A cross-sectional and analytical study was performed during the 2012 school year with university students with physical education and sports majors from a public university in Guadalajara, Mexico. Total enrollment was 976 students (63% men and 37% women) from which a simple random selection with an expected prevalence of 64.5% (Marty, Lavín, Figueroa, Larraín, & Cruz, 2005), a 70% accepted minimum frequency and with 99% precision (Lwanga & Lameshow, 1991), which produced 527 individual interviews of university students.

The selection of this university population was carried out by random and proportional numbers, taking into consideration gender and scholastic cycles. The list of students registered for the 2012 school year was used to select those who would voluntarily answer the surveys under an informed consent status. The investigative protocol and informed consent form were reviewed and approved with reference number IISO/CI/11/2012-2013 in adherence to the Helsinki Declaration of 2008 in terms of the ethics of investigation of human beings.

The Academic Stress Inventory was used (ASI; Polo, Hernández, & Pozo, 1999) validated by the Spanish Society of Anxiety and Stress. This is a questionnaire with eleven situations which were considered potential stress generators in students in the academic arena. Each one of the situations offers a scaled answer of 5 points (1 no stress, 5 high stress) where each participant gives a value according to his or her perception of whether or not it produces stress.

The eleven situations considered to be potential stress generators are: test taking (EA1), oral presentations (EA2), class participation (EA3), seeking help from tutors (EA4), academic overload (EA5), overly large class size (EA6), lack of time (EA7), obligatory assignment (EA8), homework (EA9), group projects (EA10), and competition among classmates (EA11). The reliability in terms of internal consistency corresponds to an alpha coefficient by Cronbach of .90, which is considered satisfactory. In order to establish association, the score is converted from ordinal values into a cardinal level of academic stress in the following way: if the value was 1-2 it was considered low level; a value of 3, moderate level; and between 4-5, high level.

The Stress Symptom Inventory (SSI), a questionnaire which was developed and approved by Lipp and Guevara (1994), contains a list of 42 psycho-physiological symptoms characteristic of chronic stress which is based on a three phase model developed by Selye (alarm, resistance, and exhaustion). In 1988, Dominguez adapted it for use in Mexico by means of a content validation and reported an alpha internal consistency rating by Cronbach of .94, which indicates an acceptable degree of reliability. The classification of high, moderate, and low levels of chronic stress was carried out by mean and standard deviation (SD). Furthermore, high level is considered two and three SD above mean, medium level ± 1 SD and low level stress 2 and 3 SD below mean (Pozos-Radillo, Torres, Aguilera, Acosta, & González, 2008). Variables gender and age (in five year increments) are also presented.

Pearson's analysis of correlation was used to pinpoint the information about the predictive value and, in order to determine the validity of the measurements obtained, we carried out an analysis of regression by steps for the ASI and SSI by simultaneously introducing the equation of measurement to the level of significance (p < .05). In these analyses, the variables appear in order in the equation according to the percentage of the explained variance.

Afterwards, a hierarchical analysis of multiple regression was applied to determine the predictive value of the situations of the ASI in which the Introduce method was used with the ASI situations. With this procedure, an incremental value of prediction of the variable included in third place (ASI) was obtained, once the effect of the second and the first was controlled.

In order to explore the statistically significant difference of gender and age in chronic stress (SSI), as it relates to academic stress situations (ASI), the bi-variant associations were tracked through an analysis of the contingency charts. In these contingency charts the categories of the ASI and SSI (high, moderate and low) were evaluated and were transformed into dichotomies, with values of “0” or “1”; a high level rating was considered as risk. In order to carry out the association of risk calculation the Odds Ratio (OR) was estimated with a CI of 95% and a level of significance of p < .05. The data was tabulated and processed using SPSS (Statistical Package for Social Sciences), Version 15 for Windows XP, with university license.

In order to carry out this study, 527 students from a public university were interviewed, of whom 311 (59%) were women and 216 (41%) men; the age range was 18-33 years with an average age of 21.07 (± 1.80) years. The descriptive analysis of the SSI, according to the levels present, showed that 35.3% (186) of the students showed a high level of chronic stress, 44.8% (236) moderate levels, and 19.9% (105) low levels. After the analysis of the situations of the ASI, 32.8% (173) showed high levels of stress when taking a test; 25.4% (134) from competitiveness among classmates; 18.6% (98) from lack of time; 18% (95) from oral presentations; 17.8% (94) from overly large class size; 17.5% (82) from seeking help from tutors; 17.1% (90) from homework; 16.7% (88) from obligatory assignments; 15.9% (84) from group projects; and 15.5% (82) from class participation, academic overload or both. The breakdown and description of this data, according to gender, is shown in Table 1; the age distribution is shown in Table 2.

Table 3 indicates the correlations between the SSI scores and the ASI situations. The analysis reveals that the positive correlations, which indicated a moderate strength, were situation EA3 (r = .63, p < 01), followed by EA8 (r = .49, p < .01); the situation EA1 showed a weak correlation (r = .21, p < .01). Other situations presented a non- significant correlation with r less than .20: r = 12 for EA2, r = .13 for EA4, r = .16 for EA5, r = .14 for EA6, r = .11 for EA7, r = .10 for EA9, r = 16 for EA10. Situation EA11 yielded r = .01, which was considered a very low correlation and was therefore not included.

The correlation coefficient obtained by analysis of chronic stress through regression by steps, initially revealed that the situations of classroom participation (EA3), obligatory homework (EA8), and test taking (EA1) met the predictive criteria with a significant value (R2 = 26, F = 13.3, p < .05). The EA3 situation presented a stronger correlation according to the adjusted R2; in the first model, the last situation had a 21% SSI prediction, with an F = 25.2 and a p < .05. Nevertheless, when EA8 and EA1 situations were introduced the prediction only increased by 5%. The rest of the situations were excluded because they had either an insignificant or a negative correlation. The results showed a significant increase in the explanation of the ASI and SSI, as appraised by the university students. When the EA3 equation was introduced the additional predictive value related to the situations EA8 (p < .01) and EA1 (p < 05), as reported in Table 4.

In the exploration to determine the association between high levels of the SSI and the ASI situations, with the variable of gender and age (Table 5) it was found that the high level of chronic stress (SSI) was related by means of the OR with the feminine gender variables (OR = 1.47) and with the age of 25 years (OR = 6.3) The academic stress situations (ASI) that were found to have a significant association were EA1 with feminine gender (OR = 1.48) and with the age of 23 (OR = 2.09); for EA3 the association was established with the age of 18 (OR = 2.89) and for EA8 only the feminine gender was significant with an OR = 1.70. The other situations did not show any significance with the variables of age and gender.

With regard to these analyses, we have made various contrasts relative to the supposed independence, normality, and homoscedasticity. In this sense, we must review the data, which does not show multicollinearity among the predictor variables. The average of the statistical values of “tolerance” carried out for the physical, psychological, and behavioral factors was set at .79, with no value below .71. The proximity of these values and the maximum value (range 0-1) indicates the independence of the contributions of the predictor variables over the values of the eleven academic stress indicators, thereby showing that the variance of the residuals is constant, proving that the residuals were distributed normally. The average of the value (VIF = 1.24), with no value below 1.05, shows that there was no a collinearity problem.