Introduction

Standardized tests are designed with a goal of being a measure of students’ knowledge and demonstrated academic ability. They aim to collect data on the “quantity” of information learned, determined both by the quality of education and the amount of information retained by students. (Strauss) This should be the single most important factor of student performance, but  there are constantly an abundance of external factors that also affect student performance. These unintentional variables may cause collateral damage to a student’s test scores so that the results do not actually demonstrate the student’s full abilities. In standardized tests such as the SAT that are important for individual achievement, minimizing the impact these factors have is critical to improving overall performance. 

One factor that impacts performance is the confidence a student has going into the test. Confidence is a very abstract concept that furthermore can be affected by a countless number of other deeply rooted metacognitive factors such as positive future outlook, being goal oriented, and being aware of one’s own abilities. (Hagen) Even just some anecdotal evidence such as a claim of last year’s test being “very hard” can sway confidence levels. Unfortunately, a perfect test that eliminates all of these external factors is impossible to administer. It first must be established whether or not confidence levels independent of actual ability have a significant correlation with student performance before a course of action is taken accordingly.

Previous research finds confidence levels generally agree that there is some level of a direct relationship between performance of target action and confidence. They struggle though to “quantify” this relationship, so it is not known how much of an impact confidence levels have compared to other factors. Those studies are designed to find whether or not confidence levels have an impact on performance as opposed to how much of an impact they have. Furthermore, they rarely examine general patterns between confidence levels. The goal with this study is to eliminate as many external factors as possible so only confidence levels are at play during the experiment portion, while still attempting to identify possible patterns that influence confidence levels, and in turn, how confidence levels influence performance. 

Literature Review

Previous studies that seek to find a relationship between confidence and performance have come up with mixed results depending on the area of focus. Even within focus areas, there is contention overall the extent to which confidence affects performance, and studies rarely translate their results into a format that is applicable to a broader situation. Because of this, most past studies are limited to the scope of their study, whether it be the relationship between confidence and a specific sport, writing, memory, etc.. Two important factors have to be quantified in order to analytically determine a definitive relationship. They have been repeated and are even included in the title: confidence and performance. 

First establishing a precedent for what confidence, also known as self efficacy, is defined as is important to interpret the study results. Two overarching factors affect confidence levels: protective and vulnerability factors. Protective factors, also known as resilience, are factors that “buffer against negative effects of adverse life events.” Vulnerability factors are factors that lower confidence levels, such as “depression, social anxiety, eating disorder, and substance use.” These two factors are what pre-determines confidence levels. (Hagen) These factors will be important to determine in order to identify the background factors at play that determine preliminary student confidence levels. 

The first source I discovered presents the overarching theory that my hypothesis will be based on: the social cognitive theory. The social cognitive theory has many facets, but I will be specifically focusing on the self-efficacy component of the theory. Self-efficacy is defined by Bandura to be based on individual feelings of whether or not students believe they have the capabilities to succeed, in other words, the students’ confidence in themself. Of the 4 components of the social cognitive theory, self efficacy is stated to be a key element because it heavily influences the other components, such as motivation. In simpler terms, the social cognitive theory hypothesizes that self efficacy influences the overall performance of a student multidimensionally.

Studies that look for a relationship between academic performance and confidence levels generally agree that confidence levels have a direct relationship with academic performance. Within academic performance though, there is a breakdown of even more subsections. The main ways academic performance is measured are via either objective or subjective measures. Objective tests, primarily multiple choice tests (which standardized tests are) eliminate the subjectivity and make it possible to measure performance quantitatively. Studies in the past, such as one by Pajeres that examined confidence levels and writing improvement over time fail to quantify performance in a meaningful way. This study created even more inconsistencies due to the other many variables introduced. Pajere’s study was designed so that the teachers involved were focused on increasing the students’ abilities to respond to many different types of writing tasks. This means that the writing prompts and subjective teacher grading were inconsistent variables which affected the results. Pajeres admits that “assessing an individual’s writing is not an objective task. It involves an inference by the reader about the quality of the written work.”  (Pajeres) Furthermore, the design of the study failed to eliminate one of the most important external variables: actual increase in student ability. Because of the practice over the course of the 16 week period, the student’s writing skills may have improved, instead of an increase in confidence which produced an improvement. The results of the study correlated writing apprehension with negative self writing confidence. Pajeres notes that another reason for low self confidence is poor teacher feedback. Students that are told they are bad at certain things tend to have lower self confidence, and have increased apprehension. Surprisingly, there was no correlation drawn between self-writing confidence and performance, although a tentative relationship was drawn. (Pajeres) These results identify probable causes of low academic confidence and possible reforms to the educational system. But, the inconclusiveness of the study states that there is no correlation or causation between confidence and writing performance.

Studies involving motor-based activities such as sports and confidence levels have a similar dilemma as Pajeres. Stranger identifies the impact low confidence has on a player during a sports game: task irrelevant thoughts and performance worries. This means that anxiety and dejection positively resulted in cognitive interference which meant poorer performance. Similarly, reappraisal and confidence resulted in better performance. (Stranger) It was unfortunately not measured to what extent this was the case though, so yet again, there is no quantification for how much this factor affects performance compared to other factors. On the contrary though, a study by Christine conducted at Guilford College found that competitive anxiety was not related to sports performance. But, it was negatively related to confidence, which was positively related to performance. (Christine) This extra variable of competitive anxiety differs from the standpoint of academics. This was not relevant since academics are not supposed to be inherently competitive, whereas sports are. Christine still finds though that confidence is positively related to performance, which has been the consensus thus far across the board. 

One notable study finds that there is a relationship between gender, learning approach, confidence levels, and academic performance. Fuentes surprisingly found that the highest confidence groups actually had the lowest GPA’s. Interestingly enough, male students also had lower GPA’s and had a more “surface level” approach to studying, where they enjoyed it less. There is no relationship between study method and confidence levels, since it was too inconsistent to be determined. (Fuentes) The usage of GPA to measure academic performance by Fuentes is unique so far, and for the first time provides a quantitative analysis. Furthermore, this study also examines the most important external factor that influences performance: learning approach. Unfortunately, it doesn’t actually isolate this variable from confidence. A combination of confidence and learning approach determine the GPA. Because of this, the quantitative conclusion that Fuentes drew is still somewhat skewed. Nonetheless, it has been the most objective so far, and has attempted to address the most variables instead of outright ignoring them. 

Previous studies that seek to find a relationship between confidence and performance have come up with mixed results depending on the area of focus. Even within focus areas, there is contention overall the extent to which confidence affects performance, and studies rarely translate their results into a format that is applicable to a broader situation. Because of this, most past studies are limited to the scope of their study, whether it be the relationship between confidence and a specific sport, writing, memory, etc.. Existing research fails to isolate confidence as the only variable in the study acting significantly on performance. I was curious about just the effects of confidence on performance and not other factors such as natural improvement over time, which led me to design a study where the experimental groups are split based on confidence levels. 

Methods

In order to quantify the effect of confidence on performance, the data gathering instrument had three different sections. All were quantitative and the results were measured numerically in order to prevent any inconsistencies with judging the two main factors of confidence and performance. This makes an objective quantitative analysis possible, since all of the results are measured on the same scale in the same way. 

The whole experiment consisted of an online test, as well as Likert scale based questions that were administered to a class of English students who were all in the same age range and in the same difficulty level English class. This ensured that confounding variables such as ability levels and experience were minimized. Students were also all assigned a unique number so they could maintain anonymity for their self evaluation and test score. 

The first section of the online questionnaire was a series of questions that utilized the Likert scale. The Likert scale is a commonly employed psychometric scaling method that assesses attitudes and views of the participant. These questions with the Likert scaleasked students to evaluate aspects of their testing experience on a scale from 1-5, with 1 being “Strongly Disagree” and 5 being “Strongly Agree”. The questions covered a large scope; possible causes for success and past performance on standardized tests were all accounted for. The answer to these self-evaluation questions become the baseline to measure how accurate a student's performance was to what they believed it to be. Furthermore, it could provide insight on possible causes of low confidence. The numbering system was utilized to both anonymize the test and sort students into the three groups. Numbers were given completely randomly, and nobody except the students themselves knew who was what number. 

The questions in the pre-questionnaire that students were asked to answer on the likert scale were:

I perform well on standardized tests as a whole.
I have studied extensively for standardized tests in the past such as the SAT.
I am anxious before standardized tests because I am uncertain about my abilities.
I generally perform better in topics I am more comfortable in.
I am confident with my standardized test math section abilities.
I am confident with my standardized test English section abilities.

The next section of the questionnaire was a standardized test with difficulty 1 questions selected from the SAT. Difficulty 1 questions were intentionally selected in hopes that confidence levels would be measured, as opposed to academic prowess. The previous Likert scale questions made students hyper aware of their own confidence with various sections, which should increase the impact their confidence levels have when they take the standardized test portion. Based on their assigned number, the students were then separated into three groups. One group took their more confident section (between the writing and math sections), one group took their less confident section, and the control group took both sections. The selections were based on their randomly assigned numbers, and not any previous questions or answers. What did correspond to their previous answer is what section the first two groups did. The group that did the section they were more confident in should have done the section they selected with a higher degree of confidence on the Likert scale pre-test. For example, if a student answered “4” for confidence on math, and “3” for confidence on english, and they were assigned to do their more confident section, then they would do their math section. 

The last section was a post-evaluation with a format very similar to the first section. This section asked students to evaluate their confidence levels after the test, and rate how they feel they performed with respect to their confidence level on the section. 

The questions in the post-questionnaire are as followed:

I still feel nervous because of the test. 
I feel I did poorly on the section I was less confident in.  (Only do this question if your number is 9-16 or 17 and above)
I feel I did poorly on the section I was more confident in. (Only do this question if your number is 1-8 or 17 and above)

The pre and post-evaluations are used to gain more insight into each individual. The last few questions were used as a gauge to see how students felt after taking the test sections. Anxiety levels, past performance, and predicted performance (from the post test) can be compared against their actual results to look for any other notable patterns. Some of the questionnaire results do not reveal any impactful findings though or are repetitive with previous ones, so those will not be discussed. 

Results

Before results are presented, it is important to note that we will be referring to the group that took only the section they were less confident in as the “less confident” group, and vice versa with the “more confident” group. This does NOT mean that the “less confident” group is necessarily less confident when taking standardized tests. In other words, students in the more and less confident group are sorted randomly, completely independent of their confidence levels in respective sections and overall confidence with standardized tests. 

The experiment was run in an English 4 Honors class with 24 high school seniors. Of the 24 seniors, 8 did the section they were more comfortable in, 8 did the section they were less comfortable in, and 8 did both sections as a control group. Students were allotted seven minutes for eight SAT difficulty one problems as mentioned, although many submitted early, some within minutes. 

Once data was collected, a T test based on mean was conducted to compare the first two experimental groups against the control group. The T test seeked any statistically significant difference between each confidence group. A T test was used because it serves as a direct comparison between experimental groups against the control groups since all groups have equal numbers of participants. 

The p-value of the two tail test was 4x10^-8 which corresponds to high significance as seen in the highlighted boxes in the t test above. This means that the score difference between the less confident group which did worse than the more confident group (which were both compared to the control) showed statistical significance in doing so because of the independent variable, which is confidence levels. It is also interesting to note that the hypothesized mean difference (the mean scores for the control group that did both sections) was 93.75, compared to the 92.1875 mean for the more confident group and the 84.375 for the less confident group. It is unknown why the control group actually scored higher than both other groups, but it does not meaningfully affect the results of the t-test. 

“I am anxious before standardized tests because I am uncertain about my abilities.”

“I have studied extensively for standardized tests in the past such as the SAT.”

The online questionnaire portion of the research tool first yielded an interesting trend between self efficacy and motivation levels. The statements that students were asked to answer using the likert scale are verbatim in green below their respective graphs. Self efficacy refers to the left graph, and work ethic refers to the graph on the right. As shown, lower self efficacy did not correspond to lower motivation levels. Instead, participants with lower confidence levels actually stated that they actually studied more. This deviates from the expected result presented by the social cognitive theory. 

“I perform well on standardized tests as a whole.”

Another question from the survey asked about past performance as a “second control.” It is interesting to note that the people that took the section they were less confident in believed that they performed poorly on standardized tests in the past. This group coincidentally also performed worse in general compared to the other groups which means that their past performance could indicate the presence of another factor: testing abilities. This will be discussed more in the limitations section. 

“I generally perform better in topics I am more comfortable in.”

Lastly, students reported that they believed they generally perform better in higher confidence sections. This matches the quantitative results completely, which indicates that the result of “higher confidence leads to better performance” derived from this study is true. 

Discussion

From the extremely low p-value of the two tail tests, it is undeniable that there is a positive correlation between confidence levels and performance. This matches the conventional wisdom and hypothesis that higher confidence leads to better performance. The results also affirm Bandura’s social cognitive theory which states that confidence influences performance multidimensionally. It also matches the general results of both the Pajeres and Christine studies which stated that higher confidence correlates to higher performance in writing and sports respectively. Although Pajeres’s correlation was more tentative, in conjunction with this study’s results they seem to be validated. Therefore, maintaining high confidence will generally benefit the individual and help them with performance. This is further supported by the questionnaire where participants believed that they perform better on higher confidence sections. This means that this result is proven two fold quantitatively both from the experimental results as well as the more subjective (but matching) questionnaire answers. Because of the definitive p-value and corresponding high significance result, this can be accepted without a doubt. 

The second segment of the social cognitive theory is empirically proven to not necessarily hold true, and leaves a lot more speculation. The less confident group answered that they have low self efficacy. Social cognitive theory claims that this would lower motivation, but the online questionnaire indicates that this is not necessarily the case. Although many people indicated they had low self efficacy, people still tended to study a lot, meaning that their motivation and work ethic was not negatively affected by lack of confidence. Instead, lower self confidence seemed to inspire higher levels of motivation. Due to their low levels of self efficacy, students instead looked to more controllable factors such as raw ability in order to compensate and comfort their unease. 

Instead, this result seems to match Fuentes’ results. Her results showed that people with higher confidence levels generally had lower GPA’s. Following the logic that people with lower confidence study more as shown in my questionnaire, it would make sense that people with lower confidence might develop more competence. This means that studying is used as a “coping mechanism” for low confidence individuals and suggests an opposite linkage as proposed by the social cognitive theory. In order to increase their confidence levels, low confidence individuals end up studying more than their counterparts. This could explain why in Fuentes’ study, the conventional wisdom was completely shattered.

Limitations

The most pressing limitation is the extremely small sample size. Only one trial was run with 24 people, which although it produced extremely definitive results according to the T test, is not a large enough sample size. A better sample size would have included at least 2 more classes, with some of them being from different grades as well. Generally, the larger the sample size tested on, the more definitive the results are since sampling bias is no longer a factor. If the current trend holds for a larger sample size, then the positive correlation between confidence levels and performance can be affirmed with higher confidence. 

Other limitations include the possible imbalance in inherent standardized test taking abilities. Recalling the results of the survey, coincidentally, many more of the less confident group participants said they did poorly on standardized tests in the past. This could indicate that the people that were grouped into the less confident group were also coincidentally not as good on standardized tests as the other participants. Instead of their confidence levels being the issue, this could mean that their abilities caused the poorer performance. Nonetheless, the t test correlation result was so high that it’s unlikely that the difference in ability solely caused the results. It is likely though that the ability differential did play some sort of role in the massive difference between the two groups. 

Lastly, it was obvious that the students involved were not very enthusiastic about taking the standardized test without it being counted for anything. Because of this, many rushed through and may not have tried their best which is very likely shown by some of the extremely fast submission times. This may have created some artificially low results by a few points. This most likely did not make too much of an impact though since the averages for all sections was over 80%. This lack of genuine participant effort could be avoided in the future by offering some incentive such as monetary compensation for participant time, although this study lacked the resources to be able to do so. 

Conclusions

In future studies that seek a correlation between self efficacy and resulting performance, a valuable rework to the study would be to ask all participants to take both parts of the study, without the split into 3 groups with a control group serving as the comparison. Then, students would have their more confident section compared to their less confident section. Eliminating as many extraneous variables as possible is critical for this study, and it has been reiterated that this was the goal. This might provide valuable insights since it is comparing everyone to themselves, so individual test taking skills would no longer be a limitation. 

Prior to this study, there has not been a quantitative study that successfully isolates confidence levels as the main or only variable towards success. This study succeeds in doing so by making confidence levels the only independent variable that acts on performance between the experimental groups. Although there are other limitations and possible confounding variables, they play a small role as shown by the definitive p-value, and can be eliminated in future studies that improve upon this methodology. From the results of this study, it is affirmed that confidence is important to perform well, with a very strong correlation. Thus, a takeaway from the results of this study is it is important to keep a positive mindset and be in high spirits instead of being hopeless, since simply this can change the results of something to be more favorable. But, it is important also to not be too overconfident as shown by Fuentes. Being overconfident results in very poor results. The best course of action is to make proper preparations, and once that is done, to be confident in those acquired abilities. For example, teachers can apply these results and help encourage students in classrooms by having very rigorous courses that “over prepare” students so they are confident in their own skills, as well as prepared regardless of confidence levels. This means that a good work ethic and confidence go hand in hand. As work ethic and abilities improve, self confidence will most likely end up improving too, so it is important to focus on abilities so that confidence levels naturally follow. 

Citations

Barnes, Elizabeth Christine. “Examining Trait Sport Confidence, Self-Efficacy, Competitive State Anxiety, and Subjective Sport Performance among Guilford College Student-Athletes.” JSTOR, jstor.org/stable/community.30338282. 

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall

de la Fuente*, J. (2013). Relationship between Undergraduate Student Confidence, Approach to Learning and Academic Performance: The Role of Gender. Redalyc. Retrieved from redalyc.org/pdf/175/17527003009.pdf 

DeSoto, K. A., & Roediger, H. L. (2014). Positive and Negative Correlations Between Confidence and Accuracy for the Same Events in Recognition of Categorized Lists. Psychological Science, 25(3), 781–788. jstor.org/stable/24540139

Hagen R, Havnen A, Hjemdal O, Kennair LEO, Ryum T, Solem S. Protective and Vulnerability Factors in Self-Esteem: The Role of Metacognitions, Brooding, and Resilience. Front Psychol. 2020 Jul 3;11:1447. doi: 10.3389/fpsyg.2020.01447. PMID: 32719640; PMCID: PMC7351531.

Pajares, Frank, and Margaret J. Johnson. “Confidence and Competence in Writing: The Role of Self-Efficacy, Outcome Expectancy, and Apprehension.” Research in the Teaching of English, vol. 28, no. 3, 1994, pp. 313–331. JSTOR, jstor.org/stable/40171341. 

Schunk, D. H. (2003). Self-efficacy for reading and writing: Influence of modeling, goal setting and self-evaluation. Reading and Writing Quarterly: Overcoming Learning Difficulties, 19(2), 159–172.

Stanger, N., Chettle, R., Whittle, J., & Poolton, J. (2018). The Role of Preperformance and In-Game Emotions in Cognitive Interference During Sport Performance: The Moderating Role of Self-Confidence and Reappraisal, The Sport Psychologist, 32(2), 114-124. Retrieved May 1, 2023, from doi.org/10.1123/tsp.2017-0001

Strauss, V. (2014, August 1). What do standardized tests actually test? The Washington Post.