The evolution of depression among students: Evidence for the mismatch hypothesis and the influence of sensory processing sensitivity

THE EVOLUTION OF

DEPRESSION AMONG

STUDENTS: EVIDENCE

FOR THE MISMATCH

HYPOTHESIS AND THE

INFLUENCE OF SENSORY

PROCESSING

SENSITIVITY

Lis Pernot

BACHELORPROJECT PSYCHOBIOLOGIE Supervisor: Annemie Ploeger

The evolution of depression among students: Evidence for the mismatch

hypothesis and the influence of sensory processing sensitivity

Abstract

The world is rapidly changing and the environment in which we live in is completely different from our prehistoric ancestors. Besides, the prevalence of obesity, cardiovascular disease and depression are rising as well. The prevalence of depression among students is even higher (30,6%) compared to the general population. In this research, the evolutionary mismatch hypothesis is investigated as an explanation for the occurrence of student depression. People nowadays live in a completely different environment compared to the one in which we evolved as species (i.e. hunter and gatherers). This caused so named ‘mismatch factors’ because behavioral and psychological mechanisms have had insufficient time to adapt to these rapid changes. Especially students have to deal with new environments, stressors and challenges, and are hypothesized to experience many mismatch factors. Besides, the influence of individual differences in sensory processing sensitivity (SPS) on the relationship between mismatch factors and depression is investigated as well. SPS is a trait that describes individual differences in sensitivity to environmental changes. The number of mismatch factors was highly correlated with and predicted a significant part of the variance in depression. SPS had a moderating effect on the association of mismatch factors and depression and correlated high with the mismatch factors as well as with depression. These findings are important for the comprehension of depression among students and developing effective interventions. There need to be further longitudinal and experimental research to mismatch factors and SPS regarding depression in order to explore the causality of this problem. Keywords: evolutionary mismatch hypothesis, depression, students, sensory processing sensitivity

1.Introduction

The western world nowadays is completely different from the ancestral world in which we are evolved as a species. The occurrence of mental illnesses rises, especially depression is a major health problem in the western world and is the largest contributor to non-fatal health loss in the world (World Healy Organisation, 2016). Symptoms of depression encompass having a depressed mood, losing interest and joy, changes in appetite, sleep difficulties, loss of energy, problems with concentration and suicidality (DSM5, 2013). Depression has adverse consequences and is associated with the highest number of days out of role at societal level compared to other physical and mental disorders (Kessler & Bromet, 2013). The lifetime prevalence of depression among citizens in the Netherlands is 17.9 % (Kessler & Bromet, 2013). Besides, a study investigated the prevalence of depression among students in a systematic review and included 24 articles from different countries which measured the prevalence among undergraduate university students (Ibrahim, Kelly, Adams, & Glazebrook, 2013). Ibrahim et al. (2013) reported that the weighted mean average prevalence among university students is even higher (30,6%) than the prevalence in general populations (e.g. 9 % in the US). Moreover, student depression is the most common reason why students visit the university counsellor (Benton et al., 2003). Generally, students have to deal with a new environment, various social demands, financial stressors and intellectual demands which comes together with increased psychological distress (Stallman, 2010; Vaez & Laflamme, 2008). These findings have not yet been extensively investigated while depression among students is a growing problem in modern society. Depression during student years has a negative impact on various aspects of life. It has been associated to a lower health related quality of life, a negative impact on academic

performance and career development (Arslan, Ayranci, Unsal & Arslantas, 2009; Vaez & Laflamme, 2008; Hysenbegasi, Hass & Rowland, 2005; Newman et al., 1996).

One hypothesis about depression is the evolutionary mismatch hypothesis: people in the western world suffer from depression because the environment in which they live is completely different from the one in which our species evolved (e.g., Hahn-Holbrook & Haselton, 2014). In the modern world people live in a more dense population, eat processed food and substances, are less outside in nature, have more financial stress, increased electronica usage, deal with more separation of families and have a different sleep rhythm compared to ancestral world were humans lived as hunter and gatherers (Li, van Vugt, & Colarelli, 2018). The novel environments can elicit maladaptive behaviour, because there has not been sufficient time for selection to adapt psychological and behavioural mechanisms to these changed environments (Li et al., 2008; Marlow, 2010). When past selection induced adaptive behaviour that is no longer beneficial (fitness-enhancing) and psychological mechanisms are no longer associated in the same way with the environment as in the past, it is called a mismatch (Li et al., 2008).In this way depression could be described as “a disease of modern civilization and affluence” (Koplewicz, Gurian, & Williams, 2009). Some evidence supporting this statement is that

depression has the highest prevalence in the wealthiest countries, because in these countries the environment changes more quickly as a result of rapid evolution (Kessler & Bromet, 2013).

For this study a mismatch questionnaire is developed which measures the amount of mismatch factors, the questions examine in which extend the participant lives similar to his/her nature (i.e. hunter and gatherer lifestyle; Li et al., 2018; Marlow, 2010). The mismatch factors are based on the lifestyle of the Hadza hunter-gatherers, they live in Tanzania in an area where flora and fauna have not changed much since the Plioscene began 5.3 million years ago (Marlow, 2010). There are roughly 1000 Hadza, from which 300 to 400 still live exclusively on hunting and gathering (no cattle), although they had some agriculture contact, they have not changed their lifestyle. Therefore, the Hadza are also called ‘living fossils’ nowadays (Marlow, 2010).

In this study the mismatch hypothesis is applied to students because students might have even more mismatch factors compared to the general population. University life is associated with many changes and challenges such as; moving to a new environment, coping with unfamiliar circumstances, living away from family, dealing with academic stressors, making new social relationships, having less sleep and most often having a different diet. Therefore, it is crucial to examine the mismatch hypothesis as a possible explanation for depression among students.

Furthermore, this study adds another aspect to investigate the evolutionary mismatch hypothesis in students. This aspect involves individual differences in responsiveness and sensitivity to environmental changes. These differences have been seen from pumpkin fish, mice and primates to humans (Wolf, Doorn & Weissing, 2008). Formerly, this division had been described as some being more bold, impulsive and aggressive and others more avoiding, shy, sensitive and cautiously. From an evolutionary perspective it is beneficial if the minority in a population is more sensitive (Wolf et al., 2008). This has survival advantages because these individuals will be more likely to notice possible, subtle threats in the environment (Aron & Aron, 1997).

The described responsivity strategy is further investigated and labelled as a trait named sensory processing sensitivity (SPS). This trait has been identified over 100 species and is evolutionary ancient (Wolf et al.,2008). Sensory processing sensitivity is a genetically based personality trait which is defined by being more sensitive to subtle environmental stimuli, having deeper processing of sensory information, having enhanced susceptibility to overstimulation and having stronger emotional reaction to both positive and negative stimuli. (Aron & Aron, 1997, Jagiellowicz et al., 2011; Homberg et al., 2016). Persons with a high measure of SPS are most often referred to as highly sensitive persons (HSP) and approximately 15 to 20 percent of population is highly sensitive (Aron & Aron, 1997). To measure SPS trait in human the Highly Sensitive Person Scale (HSPS) is developed which is a result of years of extensive interviews and research with highly sensitive individuals (Aron, Aron & Jagiellowicz, 2012). The HSP scale was originally designed to measure an unidimensional construct, however recent studies supporting the existence of a multiple trait structure (Lionetti et al., 2018; Smolewska, McCabe, & Woody, 2006). One global higher-order sensitivity construct with three underlying subscales: Ease of Excitation (EOE; i.e., easily overwhelmed by sensory stimulation), Aesthetic Sensitivity (AES; i.e., aesthetic awareness) and Low Sensory Threshold (LST; i.e., unpleasant arousal to external stimuli).

SPS is a psychological construct, however there is growing evidence that SPS has biological foundations. As explained above SPS is biologically ancient, as it exists as a minority across species. Furthermore, individuals with SPS differ in neural responses during fMRI studies, for example individuals with a high level SPS, are more aware to any subtle changes in visual scenes (Jagiellowicz et al., 2011; Acevedo et al., 2014). Besides, molecular genetic studies found genes associated with SPS such as the s-allele of the serotonin transporter polymorphism (5-HTTPLPR) and genes belonging to the dopamine system (Homberg, Schubert, Asan, & Aron, 2016; Licht, Mortensen & Knudsen, 2011; Chen et al., 2011). The s-allele and SPS are both associated with a greater sensitivity to environmental stimuli and are evolutionary ancient (Homberg et al., 2016). The biological basis of highly sensitive individuals is further explained in a document added to the appendix (See Appendix G). In conclusion, individuals differ in environmental sensitivity and this led to several theoretical frameworks. In particular, the famous orchid-dandelion metaphor, where ‘orchids’ present the individuals who are more sensitive (i.e. they flourish well in ideal conditions and badly in poor conditions) and ‘dandelions’ present the ones who are less sensitive to their environment (i.e. they are resilient and can grow anywhere) (Boyce & Ellis, 2005). More recently there is a growing evidence for three groups rather than two, were the majority who is less sensitive can be divided in medium sensitive and low sensitive individuals, were ‘tulips’ present the medium sensitive group (Lionetti et al., 2018). These groups differ in relative sensitivity but not in the

distribution of the three sub traits, supporting the global unidimensional construct of sensitivity (Lionetti et al., 2018).

It is important to mention that highly sensitive individuals are more sensitive to both negative and positive environmental influences. Negative consequences in terms of vulnerability to adversity (Diathesis-Stress model, Belsky & Pluess, 2009) and beneficial consequences in terms of stronger responses to positive experiences (Vantage Sesitivity model, Pluess & Belsky, 2013). Furthermore, this is supported by fMRI research, were SPS correlated with an increased response to both happy and sad facial expressions compared to neutral faces, especially in sensorimotor and empathy related areas (Acevedo et al., 2014; See Appendix G).

It has not been tested whether sensitivity differences influence the experience of mismatch factors. Therefore, this study combined the evolutionary mismatch hypothesis and SPS to understand and clarify the high number of students with depression. In earlier research, SPS is correlated with depression and appears to be an independent risk factor for the experience of psychological distress (even beyond parental experiences) (Liss, Mailloux, & Erchull, 2008; Yano et al., 2019; Baker & Moulding, 2012). The central questions in this study are: is the high prevalence of depression among university students the result of a mismatch between the current environment and the environment in which the human species evolved? And could individual differences in the experience of mismatch factors, associated with depression, being explained by being highly sensitive?

Highly sensitive persons are more sensitive and might be more vulnerable for mismatch factors as a student. The transition from living at home surrounded with family to live as a student in the city, coping with new and unfamiliar circumstances, having new friends, less sleep, other diet, academic stressors and possibly living less outside can be quite a difference to their formal life. The hypothesis is that SPS is acts as a moderator in the relationship of mismatch factors and depression (See Figure 1). Individuals with a high level of SPS who experience many mismatch factors will be more negatively affected by these mismatch factors, because students with SPS process the environment more strongly and deeply compared to others with low SPS. This could lead to depressive symptoms during their student years. If depression among students is related to high sensitivity the approach to reduce the incidence of depression should be different. For instance, SPS is an important predictor of treatment response in curing depression (Pluess & Boniwell, 2015). Therefore, the relation between HSP and depression in students is crucial to be investigated.

In this correlational study, data will be acquired by measuring symptoms of depression, SPS and the experience of mismatch factors within students. The main purpose is to test the hypothesis if the occurrence of student depression could be explained by the amount of mismatch factors someone experiences and the level of SPS. Depression will be measured by a USDI (University Student Depression Inventory), mismatch factors will be

measured by a self-developed mismatch questionnaire and SPS will be measured by a Dutch translation of the 27-item HSPS (Khawaja & Bryden, 2006; Evers et al., 2008; Aron &Aron 1997). The correlation neuroticism and introversion with SPS will be measured and taken into consideration as well, a Quick Big Five (QBF; Vermulst & Gerris, 2005) is used as a measurement for neuroticism and introversion. Students who score high on the HSPS are expected to have more depressive symptoms when they experience many mismatch factors compared to students scoring low on the HSPS. Besides, it is expected that neuroticism and introversion will have a positive correlation with SPS. Earlier research found these associations as well, because introversion and neuroticism correlate with the negative emotionality aspects of SPS (Smolewska et al., 2006; Sobocko & Zelenski, 2015).

Figure 1. Moderating effect of sensory processing sensitivity on the relationship of mismatch factors and depression. After this quantitative investigation some participants, who scored high and low on mismatch factors, depression and SPS, were invited for an in-depth interview. The aim of the interviews is to get more

understanding about the onset of student depression considering high sensitivity and the timeline of mismatch factors. These interviews might reveal patterns and insights which could contribute to a better understanding of the underlying processes of student depression.

2. Method

2.1 Quantitative measurements

Sample size justification

For a justification of the sample size a power analysis is done with the programme G*Power 3.1.9.4 (Erdfelder, Faul, & Buchner, 1996). For a linear multiple regression with two predictors, a power of 0.8, an alpha level of 0.05 and a medium effect size, a minimum total sample size of 68 participants was needed. Nonetheless, if more exploratory predictors (different mismatch factors) are investigated a larger sample size is needed. In this study 147 participants filled out the questionnaire, however only 106 participants completed the questionnaire.

Participants

The respondents were recruited via social media and had to be currently a student with a minimum age of 16 years old. The participants were recruited within 18 days in October in Amsterdam. The sample was based on voluntary participation and a snowball approach is used to distribute the survey. In total 106 Dutch students between the 17 and 30 years (M = 22.07, SD = 2.51) completed the questionnaires, from which 33 were men and 73 women. Most of the participants were wo-bachelor students (n = 56), some hbo-bachelor students (n = 27), some master students (n = 19) and four mbo students.

Instruments

The survey took approximately 25 minutes in total and consisted of six questionnaires; a sleep questionnaire (PSQI), a food questionnaire, a mismatch questionnaire, a Highly Sensitive Person Scale (HSPS), a Quick Big Five

test (QBF) and a student depression inventory (USDI) and some remainder questions. The data from the first two questionnaires were not used in this study.

Depression

A Dutch translated version of the University Student Depression Inventory (USDI, See Appendix B) was used to measure the symptoms of depression among students (Khawaja & Bryden, 2006). The USDI contained of 30 items and the participant had to rate each statement on a 5-point scale from 1 (‘not at all’) to 5 (‘all the time’). The USDI consisted of three subscales: lethargy (L), cognitive/emotional (CM) and academic motivation (AM). All the items scored in the same direction and the total score range from a minimum of 30 to a maximum of 150 points. The Cronbach’s alpha, a measure of internal consistency, of the 30-item Dutch translated USDI is .95 (Heerink, 2019). A Psychometric investigation of the English original USDI revealed a good reliability and divergent- and convergent validity (Khawaja & Bryden, 2006). Besides, a Persian translation of the USDI showed good concurrent and discriminant validity (Habibi et al., 2014).

Mismatch factors

For this study a special mismatch questionnaire was developed. This questionnaire ought to be a measurement of the amount of mismatch factors each participant experiences. Mismatch factors represent the different way of living nowadays compared to the ancestral world were humans evolved (Hahn-Holbrook & Haselton, 2014). The questions investigated in which extend the participant lives similar to his/her nature (i.e. hunter and gatherer lifestyle; Li et al., 2018; Marlow, 2010). The questionnaire included 77 items divided over 14 mismatch themes. These themes were based on the lifestyle of the Hadza hunter-gatherers who live in Tanzania and are one of the last remaining hunter-gatherers in the world (Marlow, 2010).The first theme is sleep rhythm which covered sleep habits related to hunter- gatherer lifestyle, the second theme covered a lack of movement and questioned if someone had a hard time to sit still. The third theme questioned food habits (processed food in particular), the fourth theme examined the lack of being outside and the exposure of sunlight and herbage. The fifth theme covered the lack of close kinship (e.g. family, having a partner, friends), the sixth theme examined in which extend the participant is attached to materialism and luxury, the seventh theme questioned whether the participant is a perfectionist and performance oriented. The eight theme examined if the participant is rushed and experiences a lack of time, the ninth theme covered in which extend the participant experiences a lack of freedom and does not have the feeling that he/she can do what he/she wants. The tenth theme included questions about a lack of cheerfulness (e.g. laughing, dancing, singing), the eleventh theme questioned social media use and the fear of missing out, the twelfth theme investigated if the participants worry a lot during the week about, for example deadlines, finances and world problems. The thirteenth theme covered the youth of the participant and in which extend this was deviant from the rearing of hunter and gatherers. The last theme consisted of questions about remainder unhealthy habits and substance use such as, drugs, alcohol, beta blockers etcetera.Each item can be answered by either ‘yes’ or ‘no’, the direction of the mismatch scoring differs, and it depends on the question which answer is the mismatch. For example: ‘do you go at least once a day in nature (park, forest, dunes, etc.)?’, if the answer is no, the participant gets a mismatch point for this question. Therefore, for every question the participant gets one or zero points. All the mismatch points will be summed across all the themes, so every participant gets a score for every theme and a score of the total amount of mismatch factors. The mismatch score can run from 0 to a maximum of 77 points (for woman) or 76 points (for men) due to a question for woman about using the birth control pill. However, this question is removed from the total mismatch score for the analysis and therefore the total score is 76 for every participant. The higher the score the more mismatch factors are experienced by the participants in their daily life. The complete questionnaire is added to theappendix (See appendix A).

Sensory Processing Sensitivity

A Dutch translation of the Highly Sensitive Person Scale (HSPS, Aron & Aron, 1997), consisting of 27 items, was used to measure the level of sensory processing sensitivity (SPS) in the participant. Participants were asked to rate their agreement with 27 statements on a Likert Scale rating from 1 (‘not at all’) through 4 (‘moderately’) to

7 (‘extremely’). These items refleced a broad range of sensitivity from ‘Are you easily overwhelmed by strong sensory input?’ to ‘Do you have a rich, complex inner life ?’. All items were scored in the same direction, a higher mean score indicates a higher level of SPS. The dutch translated HSPS is added in the appendix (See Appendix C). The original version has a Cronbach's aplha of .87 and a good discriminant and convergent validity (Aron &Aron 1997). In a psychometric evaluation of the English original HSPS a Cronbach's alpha of .89 was found (Smolewska et al., 2006). The mean score was taken across 27 items for every particpant and this score will be between 1 and 7. A higher score on the HSPS implicated a higher measure of SPS in the participant. There had been done a lot of research on the unidimensional model of the HSPS of Aron & Aron (1997). Smith, Sriken & Erford (2019) did a psychometric analysis of all the 29 articles who used the HSPS. There is evidence to support both unidimensional and 3-factor models underlying the HSPS items. Furthermore, Lionetti et al. (2018) did a confirmatory factor analysis of the HSPS and found a general sensitivity construct underlying all items. Besides, there is evidence that this sensitivity construct has three underlying subscales; Ease of Excitation (EOE; i.e., easily overwhelmed by sensory stimulation), Aesthetic Sensitivity (AES; i.e., aesthetic awareness) and Low Sensory Threshold (LST; i.e., unpleasant arousal to external stimuli) (Lionetti et al., 2018; Smolewska et al., 2006). In this study the separate subscales are not included in our analyses, however a factor analysis is done to explore which sensitivity factor model fits the best to our data. Taken the studies of the original HSPS together, there had been found a Cronbach’s alpha of 0.874 and a good concurrent validity (Smith et al., 2019). This was good enough to use it as a screening-level instrument (≥.80).

Many of the HSPS items involved negative phrasing and negative affect (Aron & Aron, revised in 2018). Earlier research found associations of the negative emotionally aspects of SPS with less desirable personality traits (Smolewska et al., 2006; Sobocko & Zelenski, 2015). This captured a facet of SPS that these persons are more bothered by certain things. However, this negative affect could also be the result of trait negative affectivity (neuroticism). Therefore, a Quick Big Five test was used to measure neuroticism and introversion. Some studies found a difference in controlling for this aspect and others did not, most often studies used a Big-5 Neuroticism scale or a three-item measure that is used a lot in SPS research (Aron, Aron & Davies, 2005; Aron, Ketay & Hedden, 2010). However, this three-item measure overlaps with the USDI and the Big-5 Neuroticism scale was too long to include in the survey (Goldberg, 1992). Besides, another variable which may be needed to control for is introversion (Aron & Aron, 1997; Aron et al., 2005; Aron et al., 2010). Both neuroticism and introversion are often associated with SPS, however in this study it was desired to only measure the effect of SPS on individual differences. Therefore, the subscales neuroticism and introversion from the Quick Big Five (QBF; Vermulst & Gerris, 2005) were included in the analysis and the correlation between these subscales and SPS will be examined.

Quick

Big Five (neuroticism and introversion)

Neuroticism and introversion were measured with the Quick Big Five (QBF; Vermulst & Gerris, 2005) as a control for the HSPS measurements. The QBF determined the personality characteristics according to the Big Five model of Goldberg (1992). The questionnaire consisted of 30 adjectives reflecting the traits extraversion, conscientiousness, agreeableness, openness and emotional stability. Each trait consisted of 6 adjectives and participants had to rate on a 7-point Likert scale in what extend they believed the adjectives were present in themselves, ranging from 1 (‘completely untrue’) to 7 (‘completely true’). The score of emotional stability was reversed and used as a measure for neuroticism, the score of extraversion is reversed as well and used as a measure for introversion. The Dutch version of the QBF is a valid and reliable (α ≥ 0.75) measure of the Big Five dimensions (Vermulst & Gerris, 2005). The Cronbach’s alpha of the measurement of emotional stability was .84 and for introversion .88.

Remainder questions

In the beginning of the questionnaire participants were asked to mention which level of education they occupy. In the end the participant had to mention if they are diagnosed with any mental illness and specify which mental disorder. Besides, the participant had to answer if they suffer from headache, stomach ache, allergies or

eczema more than once a month. These symptoms were interrogated because they could be due as a

consequence of a high amount of mismatch factors or a high level of SPS. Therefore, this could be used as extra exploratory variables in this research.

Procedure

A link to the online survey was made available to friends and added in student Facebook groups (Qualtrics, Provo, UT). Before the survey started, the participants had to confirm they were currently student and 16 years or older. Besides, they had to agree with the inform consent and the right to withdrawal from this research at any moment. The sequence of the surveys was as follows; as first some demographic questions (age, gender, level of education), a sleep questionnaire (PSQI), a food questionnaire, then the mismatch questionnaire, the HSPS, the QBF and as last the USDI. Some final questions were added to ask for any current mental diagnoses or ailments. Furthermore, in the end the participant was asked to state if he/she is willing to cooperate to a second part of this study which includes in-depth interviews.

Data analysis

Before the analysis, the dataset was checked for missing data. Participants with unfinished questionnaires were removed and the data of 106 participants remained. First, the normality of the data was checked for all the variables by performing a Shapiro Wilk test and the internal consistencies were measured for each instrument including the subscales. A principal component analysis was done for the HSPS, USDI and the mismatch questionnaire. For every participant a mean score of HSPS, a sum score of the USDI and a sum score of the total amount of mismatch factor from the mismatch questionnaire is calculated. A sum score of each separate mismatch theme and a sum score for neuroticism and introversion from the QBF were calculated as well. Firstly, the hypothesis was tested by using a linear regression were the mismatch score explained the depression variable. A partial correlation was done to analyse the effect of SPS on the relationship of mismatch factors and depressive symptoms. The contribution of different mismatch themes on depressive symptoms were examined as well. The correlation of neuroticism and introversion with HSPS score will be taken into account too. The variables will be checked for the variables age, gender and education. After that, the impact of remainder exploratory variables on depression, mismatch factors and SPS will be evaluated mostly by comparing means using independent sample t-test.

2.2 Qualitative measurements

Interviews

Interviews were conducted at various places (i.e. at home, in a café and via an online meeting) and lasted approximately 45 minutes each. Each interview was recorded, and any observations made during the interview were written down. The interviews had a semi-structured form which means that key questions are

predetermined and defined the areas needed to be explored (Gill, Stewart, Treasure, & Chadwick, 2008). However, the interviewer was free to ask for clarification and to go more in depth into the details. Therefore, some follow-up questioning was used during the interview. Besides, probing marks (e.g. ‘what do you mean by that’), emotional reflections (e.g. ‘annoying?’) and in-between summaries are used to clarify and verify by the interviewee. The interview format was flexible and allowed to investigate information and patterns which were important for the participants and would normally not be found by using only quantitative data. A pilot interview was done with a colleague to practice the interview questions and test the interview settings (Gill et al., 2008). Sometimes triangulation (duplicate questions) were used to validate the interview (Griffee, 2005).

Procedure

First, an email was sent to the participant of interest with an invitation for an interview. Some information about the purpose and the format of the interview was given and an indication about the time the interview would take. Besides, the terms of confidentiality and anonymity were addressed to the interviewee and an approval for recording audio data. In this way, a rapport and comfortable interaction was built with the respondent in advance of the interview (McGrath, Palmgren, & Liljedahl, 2018; McCracken, 2016). Before the start of the interview the purpose and format of the interview were mentioned again, and the interviewee was allowed to not answer certain questions or stop with the interview at any moment. The interview guideline is added to the appendix (See appendix E). The interview protocol began with some simple biographical questions (e.g. home environment, job, education, family, hobbies). After this background information, the themes: mismatch factors, depression and SPS will be discussed more in depth. Sometimes the scores on certain variables (HSPS and mismatch factors) were mentioned to the interviewee. The timeline of mismatch factors and depression and interaction between these two will be questioned as well. Each interview was closed by asking if the interviewee wants to say or add anything else.

Interviewees

Interviewees were students who completed the online survey and were selected based on their results on the questionnaires. Participants who scored high and low on depressive symptoms were asked for an interview as well as participants with a various amount of mismatch factors.

For the qualitative part of the interview, five participants were interviewed according to the interview guideline (See Appendix E) and answered some biographical questions and questions about depression, mismatch factors and high sensitivity. One extra male participant who scored very high on the HSPS was interviewed by a colleague for her interview, however she asked a few questions about being highly sensitive and reported the answers to me. Four of the six interviewees are likely to be highly sensitive and scored between 5.1 and 6.4 on the HSPS. Besides, one man is interviewed who had the lowest score on the HSPS. Two of the interviewees scored above 90 on the USDI, indicating they experience depressive symptoms high above average. One interviewee experienced many mismatch factors and one interviewee scored low on all the three

measurements (USDI, MM and HSPS). These six interviewees were representative for the data and encompass various outcomes of the quantitative data. The lowest part on the three measurement scales as well as the middle and the higher scores were represented. In order to secure the anonymity of the interviewees pseudonyms are used. Table 2 shows the descriptives and results for every interviewee, table 1 displays the score distribution of the total research sample for every instrument.

Table 1.

Descriptive results for each instrument in the research sample.

-2SD -SD Mean +SD +2SD USDI 29.18 46.78 64.37 81.96 99.56 Mismatch questionnaire 18.79 24.92 31.05 37.18 43.31 HSP Scale 2.50 3.42 4.35 5.28 6.20

Note: -2SD = the mean minus 2 standard deviations, +SD = the mean plus one standard deviation.

Table 2.

Information interviewees.

Anne Bart Merel Tessa Jasper Elianne

Gender V M V V M V

Age 20 24 25 18 25 23

USDI 93 43 90 58 91 74

HSP Scale 5.11 2.70 6.37 3.93 6.00 5.93

questionnaire Sleep deviations 3 1 2 3 4 2 Lack of movement 4 3 3 4 1 4 Lack of being outside 2 2 2 2 0 4 Processed food 1 0 1 1 4 0 Lack of close kinship 2 2 2 1 0 1 Materialism 2 0 2 1 2 1 Perfectionism 4 4 5 5 4 5 Lack of time 5 1 5 5 4 5 Lack of freedom 2 1 1 0 0 1 Lack of cheerfulness 2 2 2 0 1 1 Social media (FOMO) 5 1 2 3 1 3 Worries 5 3 3 0 5 3 Deviant youth 0 0 1 0 0 1 Unhealthy habits (substance use) 1 1 1 0 3 1

Note: Gender, age and scores on the three instruments are reported. Besides the scores on the separate mismatch themes are mentioned as well (for each theme a maximum of five mismatch factors could be achieved, except for unhealthy habits nine could be achieved).

Data analysis

A deductive thematic content analysis was used to analyse the interviews (Braun & Clarke, 2006). The first step in the qualitative data analysis was to make a transcript of the recorded audio data and taken notes. Afterwards the transcripts were read to familiarize with the data (Griffee, 2005). Relevant fragments in the transcript were coded (e.g. something that surprised the researcher, something which was relevant to the interviewee, something which was similar to existing theories or concepts). A recapitulation was made of the interviews and categories were made based on the key themes discussed in the interviews. The final step was the

interpretation of the data and integrating the pre-made hypothesis about the influence of SPS and mismatch factors to develop student depression.

3. Results

3.1 Quantitative results

Descriptive results

Due to missing data, the data of 106 students between the 17 and 30 years (M = 22.07, SD = 2.51) were analysed, from which 33 were men and 73 women. The sample characteristics are listed in Table 3.

Table 3.

Sample characteristics of the total scores of the instruments used in this study.

Minimum Maximum M SD USDI 33.00 111.00 64.37 17.59 Mismatch questionnaire 19.00 46 31.04 6.13 HSPS 2.26 6.48 4.35 0.93 Neuroticism 6.00 39.00 24.11 7.28

Introversion 12.00 36.00 22.63 5.55

Note: M = mean, SD = standard deviation, USDI = University Student Depression Inventory, HSPS = Highly Sensitive Person Scale. Neuroticism and Introversion are reversed subscales from the Quick Big Five.

Student depression (USDI)

The sum scores of the 30-item USDI were calculated. Normality was tested with a Shapiro-Wilk test, the scores on the USDI were not normally distributed (W(106) = .97, p = .032), with an alpha level of .05. The scores on the USDI between men and women were not significantly different (See Table 10). The assumptions for equal variances were met, measured with a Levene’s test (F = 1.97, p = .163).Therefore, the sample is treated as whole for the statistical analysis. Besides, there was investigated if depression scores differ between groups made on level of education. There was only found a difference between master students and wo-bachelor students, bachelor students had significantly a higher score on depressive symptoms compared to master students (p = .009, See Table 10). There was a good reliability, measured by internal consistencies, of the whole USDI scale as well as the subscales; cognitive/emotional (CM), lethargy (L) and academic motivation (AM) (See Table 4).

Table 4.

Internal consistencies for the Dutch translated USDI and the subscales. Cronbach’s alpha USDI total .93 Cognitive/emotional .84 Lethargy .75 Academic motivation .77

Mismatch factors

One question, about the birth control pill, was excluded from the analysis of the mismatch score, because otherwise men and women had a different total score and there is found no difference in depression scores between women who take the pill and women who did not (p =.714, See Table 6). For every participant the total score of 76 items from the mismatch questionnaire was measured, which indicates the amount of mismatch factors. The scores on the mismatch questionnaire were normally distributed (W(106) = .98, p = . 181). The scores of men and women were not significantly different (See Table 10). The assumptions for equal variances were met, measured with a Levene’s test (F = 0.58, p = .450). Besides, there was no difference in mismatch scores between groups based on level of education (See Table 10).

Mismatch factors had a significant positive association with the USDI (r(104) = .50, p < .001, See Table 5). The contribution of the 14 different themes of the mismatch questionnaire to the USDI score were analysed as well (See Table 5). As first a multiple linear regression was done with the 14 mismatch themes as predictors of the USDI scores, however these predictors were not normally distributed and the assumptions for doing a multiple regression were violated. Therefore, there is chosen to only report the correlations between the mismatch themes and the USDI. It has been found that especially lack of freedom, sleep deviations, lack of time and worries had the strongest positive correlation with the USDI, especially lack of freedom had the highest correlation. These results were in accordance with the contribution of the mismatch themes measured with a multiple regression, regarding R squared scores (due to violated assumptions and a lack of power for this analysis, these results are not reported). The correlation of mismatch factors with depressive symptoms (r(104) = .50, p < .001) were diminished when controlling for lack of freedom and sleep (r(102) = .29, p = 0.002). The relation of depressive symptoms and mismatch factors resolved if the correlation is controlled for lack of freedom, sleep and worries (r(101) = .199, p = .044), and even more when including lack of time (r(100) = .17, p = .098). The total amount of mismatch factors explained equally the three different subscales of the USDI (See

Table 7), the three subscales were normally distributed as well (p < .01). The reliability of the total mismatch questionnaire is reasonable, however the internal consistencies of the separate themes are low (See Table 6).

Table 5.

Spearman rank correlation of the mismatch themes with the USDI.

USDI USDI

Total mismatch scale .50**

Sleep deviations .24* Lack of time .27**

Lack of movement -.10 Lack of freedom .47**

Processed food .25** Lack of cheerfulness .18

Being outside .13 Worries .27**

Lack of close kinship .17 Social media (FOMO) .12

Materialism .08 Deviant youth .18

Perfectionism .23* Unhealthy habits

(substance use)

.11

Note: * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

Table 6.

Internal consistencies for the Mismatch questionnaire and separate mismatch themes.

Cronbach’s alpha Cronbach’s alpha

Total mismatch scale .67

Sleep deviations .08 Lack of time .70

Lack of movement .47 Lack of freedom .52

Processed food .27 Lack of cheerfulness .39

Lack of being outside .44 Worries .44

Lack of close kinship .22 Social media (FOMO) .79

Materialism .45 Deviant youth .40

Perfectionism .66 Unhealthy habits

(substance use)

.37

Table 7.

Linear regression of the Mismatch factors in predicting the subscales of the USDI; lethargy (L), cognitive/emotion (CE) and academic motivation (AM).

Mismatch factors predictors

t p-value β F df p-value R Squared

L 5.40 .000** .47 29.18 1, 104 .000** .22

CE 4.76 .000** .42 22.69 1, 104 .000** .18

AM 5.25 .000** .26 27.55 1, 104 .000** .21

Note: * Significance at the 0.05 level (2-tailed). ** Significant at the 0.01 level (2-tailed).

Sensory processing sensitivity (SPS)

The mean scores on the HSPS were normally distributed after testing with a Shapiro-Wilk test (W(106) = .99, p = .703). Men scored significantly lower on the HSPS than women (p < .001, See Table 10). The assumptions for equal variances were met, measured with a Levene’s test (F = 0.08, p = .779). The 27-item HSPS had a high internal consistency with a Cronbach’s alpha of .91. Lionetti et al. (2018) provided cut-off score to categorize individuals into the three different sensitivity groups: i.e. low sensitive (mean SPS scores below 3.71), medium sensitive (mean SPS score between 3.71 and 4.66) and high sensitive (mean SPS score above 4.66). Taking these approximate cut-off scores into account, 38 % of our sample should be highly sensitive.

As control variables for SPS, neuroticism and introversion from the QBF were measured for each participant. The maximum possible score for each aspect of the QBF was 42 points (See Table 3). The scores of neuroticism were normally distributed (W(106) = .98, p = .192) as well as the scores for introversion W(106) = .98, p = .115,

See Table 3). Cronbach’s alphas for neuroticism and introversion items were .87 and .90, respectively. There was found a relatively high correlation of the total HSPS and neuroticism (r(104) = .62, p < .001). Besides, HSPS had a weak correlation with introversion (r(104 = .24, p < .001, See Table 8).

Table 8.

Correlations between HSPS, neuroticism and introversion.

HSPS Neuroticism Introversion

HSPS 1.00 -

-Neuroticism .62** 1.00

-Introversion .24* .38** 1.00

Note: Introversion and neuroticism are subscale measurements of the Quick Big Five (QBF). * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). Cronbach’s alphas are shown in the diagonal.

Bivariate correlations between depression, mismatch factors

and SPS

Correlations between variables were tested with a Pearson correlation except for the USDI, because these scores were not normally distributed, a Spearman’s rank correlation is used. Depressive symptoms were strongly correlated with the amount of mismatch factors a participant experienced (r = .50, p < .001, See Table 9). Besides, a positive correlation was found between SPS and the USDI (r = .37, p < .001, See Table 9), however the significant positive correlation between SPS and depression was a lot less when controlling for mismatch factors, r(103) = .22, p < .025. SPS had a significant positive correlation with the mismatch score as well (r(104) = .40, p < .001, See Table 9).

Table 9.

Correlations between mismatch factors, USDI and HSPS

Mismatch factors USDI HSPS

Mismatch factors 1.00 -

-USDI .50** 1.00

-HSPS .40** .37** 1.00

Note: USDI = University Student Depression Inventory, HSPS = Highly Sensitive Person Scale. The correlations are tested with a Pearson correlation except for USDI, a Spearman rank correlation is used because the assumption of normality is violated. * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). Cronbach’s alphas are shown in the diagonal.

Answering the main hypothesis

A linear regression was done to examine the explained variance of depressive symptoms due to mismatch factors. The residuals of this model were normally distributed (W = .98, p = .229) and the independent variables were not highly correlated with each other (VIF = 1.00). Besides, homoscedasticity was confirmed by analysing the residuals plot and tested by a Breusch-Pagan test (BP(2) = 3.88, p = .143). The amount of mismatch factors experienced by the participant, significantly predicted depression scores, .47, t(104) = 5.40, p < .001. Mismatch factors also explained a significant proportion of variance in depression scores, R² = .22, F(1,104) = 29.17, p < .001. The correlation between mismatch factors and depression was controlled for age, however the impact of age in a partial correlation was very small (r(103) = .47, p < .001).

Additionally, a linear regression was done for the explained variance of depression by SPS. SPS explained a significant proportion of the variance of the scores of depressive symptoms as well, R² = .13, F(1,104) = 16.10, p < .001. A partial correlation was run to determine the influence of SPS on the relationship between the experienced mismatch factors and depressive symptoms. There was a positive significant correlation between the depression scores and mismatch factors whilst controlling for SPS, r(103) = .38, p < .001. This controlled

correlation was lower than the zero-order correlations between depression scores and mismatch factors (r(104) = .47, p < .001), indicating a moderating effect of SPS.

Additional group differences and controls

Some independent t-test were done to investigate possible difference between groups based on some variables (e.g. living at home, religion, living in the city) which might be accounted to increased depression or mismatch factors (See Table 10). The assumptions for equal variances were checked with a Levene’s test. There was no significant difference in either mismatch factors or depression scores between students who live with their parents/caregivers or students who live by themselves (p = .179; p = .366, See Table 10). Students who live in the city did not significantly differ with students who live by themselves in their scoring on the mismatch questionnaire (p = .072) as well as depression scores (p = .826, See Table 10).

Furthermore, students who were religious or spiritual did not significantly differ from students whom were not, in either mismatch scores as depression scores (p = .198; p = .408, See Table 10). Individuals who had a

stomach-ache regularly did not differ in their scores on the USDI or mismatch questionnaire, however individuals who had a headache regularly did experience more mismatch factors and no such difference was found in the USDI scores (See Table 10).

Table 10.

Independent samples t-test comparing means within variables for each questionnaire (USDI/MM/HSPS). Variable USDI/MM/

HSPS

Levene’s Test Independent Samples T-test

Sample (n) M SD F Sig. t df SE

diff.

Sig. Gender USDI Men (33)

Women (73) 62.88 65.05 19.41 16.81 1.97 .163 -0.58 104 3.70 .560 MM Men (33) Women (73) 30.12 31.45 5.87 6.24 0.58 .450 -1.04 104 1.29 .303 HSPS Men (33) Women (73) 3.88 4.57 0.88 0.87 0.08 .779 -3.72 104 0.18 .000**

Education USDI HBO (27)

WO Bachelor (56) 65.93 65.96 16.67 17.29 0.02 .892 -0.01 81 4.01 .992 MM HBO (27) WO Bachelor (56) 30.52 32.16 7.04 5.72 1.68 .198 -1.14 81 1.45 .260 USDI WO Bachelor (56) Master (19) 65.96 54.00 17.29 15.34 0.23 .635 2.68 73 4.47 .009** MM WO Bachelor (56) Master (19) 32.16 29.37 5.72 5.90 0.01 .922 1.83 73 4.21 .072 Living with parents USDI Yes (42) No (64) 66.29 63.11 16.63 18.22 0.87 .353 -0.91 104 3.50 .366 MM Yes (42) No (64) 30.05 31.69 5.86 6.26 0.09 .763 1.35 104 1.12 .179

Religious USDI Yes (15) No (91) 67.87 63.79 16.03 17.86 0.45 .503 -0.83 104 4.91 .408 MM Yes (15) No (91) 32.93 30.73 5.64 6.18 0.34 .562 -1.30 104 1.70 .198 HSPS Yes (15) No (91) 5.07 4.23 0.59 0.92 3.72 .056 -3.39 104 0.25 .001** Living in the city USDI Yes (76) No (30) 64.61 63.77 17.56 17.96 0.03 .870 0.22 104 3.81 .826 MM Yes (76) No (30) 31.71 29.33 6.12 5.92 0.01 .936 1.82 104 1.31 .072

Birth control pill USDI Yes (46) No (31) 64.57 66.00 16.85 16.65 0.02 .904 -.37 75 3.90 .714 Stomach-ache HSPS Yes (44) No (62) 4.62 4.16 0.93 0.88 0.01 .936 2.57 104 0.18 .012* USDI Yes (44) No (62) 64.73 64.11 18.56 17.03 0.17 .679 -0.18 104 3.48 .860 MM Yes (44) No (62) 31.45 30.74 0.82 0.86 0.63 .428 -0.59 104 1.21 .558 Headache HSPS Yes (48) No (58) 4.66 4.10 0.82 0.93 2.17 .144 3.30 104 0.17 .001** USDI Yes (48) No (58) 67.79 61.53 18.45 16.48 1.07 .305 -1.84 104 3.40 .068 MM Yes (48) No (58) 32.92 29.48 5.91 5.92 0.25 .619 -2.98 104 1.15 .004** Allergies/ eczema HSPS Yes (18) No (88) 4.43 4.34 0.92 0.93 0.00 .962 0.37 104 0.24 .715

Note: USDI= university student depression inventory, MM= Mismatch questionnaire, HSPS= highly sensitive person scale. Stomach ache, headache and allergies/eczema are considered “yes” if the participant experience these symptoms more than once in a month. M = Mean, SD = standard deviation, SE diff = standard error difference, df = degrees of freedom, sig. = significance (p-value). * T-test is significant at the 0.05 level (2-tailed). ** T-test is significant at the 0.01 level (2-tailed).

Associations with SPS

Students who were religious or spiritual did score significantly higher on the HSPS compared to non-religious students (p = .001, See Table 10). Besides, participants who had stomach ache more often than once a month scored higher on the HSPS as well (p = .012, See Table 10). This was the same for participants who had a headache more often than once a month (p = .001, See Table 10).

Factor analysis

An initial principle component analysis was done with the data of the 27 HSP items. There is a good collinearity between the items following from the correlation matrix. The Kaiser-Meyer-Olkin measure of sampling

adequacy was .85, above the recommended value of .60 and the Barlett’s test of sphericity was significant (χ2 (351) = 1256.50, p < .001). The communalities were all above .40, indicating that a good proportion of each item variance can be explained by the factors.

Taken everything together, a factor analysis was suitable for the 27 items. Initially, six factors emerged with an eigenvalue bigger than 1.0, however the first three factors explained already 47% of the variance, the last three 13 %. A rotation was needed because the first factor explains 32% of the variance. Based on previous theories, a three-factor solution was chosen with a Direct Oblimin rotation (Smolewska et al., 2016). In the component correlation matrix, the components correlated with each other, especially component one and three (See Appendix Table D3). Because of the sample size and rotation used in this study, the structure matrix is used to analyse the different factors. As mentioned in the introduction there is a growing evidence for three subscales Ease of Excitation (EOE), Low Sensory Threshold (LST) and Aesthetic Sensitivity (AES) (Smolewska et al., 2016). However, in the structure matrix of our sample this division was not so clear, most items (13) loaded on the first component and this was a combination of EOE and LST items (See Appendix Table D4). However, it was obvious that the second component belongs to the subscale AES because it loaded only AES items. The third component loaded eight items, belonging to EOE and one item belonging to the LST subscale. Therefore, in this sample the AES subscale was clearly distinguishable from the other two subscales, EOE and LST were a bit more similar to each other. This was seen in the component correlation matrix as well, component two (AES) correlates the lowest with the other two components (See Appendix Table D3).

USDI

There was a good collinearity between the items following from the correlation matrix. The Kaiser-Meyer-Olkin measure of sampling adequacy was .81, above the recommended value of .6 and the Barlett’s test of sphericity was significant (χ2 (435) = 1743.80, p < .001). The communalities were all above .60, indicating that a good proportion of each item variance can be explained by the factors. Initially, eight factors with an eigenvalue bigger than 1.0 emerged from the factor analysis. The first three factors explained 48% of the variance and the last five factors 22 %. Based on the assumed three subscales of the USDI, a three-factor solution was chosen with a Direct Oblimin rotation (Khawaja & Bryden, 2006, Habibi et al., 2014). An oblique rotation was chosen because the components correlated with each other (See Appendix Table D1). After the rotation the sum of squared loadings were around 6.0 for the three factors. Because of the sample size and rotation used in this study, the structure matrix was used to analyse the different factors. The three subscales: lethargy (L), cognitive/emotional (CM) and academic motivation (AM) should be distinctive from each other (Khawaja & Bryden, 2006). However, in the structure matrix the items from lethargy were equally distributed between the three components and the items of CM are loaded on component one and three (See Appendix Table D2). Only all items of AM loaded on component two. Therefore, the subscale AM could be distinguishable from L and CM in this sample, L and CM were harder to distinguish from each other.

Mismatch questionnaire

A principle factor analysis was done with the 76 items of the mismatch questionnaire. There was not a good collinearity between the items following from the correlation matrix. The Kaiser-Meyer-Olkin measure of sampling adequacy was .32, under the recommended value of .60, indicating the sample was not very adequate for a factor analysis. Barlett’s test of sphericity was significant (χ2 (2926) = 3996,12, p < .001) and

communalities were all above .60. Initially, 28 factors with an eigenvalue bigger than 1.0 emerged from the factor analysis explaining 77% of the variance. In the scree plot there was a clear ‘leveling off’ after 13 factors. The first 13 factors had an eigenvalue above 2.0 and explain 51% of the variance. The questionnaire was based on 14 themes (13 transparent themes and one with many items related to unhealthy habits and some

remainder items), therefore a thirteen-factor solution was done with a Varimax rotation. There was chosen for a Varimax rotation because the correlations between the components in the component correlation matrix were very low (r < .01, See Appendix Table D5).

In the appendix the rotated component matrix is added, the items were put in order of factor loadings from high to low beginning with component one (See Appendix Table D6). Component one could be identified as the theme lack of time, however it is noteworthy to mention that an item of the lack of freedom loaded on this component as well (See Table 11). The items of the social media use and two sleeping items (related to electronica use) loaded on component two, this component could be identified as the social media mismatch theme. The third component could be named as a lack of freedom component. The fourth component almost loaded all items about perfectionism and the fifth component could be seen as the deviant youth component. Almost all the items about substance use loaded on factor six and the seventh component loaded the items about living close to your family and some other items about living in the city and worrying. Component eight and ten consisted of items of a variety of themes and could be seen as two components with remaining items. Component nine could be seen as the materialism component and the eleventh component loaded items about a lack of being outside and a lack of movement. The lack of sleep and healthy food items loaded the most on component twelve. The last component thirteen consisted of different items but most apparent were two items about worrying about finances and finding a job. This interpretation of the components from the output of the factor analysis is displayed in Table 11. In the second row of Table 11 the potential name (inspired by the existing mismatch themes) is given and the third row shows which items loaded the most on this component. In the fourth and the fifth row items are discussed which are loading unexpected to that particular component.

Table 11.

Clarification of the components of the rotated component matrix from the mismatch questionnaire

Componen t from matrix Specified (i.e. identified as) Included items of several themes

Oddities (i.e. unexpected loading of an item with a component)

Explanation

1 Lack of time - four items of a lack of time

- one item of lack of freedom

- one item of worrying (deadlines)

Heb je het gevoel dat je altijd kunt vertrekken als dingen je tegen gaan staan? (.44)

This item from the theme lack of freedom has de most divided answers (61

answered no) compared to the other four lack of freedom items (maximum of 18 answered no).

Ben je religieus en/of spiritueel? (-.24)

Apparently because of the limited distribution in answers (only 15 participants were religious/spiritual).

2 Social media use - four items of a social media use

- two items of a lack of sleep (related to electronical use) - one remainder item (being religious/ spiritual)

Ben je religieus en/of spiritueel? (.139)

This item has a low factor loading and as mentioned above it has a limited distribution in answers

3 Lack of freedom

and joy

- three items of a lack of freedom

- three items of a lack of cheerfulness (laughing, social activities, dancing) - one item of materialism

It seems like having freedom and experiencing joy (laughing, social activities, dancing) loading together on component three

4 Perfectionism - five items of

perfectionism - one item of lack of time (full planned agenda)

Heb je de neiging om je agenda vol te plannen? (.40)

Maybe planning a full agenda is a characteristic of perfectionism, because the other items of a lack of time reflect more the feeling of having too less time and being rushed.

5 Deviant youth - four items of deviant youth

- one item of lack of close kinship

- one item of substance use (Ritalin) Gebruik je momenteel Ritalin of andere stimulerende medicijnen? (.55) A limited distribution of answers, only 2 participants answered ‘yes’.

6 Substance use - four items of

substance use - two items of processed food (sugar drinks and fried/ salty food)

- one item of a lack of cheerfulness (singing)

- one item about worrying (finishing study) 7 Lack of close kinship (lack of family bonding)

- one item of a lack of close kinship (living close to your family) - two items from remaining questions (living by your parents and living in the city) - one item about worrying (world situation)

- one item about lack of movement (problems sitting still)

Heb je moeite met stilzitten? (.22)

This items loads very low so the contribution on this component is small.

8 Leftovers - two items of

substance use (antidepressants and painkillers)

- one item of a lack of freedom

- one item of a lack of being outside (having plants inside)

- one item of a lack of close kinship (having a partner)

9 Materialism - three items of

materialism - one item of social media use (FOMO) - one item of a lack of movement (long periods of sitting still) - one item of a lack of sleep (napping)

Doe je minstens drie keer per week een dutje overdag? (.18)

Very low factor loading, small contribution to this component.

10 Leftovers - two items of

substance use (beta blockers and sleeping pills)

- one item of

processed food (sweet snack)

- one item of a lack of cheerfulness (listening to music)

- one item of deviant youth (having siblings)

11 Lack of being

outside and lack of movement

- three items of a lack of being outside - three items of a lack of movement

12 Lack of sleep and a

lack healthy food

- two items of a lack of sleep

- two items of

Heb je één of meer goede vrienden met wie je persoonlijke zaken kunt

A limited distribution of answers, only 2 participants answered ‘no’ to this item.

processed food (only items about a lack of healthy foods) - one item of a lack of close kinship (discuss personal problems) - one item of a lack of being outside (daylight while studying)

bespreken? (.49)

13 Worrying and

leftover items

- two items of worrying (finances and finding a job)

- one item of materialism

- one item of deviant youth (breastfeeding)

Note: Be aware of the scoring of the items in terms of mismatch factors. At first sight some the clustering of some items seems incorrect, however if this is not mentioned this is due to the scoring. For example, living with your parents and living in the city are loading together. This is because an individual gets a point if he/she lives with their parents and gets zero points is he/she is living in the city (mismatch factor scoring). Therefore, it is reasonable that these items load together because most of students are living in the city without their parents.

3.2 Conclusion quantitative research

The main goal of the quantitative research was to examine if depression could be explained by the amount of mismatch factors and if sensory processing sensitivity (SPS) plays a role in this relationship. There was a significant positive correlation between the amount of mismatch factors and depressive symptoms. Besides, mismatch factors predicted depression scores and explained a significant proportion of the variance in depression scores. The mismatch themes: lack of freedom, sleep deviations, lack of time and worries had the strongest positive association with depression, especially lack of freedom shows the biggest contribution in the relationship between mismatch factors and depression.

Additionally, SPS had a moderating effect on the association between mismatch factors and depression, indicating that the level of SPS had an influence on the experience of mismatch factors and the occurrence of depressive symptoms. SPS was significantly positively correlated with depression and the amount of mismatch factors as well and appeared to explain a significant proportion of the variance of depression scores too. Besides, SPS correlated relatively high with neuroticism and had a weak correlation with introversion.

The reliabilities of the HSPS and USDI were very good. The internal consistencies of the three subscales of the USDI were high as well and the subscales (lethargy, academic motivation and cognitive/emotional) were equally explained by the mismatch factors. The reliability of the mismatch questionnaire was reasonable, however the internal consistencies of the separate mismatch themes were low.

There were no gender effects in depression scores or mismatch scores, however women scored higher on the HSPS than men. There was investigated if education level influences the outcomes on the questionnaires, it seemed that wo-bachelor had more depressive symptoms compared to wo-master students and no such differences were found by mismatch factors or SPS. Spiritual or religious participant had higher levels of SPS compared to non-religious participants, however there only were a small amount of participants who were religious or spiritual. Participants with high SPS did have a headache and stomach ache more often compared to participants with lower SPS.

In the factor analysis of HSPS a three-factor solution fits the best, aesthetic sensitivity items from the HSPS were the most distinctive from items of the other two subscales ease of excitation and low sensory threshold. The three-factor solution was in line with earlier research for a three-factor model and one global underlying SPS

factor (Smolewska et al., 2006). A three-factor solution fitted the best for the USDI as well. The items belonging to the academic motivation subscale were the most distinctive and items of the lethargy and

cognitive/emotional subscale loaded more equally over the two factors. The mismatch questionnaire fitted the best in a thirteen-factor solution, however there were some anomality’s because not all the items of the separate mismatch themes loaded on the same components in the factor solution (See 3.2.1).

3.2.1 Evaluation of the mismatch questionnaire

The output and interpretation of the factor analysis of the mismatch questionnaire (See Table 11) are important for evaluating the mismatch questionnaire as a valid measurement of mismatch factors. Most of the items of the themes perfectionism, materialism, a lack of close kinship, deviant youth, a lack of time, social media and substance use load on the components as expected. However, items of the themes processed food, a lack of sleep and worrying are spread out over other components. In the factor analysis some items of a lack of being outside and a lack of movement are loading together at a component. In the future these two themes might be put together to one theme in order to find a better model fit in the factor analysis. This applies for the items of a lack of freedom and cheerfulness as well. Furthermore, from the factor analysis appeared that items of the intake of healthy food and having enough sleep are loading together on a component which might be expected. The items of the rest of the themes were divided over various components. In the factor analysis some items loaded in a unexpected manner on certain components and it could be considered to remove these items. One example is an item of the remainder questions (‘Ben je religieus en/of spiritueel?’), this item had a very low factor loading and had a limited distribution in answers (only 15 participants were religious/spiritual). Another example is an item of a lack of freedom (‘Heb je het gevoel dat je altijd kunt vertrekken als dingen je tegen gaan staan?’) which loaded by the component considered to be a lack of time (See Table 11). This item is isolated from the other four items of a lack of freedom and this might be because of the variance in answers is more compared to the other four items. So an advise would be to increase the variance of the other items of a lack of freedom. This can be done by adding more specific items reflecting the feeling of a lack of freedom, because the other four items are almost questioning the same facet of a lack of freedom. Furthermore, one item of a lack of movement (‘Heb je moeite met stilzitten?’) is not loading together with the other items of a lack of movement. This might be because having difficulties with sitting still does more refer to concentration problems or restlessness instead of moving enough during the day. In conclusion, there might need to be some

adaptations to the mismatch questionnaire and the output of the factor analysis is valuable when the questionnaire is used for other research or in a repeated experiment.

3.3 Recapitulation of the interviews

This is a summary made of the entire interviews, the complete interviews are added to the appendix (See Appendix F).

Biographical information

Anne and Merel were doing a hbo-study, however Anne studied a lot less than before due to a burn-out and Lyme disease. Tessa and Elianne are doing a wo bachelor (medicine and psychology) and Bart almost finished his master’s in mathematics. After living a fulltime student life in Amsterdam Anne lives at home with her parents again. Elianne, Merel and Bart live by themselves in relatively quiet environments with one or a few housemates. Tessa still live with her parents and travels every day with public transport for one hour to Erasmus Medical Centre for her study medicine. Merel lives in Switzerland now because of her internship as

physiotherapist and Bart is doubting if he wants to start working or study further for his PhD. Merel, Tessa and Elianne experience a high workload and stress because of their studies. Elianne works one day in the week as a writer and has an extra pressure to do her job well besides her busy study. All the interviewees try to plan a social activity at least once a week. Bart and Merel are very sportive during the week in contrast to Tessa who does not exercise at all.