Fear of Failure in Swedish 9th Grade Students and its effect on their decision to study STEM

(1)

Fear of Failure in Swedish 9th Grade Students and its effect on their decision to study STEM

A quantitative study

Författare: Melanie Montes Handledare: Mattias Lundin Examinator: Karina Adbo Termin: HT20

Ämne: Kompletterande Pedagogisk Utbilding, Biologi och Naturkunskap

Nivå: Avancerad Kurskod: 2KP60E

M.A. thesis in secondary education for

upper secondary school

(2)

(3)

Abstract

Sweden faces a recruitment problem in the fields of science, technology, engineering, and mathematics (STEM). It has been suggested that imposter phenomenon, a feeling of fraudulence and inability to internalize success, may be the reason why many individuals, especially women, leave STEM fields. Fear of failure is theorized to be the main motive underlying imposter phenomenon. In Sweden, students are asked to choose a direction of study already when entering upper secondary school. In the current study, a survey of 1045 ninth grade students from 27 schools throughout Sweden aimed to answer five research questions: (1) How prevalent is fear of failure in grade 9 students in Sweden, (2) How does the prevalence of fear of failure differ between male and female students, (3) and between students with differing levels of achievement, (4) How do students’ fear of failure differ in relation to STEM subjects and other subjects, and (5) Are students with a higher fear of failure less likely to pick a STEM program for their continued education? The results showed that the average Swedish grade 9 student did not experience a high fear of failure, but over one fourth of the students surveyed at least partially agreed with the statements in the Performance Failure Appraisal Inventory, which puts them at risk for suffering negative mental health effects.

Females experienced a significantly higher fear of failure, which may be rooted in gender stereotype consciousness. Perceived risk of failure in STEM programs was not a significant predictor for whether students picked STEM programs, and surprisingly, fear of failure was significantly higher in those that would choose to study STEM. It seems that interest in a program and future career opportunities outweighs any potential risks of failure. If students with a high fear of failure are choosing STEM programs, it is important for educators to break them out of the

“imposter cycle” of overpreparing and then being unable to internalize their successes. One way to do this is to teach students about the early struggles that famous scientists faced in producing the information they are learning, in order to teach them that failure, or hard work and effort, does not signal a lack of

intelligence. The best predictor for students that would choose STEM programs was whether they found mathematics and science classes easy, so sparking interest and demystifying these subjects is one of the keys to the STEM recruitment problem.

Keywords

Imposter phenomenon, STEM attrition, Gender, gymnasieval

Acknowledgements

A big thank you to all the principals, teachers and students who made this study possible by sending along and answering my survey. Special thanks to Fredrik Engblom, Björn Palmqvist, and the students at Vasaskolan in Kalmar for helping me conduct a pilot study of the survey. Thank you as well to my supervisor, Mattias Lundin, for the interesting feedback and discussions. To my friends and colleagues in the KPU program, thanks for making the past year and a half so much more enjoyable.

And finally, to my partner Fredrik, thanks for all your patience in proofreading my work!

(4)

Appendix

Appendix 1: Survey

(6)

1(35)

1 Introduction

In Sweden there is a serious recruitment problem within science and engineering (SCB, 2019a). Each year Statistics Sweden (SCB), the official statistics agency of Sweden, gathers data from employers on how their current employment needs are being met and on whether they foresee a need for more workers in coming years. In their latest report they found that currently the largest deficit is in healthcare professionals (SCB, 2019a). Programming and IT were the fields that were expected to have the highest demand for graduates within the next 3 years, followed by civil engineers (SCB 2019b). A 2019 report from the Swedish Public Employment Service similarly found that 11 of the 15 jobs predicted to be most in demand for college graduates in the coming 5 years required a science or technology program at the upper secondary school level (Arbetsförmedlingen, 2019). Despite growing needs in the fields of Science, Technology, Engineering and Mathematics (STEM), fewer students are choosing science and technology programs at the upper secondary school level.

According to statistics from the Swedish National Agency for Education (Skolverket, 2020a), the social sciences program is consistently the most popular and interest in economics programs is steadily increasing, while the number of students who have applied for natural sciences has decreased over the past five years, and the technology program has plateaued under nine percent of applicants to upper secondary programs (Fig.1).

Figure 1. Percent applicants to different upper secondary programs each school year beginning in 2011, when the current system was put into place. The four most popular college-preparatory programs are pictured. Programs are abbreviated as follows: NA= Nature sciences, TE= technology, SA= social sciences, EK = economics. Percentage is calculated from the total number of students who applied to any upper secondary school program each year (both trade and college-preparatory programs).

Statistics obtained from Skolverket (2020a).

0,0 2,0 4,0 6,0 8,0 10,0 12,0 14,0 16,0 18,0 20,0

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Percentage of Applicants

Year

NA TE SA EK

(7)

2(35)

Imposter phenomenon (IP) is theorized to be a contributing factor to STEM attrition, or why individuals choose to leave STEM fields at the university and post-graduate levels (Nelson et al., 2019). IP is a strong feeling of fraudulence and an inability to recognize one’s own success as a result of competence or talent (Clance and Imes, 1978). It is theorized by some that IP is more prevalent in females, and that this may explain the lack of women at higher levels in the research world (Tao and Gloria, 2019). It is also thought that IP may be more common in academically gifted students (Lee et al., 2020), and that there may be a link between IP and academic achievement.

Fear of failure is closely associated with IP and often cited as one of its root causes (Nelson et al. 2019). While IP and fear of failure has been intensively studied at the university level, studies at younger ages are less common. Sweden’s education system is unique from, for example, the United States’ in that students are asked to choose a direction of study at an earlier age (grade 9), before entering upper secondary school.

The object of my study will be to explore how common fear of failure is in Swedish grade 9 students, and how it therefore may affect students’ decisions to study a STEM program in upper secondary school. I will also explore whether the phenomenon is more prevalent in female students, or high-achieving students. If fear of failure does begin at an early age, it could lead certain students to already opt out of STEM subjects before upper secondary school.

2 Background

2.1 Imposter phenomenon

The imposter phenomenon (IP) was first described in a clinical study by Clance and Imes in 1978, in a group of high achieving women who maintained that they were not intelligent despite having achieved advanced degrees, academic honors and professional status. IP can be defined as a powerful feeling of fraudulence due to an inability to internalize one’s own sense of competency or talent (Clance and Imes, 1978). Individuals with IP therefore often attribute their success to external factors such as luck, charm, networking, error, or because they fill a certain quota (Chakraverty, 2018). They feel that others mistakenly see them as competent and therefore feel they must continually perform at a high level to avoid being “found out” (Clance and Imes, 1978). IP can lead to feelings of inadequacy, fear of being

“discovered”, and anxiety (Clance and Imes, 1978; Langford and Clance, 1993). IP has also been significantly linked to depression (McGregor, Gee and Posey, 2008;

Sonnak and Towell, 2001) and to feelings of shame (Cowman and Ferrari, 2002). In other words, it can be detrimental to mental health and should be taken seriously. A consistent positive correlation between IP and introversion has also been seen, and this tendency to shy away from the world may give rise to feelings that one is not seen for who one really is (Langford and Clance, 1993). Introverts more often describe themselves as shy, anxious and low in self-confidence, as do those with IP (Langford and Clance, 1993).

Imposterism has been linked to low academic self-concept (self-esteem) (Cokley et al., 2015), but has been found to be a distinct concept from self-esteem (Langford and Clance, 1993; Cromwell, 1989). Repeated success does not weaken individuals

(8)

3(35)

experience of IP, but rather many students with IP find themselves locked in what Clance calls the “Imposter Cycle” (Clance and Imes, 1978). When faced with an achievement-related task, individuals experience intense anxiety, self-doubt and worry which leads to extreme over-preparation. When this leads to good results, the individuals feel temporary relief and elation, which reinforces the behavior (Clance and Imes, 1978). Over time these individuals come to falsely believe that a break in this cycle would lead to certain failure. It is therefore thought that IP could be especially prevalent in high-achieving students and may detract from their quality of life (Cromwell, 1989). In emerging adults, IP also seems to be most prevalent in individuals who are highly motivated and perfectionistic, and who have trouble self- validating and therefore need external praise (Lane, 2015), characteristics which one could expect to see more often in high-achieving students. Lee et al. (2020) have similarly seen an empirical link between perfectionism and IP, and found that it is more common in honors students than non-honors students at the college level.

2.2 Fear of failure

Fear of failure has been theorized as the main motive underlying IP in most individuals (Clance and O’Toole, 1988), and empirically has proven to be strongly correlated to IP (Fried-Buchalter, 1992; Nelson et al. 2019; Jöstl et al., 2012).

Individuals with this “terror of failure” are “very afraid of the shame and humiliation associated with ‘looking foolish’.” (Clance and O’Toole, 1988). Shame can be described as a “devastating emotion”, in which one experiences a desire to

“disappear”, and in which an individual feels that one’s entire self is defective and that these deficiencies are being judged by a real or imaginary audience (McGregor and Elliot, 2005). Shame has been empirically identified as the core emotion behind fear of failure, and individuals with high fear of failure reported experiencing more shame upon failure than individuals with a low fear of failure (McGregor and Elliot, 2005).

When faced with a project, fear of failure leads to intense anxiety and can cause individuals to procrastinate or over-prepare (Clance and Imes, 1978) or avoid the project altogether to prevent feelings of shame (Nelson et al., 2019; Cowman and Ferrari, 2002). This avoidance behavior can range from small activities, such as not answering questions in class out of fear of sounding unintelligent, to major life decisions, such as whether to pursue a certain career path (Lane, 2015). Individuals with IP will set goals that are far below their actual abilities to avoid the negative feelings associated with IP and failure (King and Cooley, 1995), or may even self- handicap in order to blame their lack of success on an external factor (Cowman and Ferrari, 2002). A negative correlation has also been seen between fear of failure and self-efficacy, which has been identified as a predictor for career choice (Nelson et al., 2019).

The Dweck model states that people’s achievement-related behavior falls into two categories: those that believe intelligence is malleable and those that believe it is fixed (Elliott and Dweck, 1988). Individuals in the first category are more motivated by

“learning goals” (aim to increase knowledge) and react to failure in a resilient way, with renewed effort and without any feelings of inadequacy (Elliott and Dweck, 1988). On the other hand, individuals who see intelligence as a fixed entity are motivated by “performance goals” (aim to prove one’s intelligence) and upon failure

(9)

4(35)

react with feelings of shame and inadequacy, blame themselves, and withdraw from the task (Elliott and Dweck, 1988). A strong correlation has been seen between this

“helpless reaction” when faced with failure and imposter feelings (Langford and Clance, 1993). Individuals motivated by learning goals aim to master a certain task (“task goals”), while individuals with performance goals aim to outperform others, which means that they may see high effort as a sign of low ability, if others can achieve the same performance with less effort (Kumar and Jagacinski, 2006). In this way even successes can be seen as failures, if they require too much effort and preparation. Fear of failure can lead individuals to be motivated by “ability-avoid goals”, where they try to avoid situations where they can fail relative to others (Kumar and Jagacinski, 2006).

In the current study, fear of failure is measured as a proxy for IP for several reasons.

Two different tools have been constructed for measuring IP: the Harvey IP scale (Harvey, 1981) and the Clance Imposter Scale (Clance and O’Toole, 1988). The Harvey IP scale is a 14-item survey that has been used in the majority of studies on IP in adolescents, but it has faced some criticism for poor construct validity (Hellman and Caselman, 2004), and has been found to be less sensitive that the Clance Imposter Scale (Holmes et al., 1993). The imposter phenomenon is most often described in people who have reached some level of achievement, such as a certain high-level career or degree, and the Clance Imposter Scale includes questions about whether individuals feel they deserve these achievements. Children in grade 9 may not yet have experienced such situations with IP per say, but they may still have a fear of failure that will lead to imposter feelings later in their academic careers, or that already manifests itself in avoidance behavior. In contrast to these IP scales, a measure for fear of failure with high construct validity, and that does not assume a certain level of achievement, already exists (The Performance Failure Appraisal Inventory; Conroy, Willow & Metzler, 2002).

2.3 Role of gender in the imposter phenomenon

When Clance and Imes first described IP in 1978, they thought it was unique to high- achieving women. They believed that sex-role socialization could be a cause, as women grew up learning that they are less intelligent or competent than men (1978).

This meant that men would be more likely to attribute their success to ability, while women would be more likely to attribute success to other factors such as effort (Clance and Imes, 1978). It was later found that IP occurs in men as well, and now research as to whether it is more common in women is contradictory with many studies finding equivalence between genders (e.g. Blondeau and Awad, 2018;

Caselman, Self and Self, 2006; Cromwell, 1989), and others finding that it is more prevalent in females (e.g. McGregor, Gee and Posey, 2008; Kumar and Jagacinski, 2006, King and Cooley, 1995; Jöstl et al., 2012), and one even early study finding higher occurrence in men among university faculty (Topping and Kimmel, 1985).

Even when the prevalence of IP is equal between genders, it may affect the genders differently. Based on their work with highly successful women in a clinical setting, Clance and O’Toole (1988) suggested that regardless of how prevalent IP is in men versus women, women are more powerfully limited by IP, because they do not receive the same support from mentors and society in general as men do. Cokley et al. (2015) found that both men and women who believed that they are judged based on their

(10)

5(35)

gender rather than on performance alone (higher Gender stigma consciousness) are at a higher risk of experiencing IP (Cokley et al., 2015). This relationship was stronger in women, who were more likely to be concerned that gender stereotypes are used to judge their competence (Cokley et al., 2015). If these stereotypes about women’s competence in a certain field are internalized, it can lead to greater imposter feelings and may make women more likely to pursue gender-typed fields (Cokley et al., 2015).

The authors observed that race and socioeconomic status could also play into this relationship, although the interaction between race and gender was not tested.

King and Cooley (1995) found a significant relationship between grade point average (GPA) and IP, as well as in the number of hours spent on academic endeavors and IP.

When looking at genders separately, however, these relationships were only significant in women (King and Cooley, 1995). Cokley et al. (2015) similarly found that although levels of achievement between genders in undergraduate students doesn’t differ significantly, the link between IP and GPA was significant in women but not in men. IP drives women to work harder, which in turn leads to a higher GPA (Cokley et al., 2015).

Women are also more likely to attribute their success to effort, whereas men attribute it to their ability (Cokley et al., 2015). For example, if a female student scores well on a math test she is more likely to attribute her result to hard work studying and preparing for the exam, while a male counterpart is more likely to say that he scored high because he is “good at math” or intelligent. Interestingly, in interviews with teachers and study counselors, Perander, Londen and Holm (2020) found that they also most often attributed female students’ success to effort and male students’

success to ability, suggesting that students may be internalizing these gender biases from the adults around them. When boys failed it was often attributed to “laziness”

and a lack of effort, while girls’ conscientiousness with regards to their studies was seen as signs of uncertainty and low self-belief (Perander, Londen and Holm, 2020).

By pointing out that girls work hard to get high grades while boys achieve them effortlessly, teachers are promoting the idea that effort is a compensation for lack of ability, and therefore may be strengthening IP in female students (Perander, Londen and Holm, 2020).

Kumar and Jagacinski (2006) found that IP was more prevalent in women than in men, and also that IP related to Dweck’s achievement goals in different ways between the genders. Both men and women with IP were motivated by fear of failure and

“ability-avoid goals”, but in women IP was also negatively correlated to “task goals”, which means that success in individual tasks did not produce feelings of competence in women with imposterism (Kumar and Jagacinski, 2006). This suggests that it could be harder to counter-act imposter feelings in women, as Clance and O’Toole suggested. Women with IP were also more likely to see intelligence as a fixed entity (Kumar and Jagacinski, 2006), and are therefore more vulnerable to the helpless response described by Dweck (Elliott and Dweck, 1988).

According to the Swedish curriculum for the compulsory school, “The inviolability of human life, individual freedom and integrity, the equal value of all people, equality between women and men, and solidarity between people are the values that the education should represent and impart.” (Skolverket, 2018, p. 5). One of the goals of the school is to promote equality between genders, and the curriculum states that,

“The school also has a responsibility to combat gender patterns that limit the pupils’

(11)

6(35)

learning, choices and development. How the school organizes education, how pupils are treated and what demands and expectations are made of them all contribute to shaping their perceptions of what is female and what is male.” (Skolverket, 2018, p.

7). If IP or fear of failure differs between genders, it is therefore part of Swedish educators’ directive to actively counteract this trend.

2.4 Imposter phenomenon in adolescents

From early on, researchers have postulated that IP may have its root in early family dynamics (Clance and Imes, 1978; Langford and Clance, 1993) and this has found some support empirically (Li, Hughes, and Thu, 2014; Sonnak and Towell, 2001).

Family elements that may foster IP in children could be: (1) that children believe their talents are atypical for their family, background, gender or race, (2) children receive inconsistent feedback from their teachers and family members, (3) family members don’t recognize or praise the children’s achievements and talents, or (4) parents convey to their children that it is important to be intelligent or successful with little effort (King and Cooley, 1995). It is also thought that families in which there is much conflict and a lack of active support for children could lead to an excessive concern with getting praise from others and a fear of criticism (Langford and Clance, 1993).

Families in which the parent and child role are reversed could also lead to IP (Castro, Jones and Mirsalimi, 2004).

While a wealth of studies on IP at the university level for a number of different disciplines exists (as reviewed in Nelson et al., 2019 and Chakraverty, 2018), studies of IP in younger individuals are much rarer. When Chakraverty (2018) asked STEM graduate students and post-docs to describe a time they felt like an imposter, a small number (n=6) mentioned already feeling like imposters in high school, but the strong majority (n=90) described events linked to their PhD training. Only a few studies have looked at IP specifically in adolescents. Both Cromwell (1989) and Caselman, Self and Self (2006) found that levels of IP in high school students mirrored those found in young adults, although Cromwell only surveyed honors English students.

Caselman, Self and Self (2006) and Cromwell (1989) found no difference in IP scores between genders in high school students. However, Caselman, Self and Self (2006) did observe a difference in self-perceptions and social support deficits between the genders, where females were more dependent on support from parents, teachers and classmates to buffer them from IP. Cromwell (1989) also saw no significant correlation between grade point average and IP. Parker et al. (2005) on the other hand, did observe a positive correlation between GPA and IP in high school juniors and seniors.

A German study of secondary school students in the fifth grade found that fear of failure negatively impacted performance in math for girls but not for boys, and not for performance in German language (Wach et al., 2015). Girls also had lower self- efficacy in math, which it was suggested could stem from gender stereotypes (Wach et al., 2015).

2.5 STEM

STEM is an abbreviation commonly used to encompass all fields related to science, technology, engineering and math. In Sweden the abbreviation NO (nature-oriented)

(12)

7(35)

is often used for that which we in English call “science”, i.e. biology, chemistry, and physics.

2.5.1 The STEM gender gap

A well-established gender gap exists in STEM, where women are underrepresented in the number of STEM degrees and in the science and engineering workforce (White and Massiha, 2016). Many different factors have been suggested to explain the attrition of women in STEM. Women exhibit a greater lack of self-confidence, even when their performance is equal or superior to males, and women more often internalize failures and credit others with successes, while the opposite is seen in men (White and Massiha, 2016). Assouline et al. (2006) found that in gifted school children (grades 3-11) a higher percentage of girls attributed success in science and mathematics to effort, while boys attributed it to innate ability. Boruchovitch (2004) found that in students aged 8-16 years old, girls more often attributed failure on mathematics exams to task difficulty, while boys more often attributed it to external factors, such as not being liked by their teachers or being nervous, though overall effort or lack of effort were the most important attributions for success or failure, respectively.

It is also thought that gender stereotypes may play a role in establishing the STEM gender gap. Gender stereotypes surrounding STEM have deep historical roots and likely are founded in early school systems where women were trained to be housewives or low skilled workers and men were taught to be scientists, mathematicians and engineers (Collins, Joseph and Ford, 2019).The fact that many STEM fields are traditionally white male-dominated fields may also attribute to

“chilly STEM climates” which discourage a sense of belonging for other groups (Collins, Joseph and Ford, 2019; Tao and Gloria, 2019). Gender and race stereotypes are also perpetuated by educators whose evaluation of students is often subconsciously colored by preconceived expectations, for example that males are more gifted in mathematics and females in language arts (Collins, Joseph and Ford, 2019). Research has also shown that parents are more likely to attribute a male child’s competence in mathematics to natural talent and a female child’s to effort, and that this may influence the child’s self-perceptions (King and Cooley, 1995). It has been found that men and women who are more aware of gender stereotypes (higher gender stigma consciousness) experience higher levels of IP, which may lead to differences in career choice, especially for women (Cokley et al., 2015). In addition to lower retention rates, females also seem to get lower grades than males in undergraduate STEM courses, which may lead them to leave STEM programs (Ballen, Salehi and Cotner, 2017). This difference can be explained by a difference in test anxiety, which in itself may be rooted in stereotype threat (Ballen, Salehi and Cotner, 2017).

2.5.2 STEM recruitment in Swedish upper secondary schools

In a longitudinal study of Swedish middle school students, Lindahl (2003) explored students’ attitudes towards STEM subjects and how it affected their choice of upper- secondary school program. Interestingly, Lindahl found that many students already had an idea of what they wanted to do in grade 5, and these future plans did not often change throughout middle school (2003). Even among students who received good grades in mathematics and science, showed conceptual understanding of these

(13)

8(35)

subjects, or reported an interest in NO, the majority chose a social science program for upper secondary school, perhaps because their interest in these subjects was even greater (Lindahl, 2003). This was before the latest curriculum reform, so the choice of programs was slightly different than in the current study. Students’ fear of failure was not explored.

About 75% of Swedish students in their final year of upper secondary school preparatory programs in 2019 had plans to continue studying at a college or university within the next three years (SCB, 2020). This figure was higher for women (ca. 70%) than for men (ca. 50%) (SCB, 2020). However, a difference between genders was seen in which majors were most commonly chosen, with women more often choosing social sciences and healthcare, and men more often choosing technical and IT fields (SCB, 2020). The male dominance in technical majors is not surprising considering that the large majority of students choosing the technology program at the upper secondary level are also male (Skolverket, 2020a). In order to increase the number of females in the technical fields, which are likely to be in high demand in coming years, the change must therefore occur already in their choice of upper secondary program.

2.5.3 Imposter phenomenon and fear of failure in STEM

Maltese, Melki, and Wiebke (2014) studied how STEM interest was generated and maintained at different levels of schooling. They found that for both STEM and non- STEM students, interest was most often sparked at an early age, in elementary school.

The most important factor for determining whether interest in STEM persisted in middle and high school students was own passion or interest in the field (20% of individuals in middle school, 22% in high school), but was closely followed by good grades (17% middle- and 15% high school). Students who received an A in their first STEM course in college were also more likely to complete a STEM major than those who received B, C, D or F (Maltese, Melki and Wiebke, 2014). Of all college students surveyed who had a non-STEM major, 43% had entered college with the original intention of studying STEM, but then left those fields (Maltese, Melki and Wiebke, 2014).

One could argue that researchers in STEM fields must deal with a high level of failure in their everyday work in the form of failed experiments and designs (Simpson and Maltese, 2017). Interviews with STEM professionals show that their experiences with failure have shaped their outlook on failure and their career trajectory (Simpson and Maltese, 2017). IP and fear of failure have therefore been suggested as a reason for why many individuals, and particularly women, choose to leave STEM fields (Nelson et al., 2019). For example, Ivie and Ephraim (2011) found that IP was more prevalent in female astronomy graduate students than males, and that this could explain the higher number of females leaving the field.

Nelson et al. (2019) found that fear of failure is a significant contributor to university students’ decision not to major in a STEM discipline. As fear of failure grows in STEM students, their self-efficacy disintegrates, which causes them to doubt their own abilities in relation to STEM tasks; this in turn causes them to reconsider further education or a career in STEM (Nelson et al., 2019). Females in particular have previously been found to have a significant fear of not being valued in the field of engineering, and to have lower feelings of self-efficacy in regard to mathematics than males, which could factor into their decision to study science or mathematics (as

(14)

9(35)

reviewed in Nelson et al., 2019). Jöstl et al. (2012) similarly found that one third of Austrian doctoral students surveyed suffered from moderate to strong IP, which negatively affected research self-efficacy, especially for women, an important indicator for success in university careers.

Tao and Gloria (2019) studied a group of 224 women completing a STEM doctoral degree and found that higher levels of IP were linked to a lower chance of finishing their PhD degree. This effect was mediated by the subjects’ self-efficacy or perceptions of the doctoral environment, so that women with lower self-efficacy or in an environment that they perceived to be male dominated were more likely to suffer from IP and to not complete their degrees (Tao and Gloria, 2019). Women with female mentors and peers were more likely to persist in STEM (Tao and Gloria, 2019). The study looked solely at women.

Surprisingly, Blondeau and Awad (2018) found that while IP was equally prevalent in male and female STEM undergraduates, suffering from IP was more likely to cause men to choose a field outside of STEM after graduation than women. The study surveyed undergraduates who were in math-intensive majors, and the authors suggested that the observed pattern could be because women may experience IP from a younger age and would have either chosen a different major already before entering university or will have developed coping strategies that allow them to continue in the field despite their imposter feelings and possible negative mental health affects (Blondeau and Awad, 2018).

The majority of studies on IP and STEM attrition have been conducted at the university or post-doctoral level and in the United States. Sweden’s education system is unique from the United States’ in that students are asked to choose a direction of study at an earlier age, before upper secondary school. It is therefore interesting to explore whether the same dynamic that leads university students to avoid STEM majors in the USA exists in Swedish adolescents facing the decision of upper secondary school program at a younger age. Does their fear of failure in certain subjects lead to avoidance behavior?

2.6 Aim and Research Questions

The aim of the current study is to explore whether fear of failure is contributing to the STEM recruitment problem in Sweden by affecting grade 9 student’s choice of upper-secondary program, and whether this effect is stronger in female and/or high achieving students.

The study aims to answer the following questions:

(1) How prevalent is fear of failure in grade 9 students in Sweden?

(2) How does the prevalence of fear of failure differ between male and female students?

(3) How does the prevalence of fear of failure differ between students with differing levels of achievement?

(4) How do students’ fear of failure differ in relation to STEM subjects and other subjects?

(5) Are students with a higher fear of failure less likely to pick a STEM program for their continued education at the secondary school level?

(15)

10(35)

3 Methods

This study was mostly quantitative in nature and aimed to find statistically significant differences in a representative sample of Swedish grade 9 students. Participants in grade 9 were recruited to take a self-report survey that included questions about fear of failure, perception of different subjects and choice of upper secondary program.

3.1 Participants

Grade 9 students were found by contacting schools throughout Sweden. In total 668 different schools were contacted. A small fraction was found through Facebook groups, but the large majority were contacted via e-mail using information available on the Swedish National Agency for Education’s webpage (Skolverket, 2020b).

Schools were contacted based on alphabetical order (filter: schools with grade 9, beginning with A-L), so that they were geographically random. In total 27 different schools participated, and their locations were noted (Fig. 2). Although sampling was not intentionally stratified, the higher representation of schools in the south of Sweden and near big cities is loosely representative of how Sweden’s population is distributed, and both more urban and rural municipalities were included. All types of schools were included, regardless of whether they were public or private, or had a special “profile”.

The only exception was International Baccalaureate (IB) schools, which were excluded due to the fact that it is assumed that the majority of IB students would also pick an IB program for upper secondary school, and these programs were not included in the survey. Since information about school or location was not collected in the survey itself, it is possible that teachers in other locations throughout Sweden gave out the survey without notifying me. Teachers were asked to give out the survey during class to increase the likelihood that all types of students would answer and not just the high achieving ones. Participants were asked to anonymously complete a short online survey (see Appendix).

Response frequency among students who were given the survey was 96.4%, with the remaining students declining to participate. In total the sample analyzed consisted of 1045 grade 9 students, of which 53.9% were female (n=563), 44.0% were male (n=

460), and 2.1% were other (n=22).

(16)

11(35) Figure 2. Map of school locations. A total of 27 different schools participated from 26 different municipalities in Sweden. Municipalities from south to north: Kävlinge, Eslöv, Tollarp, Karlskrona, Kalmar, Sävsjö, Härryda, Gothenburg, Borås, Bräcke, Orust, Norrköping, Oxelösund, Ekerö, Stockholm, Solna, Upplands-Bro, Vallentuna, Västerås, Enköping, Håbo, Gävle, Östersund, Luleå, Boden, and Pajala.

(17)

12(35)

3.2 Measures

3.2.1 Fear of Failure

In the current study a Swedish-translated version of the Performance Failure Appraisal Inventory (PFAI; Conroy, Willow and Metzler, 2002) was employed. The PFAI is a survey that has been thoroughly tested and construct validity, cross-validity, and external validity when compared to similar constructs is considered strong (Conroy, Willow and Metzler, 2002), as well as temporal stability (Conroy, Metzler and Hofer, 2003).

The PFAI is based on the cognitive—motivational—relational theory and asserts that for fear of failure to present itself two processes must occur: (1) the individual must feel that failure is possible and (2) the individual must feel that failure will lead to negative consequences. The PFAI is a tool that measures individuals’ beliefs in specific negative consequences of failing, which fall into five categories: (a) experiencing shame and embarrassment, (b) devaluing one’s self-estimate, (c) having an uncertain future, (d) having important others lose interest, and (e) upsetting important others (Conroy, Willow and Metzler, 2002). In the current study the short form of the survey (PFAI-S) was employed, which includes one item from each of these five categories (Conroy, Willow and Metzler, 2002), in order to minimize survey length and increase participation. These five items are listed in Table 1. The PFAI-S has empirically shown no significant difference in validity from the long form (Conroy, Willow and Metzler, 2002; Conroy, Metzler and Hofer, 2003). Each item on the PFAI begins with either the phrase “When I am failing...” or “When I am not succeeding...” followed by a perceived negative consequence, to which participants respond on a Likert-type scale. In the current study, participants had the options from 1 (Don’t agree at all) to 5 (Agree totally), which was later converted to the -2 to +2 scale employed by Conroy, Willow and Metzler (2002) for analysis. The hierarchal nature of the survey allowed each of these five categories to be analyzed separately, as well as summed together as a measure of generalized fear of failure (GFOF). No cut off for what constitutes “high fear of failure” has been previously defined, so the measure is most useful in finding relative levels between individuals or groups.

Table 1. The five specific negative consequences of failure that individuals may fear, and the five items from the PFAI-S corresponding to these categories, which were employed in the current study.

Fear of... PFAI-S Item

(a) Shame and embarrassment When I am failing, I worry about what others think about me.

(b) Devaluing self-estimate When I am failing, I am afraid that I might not have enough talent or intelligence.

(c) Uncertain Future When I am failing, it upsets my “plan” for the future.

(d) Important others losing interest When I am not succeeding, people are less interested in me.

(e) Upsetting important others When I am failing, important others are disappointed.

(18)

13(35)

In addition to the five items from the PFAI-S, one additional item about fear of failure was added to the survey: “I get stressed when I am faced with something new, because I am afraid that I will fail.”. This item was added in order to gauge whether fear of failure leads to anxiety or helplessness in new situations for the students, but was not included in calculations of GFOF.

3.2.2 Gender and Achievement

In order to answer research question (2), “How does the prevalence of fear of failure differ between male and female students?” demographic data was collected about the students’ gender (female, male or other). No other demographic data was collected.

In order to answer research question (3), “How does the prevalence of fear of failure differ between students with differing levels of achievement?”, achievement was measured by students’ own perceptions of their grades. As not all students are aware of their grade point average, students were instead asked to answer, “Which of the following describes your grades in lower secondary school best?”, with the options

“Mostly A”, “Mostly B or C”, “Mostly D or E”, “Mostly F”, or the option to write in another description. Thirteen percent chose to write in a different description, but this did not affect analyses as the main motivation was to separate the highest achieving students. For the purposes of the study, students who chose “Mostly A” were considered to be “high achieving” and all others were “not high achieving”.

3.2.3 Perception of different subjects

In order to answer research question (4), “How do students’ fear of failure differ in relation to STEM subjects and other subjects?”, students’ were asked to pick in which three school subjects it was hardest or easiest for them to succeed, from a list of the 17 most common school subjects.

3.2.4 Choice of upper secondary school program

Upon entering upper secondary school, students in Sweden must choose between 18 different national programs, of which 12 are trade programs and six are preparatory programs for further education at a college or university. The six preparatory programs are: economics (EK), arts (ES), humanities (HU), natural sciences (NA), social sciences (SA), and technology (TE). In order to major in STEM in their continued education, students must choose NA or TE at the upper secondary level, and it is therefore these two programs that are considered STEM programs in the following study.

In order to answer the final research question, “Are students with a higher fear of failure less likely to pick a STEM program for their continued education at the secondary school level?”, students’ perception of different programs was first measured by asking in which upper secondary programs they felt there was a big risk of failure. They were able to pick from one to five (1-5) different programs from the list of 18. Additionally, they were asked to pick in which one of the six preparatory programs there was the highest risk of failure.

Finally, students were asked whether or not (yes/no) they would choose NA or TE programs if they knew they would be accepted into their first choice. The aim here was to gauge interest in NA and TE programs and therefore not to disqualify students

(19)

14(35)

with grades that are lower than what is required for most NA or TE programs. This question was complemented with the open-ended question of “Why/why not?”.

3.2.5 Open-ended comments

At the end of the survey students had the option to add any additional comments, which a small fraction of the students did. Some comments were on the quality of the survey while others were more general thoughts on fear of failure, stress and anxiety in school, and the decision process for upper secondary school program.

3.3 Data Analysis

3.3.1 Statistical Analyses

Statistical analyses were performed using SPSS version 27 (IBM Corp.).

In order to answer the first research question, “How prevalent is fear of failure in grade 9 students in Sweden?”, a Generalized Fear of Failure (GFOF) score was calculated for each individual by calculating the mean of the five items from the PFAI-S. The five-item scale had a good Cronbach’s alpha of a=0.785, indicating that the scale is reliable and has internal validity. Descriptive statistics were generated for each of the six items relating to fear of failure separately, and for GFOF as a whole.

A Shapiro Wilk test for normality tests the null hypothesis that data is normally distributed. It showed that the GFOF data was significantly different from a normal distribution (P= 0.000). The skew of GFOF was 0.585, indicating moderately right- skewed data (normally distributed data has a skew of 0, skew > 1 is considered high).

In smaller datasets, a non-normal (skewed) distribution calls for the use of non- parametric tests in further analyses. However, the size of the dataset in the current study was sufficiently large to allow for parametric analyses despite the skewed nature of the data.

Students’ perceptions of different subjects and upper secondary programs was analyzed using descriptive statistics.

The remaining research questions were analyzed using inferential statistics, which means that you test hypotheses in order to draw conclusions and make predictions for the entire population (in this case 9^th grade students) based on a sample from that population. The different relationships which were tested are summarized in the model in figure 3. Fear of failure could lead directly to an avoidance of STEM programs, or this effect could be mediated by grades or a perceived higher risk of failure in these programs. Gender may have an effect on prevalence of fear of failure or could by linked to achievement through an unmeasured mechanism. The GFOF score was used in the following tests.

(20)

15(35) Figure 3. Model of possible relationships between fear of failure and upper secondary school program choice which were tested.

Due to the low frequency of participants who gave their gender as “other”, only respondents with gender “male” or “female” could be included in tests that had gender as a variable. Chi-square tests were employed to see whether there were any significant relationships between the categorical variables of gender and

achievement. Cramér’s V was used to assess the strength of the correlation between gender and achievement. Cramérs V ranges from a value of 0 to 1, where 0 indicates that the variables are totally independent of one another, and 1 indicates that they are perfectly associated.

Independent-Samples T tests were used to test for a significant effect of gender or achievement on fear of failure, with the null hypotheses that there was no difference between males and females, and no difference between high achieving students (those who answered that they had “mostly A”) and others (all who did not answer

“mostly A”). A two-way ANOVA was also employed to test whether there was an interaction between these two factors and their effect on fear of failure.

A binomial logistic regression was used to test which factors were good predictors for program choice (whether the students would pick NA or TE if given the

opportunity: No: 0; Yes:1). The first model tested included the following variables:

gender (male: 0; female 1), achievement (high achieving? No: 0; Yes :1), GFOF, and perceived risk of failure in NA or TE (Whether they picked at least one of these programs from the list of 18 as high risk of failure. No:0; Yes:1). A second model was then tested with students’ perceptions of mathematics and NO subjects as an additional variable. The following two factors were added: whether students picked math and/or a NO subject as one of their three easiest subjects (No:0; Yes:1), and whether they picked math and/or NO as one of their three hardest subjects (No:0;

Yes:1).

3.3.2 Open-ended comments

After reading through the responses to get an idea of which answers were most common, open ended comments on why or why not students would apply to NA/TE

(21)

16(35)

programs were coded under the following categories: interest (I), difficulty or risk of failure in the program (D), time and effort (E), find subjects easy or know they will succeed (Y), future job opportunities (F), reputation (if the student has heard good or bad things about the program) (R), proof of intelligence (if they get in, it shows that they have good enough grades or are intelligent enough) (P), stress or mental health (S), have good enough grades to get in (G), mathematics (M), intrinsic abilities (or shortcomings) (A), no answer/I don’t know/unable to interpret (N), or other (O). Some responses fit into more than one category.

3.4 Ethical Considerations

As researchers we must always find a balance between our responsibility to inform society and individuals’ right to privacy (Vetenskapsrådet, 2017). There are four main ethical considerations that were kept in mind as the study was conducted:

participants rights to be informed, consensual participation, confidentiality, and how data is utilized (Vetenskapsrådet, 2011). The data was collected anonymously and no identifying information about the participants was collected. Participants had the option to include their e-mail for future interviews, but this information was

immediately separated from the remaining data. No sensitive data, for example about medical conditions or ethnicity, was collected. All participants were fully informed about the purpose of the study and what the results would be used for. The data will not be used for any other purposes than those relayed to the participants.

The participants gave their consent to be included and had the option not to participate.

The confidential nature of this and similar studies can also lead to certain ethical dilemmas. For example, some of the comments received in the survey showed signs of extreme psychological distress, which begs the question at which point, if any, it is OK to break with confidentiality and inform a teacher that they have a student displaying intentions of self-harm. I received one comment that said, “Life has no meaning and is only suffering. There is no point, but we are forced to do it anyways.” In this instance, as no information about school or other identifying characteristics were given, it was not even an option to report the comment to anyone.

4 Results

4.1 Prevalence of Fear of Failure in Grade 9 students

Students were asked to answer the five item PFAI-S with scores ranging from -2 (Do not agree at all) to +2 (Agree totally). The General Fear of Failure (GFOF) for each individual was calculated as the mean of these five items. The GFOF data was significantly different from a normal distribution (P= 0.000) and was moderately skewed to the right (skewness= 0.585) (Fig. 4). The mean GFOF for all students was x̅ = -0.61 with a standard deviation of σ^x= 0.908.

(22)

17(35) Figure 4. (a.) Histogram of GFOF with a normal curve superimposed. The distribution differs significantly from that which is expected in a normal distribution (P=0.000).

All together this suggests that in the average grade nine student, fear of failure is low to non-existent. That being said, 27.4% of students (n=286) had a score of 0 or greater, suggesting that on average they “partially agreed” to each statement, and 6.5% (n=68) had a score of 1 or greater, suggesting they experienced a high degree of fear of failure.

Analyzing the items of the PFAI-S individually gives a clearer picture of which consequences of failure students fear most often (Fig. 5).

(23)

18(35) Figure 5. Distribution of answers for each of the five items in the PFAI-S (a-e) and the additional item of stress upon facing something new (f). Students answered on a scale from -2 (don’t agree) to +2 (agree totally) for each item, as shown along the bottom x-axis

Fear of important others losing interest was the most skewed to the right, suggesting that this fear was least relevant for grade 9 students. However, removing this item from GFOF did not change any of the results in the following analyses. Fear of upsetting important others and fear of experiencing shame and embarrassment were also considerably right-skewed. The fears most prevalent in grade 9 students were fear of having an uncertain future and fear of devaluing one’s self-estimate. In addition to the items from the PFAI-S, students were asked whether they agreed to the statement, “I get stressed when I am faced with something new, because I am afraid that I will fail.”. The answers for this item were much more evenly distributed, suggesting that it measured a different phenomenon than the PFAI-S.

4.1.1 Differences in Fear of Failure between differing Genders and Levels of Achievement

A Chi square found no significant difference between genders in achievement (P=

0.092), and a Cramér’s V of 0.053 indicates close to no correlation between the variables. An independent-samples T test showed that GFOF differed significantly between genders (t= 7.929, df=1021, P=0.000), with females experiencing significantly higher levels of GFOF (Fig. 6).

(24)

19(35) Figure 6. Distributions of GFOF for males (top) versus females (bottom). Females were significantly higher (P= 0.000).

Fear of failure was also significantly higher is students who were high achieving (grades described as “mostly A”) than in others according to an independent- samples T test (t=2.511 df= 1043 P= 0.012). However, when both factors were entered in a two-way ANOVA, only gender was significant (Table 2) and no significant interaction between gender and achievement was found (P=0.197). In other words, the effect of gender did not change based on achievement (Fig. 7), but achievement was no longer significant when the effect of gender was taken into account.

Table 2. Two-way ANOVA analysis of the interaction of gender and achievement and their effects on GFOF. Significant values in bold.

Mean Square

F Significance

Gender 25.865 33.638 .000

Achievement 2.083 2.709 .100

Gender * Achievement 1.280 1.664 .197

(25)

20(35) Figure 7. A two-way ANOVA showed a significant difference between genders regardless of achievement. No significant interaction of gender x achievement was found.

Figure 7 shows that females had a significantly higher level of GFOF than males regardless of their level of achievement.

4.2 Students’ perceived risk of failure in different school subjects and upper secondary programs

All subjects were represented among the students answers for both the hardest and easiest subjects to succeed in for them personally. When asked to pick three school subjects which were hardest to succeed in, the most common answer was mathematics (42.9%), followed by physics (34.3%), and chemistry (29.0%) (Fig. 8).

Figure 8. Percentage of students who picked a subject as “hardest to succeed in” when asked to pick three from a list, organized from highest to lowest along the vertical axis. Shaded bars are subjects which are considered to be STEM subjects in the current study.

(26)

21(35)

When asked which subjects were easiest to succeed in, most students picked physical education and health (42.9%), English (40.6%), and art (28.4%).

Figure 9. Percentage of students who picked a subject as “easiest to succeed in” when asked to pick three from a list, organized from highest to lowest along the vertical axis. Shaded bars are subjects which are considered to be STEM subjects in the current study.

Mathematics was higher up on the list than other STEM subjects, with 18.4% of students answering that math was one of the easiest subjects for them.

When asked to pick which upper secondary school program had the highest risk of failure from a list of all 18 programs (option to pick from 1-5 different programs), the Natural Sciences program was picked by the majority of students (62.5%) (Fig. 10a).

The same held true when asked to pick which single program had the highest risk of failure of the college preparatory programs (Fig. 10b), with 55.8% picking Natural Sciences.

Figure 10. Upper secondary programs that students perceive have the highest risk of failure. (A) Students picked which programs had the highest risk from a list of all 18 programs (picked 1-5 options). Shaded

(27)

22(35) bars are the college preparatory programs, and all others are trade school programs. (B) Programs with the highest perceived risk out of the six college preparatory programs. Students were asked to pick one.

In both cases Economics was the second most common answer. In general, college preparatory programs were deemed to have a higher risk of failure than trade programs, although certain trade school programs were deemed riskier than the arts and humanities programs.

4.3 Are students with higher fear of failure less likely to pick STEM?

In total 34.7% of students said they would pick a STEM program (NA or TE) if they knew that they would get into the program. A binomial logistic regression tests whether adding factors to a model significantly increases the ability of the model to predict the observed outcomes (in this case to predict whether students picked STEM or not). The binomial logistic regression showed that the combination of the factors gender (P=.000), achievement (P=.000) and GFOF (P=.020) were significant predictors for whether students would choose STEM if given the opportunity (Omnibus test: P= 0.000). However, the classification table (Table 3) revealed that the model was very efficient in predicting students that would not pick STEM (95.8%

correct), but not very effective in predicting which students would pick STEM (only 8.6% correct). Whether students thought NA or TE was a high-risk program did not significantly predict whether they would choose to study one of those programs, perhaps because such a large percentage picked these programs as highest risk regardless of their choice in program.

Table 3. Classification Table for the binomial logistic regression based on the variables gender, achievement, GFOF and perceived risk in NA/TE.

When students’ perceptions of math and/or NO subjects was added to the model, it became much more efficient in correctly predicting students that would pick NA/TE (Table 4).

Table 4. Classification Table for the binomial logistic regression based on the variables of gender, achievement, GFOF, perceived risk in NA/TE, and perceived difficulty of math and/or NO.

(28)

23(35)

The parameters added to the model were whether students picked math and/or an NO subject as one of their hardest subjects (Yes/No), and whether they picked math and/or NO as one of their easiest subjects (Yes/No). The model was a significantly good predictor of program choice (Omnibus test, P= .000). The results of the second binomial logistic regression, which more often predicted cases correctly, are summarized in Table 5.

Table 5. Summary of the binomial logistic regression that best predicted whether students would pick NA/TE for their upper secondary program if given the choice. The regression weights (B) and Standard Error (S.E.) used to calculate Wald’s Z value (Wald) are listed, as well as the significance of the Z value. Significant values are marked with an *. Exp(B) is the odds ratio of picking NA/TE.

Parameter (test category) B S.E. Wald Significance Exp(B) Gender (female) -.386 .1516 6.464 .011* .680 Achievement (high) .386 .2258 2.929 .087 1.472

GFOF .217 .0820 7.036 .008* 1.243

Math and/or NO hardest (yes) -.975 .1699 32.919 .000* .377 Math and/or NO easiest (yes) 1.215 .1531 62.960 .000* 3.369 Perceived Risk in NA/TE (yes) .004 .1601 .001 .982 1.004 The value of B shows whether the test category (in parentheses) was more likely (positive value) or less likely (negative value) to pick STEM. For example, the negative B-value for gender shows that females were less likely to pick STEM.

Exp(B) shows the odds ratio of picking to study STEM, so that females were 0.680 times as likely to pick STEM as males. Gender was a significant predictor (P=0.011), with females less likely to pick NA/TE. GFOF was positively correlated to the likelihood to pick NA/TE (P=0.008), so that students with a higher GFOF were actually more likely to pick STEM. The most significant predictors were student’s perceptions of different subjects. Students who picked math and/or NO subjects as one of their hardest subjects were significantly less likely to pick NA/TE (P=0.000).

Students who found math and/or an NO subject easy were more than 3 times as likely to pick NA/TE, a significant difference (P= 0.000). When perception of subjects was added to the model, achievement was no longer found to be a significant predictor.

Perceived risk of failure in NA/TE was still not significant.

5 Discussion

5.1 Is fear of failure a problem in grade 9 students?

If we only look at the mean value for Generalized Fear Of Failure we could conclude that the average grade 9 student does not suffer from more than a low degree of fear of failure, but we would not be getting the full picture. There is no empirically determined cut off for what is considered a high fear of failure, and what level of

(29)

24(35)

prevalence is “too much” is also up for discussion. More than one-fourth of grade 9 students surveyed had a score of at least zero, indicating that they at least partially agreed with the items in the survey. High fear of failure was experienced by 6.8%.

Fear of failure and IP has been linked to mental health problems such as anxiety and depression (McGregor, Gee and Posey, 2008). Considering this, one could argue that even 6.8% of students is too many. The following comment from one student is long, but it paints a good picture of the ramifications a high fear of failure can have on mental health and the motivation to learn:

“Sometimes I ask myself if the point of school is to make you feel bad. Has it always been this way or is our generation more stressed than other generations? In that case, if students have always felt bad about school, why doesn’t anyone do anything about it? From my point of view, the school should talk more about how to deal with anxiety and what you should do to not put such high expectations on yourself. Personally, I have unrealistic expectations for myself and have had a hard time since the third grade. I’ve been to several psychologists and counselors to deal with the anxiety that comes from these high expectations and almost constantly have a feeling of failure. They have helped me with my anxiety attacks, but I still have sky high expectations for myself. It’s a pity because I LOVE to learn new things, and in my free time I listen to science podcasts, prefer documentaries to movies and read as many non-fiction books and fiction. I simply love to learn and want to learn, but school suffocates my curiosity and interest, since there is so much pressure to learn and you get anxiety.”

The student pinpoints one of the main issues, which is that failure can mean different things to different students. For some students, passing classes is not enough, and anything other than the highest grades is seen as failure. This student has a high self- awareness, but other students may not be as self-aware or active in seeking the help they need.

Of the five items from the PFAI-S, fear of devaluing one’s self estimate was the most prevalent among students. The statement for this item was, “When I am failing, I am afraid that I might not have enough talent or intelligence”. It seems that the students to a large degree see success or failure as something that depends on intrinsic attributes, such as intelligence, rather than something external which they have the power to control. This was also apparent in students’ comments on why or why not they would pick STEM as an upper secondary program. Some students said they would not pick NA or TE because, “I’m stupid,” or “I’m awful,” or “I suck.” On the other hand, some students who said they would pick NA or TE, said it was because

“I am smart” or “because only smart people are in that program” and saw their admittance to the program as proof of their intelligence: “If I get in, then I’ll know I have what it takes and don’t need to worry about being too ‘dumb’.” Students who see intelligence as a fixed entity rather than something malleable are more likely to be motivated by “performance goals” and react to failure with feelings of shame and inadequacy (Elliott and Dweck, 1988). Individuals with performance goals aim to outperform others, which means that they may also see high effort as a sign of low ability and as a type of failure, if others can achieve the same performance with less effort (Kumar and Jagacinski, 2006). When faced with failure these individuals are more likely to react helplessly and give up (Langford and Clance, 1993).

A moderate number of students also agreed with the statement, “When I am failing, it upsets my ‘plan’ for the future”, indicating a fear of having an uncertain future. This fear was probably extra salient around this time period when students are in the process of picking an upper secondary program. It was apparent that students in grade

Fear of Failure in Swedish 9th Grade Students and its effect on their decision to study STEM