Student health behavior and academic performance

Ethical research statement

The research protocol and its amendment were approved by an ethical review board (Institutional Review Board) at Florida Gulf Coast University (FGCU) prior to data collection (FGCU IRB 2018-17, March 30, 2018). All researchers involved in the study were trained in ethical data collection through the Collaborative Institutional Training Initiative (CITI). Data collection followed all laws relevant to the survey of university student populations.

Data collection

Data were collected over a 2-year period between April 1, 2018 and January 31, 2020 using an anonymous online survey. Students at all colleges at a regional state university in southwest Florida were invited via email to participate in the survey. The first page of the survey consisted of an approved online survey consent form; in other words, consent was obtained. Participation was voluntary and participants did not receive any compensation or extra credit. Some part of the data unrelated to current research purpose was published elsewhere (Reuter, Forster & Brister, 2020).

The survey consisted of five groups of questions around health and wellness, requesting information–among other topics–regarding demographic information, such as gender, age, ethnicity/race, year at school, and current overall grade point average (GPA), eating and sleeping habits, and drugs and alcohol consumption (see Appendix 1). Most of the questions were modeled after questions used in the 2017 Standard High School Youth Risk Behavior Survey (Centers for Disease Control & Prevention, 2017).

Data analyses

For questions with categorical answers, data are presented as a percentage of the total participant pool, or a portion of this pool. For questions with quantitative answers, data are presented as means with standard deviations. Sample sizes are indicated as they vary for different analyses due to the voluntary nature of the survey.

To examine differences in self-reported GPA across 28 different health behaviors and habits, Standard Least Squares Models were used (Restricted Maximum Likelihood Method, REML), with individuals as replicates. GPA differed significantly by cohort (Wilcoxon Rank Sums Test, Chi-square = 18.9057, DF = 3, p = 0.0003) and was therefore included as a random effect in all tests. GPA differed by biological sex (Wilcoxon Rank Sums Test, Chi-square = 4.5894, DF = 1, p = 0.0322) and biological sex of students was therefore included as a random effect in all tests.

Student’s T-Test All Pairwise Comparisons (Least Squares Means) were used as post-hoc tests. The dependent variable in all tests was student self-reported GPA (a continuous numerical variable ranges from 0.00 to 4.00). Independent variables are listed in Table 1 below or can be found on the survey in Appendix 1. Given the number of statistical analyses performed (41 tests) we have used the Benjamini–Hochberg Procedure (with a false discovery rate of 5%) to determine significance (tests with a p-value < 0.0189 are being considered significant using this approach). Statistical analyses were performed using JMP software program Version 15 (JMP^®; SAS Institute Inc., Cary, NC, USA).

Study population

Of 761 students who participated in the online survey, 147 respondents were excluded from analyses for a number of reasons, including: (a) respondents who indicated an age younger than 18 years of age or older than 25 years of age or failed to provide information on their current GPA; (b) part-time students, as the sample sizes was too small (n = 15) and they had a lower GPA than full-time students (Wilcoxon Rank Sums Test, Chi-square = 4.61, DF = 1, p = 0.0317); (c) non-traditional students (graduate, non-degree seeking, and second-degree seeking students); and (d) students who failed to provide information about their biological sex. As such, 614 students remained in the dataset.

The average current GPA for all 614 respondents was 3.39 ± 0.51 (range: 0.40–4.0; median GPA = 3.5). Freshmen reported the highest average GPA (3.45 ± 0.64), followed by seniors (3.45 ± 0.42), juniors (3.36 ± 0.39), and sophomores (3.36 ± 0.48).

The average GPA for female students was 3.42 ± 0.52 compared with 3.34 ± 0.49 for male students.

Health behaviors and self-reported GPA

Differences in self-reported GPA across 28 different health behaviors and habits were examined (Table 1).

Sleep

Average self-reported GPA differed by the number of hours slept on average per night (Fig. 1). Students who slept 4 h or less per night had lower GPAs than students who slept 6 h per night (p-value = 0.0062), 7 h per night (p-value = 0.0025), and 8 h per night (p-value = 0.0004, Student’s T-Test All Pairwise Comparisons). Students who slept 9 h per night, had lower GPAs than students who slept 8 h per night (p = 0.0006, Student’s T-Test All Pairwise Comparisons).

There was no significant difference in the average GPA of students based on when they went to sleep. Nine out of ten respondents went to bed between 10 pm and 2 am (87.3%). Students going to bed between 8 pm and 10 pm (4.1%) had a self-reported GPA of 3.46 (±0.33), students going to bed between 10 pm and midnight (44.1%) reported an average GPA of 3.45 (±0.47), and the average GPA for students who went to bed between midnight and 2 am (43.2%) was 3.37 (±0.53).

Almost equal numbers of students got up before 6 am (8.5%) or liked to sleep in and got up after 10 am on average (8.6%). The rest preferred to get up between 6 am and 8 am (42.0%) or between 8 am and 10 am (40.9%). But, there was no significant difference in the average GPA of students based on when they woke up.

Working

Self-reported GPA for respondents differed depending on whether or not respondents worked, with a negative association between hours worked on average per week and self-reported GPA. Students who did not work/worked zero hours per week had a higher GPA than students who worked 20–30 h per week (p = 0.0188), 30–40 h per week (p = 0.0094), and more than 40 h per week (p = 0.0014). Students who did not work or only up to 10 h per week reported average GPAs of above 3.45, whereas students who worked 10 h or more per week had GPAs of 3.38 or less (Fig. 2).

Physical activity

One-sixth of respondents considered themselves as not physically active at all (17.0%). Their self-reported GPA was 3.28 ± 0.64 as compared to an average GPA of 3.43 ± 0.48 for respondents who were physically active (83.0%). There was a significant difference in the average GPA of students based on the number of days they were physically active in the week prior.

In addition, respondents who engaged in exercises to strengthen or tone their muscles, such as push-ups, sit-ups, or weight lifting, reported higher GPAs than students who did not engage in such exercises.

Eating habits

Consumption habits for vegetables, fruits, fruit juice, or green salad did not change among students with different GPAs. There was also no significant difference in the average GPA of students based on the number of times milk, soda, diet soda, or sports drinks were consumed in the last week either. Nevertheless, the difference in average GPA between respondents who did and those who did not consume energy drinks was significant with a p-value of 0.0084.

Self-reported GPA differed based on the days that students had eaten breakfast in the past seven days (Fig. 3). Students who ate breakfast seven days a week/everyday were more likely to have higher GPAs than students who never ate breakfast (zero days of the last seven), ate breakfast one day per week, ate breakfast 2 days per week, or ate breakfast three days per week (p < 0.0075).

Self-reported GPA differed based on the number of times that students had eaten fast food in the past 7 days (Fig. 4). Students who ate no fast food had higher GPAs than students who reported eating fast food 7 to 10 times (p < 0.0001).

Alcohol, smoking, and drugs consumption

The self-reported GPAs for students who had consumed alcoholic drinks, students who had tried cigarette smoking, and students who had used electronic vaping products were lower versus students who had not engaged in any of these behaviors before.

There was also a significant difference in self-reported GPA of students based on whether or not they had ever used marijuana at all, with students who had never used marijuana reporting higher GPAs than those who had used it at least once.

Conversely, there was no significant difference in the average GPA of students based on whether or not they considered themselves to be a smoker, the number of times they had used an electronic vaping product in the last week, the number of alcoholic beverages they had consumed in the last 30 days, and the number of times they had used marijuana in the past 7 days.

There was also no significant difference in the average GPA of students based whether or not they had ever taken prescription pain medication without a prescription or differently than prescribed, and whether or not they had ever consumed drugs other than marijuana.

The most commonly named drugs were hallucinogens (n = 40) such as LSD and DMT, cocaine (n = 34), ecstasy/molly/MDA (n = 24), and mushrooms (n = 17). Surprisingly, Xanax^® and ADHD medications, such as Adderall^®, were listed by five respondents each only, although they are reportedly used more frequently (National Survey on Drug Use & Health, 2017).

Sleep

Our findings that students who sleep less than 4 h or more than 9 h per night have lower GPAs than students who sleep 6, 7, or 8 h per night, are similar to what Vedaa et al. (2019) reported for Norwegian students. Chen & Chen (2019) reported that a lack of sleep has a negative association with GPA for college freshmen.

However, our data did not confirm the results from other studies that indicated the timing of sleep to be a better predictor for academic performance than actual hours slept (Eliasson, Lettieri & Eliasson, 2010; Genzel et al., 2013). For example, Eliasson, Lettieri & Eliasson (2010) reported that students with earlier bed times and earlier wake times had higher GPAs, and concluded that timing of sleep is a better predictor for academic performance than actual sleep hours. In our study, however, it did not matter what time students went to bed as long as they got a sufficient amount of sleep. For instance, students who went to bed between midnight and 2 am reported an average GPA of 3.24 ± 0.60 if they slept 5 h or less but an average GPA of 3.47 ± 0.60 if they slept 8 h per night.

Working

There is a consensus among published studies that the number of hours that full-time students work negatively affects their academic performance (Miller, Danner & Staten, 2008; Pike, Kuh & Massa-McKinley, 2008; Darolia, 2014; Tessema, Ready & Astani, 2014; Andemariam et al., 2015). In line with our study, Tessema, Ready & Astani (2014) reported lower GPAs for students working more than 11 h per week, and students working over 31 h per week had the lowest GPA in the study by Andemariam et al. (2015).

Additionally, the positive effect on average GPA we found for students working fewer than 10 h per week (Fig. 2), is supported by the studies of Darolia (2014), Andemariam et al. (2015), and Tessema, Ready & Astani (2014).

Physical activity

We found a positive effect for both strength training and physical activity in general on average self-reported GPA. Respondents who engaged in exercises to strengthen or tone their muscles, such as push-ups, sit-ups, or weight lifting, on four days or more during the past seven days had a higher GPA at 3.51 ± 0.44 than students exercising 1-3 days (3.43 ± 0.44) and students who did not exercise at all (3.28 ± 0.64).

However, Wald et al. (2014) reported a modestly higher GPA for physical activity only and no association between GPA and strength training. Trockel, Barnes & Egget (2000) did not find an association between GPA and either exercise or strength training; nor did Flueckiger et al. (2014). Contrary to that, Bellar et al. (2014) reported that higher levels of physical activity are positively associated with higher GPAs. Roddy, Phle-Krauza & Geltz (2017) showed that students who used a university’s recreation center frequently and regularly were the students with the higher GPAs as compared to their peers who visited irregularly. However, they did not look at the different types of exercise and how they affected GPA.

Eating habits

We found a positive association between eating breakfast and GPA as well as a negative association between the number of times fast food was consumed per week and GPA. The positive effect of eating breakfast regularly has been reported in studies from different countries (Trockel, Barnes & Egget, 2000; Phillips, 2005; Rampersaud et al., 2005; Adolphus, Lawton & Dye, 2013; Chawla et al., 2019; Reuter, Forster & Brister, 2020). Kobayashi (2009), Deliens et al. (2013) and Reuter, Forster & Brister (2020) also found a negative association between fast food consumption and academic performance.

In agreement with previous studies (Petit & DeBarr, 2011; Trunzo et al., 2014; Buchanan & Ickes, 2015; Champlin, Pasch & Perry, 2016), our study showed a significant negative association between energy drink consumption and GPA. Interestingly, 84.3% of our respondents reported not having consumed energy drinks at all over the past seven days. On the other hand, Petit & DeBarr (2011) reported that more than half the students in their study drank at least one energy drink in the past seven days, illustrating how this health behavior has changed over the last 10 years.

Alcohol, smoking, and drugs consumption

A negative relationship between alcohol consumption and GPA was reported by Conway & DiPlacido (2015), Piazza-Gardner, Barry & Merianos (2016) as well as Meda et al. (2017). However, we did not find an association between the number of alcoholic beverages consumed in the past 30 days and GPA. But, students in our study who indicated having consumed alcohol at least once so far reported lower GPAs than students who had never had an alcoholic drink.

Smoking clearly is not as popular with current college students as it used to be. Only one-fifth (19.5%) of students who answered the question “Have you ever tried cigarette smoking?” answered ‘yes’ and only 10 students (1.7%) considered themselves to be smokers. Still, students who had tried cigarette smoking reported lower GPAs than students who had never smoked. Phillips et al. (2015) and Meda et al. (2017) also reported that smoking was a negative predictor of cumulative GPA.

Vaping was much more common than smoking among our study population; one-third of respondents (35.9%) answered ‘yes’ to the question “Have you ever used an electronic vapor product?” and 15.1% of respondents had vaped during at least 1 day over the last 7 days. The fact that in our study only 22.5% of seniors but 39.5% of freshmen admitted to having used electronic vaping products demonstrates that vaping is still on the rise. Study respondents who reported having vaped before had lower GPAs than respondents who had never vaped.

The only drug used by a substantial number of respondents in this study was marijuana with 44.2% of participants responding with ‘yes’ to the question “Have you ever used marijuana?”. While we did not find an association between marijuana use frequency (i.e., number of days marijuana used in the past 7 days) and GPA, our data show that students who answered the question with ‘yes’ reported lower GPAs compared with students who answered with ‘no’. Arria et al. (2015) did not find an association between marijuana use and GPA, whereas Meda et al. (2017), Suerken et al. (2016), Souza, Hamilton & Wright (2019), and Wallis et al. (2019) all reported a negative relationship between marijuana use and academic success.

We did not find an association between pain killer abuse (i.e., taking prescription pain medication without prescription or differently than prescribed) or the use of drugs other than marijuana, such as cocaine, heroin, methamphetamines, ecstasy, hallucinogenic drugs, or synthetic marijuana, and GPA. Pain killer abuse among our study population was at about the same level as reported for the age group 18–25 in the 2017 National Survey on Drug Use and Health (NSDUH) at 7.7% in our study vs. 7.2% in the NSDUH.

The three main limitations of our study were: (1) participant selection, (2) reliance on self-reported health behaviors and habits, and (3) reliance on self-reported GPAs. Although we invited students from all colleges across the university to participate, our final study population with 78.5% female students and 21.5% male students does not reflect the demographics of the student body at our university (53% female students and 47% male students) (Florida Gulf Coast University, 2020). Research has shown that female students are more interested in or worried about health-related issues and diet, and are more likely to participate in online surveys in general (Rozin, Bauer & Catanese, 2003; Smith, 2008).

Participants may have intentionally or unintentionally provided incorrect information about their lifestyle choices or their current GPA. For example, students may have given us the GPA they were hoping to achieve at end of the current semester instead of the actual GPA for completed classes. Also, students may have wanted to appear to be making better health behavior choices or may not remember accurately how often they had consumed fast food, for example. Moreover, some of the behaviors and habits included in the survey were not defined and participants may have interpreted them differently. For example, some participants may have considered walking to class as being physically active, while others may have set the bar higher. Additionally, we cannot rule out that students may have participated more than once. However, there is no reason to believe that inaccuracies such as these or repeat responses had a substantial impact on the findings of this study due to the number of participants being higher than 600.

Our survey was not designed to look into factors that affect student health behaviors and habits; it was designed to look at whether or not the included behaviors impact students’ academic performance. Therefore, the information received from participants is not suitable to make inferences as to potential mechanisms underlying the findings of our study. These questions should be addressed in future studies.