https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf

Chand, V. (2011). Elite positionings towards Hindi: Language policies, political stances and

language competence in India. Journal of Sociolinguistics, 15(1), 6-35.

Chew, B., Hassan, N., & Sherina, M. (2015). Determinants of medication adherence among

adults with type 2 diabetes mellitus in three Malaysian public health clinics: A cross-

sectional study. Patient Preference and Adherence, 9, 639-648.

Clevenger, K. A., Pfeiffer, K. A., Yee, K.E., Triplett, A. N., Florida, J., & Selby, S. (2018).

Mindfulness and children’s physical activity, diet, quality of life, and weight status.

Mindfulness, 9(1), 221-229. 74

Comrey, A. L., & Lee, H. B. (2013). A first course in factor analysis. Psychology Press.

Cuffee, Y. L., Hargraves, J. L., Rosal, M., Briesacher, B. A., Schoenthaler, A., Person, S., … &

Allison, J. (2013). Reported racial discrimination, trust in physicians, and medication

adherence among inner-city African Americans with hypertension. American Journal of

Public Health, 103(11), e55-e62.

D’Anna, L. H., Ponce, N. A., & Siegel, J. (2010). Racial and ethnic health disparities: Evidence

of discrimination’s effects across the SEP spectrum. Ethnicity & Health, 15(2), 121-143. de Bruin, E. I., Topper, M., Muskens, J. G., Bögels, S. M., & Kamphuis, J. H. (2012).

Psychometric properties of the Five Facets Mindfulness Questionnaire (FFMQ) in a

meditating and a non-meditating sample. Assessment, 19(2), 187-197.

Desrosiers, A., Vine, V., Klemanski, D. H., & Nolen‐Hoeksema, S. (2013). Mindfulness and

emotion regulation in depression and anxiety: Common and distinct mechanisms of

action. Depression and Anxiety, 30(7), 654-661.

Devarajooh, C. & Chinna, K. (2017). Depression, distress and self-efficacy: The impact on

diabetes self-care practices. Plos One, 12(3), 1-16.

Di Benedetto, M., & Swadling, M. (2014). Burnout in Australian psychologists: Correlations

with work-setting, mindfulness and self-care behaviours. Psychology, health & medicine,

19(6), 705-715.

Diamantopoulos, A., & Sigauw, J. A. (2000). Introducing LISREL: A guide for the uninitiated.

Sage Publications.

Difallah, D., Filatova, E., & Ipeirotis, P. (2018, February). Demographics and dynamics of

mechanical turk workers. In Proceedings of the eleventh ACM international conference

on web search and data mining (pp. 135-143). 75

Dipnall, J. F., Pasco, J. A., Meyer, D., Berk, M., Williams, L. J., Dodd, S., & Jacka, F. N. (2015).

The association between dietary patterns, diabetes and depression. Journal of Affective

Disorders, 174, 215-224.

Edwards, M. B. & Cunningham, G. (2013). Examining associations of perceived community

racism with self-reported physical activity levels and health among older racial minority

adults. Journal of Physical Activity and Health, 10(7), 932-939.

Egede, L.E. & Osborn, C. Y. (2010). Role of motivation in the relationship between depression,

self-care, and glycemic control in adults with type 2 diabetes. The Diabetes Educator,

36(2), 276-283.

Fanning, J., Osborn, C. Y., Lagotte, A. E., & Mayberry, L. S. (2018). Relationships between

dispositional mindfulness, health behaviors, and hemoglobin A1c among adults with type

2 diabetes. Journal of Behavioral Medicine, 41(6), 798-805.

Field, A. (2011). Discovering statistics using SPSS: (And sex and drugs and rock'n'roll). Sage

Publications.

Fekete, E., Geaghan, T. R., & Druley, J. A. (2009). Affective and behavioural reactions to

positive and negative health-related social control in HIV+ men. Psychology and

Health, 24(5), 501-515.

Fekete, E. M., Stephens, M. A. P., Druley, J. A., & Greene, K. A. (2006). Effects of spousal

control and support on older adults' recovery from knee surgery. Journal of Family

Psychology, 20(2), 302.

Fisher, L., Mullan, J. T., Arean, P., Glasgow, R. E., Hessler, D., & Masharani, U. (2010).

Diabetes distress but not clinical depression or depressive symptoms is associated with 76

glycemic control in both cross-sectional and longitudinal analyses. Diabetes Care, 33(1),

23-28.

Fisher, L., Hessler, D. M., Polonsky, W. H., & Mullan, J. (2012). When is diabetes distress

clinically meaningful?: Establishing cut points for the Diabetes Distress Scale. Diabetes

Care, 35(2), 259-264.

Foynes, M. M., Smith, B. N., & Shipherd, J. C. (2015). Associations between race-based and

sex-based discrimination, health, and functioning: A longitudinal study of Marines.

Military Service and Deployment, 53, S128-S135.

Forsyth, J. M., Schoenthaler, A., Ogedegbe, G., & Ravenell, J. (2014). Perceived racial

discrimination and adoption of health behaviors in hypertensive Black Americans: The

CAATCH trial. Journal of Health Care for the Poor and Underserved, 25(1), 276-291.

Franks, M. M., Sahin, Z. S., Seidel, A. J., Shields, C. G., Oates, S. K., & Boushey, C. J. (2012).

Table for two: Diabetes distress and diet-related interactions of married patients with

diabetes and their spouses. Families, Systems, & Health, 30(2), 154-165.

Fuller-Rowell, T. E., Curtis, D. S., Chae, D. H., & Ryff, C. D. (2018). Longitudinal health

consequences of socioeconomic disadvantage: Examining perceived discrimination as a

mediator. Health Psychology, 37(5), 491-500.

Garcia-Perez, L. E., Álvarez, M., Dilla, T., Gil-Guillén, V., & Orozco-Beltrán, D. (2013).

Adherence to therapies in patients with type 2 diabetes. Diabetes Therapy, 4(2), 175-194.

Gee, G. & Walsemann, K. (2009). Does health predict the reporting of racial discrimination or

do reports of discrimination predict health? Findings from the National Longitudinal

Study of Youth. Social Science & Medicine, 68(9), 1676-1684. 77

Gentil, L., Vasiliadis, H., Berbiche, D., & Preville, M. (2017). Impact of depression and anxiety

disorders on adherence to oral hypoglycemic in older adults with diabetes mellitus in

Canada. European Journal of Ageing, 14(2), 111-121.

Gilbert, D. & Waltz, J. (2010). Mindfulness and Health Behaviors. Mindfulness, 1(4), 227-234.

Goldberg, S. B., Tucker, R. P., Greene, P. A., Davidson, R. J., Kearney, D. J., & Simpson, T. L.

(2019). Mindfulness-based cognitive therapy for the treatment of current depressive

symptoms: A meta-analysis. Cognitive Behaviour Therapy, 48(6), 445-462.

Gonzales, K. L., Harding, A. K., Lambert, W. E., Fu, R., & Henderson, W. G. (2013). Perceived

experiences of discrimination in health care: A barrier for cancer screening among

American Indian women with type 2 diabetes. Women’s Health Issues, 23(1), e61-e67.

Gonzales, K. L., Lambert, W. E., Fu, R., Jacob, M., & Harding (2014). Perceived racial

discrimination in health care, completion of standard diabetes services, and diabetes

control among a sample of American Indian women. The Diabetes EDUCATOR, 40(6),

747-755.

Gonzalez, J. S., Delahanty, L. M., Safren, S. A., Meigs, J. B., & Grant, R. W. (2008).

Differentiating symptoms of depression from diabetes-specific distress: Relationships

with self-care in type 2 diabetes. Diabetologia, 51(10), 1822-1825.

Gonzalez, J. S., Shreck, E., Psaros, C., & Safren, S. A. (2015). Distress and type 2 diabetes-

treatment adherence: A mediating role for perceived control. Health Psychology, 34(5),

505-513.

Grabovac, A. D., Lau, M. A., &Willett, B. R. (2011). Mechanisms of mindfulness: A Buddhist

psychological model. Mindfulness, 2(3), 154-166. 78

Gunn, J. M., Ayton, D. R., Densley, K., Pallant, J. F., Chondros, P., Herrman, H. E., & Dowrick,

C. F. (2012). The association between chronic illness, multimorbidity and depressive

symptoms in an Australian primary care cohort. Social Psychiatry and Psychiatric

Epidemiology, 47(2), 175-184.

Hagiwara, N., Alderson, C. J., & McCauley, J. M. (2015). “We get what we deserve”: The belief

in a just world and its health consequences for Blacks. Journal of Behavioral Medicine,

38(6), 912-921.

Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery, and

Psychiatry, 23(1), 56.

Han, T. Y. & Cotter, K. A. (2015). Work-related stress factors and health management among

Korean workers with diabetes. Journal of Managerial Psychology, 30(4), 470-486.

Harwell, T. S., Dettori, N., McDowall, J. M., Quesenberry, K., Priest, L., Butcher, M. K., ... &

Gohdes, D. (2002). Do persons with diabetes know their (AIC) number? The Diabetes

Educator, 28(1), 99-105.

Hayes, A. F. (2017). Introduction to Mediation, Moderation, and Conditional Process Analysis:

A Regression-Based Approach. Guilford publications.

Himmelstein, M. S., Young, D. M., Sanchez, D. T., & Jackson, J. S. (2015). Vigilance in the

discrimination-stress model for Black Americans. Psychology & Health, 30(3), 253-267.

Hu, L. T. & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:

Conventional criteria versus new alternatives. Structural Equation Modeling: A

Multidisciplinary Journal, 6(1), 1-55.

Huang, M. F., Courtney, M., Edwards, H., & McDowell, J. (2010). Validation of the Chinese

version of the problem areas in diabetes (PAID-C) scale. Diabetes Care, 33(1), 38-40. 79

Hunter, M. (2007). The persistent problem of colorism: Skin tone, status, and inequality.

Sociology Compass, 1(1), 237-254.

International Diabetes Federation (2019). International Federation Diabetes Atlas.

https://diabetesatlas.org/upload/resources/material/20200302_133351_IDFATLAS9e-

final-web.pdf

Johnson, S. B., & Marrero, D. (2016). Innovations in healthcare delivery and policy:

Implications for the role of the psychologist in preventing and treating

diabetes. American Psychologist, 71(7), 628.

Jordan, C., Wang, W., & Donatoni, L. R. (2014). Mindful eating: Trait and state mindfulness

predict healthier eating behavior. Personality and Individual Differences, 68, 107-111.

Kang, H. J., Kim, S. Y., Bae, K. Y., Kim, S. W., Shin, I. S., Yoon, J. S., & Kim, J. M. (2015).

Comorbidity of depression with physical disorders: research and clinical implications.

Chonnam Medical Journal, 51(1), 8-18.

Katon, W., Russo, J., Lin, E. H. B., Heckbert, S. R., Karter, A. J., Williams, L. H., … & Von

Korff, M. (2009). Diabetes and poor disease control: Is comorbid depression associated

with poor medication adherence or lack of treatment intensification? Psychosomatic

Medicine, 71(9), 965-972.

Katon, W. J., Russo, J. E., Heckbert, S. R., Lin, E. H. B., Ciechanowski, P., Ludman, E., … &

Van Korff, M. (2010). The relationship between changes in depression symptoms and

changes in health risk behaviors in patients with diabetes. International Journal of

Geriatric Psychiatry, 25(5), 466-475. 80

Kilbourne, A. M., Reynolds, C. F., Good, C. B., Sereika, S. M., Justice, A. C., & Fine, M. J.

(2005). How does depression influence diabetes medication adherence in older patients.

American Journal of Geriatric Psychiatry, 13(3), 202-210.

Kirkman, M. S., Rowan-Martin, M. T., Levin, R., Fonseca, V. A., Schmittdiel, J. A., Herman, W.

H., & Aubert, R. E. (2015). Determinants of adherence to diabetes medications: Findings

from a large pharmacy claims database. Diabetes Care, 38(4), 604-609.

Kline, R. (2011). Principles and practice of structural equation modeling (3rd ed). Guilford

Press.

Korb‐Savoldelli, V., Gillaizeau, F., Pouchot, J., Lenain, E., Postel‐Vinay, N., Plouin, P. F., ... &

Sabatier, B. (2012). Validation of a French version of the 8‐item Morisky medication

adherence scale in hypertensive adults. The Journal of Clinical Hypertension, 14(7), 429-

434.

LeBron, A. M. W., Valerio, M. A., Kieffer, E., Sinco, B., Rosland, A., Hawkins, J., … &

Spencer, M. (2014). Everyday discrimination, diabetes-related distress, and depressive

symptoms among African Americans and Latinos with diabetes. Journal of Immigrant

Minority Health, 16(6), 1208-1216.

Lee, E., Lee, Y. W., Lee, K., Kim, Y. S., & Nam, M. (2014). Measurement of diabetes-related

emotional distress using the Problem Areas in Diabetes scale: Psychometric evaluations

show that the short form is better than the full form. Health and Quality of Life

Outcomes, 12(1), 142.

Lee, H., Chapa, D., Kao, C., Jones, D., Kapustin, J., Smith, J., … & Friedmann, E. (2009).

Depression, quality of life, and glycemic control in individuals with type 2 diabetes.

Journal of the American Academy of Nurse Practitioners, 21(4), 214-224. 81

Lee, D. B., Peckins, M. K., Heinze, J. E., Miller, A. L., Assari, S., & Zimmerman, M. A. (2018).

Psychological pathways from racial discrimination to cortisol in African American males

and females. Journal of Behavioral Medicine, 41(2), 208-220.

Lee, D. B., Kim, E. S., & Neblett, E. W. (2017). The link between discrimination and telomere

length in African American adults. Health Psychology, 36(5), 458-467.

Lentz, T. A. & Brown, C. (2018). Mindfulness and health behaviors in college students: The

moderating role of sleep. Journal of American College Health, 78(2), 169-183.

Loprinzi, P. D., Franz, C., & Hager, K. K. (2013). Accelerometer-assessed physical activity and

depression among US adults with diabetes. Mental Health and Physical Activity, 6(2),

79-82.

Loucks, E. B., Gilman, S. E., Britton, W. B., Gutman, R., Eaton, C. B., & Buke, S. L. (2016).

Associations of mindfulness with glucose regulation and diabetes. American Journal of

Health Behaviors, 40(2), 258-267.

Lustman, P. J., Clouse, R. E., Ciechanowski, P. S., Hirsch, I. B., & Freedland, K. E. (2005).

Depression-relation hyperglycemia in type 1 diabetes: A mediational approach.

Psychosomatic Medicine, 67, 195-199.

Lyles, C. R., Karter, A. J., Young, B. A., Spigner, C., Grembowski, D., Schillinger, D., & Adler,

N. (2011). Provider factors and patient-reported healthcare discrimination in the Diabetes

Study of California (DISTANCE). Patient Education and Counseling, 85(3), e216-e224.

Manan, M. M., Husin, A. R., Alkhoshaiban, A. S., Al-Worafi, Y. M. A., & Ming, L. C. (2014).

Interplay between oral hypoglycemic medication adherence and quality of life among

elderly type 2 diabetes mellitus patients. Journal of Clinical and Diagnostic Research,

8(12), JC05-JC09. 82

Mardia, K. V. (1970). Measures of multivariate skewness and kurtosis with applications.

Biometrika, 57, 519-530.

Mardia, K. V. (1985). Mardia’s test of multinormality. In S. Kotz & N. L. Johnson (Eds.),

Encyclopedia of Statistical Sciences (Vol. 5, pp. 217-221). New York: Wiley.

Masuda, A., & Tully, E. C. (2012). The role of mindfulness and psychological flexibility in

somatization, depression, anxiety, and general psychological distress in a nonclinical

college sample. Journal of Evidence-Based Complementary & Alternative Medicine,

17(1), 66-71.

Mekawi, Y., & Todd, N. R. (2018). Okay to say?: Initial validation of the Acceptability of

Racial Microaggressions Scale. Cultural Diversity and Ethnic Minority

Psychology, 24(3), 346.

Mishra, N. (2015). India and colorism: The finer nuances. Washington University Global

Studies Law Review, 14(4), 725.

Moody-Ayers, S. Y., Stewart, A. L., Covinsky, K. E., & Inouye, S. K. (2005). Prevalence and

correlates of perceived societal racism in older African-American adults with type 2

diabetes mellitus. Journal of the American Geriatric Society, 53(12), 2202-2208.

Moor, K. R., Scott, A. J., & McIntosh, W. D. (2013). Mindful eating and its relationship to body

mass index and physical activity among university students. Mindfulness, 4(3), 269-274.

Moore, L. V., Roux, A. V. D., Evenson, K. R., McGinn, A. P., & Brines, S. J. (2008).

Availability of recreational resources in minority and low socioeconomic status areas.

American Journal of Preventive Medicine, 34(1), 16-22. 83

Morisky, D. E., Ang, A., Krousel‐Wood, M., & Ward, H. J. (2008). Predictive validity of a

medication adherence measure in an outpatient setting. The Journal of Clinical

Hypertension, 10(5), 348-354.

Morisky, D. E., Green, L. W., & Levine, D. M. (1986). Concurrent and predictive validity of a

self-reported measure of medication adherence. Medical Care, 67-74.

Murphy, M. J., Mermelstein, L. C., Edwards, K. M., & Gidycz, C. A. (2012). The benefits of

dispositional mindfulness in physical health: A longitudinal study of female college

students. Journal of American College Health, 60(5), 341-348.

Nadal, K. L. (2011). The racial and ethnic microaggressions scale (REMS): Construction,

reliability, and validity. Journal of Counseling Psychology, 58(4), 470.

Nadal, K. L., Griffin, K. E., Wong, Y., Davidoff, K. C., & Davis, L. S. (2017). The injurious

relationship between racial microaggressions and physical health: Implications for social

work. Journal of Ethnic & Cultural Diversity in Social Work, 26(1-2), 6-17.

Nadimpalli, S. B. & Hutchinson, M. K. (2012). An integrative review of relationships between

discrimination and Asian American health. Journal of Nursing Scholarship, 44(2), 127-

135.

Nadimpalli, S. B., Cleland, C. M., Hutchinson, M. K., Islam, N., Barnes, L. L., Van Devanter, N.

(2016). The association between discrimination and the health of Sikh Asian Indians.

Health Psychology, 35(4), 351-355.

Nadimpalli, S., Keita, A., Wang, J., Kanaya, A., Kandula, N., Gans, K. M., & Talegawkar, S.

(2017). Are experiences of discrimination related to poorer dietary intakes among South

Asians in the MASALA study? Journal of Nutrition Education and Behavior, 49(10),

872-877. 84

Nakahara, R., Yoshiuchi, K., Kumano, H., Hara, Y., Suematsu, H., & Kuboki, T. (2006).

Prospective study on influence of psychosocial factors on glycemic control in Japanese

patients with type 2 diabetes. Psychosomatics, 47(3), 240-246.

Naiker, K., Overland, S., Johnson, J. A., Manuel, D., Skogen, J. C., Sivertsen, B., & Colman, I.

(2017). Symptoms of anxiety and depression in type 2 diabetes: Associations with

clinical diabetes measures and self-measurement outcomes in the Norwegian HUNT

study. Psychoneuroendocrinology 84, 116-123.

Nichols, G. A., Hillier, T. A., Javor, K., & Brown, J. B. (2000). Predictors of glycemic control in

insulin-using adults with type 2 diabetes. Diabetes Care, 23(3), 273-277.

Osborn, C. Y., Satterwhite Mayberry, L., Wagner, J. A., & Welch, G. W. (2014). Stressors may

compromise medication adherence among adults with diabetes and low socioeconomic

status. Western Journal of Nursing Research, 36(9), 1091-1110.

Ophir, Y., Sisso, I., Asterhan, C. S., Tikochinski, R., & Reichart, R. (2020). The Turker blues:

Hidden factors behind increased depression rates among Amazon’s Mechanical Turkers.

Clinical Psychological Science, 8(1), 65-83.

Papathanasiou, A., Koutsovasilis, A., Shea, S., Philalithis, A., Papavasiliou, A., Melidonis, A., &

Lionis, C. (2014). The Problem Areas in Diabetes (PAID) scale: Psychometric evaluation

survey in a Greek sample with type 2 diabetes. Journal of Psychiatric and Mental Health

Nursing, 21(4), 345-353.

Park, C. L. & Iacocca, M. O. (2014) A stress and coping perspective on health behaviors:

Theoretical and methodological considerations, Anxiety, Stress & Coping, 27(2), 123-

137. 85

Park, H., Park, C., Quinn, L., & Fritschi, C. (2015). Glucose control and fatigue in type 2

diabetes: The mediating roles of diabetes symptoms and distress. Journal of Advanced

Nursing, 71(7), 1650-1660.

Park, M., Quinn, L., Park, C., & Martyn-Nemeth, P. (2018). Pathways of the relationships among

eating behavior, stress, and coping in adults with type 2 diabetes: A cross-sectional study.

Appetite, 131, 84-93.

Pascoe, E. A. & Richman, L. S. (2009). Perceived discrimination and health: A meta-analytic

review. Psychological Bulletin, 135(4), 531.

Peek, M. E., Wagner, J., Tang, H., Baker, D. C., & Chin, M. H. (2011). Self-reported racial

discrimination in health care and diabetes outcomes. Medical Care, 49(7), 618-625.

Penner, L. A.,, Dovidio, J. F., Edmondson, D., Dailey, R. K., Markova, T., Albrecht, T. L., &

Gaertner, S. L. (2009). The experience of discrimination and Black-White health

disparities in medical care. Journal of Black Psychology, 35(2), 180-203.

Piette, J. D., Bibbins-Domingo, K. B., & Schillinger, D. (2006). Heath care discrimination,

processes of care, and diabetes patients’ health status. Patient Education and Counseling,

60(1), 41-48.

Plakas, S., Mastrogiannis, D., Mantzorou, M., Adamakidou, T., Fouka, G., Bouziou, A., ... &

Morisky, D. E. (2016). Validation of the 8-item Morisky medication adherence scale in

chronically ill ambulatory patients in rural Greece. Open Journal of Nursing, 6(03), 158.

Pintaudi, B., Lucisano, G., Gentile, S., Bulotta, A., Skovlund, S.E., Vespasiani, G., … &

Nicolucci, A. (2015). Correlates of diabetes-related distress in type 2 diabetes: Findings

from the benchmarking network for clinical and humanistic outcomes in diabetes

(BENCH-D) study. Journal of Psychosomatic Research, 79(5), 348-354. 86

Polonsky, W. H., Anderson, B. J., Lohrer, P. A., Welch, G., Jacobson, A. M., Aponte, J. E., &

Schwartz, C. E. (1995). Assessment of diabetes-related distress. Diabetes care, 18(6),

754-760.

Polonsky, W. H., Fisher, L., Earles, J., Dudl, R. J., Lees, J., Mullan, J., & Jackson, R. A. (2005).

Assessing psychosocial distress in diabetes. Diabetes Care, 28(3), 626-631.

Potter, L., Wallston, K., Trief, P., Ulbrecht, J., Juth, V., & Smyth, J. (2015). Attributing

discrimination to weight: Associations with well-being, self-care, and disease status in

patients with type 2 diabetes mellitus. Journal of Behavioral Medicine, 38(6), 863-875.

Prochaska, J. J., Sallis, J. F., & Long, B. (2001). A physical activity screening measure for use

with adolescents in primary care. Archives of Pediatrics & Adolescent Medicine, 155(5),

554-559.

Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the

general population. Applied Psychological Measurement, 1(3), 385-401.

Raskin, A., Schulterbrandt, J., Reatig, N., & McKeon, J. J. (1969). Replication of factors of

psychopathology in interview, ward behavior and self-report ratings of hospitalized

depressives. Journal of Nervous and Mental Disease, 148(1), 87.

Reddy, J., Wilhelm, K., & Campbell, L. (2013). Putting PAID to diabetes-related distress: The

potential utility of the problem areas in diabetes (PAID) scale in patients with diabetes.

Psychosomatics, 54(1), 44-51.

Reynolds, K., Viswanathan, H. N., O'Malley, C. D., Muntner, P., Harrison, T. N., Cheetham, T.

C., ... & Morisky, D. E. (2012). Psychometric properties of the osteoporosis-specific

Morisky medication adherence scale in postmenopausal women with osteoporosis newly

treated with bisphosphonates. Annals of Pharmacotherapy, 46(5), 659-670. 87

Richardson, L. K., Egede, L. E., Mueller, M., Echols, C. L., & Gebregziabher, M. (2008).

Longitudinal effects of depression on glycemic control in veterans with type 2 diabetes.

General Hospital Psychiatry, 30(6), 509-514.

Roberts, K. C. & Danoff-Burg, S. (2010). Mindfulness and health behaviors: Is paying attention

good for you? Journal of American College Health, 59(3), 165-173.

Rogvi, S., Tapager, I., Almdal, T. P., Schiotz, M. L., & Willaing, I. (2012). Patient factors and

glycaemic control- Associations and explanatory power. Diabetes Medicine, 29(10),

e382-e389.

Ryan, A. M., Gee, G. C., & Griffith, D. (2008). The effects of perceived discrimination on

diabetes. Journal of Health Care for the Poor and Underserved, 19(1), 149-163.

Sakraida, T. J. & Weber, M. T. (2016). The relationship between depressive symptoms and self-

management behaviors in patients with t2dm and stage 3 ckd. Perspectives in Psychiatric

Care, 52(4), 273-282.

Sanders-Philips, K., Kliewer, W., Tirmazi, T., Nebbitt, V., Carter, T., & Key, H. (2014).

Perceived racial discrimination, drug use, and psychological distress in African American

Youth: A pathway to child health disparities. Journal of Social Issues, 70(2), 279-297.

Schmidt, S. (2011). Mindfulness in east and west–is it the same? In Walach, H., Schmidt, S.,

Jonas, W. B. (Eds.), Neuroscience, consciousness and spirituality (pp. 23-38). Springer.

Schmitt, A., Reimer, A., Kulzer, R., Haak, T., Gahr, A., & Hermanns, N. (2015). Negative

association between depression and diabetes control only when accompanied by diabetes-

specific distress. Journal of Behavioral Medicine, 38(3), 556-564. 88

Schreiber, J. B., Nora, A., Stage, F. K., Barlow, E. A., & King, J. (2006). Reporting structural

equation modeling and confirmatory factor analysis results: A review. The Journal of

Educational Research, 99(6), 323-338.

Seligman, H. K., Jacobs, E. A., Lopez, A., Tschann, J., & Fernandez, A. (2012). Food insecurity

and glycemic control among low-income patients with type 2 diabetes. Diabetes Care,

35(2), 233-238.

Shariff-Marco, S., Klassen, A. C., & Bowie, J. V. (2010). Racial/ethnic differences in self-

reported racism and its association with cancer-related health behaviors. American

Journal of Public Health, 100(2), 364-375.

Sittner, K. J., Greenfield, B. L., & Walls, M. L. (2018). Microaggressions, diabetes distress, and

self-care behaviors in a sample of American Indian adults with type 2 diabetes. Journal

of Behavioral Medicine, 41(1), 122-129.

Slonim, J., Kienhuis, M., Di Benedetto, M., & Reece, J. (2015). The relationships among self-

care, dispositional mindfulness, and psychological distress in medical students. Medical

Education Online, 20(1), 27924.

Snoek, F. J., Pouwer, F., Welch, G. W., & Polonsky, W. H. (2000). Diabetes-related emotional

distress in Dutch and US diabetic patients: Cross-cultural validity of the problem areas in

diabetes scale. Diabetes care, 23(9), 1305-1309.

Sohi, K. K., & Singh, P. (2016). Experiencing microaggression: Invisibility, distress, and self-

stereotyping among Northeasterners in India. Frontiers in Psychology, 7, 1995.

Soper, D. (2020). Calculator: A-priori sample size for structural equation models. Free Statistics

Calculators. Retrieved 1 May 2020, from

https://www.danielsoper.com/statcalc/calculator.aspx?id=89 89

Sumlin, L. L., Garcia, T. J., Brown, S. A., Winter, M. A., Garcia, A. A., Brown, A., & Cuevas,

H. E. (2014). Depression and adherence to lifestyle changes in type 2 diabetes: A

systematic review. The Diabetes Educator, 40(6), 731-744.

Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics: International edition.

Pearson.

Tak, S. R., Hendrieckx, C., Nefs, G., Nyklicek, I., Speight, J., & Pouwer, F. (2015). The

association between types of eating behavior and dispositional mindfulness in adults with

diabetes. Results from Diabetes MILES. The Netherlands. Appetite, 87, 288-295.

Tan, S. M. K., Shafiee, Z., Wu, L. L., Rizal, A. M., & Rey, J. M. (2005). Factors associated with

control of type 1 diabetes in Malaysian adolescents and young adults. International

Journal of Psychiatry in Medicine 35(2), 123-136.

Ting, R. Z. W., Nan, H., Yu, M. W. M., Kong, A. P. S., Ma, R. C. W., Wong, R. Y. M., … &

Chan, J. C. N. (2011). Diabetes-related distress and physical and psychological health in

Chinese type 2 diabetic patients. Diabetes Care, 34(5), 1094-1096.

Tol, A., Baghbanian, A., Sharifirad, G., Shojaeizadeh, D., Eslami, A., Alhani, F., & Tehrani, M.,

M. (2012). Assessment of diabetic distress and disease related factors in patients with

type 2 diabetes in Isfahan: A way to tailor an effective intervention planning in Isfahan-

Iran. Journal of Diabetes & Metabolic Disorders, 11(20), 1-5.

Tsafou, K. E., Lacroix, J. P. W., van Ee, R., Vinkers, C. D. W., & De Ridder, D. T. D. (2017).

The relation of trait and state mindfulness with satisfaction and physical activity: A cross-

sectional study in 305 Dutch participants. Journal of Health Psychology, 22(10), 1221-

1232. 90

Tsujii, S., Hayashino, Y., Ishii, H., & the Diabetes Distress and Care Registry at Tenri Study

Group. (2012). Diabetes distress, but not depressive symptoms, is associated with

glycaemic control among Japanese patients with type 2 diabetes: Diabetes distress and

care registry at Tenri. Diabetic Medicine, 29(11), 1451-1455.

Ulmer, C. S., Stetson, B. A., & Salmon, P. G. (2010). Mindfulness and acceptance are associated

with exercise maintenance in YMCA exercisers. Behavior Research and Therapy 48(8),

805-809.

U.S. Department of Agriculture and U.S. Department of Health and Human Services (2010).

Dietary Guidelines for Americans (7th ed.). U.S. Government Printing Office.

U.S. Department of Health and Human Services (HHS), Office of Disease Prevention and Health

Promotion (2008). Physical activity guidelines for Americans.

Unnikrishnan, R., Anjana, R. M., & Mohan, V. (2016). Diabetes mellitus and its complications

in India. Nature Reviews Endocrinology, 12(6), 357.

Van Bastelaar, K. M. P., Pouwer, F., Geelhoed-Duijvestijn, P. H. L. M., Tack, C. J., Bazelmans,

E., Beekman, A. T., … & Snoek, F. J. (2010). Diabetes-specific emotional distress

mediates the association between depressive symptoms and glycaemic control in type 1

and type 2 diabetes. Diabetic Medicine, 27(7), 798-803.

Verma, S., & Acharya, S. S. (2018). Social identity and perceptions about health care service

provisioning by and for the in India. Social Identities, 24(3), 327-338.

Wagner, J., & Abbott, G. (2007). Depression and depression care in diabetes: Relationship to

perceived discrimination in African Americans. Diabetes Care, 30(2), 364-366. 91

Wagner, J. A., Tennen, H., Feinn, R., & Osborn, C. Y. (2015). Self-reported discrimination,

diabetes distress, and continuous blood glucose in women with type 2 diabetes. Journal

of Immigrant Minority Health, 17(2), 566-573.

Walters, K., Christakis, D. A., & Wright, D. R. (2018). Are Mechanical Turk worker samples

representative of health status and health behaviors in the US?. PloS one, 13(6).

Wardian, J. L., Tate, J., Folaron, I., Graybill, S., True, M., & Sauerwein, T. (2018). Who’s

distressed? A comparison of diabetes-related distress by type of diabetes and

medication. Patient Education and Counseling, 101(8), 1490-1495.

Weinger, K., & Jacobson, A. M. (2001). Psychosocial and quality of life correlates of glycemic

control during intensive treatment of type 1 diabetes. Patient Education and

Counseling, 42(2), 123-131.

Wessling, K., Huber, J., & Netzer, O. (2017). MTurk character misrepresentation: Assessment

and solutions. Journal of Consumer Research, 44(1), 211-230.

Widmar, N. J. O., Byrd, E. S., Dominick, S. R., Wolf, C. A., & Acharya, L. (2016). Social

desirability bias in reporting of holiday season healthfulness. Preventive Medicine

Reports, 4, 270-276.

Zagarins, S. E., Allen, N. A., Garb, J. L., & Welch, G. (2012). Improvement in glycemic control

following a diabetes education intervention is associated with change in diabetes distress

but not change in depressive symptoms. Journal of Behavioral Medicine, 35(3), 299-304.

Zhou, H., Zhu, J., Liu, L., Li, F., Fish, A. F., Chen, T., & Lou, Q. (2017). Diabetes-related

distress and its associated factors among patients with type 2 diabetes mellitus in China.

Psychiatry Research, 252, 45-50.

92

APPENDIX A. TABLES

Table A1 Participant Demographics

Total India USA Sample Characteristic Sample Sample

Age in years, M(SD) 32.37 (9.37) 30.15(6.93) 36.97 (11.92)

Gender n (%) Male 206 (61.1) 141 (64.4) 55(55.7) Female 126 (37.4) 105 (35.6) 45 (43.4) Declined to respond 5 (1.5) 0 (0) 5(4.7)

Race n (%) White 70 (20.8) 6 (2.7) 61 (57.5) Black/African American 26 (7.7) 1 (0.5) 24 (22.6) Native American 11 (3.3) 10 (4.6) 1 (0.9) Asian 210 (62.3) 197 (90.0) 9 (8.5) Native Hawaiian 1 (0.3) 0 (0) 1 (0.9) Two or more races 10 (3.0) 5 (2.3) 5 (4.7) Other 4 (1.2) 0 (0) 3 (2.8) Declined to respond 5 (1.5) 0 (0) 2 (1.9)

Ethnicity n (%) Hispanic 88 (26.1) 64 (29.2) 23 (22.6) Non-Hispanic 245 (72.7) 155 (70.8) 82 (77.4)

Country n (%) India 219 (65.6) USA 106 (31.7) Canada 3 (0.9) Russia 2 (0.6) Kenya 1 (0.3) Venezuela 1 (0.3) Japan 1 (0.3) China 1 (0.3)

Highest Education Level n (%) High school graduate or equivalent 9 (2.7) 1 (0.5) 8 (7.5) Some College 42 (12.5) 22 (10.0) 20 (18.9) 2 year degree 33 (9.8) 18 (8.2) 14 (13.2) 93

Total India USA Sample Characteristic Sample Sample Highest Education Level n (%) 4 year degree 186 (55.2) 138 (63.0) 42 (39.6) Master’s degree 54 (16.0) 35 (16.0) 17 (16.0) Professional degree 8 (2.4) 5 (2.3) 3 (2.8) Doctoral Degree 1 (0.3) 0 (0) 1 (0.9) Declined to respond 4 (1.2) 0 (0) 1 (0.9)

Yearly household income n (%) $10,000 - $19,000 4 (3.8) $20,000 - $29,999 13 (12.3) $30,000 – 39,999 14 (13.2) $40, 000- $49,999 13 (12.3) $50,000 - $59,999 18 (17.0) $60,000 - $69,999 8 (7.5) $70,000 - $79,999 15 (14.2) $80,000 - $89,999 5 (4.7) $90,000 - $99,999 8 (7.5) $100,000 - $149,999 6 (5.7) More than $150,000 1 (0.9) Declined to respond 1 (0.9)

Note. Data on income only includes participants from the United States. Total sample n = 334; India sample n = 219; USA sample n = 106. 94

Table A2 Comorbid Chronic Illnesses

Illness n (%) Total India USA Hypertension 58 (17.2) 30 (13.7) 27 (25.5) High Cholesterol 52 (15.4) 24 (11.0) 26 (24.5) Arthritis 38 (11.3) 20 (9.1) 16 (15.1) Heart Disease 31 (9.2) 20 (9.1) 9 (8.5) Chronic Kidney Disease 51 (15.1) 47 (21.5) 2 (1.9) Heart Failure 20 (5.9) 15 (6.8) 5 (4.7) Depression 53 (15.7) 26 (11.9) 26 (24.5) Chronic Obstructive Pulmonary Disease 20 (5.9) 12 (5.5) 7 (6.6) Note. Total sample n = 334; India sample n = 219; USA sample n = 106. 95

Table A3 Diabetes Type and Medical Maintenance

Characteristic Total India USA Diabetes type n (%) Type 1 Diabetes 170 (50.4) 123 (56.2) 42 (39.6) Type 2 Diabetes 131 (38.9) 66 (30.1) 61 (57.5) I’m not sure 31 (9.2) 30 (13.7) 1 (0.9) Declined to respond 5 (1.5)

Diabetes duration and last physician appointment Time (SD) Months since Diagnosis 42.52 (66.30) 20.52 (66.30) 80.59 (95.35) Weeks since last physician appointment 13.78 (25.11) 13.58 (23.57) 14.29 (28.46)

Type of medication n (%) Oral only 161 (47.8) 105 (47.9) 49 (46.2) Insulin only 108 (32.0) 75 (34.2) 29 (27.4) Both oral and insulin 43 (7.4) 32 (14.6) 11 (10.4) No medications 25 (7.4) 7 (3.2) 17 (16.0)

Frequency of Self-monitoring Blood Glucose n (%) Never 26 (7.7) 11 (5.0) 13 (12.3) 1-2 times a week 128 (38.0) 92 (42.0) 32 (30.2) 3-4 times a week 116 (34.4) 89 (40.6) 24 (22.6) Daily 50 (14.8) 18 (8.2) 29 (27.4) More than once per day 17 (5.0) 9 (4.1) 8 (7.5) Note. Total sample n = 334; India sample n = 219; USA sample n = 106. 96

Table A4 Frequency of Self-reported Health Behaviors

Some of All of Never the time Often the time Health Behavior (%) (%) (%) (%) Ate 5-7 servings of fruits and vegetables* 9.8 26.4 46.9 16.9 Ate high-fat foods 7.7 42.7 32.9 16.6 Ate lean proteins* 7.1 30.6 45.7 16.6 Ate whole grains 4.2 30.0 46.3 19.6 Ate fast food 16 36.3 29.5 18.2 Consumed low-fat or non-dairy substitute* 6.8 38.3 42.4 12.5 Ate refined grains 7.1 35.6 40.9 16.3 Drank 7-8 glasses of water 7.7 24.6 38.9 28.8 Consumed sugary drinks 16.6 28.2 36.5 18.7 Drank alcohol 20.8 32.6 29.1 17.5 Engaged in light physical activity 4.7 31.8 44.2 19.3 Engaged in moderate physical activity* 8.3 33.2 35.9 22.6 Engaged in vigorous physical activity* 13.9 28.5 37.7 19.9 Note. Total sample n = 334. 97

Table A5 Descriptives for Study Measures

Scale M SD Skewness Kurtosis alpha REMS Total 2.42 1.24 -0.38 -0.85 0.985 Assumptions of Inferiority 2.35 1.35 -0.42 -0.89 0.947 Second-class citizenship 2.32 1.38 -0.36 -0.93 0.943 Microinvalidations 2.36 1.30 -0.37 -0.88 0.939 Exoticization 2.42 1.37 -0.38 -0.92 0.948 Environment 2.70 1.24 -0.42 -0.47 0.908 Workplace/School 2.31 1.37 -0.39 -0.94 0.917 FFMQ 3.07 0.29 1.44 4.87 0.717 Observe 3.42 0.7 -0.58 0.30 0.793 Describe 3.15 0.50 0.73 2.49 0.480 Act with Awareness 2.80 0.78 0.58 0.18 0.833 Nonjudge 2.75 0.76 0.55 0.37 0.852 Nonreact 3.27 0.7 -0.27 -0.05 0.786 PAID 51.98 21.03 -0.62 0.11 0.949 CES-D 29.57 9.91 -0.75 0.82 0.868 Health Behaviors 1.73 0.51 -0.30 0.13 0.732 Exercise Subscale 1.73 0.51 -0.25 -0.18 0.503 Diet Subscale 1.60 0.34 -0.25 0.01 0.586 MMAS (n = 301) 3.98 1.81 -0.49 -0.43 0.547 Note. REMS = Racial and Ethnic Microaggressions Scale; FFMQ = Five Facet Mindfulness Questionnaire; PAID = Problem Areas in Diabetes; CES-D= Center for Epidemiology Scale- Depression; MMAS = Morisky Medication Adherence Scale. Total sample n = 334. 98

Table A6 Descriptives for Study Measures by Country

Variable Country M SD Mindfulness India 3.09 0.14 USA 3.12 0.47 Microaggressions India 2.57 0.97 USA 1.09 1.24 Depressive Symptoms India 30.37 6.55 USA 24.66 13.23 Diabetes-related distress India 59.04 13.46 USA 38.87 25.3 Health Behaviors India 1.91 0.49 USA 1.52 0.56 Medication Adherence India 4.11 1.53 USA 2.58 2.14 Glycemic Control India 6.12 1.35 USA 6.44 1.58 Time since diagnosis India 25.31 24 USA 62.8 62.56 Time since physician’s appointment India 8.97 13.07 USA 10.56 17.8 99

Table A7 Analyses of Between-Group Differences in Study Variables and Diabetes Variables by Country and Diabetes Type

Variable V F df p MANOVA Results Country 0.78 4.318 9, 64 <.001 Diabetes type 0.196 1.733 9, 64 0.1 Country*Diabetes type 0.078 0.598 9, 64 0.794 Follow-up Univariate Analysis Mindfulness 0.052 1,72 0.822 Microaggressions 27.437 1,72 <.001 Depressive symptoms 5.42 1,72 0.023 Diabetes-related distress 15.611 1,72 <.001 Health Behaviors 7.55 1,72 0.008 Medication Adherence 10.033 1,72 0.002 Glycemic control 0.383 1,72 0.538 Time since diagnosis 10.973 1,72 0.001 Time since last physician’s appointment 0.017 1,72 0.895 χ2 Tests and Odds Ratios India USA χ2 df p Diabetes Type: Type 1 123 42 16 1 <.001 Type 2 66 59

Medication Type: Insulin 104 48 18.5 3 <.001 Oral 75 28 Insulin and Oral 32 11 No medication 7 17 Note. Sample from USA is reference group for Odds Ratios. India sample n = 219; USA sample n = 106. 100

Table A8

Correlation Matrix for Structural Regression Models 1&2

Item REMS FFMQ CESD PAID HB1 HB3 HB4 HB6 REMS 1 FFMQ -0.2 1 CESD 0.58 -0.59 1 PAID 0.74 -0.37 0.76 1 HB1 0.21 0.22 -0.13 0.04 1 HB3 0.1 0.11 -0.03 0.1 0.26 1 HB4 0.35 0.02 0.15 0.27 0.2 0.08 1 HB6 0.31 0.13 0.1 0.22 0.33 0.15 0.22 1 HB8 0.08 0.19 -0.08 0.1 0.46 0.21 0.24 0.19 HB11 0 0.26 -0.12 0.01 0.38 0.3 0.19 0.22 HB12 0.23 0.17 -0.04 0.18 0.41 0.34 0.28 0.4 HB13 0.51 -0.04 0.24 0.42 0.48 0.22 0.3 0.32 MMAS1 0.4 -0.25 0.36 0.47 0 0.02 0.04 0.17 MMAS2 0.21 -0.16 0.22 0.3 0.07 -0.08 -0.07 -0.02 MMAS3 0.41 -0.13 0.35 0.36 0.01 -0.05 0.06 0.06 MMAS4 0.32 -0.2 0.33 0.36 0.23 0.05 0.15 0.14 MMAS5 -0.23 -0.01 -0.13 -0.14 -0.13 -0.09 -0.21 -0.23 MMAS6 0.4 -0.19 0.38 0.39 0.08 0.04 0.02 0.14 MMAS7 0.25 -0.18 0.24 0.27 0.02 0.01 0.1 0.04 MMAS8 0.15 -0.09 0.17 0.19 -0.05 -0.07 0.17 -0.05 -0.14 0.12 -0.05 -0.09 -0.07 -0.01 -0.06 0.01

Mean 2.22 3.1 28.58 51.31 1.7 1.71 1.81 1.59 Standard 1.37 0.3 10.57 23.34 0.84 0.76 0.78 0.73 Deviation Skewness -0.19 0.83 -0.63 -0.62 -0.38 -0.2 0.1 0.28 Kurtosis -1.21 2.753 0.43 -0.15 -0.37 -0.24 -0.87 -0.4 Alpha 0.989 0.776 0.892 0.964 Note. * = p < .05; ** = p < .01; *** = p < .001; N = 301; REMS = Racial and Ethnic Microaggressions Scale; FFMQ = Five Facet Mindfulness Questionnaire; PAID = Problem Areas in Diabetes; CES-D= Center for Epidemiology Scale- Depression; MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. Note. Total sample n = 150. 101

Item HB11 HB12 HB13 MMAS1 MMAS2 MMAS3 MMAS4 REMS FFMQ CESD PAID HB1 HB3 HB4 HB6 HB8 HB11 1 HB12 0.34 1 HB13 0.29 0.4 1 MMAS1 -0.07 0.07 0.25 1 MMAS2 -0.02 0 0.15 0.29 1 MMAS3 -0.12 0.14 0.24 0.32 -0.05 1 MMAS4 0.11 0.11 0.24 0.24 0.19 0.22 1 MMAS5 -0.09 -0.09 -0.16 0 0.08 -0.16 -0.19 MMAS6 -0.01 0.1 0.34 0.27 0.33 0.28 0.33 MMAS7 0.09 0.1 0.07 0.2 0.1 0.42 0.21 MMAS8 -0.04 -0.08 0.04 0.28 0.23 0.11 0.31 0.07 -0.13 -0.25 -0.22 -0.05 -0.24 -0.05

Mean 1.81 1.75 1.49 0.63 0.45 0.44 0.44 Standard 0.73 0.92 0.95 0.49 0.5 0.5 0.5 Deviation Skewness -0.1 -0.13 -0.15 -0.53 0.19 0.24 0.24 Kurtosis -0.34 -0.88 -0.9 -1.74 -1.99 -1.97 -1.97 Alpha 102

Item MMAS5 MMAS6 MMAS7 MMAS8 A1c REMS FFMQ CESD PAID HB1 HB3 HB4 HB6 HB8 HB11 HB12 HB13 MMAS1 MMAS2 MMAS3 MMAS4 MMAS5 1 MMAS6 -0.27 1 MMAS7 -0.04 0.04 1 MMAS8 -0.09 0.17 0.07 1 0.07 -0.15 0.12 -0.05 1

Mean 0.79 0.49 0.52 0.47 6.34 Standard 0.41 0.5 0.5 0.27 1.08 Deviation Skewness -1.41 0.03 -0.08 -0.41 0.86 Kurtosis -0.02 -2.03 -2.02 -1.16 2.85 Alpha 103

Table A9 Correlation Matrix for Structural Regression Model 3

MMAS MMAS MMAS REMS FFMQ CESD PAID HB1 HB3 HB6 HB12 HB13 2 4 6 REMS 1 FFMQ -0.170 1 CESD 0.551 -0.503 1 PAID 0.702 -0.293 0.734 1 HB1 0.252 0.168 -0.007 0.116 1 HB3 0.203 0.081 0.074 0.143 0.175 1 HB6 0.304 0.161 0.114 0.250 0.352 0.231 1 HB12 0.241 0.129 -0.019 0.141 0.313 0.257 0.318 1 HB13 0.462 -0.028 0.205 0.316 0.349 0.310 0.268 0.268 1 MMAS2 0.205 -0.052 0.190 0.269 0.146 0.119 0.132 0.025 0.141 1 MMAS4 0.299 -0.099 0.285 0.291 0.263 0.099 0.169 0.059 0.141 0.296 1 MMAS6 0.300 -0.091 0.276 0.286 0.169 0.117 0.124 0.084 0.227 0.316 0.249 1

Mean 2.45 3.06 30.01 53.07 1.73 1.73 1.63 1.77 1.75 0.49 0.51 0.50 Standard Deviatio n 1.22 0.25 9.46 20.53 0.83 0.76 0.77 0.89 0.50 0.50 0.50 0.50 Skewnes s -0.45 0.97 -0.84 -0.69 -0.34 -0.17 0.06 -0.09 -0.10 0.03 -0.05 -0.01 Kurtosis -0.77 3.70 1.18 0.34 -0.37 -0.45 -0.44 -0.48 -0.84 -2.01 -0.20 -2.01 Alpha 0.986 .659 0.896 0.949 Note. N = 150; REMS = Racial and Ethnic Microaggressions Scale; FFMQ = Five Facet Mindfulness Questionnaire; PAID = Problem Areas in Diabetes; CES-D= Center for Epidemiology Scale- Depression; MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. Total sample = 301. 104

Table A10 Selected Model Fit Statistics for Two-Step Testing of a Structural Regression Model

2 RMSEA (90% pclose-fit Model χ M dfM p CFI SRMR CI) H0 Measurement Model Single-factor CFA 401.46 104 <.001 .098 (.088-.108) <.001 0.548 0.096 2-factor CFA 28.91 19 0.067 .059 (.000-.100) 0.332 0.943 0.040

Structural regression model Structural Model 1 (n = 150) 124.96 60 <.001 .085 (.064-.106) 0.004 0.888 0.097 Structural Model 2 (n = 150) 18.530 16 0.294 .033 (.000-.086) 0.645 0.993 0.042 Structural Model 3 (n = 301) 18.358 11 0.074 .047 (.000-.084) 0.501 0.989 0.038 No te. RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; SRMR = Standardized Root Mean Square Residual. 105

Table A11 Factor Loadings for Two-Factor Confirmatory Factor Analysis

Standard Parameter Unstandardized b p Standardized β Error Factor Loadings Health Behaviors Factor HB1 1 0.655 HB3 0.736 0.13 0 0.5 HB6 0.736 0.117 0 0.527 HB12 0.795 0.132 0 0.49 HB13 0.948 0.147 0 0.558

Medication Adherence Factor MMAS6 1 0.533 MMAS4 0.981 0.207 0 0.523 MMAS2 1.039 0.218 0 0.554 Note. MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. 106

Table A12 Measurement Error Variances of Confirmatory Factor Analysis

Parameter Unstandardized b SE p Standardized β

Measurement Error Variances HB1 0.407 0.064 0.000 0.452 HB3 0.493 0.061 0.000 0.166 HB6 0.382 0.051 0.000 0.252 HB12 0.496 0.076 0.000 0.507 HB13 0.477 0.079 0.000 0.341 MMAS2 0.202 0.028 0.000 0.284 MMAS4 0.188 0.029 0.000 0.272 MMAS2 0.122 0.041 0.003 0.308 Factor Variances and Covariance Health Behaviors 0.310 0.081 0.000 1.000 Medication Adherence 0.047 0.022 0.042 1.000 Health Behaviors x Medication Adherence 0.046 0.020 0.019 0.480

Note. MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. 107

Table A13 Maximum Likelihood Estimates for a Recursive Path Model of Glycemic Control (Model 1)

Unstandardized Standardized Parameter SE p 2 b β R SMC Direct Effects Microaggressions -> Depressive Symptoms 3.742 0.423 0.000 0.484 Microaggressions -> Diabetes-related Distress 7.527 0.953 0.000 0.441 Mindfulness -> Depressive Symptoms -17.520 1.941 0.000 -0.493 Mindfulness -> Diabetes-related Distress 2.073 4.407 0.638 0.026 Depressive Symptoms -> Diabetes-related 1.144 0.15 0.000 0.518 Distress Diabetes-related distress -> Medication 0.006 0.001 0.000 0.626 Adherence Diabetes-related distress -> Health Behaviors 0.008 0.002 0.000 0.359 Medication Adherence -> Glycemic Control -0.428 0.486 0.378 -0.093 Health Behaviors -> Glycemic Control -0.334 0.205 0.103 -0.163 Disturbance Variances Depressive Symptoms 47.39 5.49 0.000 1.000 0.573 Diabetes-related Distress 157.99 18.304 0.000 1.000 0.708 Medication Adherence 0.065 0.024 0.024 1.000 0.392 Health Behaviors 0.238 0.067 0.000 1.000 0.129 Glycemic Control 1.101 0.129 0.000 1.000 0.042 108

Table A14 Maximum Likelihood Estimates for Respecified Recursive Path Model of Glycemic Control (Model 2)

Unstandardized Standardized Parameter SE p 2 b β R SMC Direct Effects Microaggressions -> Depressive Symptoms 3.742 0.423 0.000 0.484 Microaggressions -> Diabetes-related Distress 7.527 0.953 0.000 0.441 Mindfulness -> Depressive Symptoms -17.520 1.941 0.000 -0.493 Mindfulness -> Diabetes-related Distress 2.073 4.407 0.638 0.026 Depressive Symptoms -> Diabetes-related 1.144 0.150 0.000 0.518 Distress Diabetes-related distress -> Medication 0.002 0.006 0.219 0.628 Adherence Medication Adherence -> Glycemic Control -0.646 0.525 0.770 -0.198 Disturbance Variances Depressive Symptoms 47.390 5.490 0.000 1.000 0.573 Diabetes-related Distress 157.990 18.304 0.000 1.000 0.708 Medication Adherence 0.065 0.024 0.006 1.000 0.394 Glycemic Control 1.115 0.133 0.000 1.000 0.031 109

Table A15 Indirect Effects of Selected Paths for Respecified Model

Indirect effect β b SE 90%CIbootstrap p Mediator: Depressive Symptoms Microaggressions -> Diabetes-related distress 0.251 4.28 .863 3.093 – 6.034 0.004 Mindfulness -> Diabetes-related distress -0.255 -20.038 4.222 -29.273 - -14.907 0.003

Mediator: Medication Adherence Diabetes-related Distress -> Glycemic control -0.12 -0.006 0.006 -0.016 – 0.002 0.230

Mediator: Health Behaviors Diabetes-related Distress -> Glycemic control -0.03 -0.001 0.001 -0.004 – 0.001 0.293 110

Table A16 Maximum Likelihood estimates for a recursive path model of Medication Adherence (model 3)

Unstandardized Standardized 2 Parameter SE p R SMC b β Direct Effects Microaggressions -> Depressive Symptoms 3.735 0.462 0.000 0.475 Microaggressions -> Diabetes-related Distress 7.957 0.908 0.000 0.448 Mindfulness -> Depressive Symptoms -9.029 2.473 0.000 -0.214 Mindfulness -> Diabetes-related Distress -3.086 4.379 0.481 -0.032 Depressive Symptoms -> Diabetes-related 1.135 0.100 0.000 0.523 Distress Diabetes-related distress -> Medication 0.005 0.002 0.006 0.283 Adherence Disturbance Variances Depressive Symptoms 39.452 3.869 0.000 1.000 0.288 Diabetes-related Distress 116.231 11.397 0.000 1.000 0.589 Medication Adherence 0.081 0.032 0.011 1.000 0.080 111

Table A17 Nested Model Comparisons

Path Comparison CMIN df p Total Model 19.808 9 0.011 Microaggressions -> Depressive Symptoms 0.063 1 0.802 Microaggressions -> Diabetes-related Distress 2.064 1 0.151 Mindfulness -> Depressive symptoms 8.752 1 0.015 Mindfulness -> Diabetes-related Distress 4.613 1 0.032 Depressive symptoms -> Diabetes-related Distress 4.378 1 0.036 Diabetes-related Distress -> Medication Adherence 1.933 1 0.164 112

Table A18 Maximum Likelihood estimates for a recursive path model of Medication Adherence for participants from India (model 3)

Unstandardized Standardized 2 Parameter SE p R SMC b β Direct Effects India Microaggressions -> Depressive 3.735 0.462 0.000 0.475 Symptoms Microaggressions -> Diabetes- 7.957 0.909 0.000 0.448 related Distress Mindfulness -> Depressive -9.029 2.476 0.000 -0.214 Symptoms Mindfulness -> Diabetes-related -3.086 4.383 0.481 -0.032 Distress Depressive Symptoms -> 0.978 0.119 0.000 0.433 Diabetes-related Distress Diabetes-related distress -> 0.005 0.002 0.006 0.283 Medication Adherence

USA Microaggressions -> Depressive 3.929 0.615 0.000 0.442 Symptoms Microaggressions -> Diabetes- 5.651 1.316 0.000 0.316 related Distress Mindfulness -> Depressive -19.091 2.288 0.000 -0.577 Symptoms Mindfulness -> Diabetes-related 12.022 5.446 0.027 0.181 Distress Depressive Symptoms -> 1.458 0.194 0.000 0.725 Diabetes-related Distress Diabetes-related distress -> 0.009 0.002 0.000 0.721 Medication Adherence Disturbance Variances India Depressive Symptoms 39.452 3.873 0.000 1.000 0.288 Diabetes-related Distress 116.231 11.409 0.000 1.000 0.589 Medication Adherence 0.081 0.032 0.011 1.000 0.080 USA Depressive Symptoms 60.068 9.413 0.000 1.000 0.625 Diabetes-related Distress 183.445 28.748 0.000 1.000 0.717 Medication Adherence 0.044 0.021 0.037 1.000 0.52 113

APPENDIX B. FIGURES

Figure B1 Hypothesized Path Model 114

Figure B2 Hypothesized Single-Factor Measurement Model

Note. Latent variables (Treatment Adherence and error terms) free to vary. MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. 115

Figure B3 Hypothesized Structural Regression Model

Note. Exogenous and error and disturbance terms free to vary. MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. 116

Figure B4 Two-Factor Measurement Model

Note. MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. 117

Figure B5 Structural Regression Model 1

Note. MMAS = Morisky Medication Adherence Scale; HB = Health Behaviors. 118

Figure B6 Respecified Structural Regression Model 2

Note. MMAS = Morisky Medication Adherence Scale. 119

Figure B7 Structural Regression Model 3

Note. MMAS = Morisky Medication Adherence Scale. 120

APPENDIX C. RECRUITMENT SCRIPT

This 30 minute survey is a social science survey about relationships between physical and mental health among people with chronic illness. If you are eligible, you will earn $1 for completing this survey and will be eligible to participate in two future surveys to earn $3 and $5. 121

APPENDIX D. INFORMED CONSENT

Informed Consent Form

You are invited to participate in a research study exploring predictors of health. Leah Bogusch, M.A. is responsible for this project. She is a graduate student at Bowling Green State University. Her supervisor is Dr. William O’Brien. This study is designed to increase knowledge about physical and mental health among people with chronic illness. You must be at least 18 years old to participate. This is a voluntary 30-minute survey. Expected risks are no greater than those experienced in daily life. The primary benefit of this study is the knowledge gained about health among people with chronic illness. This knowledge will be used to develop interventions to improve health.

Before beginning the survey, you will be asked questions to determine your eligibility in the study. If you are eligible, you will be asked to respond to questions about your thoughts and feelings. You will be asked questions about your mood in general and feelings related to living with chronic illness. You will be asked questions about experiences related to your race/ethnicity. You will be asked questions about healthy habits. These questions include questions about eating and exercise habits. If you are eligible for the survey and you correctly answer questions that check your attention, you will be awarded $1 to your Mechanical Turk account. You will have the option to participate in two more surveys three and six months from now. You will have the chance to earn $3 and $5 if you complete these future surveys.

Your responses are anonymous. You will be assigned an encrypted MTurk ID number. Your responses will not be connected to any identifying information. Your responses will be saved on a password- protected USB drive that will be kept in a locked office at the Bowling Green State University campus that only the primary researcher and her advisor will have access to.

Your participation is completely voluntary. You may withdraw at any time. You may skip questions at any time without penalty. Your choice to participate will not affect any relationship you may have with Bowling Green State University.

We recommend that you do not leave the online survey unattended if completing it on a public computer. This is because the internet is not 100% secure. We also suggest that you complete the survey in a private area and clear the internet browser and page history when finished with the survey. It's best for you to use your own computer rather than one controlled by someone else, such as your employer. Employers may monitor the information sent through computers that they own.

If you have any questions or concerns, you may contact Leah Bogusch. You may reach her by phone at (419)-372-4520 or by email ([email protected]). You may contact Dr. William O’Brien by phone at (419)-372-2974 or email ([email protected]). You may contact the Chair, Institutional Review Board at 122

Bowling Green State University. You may reach the Chair at (419)-372-7716 or by email ([email protected]).

Thank you for taking the time to read this information. Your completion of this survey indicates your consent to participate in this research. You may refuse to participate and discontinue without penalty. If you wish to give consent, please select the response option below and click on the “Next” button. You will be directed to the survey. If you prefer not to participate, please close the browser window at this time by clicking the X at the top right hand corner.

____I have read the informed consent. I am 18 years or older. I have been diagnosed by a medical professional with a chronic illness. I agree to participate in the study. (This box is required in order to continue. By clicking the “Next” button below, you are indicating your informed consent). 123

APPENDIX E. STUDY SCREENER

Screener Please select which of the following illnesses for which you have received a diagnosis from a medical professional:

 hypertension (high blood pressure)  high cholesterol  arthritis  Heart disease  Type 2 diabetes  Chronic kidney disease  Heart failure  Depression  Alzheimer’s dementia  Chronic Obstructive Pulmonary Disease  Type 1 diabetes 124

APPENDIX F. RACIAL AND ETHNIC MICROAGGRESSIONS SCALE

Please indicate the number of times that each of the following occurred in the past six months (0 = I did not experience this event in the past six months, 1 = I experienced this event 1 time in the past six months, 2 = I experienced this event 2 times in the past six months, 3 = I experienced this event 3 times in the past six months, 4 = I experience this event 4 times in the past six months, or 5 = I experienced this event 5 times in the past six months). 1. I was ignored at school or at work because of my race. 2. Someone’s body language showed they were scared of me, because of my race. 3. Someone assumed I spoke a language other than English. 4. I was told that I should not complain about race. 5. Someone assumed that I grew up in a particular neighborhood because of my race. 6. Someone avoided walking near me on the street because of my race. 7. Someone told me that they were color-blind. 8. Someone avoided sitting next to me in a public space (e.g. restaurants, movie theaters, subways, buses) because of my race. 9. Someone assumed that I would not be intelligent because of my race 10. I was told that I complain about race too much. 11. I received substandard service in stores compared to customers of other racial groups. 12. I observed people of my race in prominent positions at my workplace or school. 13. Someone wanted to date me only because of my race. 14. I was told that people of all racial groups experience the same obstacles. 15. My opinion was overlooked in a group discussion because of my race. 16. Someone assumed that my work would be inferior to people of other racial groups. 17. Someone acted surprised at my scholastic or professional success because of my race 18. I observed that people of my race were the CEOs of major corporations. 19. I observed people of my race portrayed positively on television. 20. Someone did not believe me when I told them I was born in the U.S. 21. Someone assumed that I would not be educated because of my race. 22. Someone told me that I was “articulate” after they assumed I wouldn’t be. 23. Someone told me that all people in my racial group are all the same. 24. I observed people of my race portrayed positively in magazines. 25. An employer or co-worker was unfriendly or unwelcoming toward me because of my race. 26. I was told that people of color do not experience racism anymore. 27. Someone told me that they “don’t see color.” 28. I read popular books or magazines in which a majority of contributions featured people from my racial group. 29. Someone asked me to teach them words in my “native language.” 30. Someone told me that they do not see race. 31. Someone clenched their purse or wallet upon seeing me because of my race. 32. Someone assumed that I would have a lower education because of my race. 125

33. Someone of a different racial group has stated that there is no difference between the two of us. 34. Someone assumed that I would physically hurt them because of my race. 35. Someone assumed that I ate foods associated with my race/culture every day. 36. Someone assumed that I held a lower paying job because of my race. 37. I observed people of my race portrayed positively in movies. 38. Someone assumed that I was poor because of my race 39. Someone told me that people should not think about race anymore. 40. Someone avoided eye contact with me because of my race. 41. I observed that someone of my race is a government official in my state. 42. Someone told me that all people in my racial group look alike. 43. Someone objectified one of my physical features because of my race. 44. An employer or co-worker treated me differently than White co-workers. 45. Someone assumed that I speak similar languages to other people in my race. 126

APPENDIX G. CENTERS FOR EPIDEMIOLOGICAL STUDIES- DEPRESSION

Below is a list of the ways you might have felt or behaved. Please tell me how often you have felt this way during the past week. Circle one number on each line (0: Rarely or none of the time [less than 1 day]; 1: Some or a little of the time [1-2 days]; 2: Occasionally or a moderate amount of time [3-4 days]; 3: All of the time [5-7 days]) 1. I was bothered by things that usually don’t bother me. 2. I did not feel like eating: my appetite was poor. 3. I felt that I could not shake off the blues even with help from my family or friends. 4. I felt that I was just as good as other people. 5. I had trouble keeping my mind on what I was doing. 6. I felt depressed. 7. I felt that everything I did was an effort. 8. I felt hopeful about the future. 9. I thought my life had been a failure. 10. I felt fearful. 11. My sleep was restless. 12. I was happy. 13. I talked less than usual. 14. I felt lonely. 15. People were unfriendly. 16. I enjoyed life. 17. I had crying spells. 18. I felt sad. 19. I felt that people disliked me. 20. I could not get “going.” 127

APPENDIX H. PROBLEM AREAS IN DIABETES

Which of the following diabetes issues are currently a problem for you? Circle the number that gives the best answer for you (0: Not a problem; 1: Minor problem; 2: Moderate problem; 3: Somewhat serious problem; 4: Serious problem). 1. Not having clear and concrete goals for your diabetes care? 2. Feeling discouraged with your diabetes treatment plan? 3. Feeling scared when you think about living with diabetes? 4. Uncomfortable social situations related to your diabetes care (e.g. people telling you what to eat)? 5. Feelings of deprivation regarding food and meals? 6. Feeling depressed when you think about living with diabetes? 7. Not knowing if your mood or feelings are related to your diabetes? 8. Feeling overwhelmed by your diabetes? 9. Worrying about low blood sugar reactions? 10. Feeling angry when you think about living with diabetes? 11. Feeling constantly concerned about food and eating? 12. Worrying about the future and the possibility of serious complications? 13. Please select 1: Minor Problem 14. Feelings of guilt or anxiety when you get off track with your diabetes management? 15. Not “accepting” your diabetes? 16. Feeling unsatisfied with your diabetes physician? 17. Feeling that diabetes is taking up too much of your mental and physical energy every day? 18. Feeling alone with your diabetes? 19. Feeling that your friends and family are not supportive of your diabetes management efforts? 20. Coping with complications of diabetes? 21. Feeling “burned out” by the constant effort needed to manage diabetes? 128

APPENDIX I. FIVE FACET MINDFULNESS QUESTIONNAIRE

Please rate each of the following statements using the scale provided. Write the number in the blank that best describes your own opinion of what is generally true for you.

1 2 3 4 5

never or very rarely sometimes often very often or rarely true always true true true true

_____ 1. When I’m walking, I deliberately notice the sensations of my body moving. _____ 2. I’m good at finding words to describe my feelings. _____ 3. I criticize myself for having irrational or inappropriate emotions. _____ 4. I perceive my feelings and emotions without having to react to them. _____ 5. When I do things, my mind wanders off and I’m easily distracted. _____ 6. When I take a shower or bath, I stay alert to the sensations of water on my body. _____ 7. I can easily put my beliefs, opinions, and expectations into words. _____ 8. I don’t pay attention to what I’m doing because I’m daydreaming, worrying, or otherwise distracted. _____ 9. I watch my feelings without getting lost in them. _____ 10. I tell myself I shouldn’t be feeling the way I’m feeling. _____ 11. I notice how foods and drinks affect my thoughts, bodily sensations, and emotions. _____ 12. It’s hard for me to find the words to describe what I’m thinking. _____ 13. I am easily distracted. _____ 14. I believe some of my thoughts are abnormal or bad and I shouldn’t think that way. _____ 15. I pay attention to sensations, such as the wind in my hair or sun on my face. _____ 16. I have trouble thinking of the right words to express how I feel about things _____ 17. Please select 2 (Rarely True) _____ 17. I make judgments about whether my thoughts are good or bad. _____ 18. I find it difficult to stay focused on what’s happening in the present. _____ 19. When I have distressing thoughts or images, I “step back” and am aware of the thought or image without getting taken over by it. _____ 20. I pay attention to sounds, such as clocks ticking, birds chirping, or cars passing. _____ 21. In difficult situations, I can pause without immediately reacting. _____ 22. When I have a sensation in my body, it’s difficult for me to describe it because I can’t find the right words. _____ 23. It seems I am “running on automatic” without much awareness of what I’m doing. _____24. When I have distressing thoughts or images, I feel calm soon after. _____ 25. I tell myself that I shouldn’t be thinking the way I’m thinking. _____ 26. I notice the smells and aromas of things. _____ 27. Even when I’m feeling terribly upset, I can find a way to put it into words. _____ 28. I rush through activities without being really attentive to them. 129

_____ 29. When I have distressing thoughts or images I am able just to notice them without reacting. _____ 30. I think some of my emotions are bad or inappropriate and I shouldn’t feel them. _____ 31. I notice visual elements in art or nature, such as colors, shapes, textures, or patterns of light and shadow. _____ 32. My natural tendency is to put my experiences into words. _____ 33. When I have distressing thoughts or images, I just notice them and let them go. _____ 34. I do jobs or tasks automatically without being aware of what I’m doing. _____ 35. When I have distressing thoughts or images, I judge myself as good or bad, depending what the thought/image is about. _____ 36. I pay attention to how my emotions affect my thoughts and behavior. _____ 37. I can usually describe how I feel at the moment in considerable detail. _____ 38. I find myself doing things without paying attention. _____ 39. I disapprove of myself when I have irrational ideas. 130

APPENDIX J. HEALTH BEHAVIORS

Please indicate how often you participated in the following health behaviors over the past week (0: Never; 1: Some of the time [1-2 times a week]; 2: Often [3-4 times a week]; 3: All of the time [Every day]) 1. Ate 5-7 servings of fruits and vegetables per day. 2. Ate high-fat foods (e.g. fried foods, cakes, cookies, pizza, butter) 3. Ate lean proteins, meats, fish, or poultry (e.g. lean ground beef, salmon, chicken breasts without the skin, buffalo, tofu). 4. Ate foods made with whole grains (e.g. whole wheat flour, quinoa, oats, brown rice). 5. Ate fast food. 6. Consumed low-fat or non-dairy products or non-dairy substitutes (e.g. yogurt, cheese, milk, or milk substitute). 7. Ate foods made with refined grains (e.g. white flour, white rice, cornmeal). 8. Drank 7-8 glasses of water per day. 9. Consumed sugary drinks (e.g. regular sodas, root beer). 10. Drank alcohol (e.g. beer, liquor, wine). 11. Engaged in light physical activity (e.g. slow walking, cooking, washing dishes) 12. Engaged in moderate physical activity (e.g. brisk walking, light bicycling, heavy cleaning) 13. Engaged in vigorous physical activity (e.g. jogging, playing basketball, carrying heavy loads). 131

APPENDIX K. MORISKY MEDICATION ADHERENCE SCALE

Please answer each question based on your personal experience with your diabetes medication. (0: No; 1: Yes) 1. Do you sometimes forget to take your diabetes pills? 2. Over the past 2 weeks, were there any days when you did not take your diabetes medicine? 3. Have you ever cut back or stopped taking your diabetes medication without telling your doctor because you felt worse when you took it? 4. When you travel or leave home, do you sometimes forget to bring along your medications? 5. Did you take your diabetes medicine yesterday? 6. When you feel like your diabetes is under control, do you sometimes stop taking your medicine? 7. Do you ever feel hassled about sticking to your diabetes treatment plan? 8. How often do you have difficulty remembering to take your diabetes medication (0: Never/Rarely; 1: Once in a while; 2: Sometimes; 3: Usually; 4: All of the time). 9. How often do you measure your blood glucose? (0: Never; 1: Some of the time [1-2 times a week]; 2: Often [3-4 times a week]; 3: Daily [Every day]; 4: More than once per day) 132

APPENDIX L. IRB APPROVAL LETTER