This study has limitations. It is a (descriptive) prevalence study and thus generalizations of the findings should be cautious. Subjective self reported data could introduce inaccuracies due to recall bias, social desirability and sociability. Students were recruited during lessons/lectures, hence those not present in the class at the time of data collection were not included in the survey. Some variables were assessed by single item measures due to respondent burden and the need that a general student health survey be conducted within a short time during the lectures. This renders the measurement of each health variable with more items unworkable. There are many other instruments that measure health behaviours,but most cover only a single aspect of health-promoting lifestyles (59–62). In addition, females were over-represented in this UK sample, hence, our data is presented categorised by gender and standardised for gender when undertaking comparisons across the participating sites. Although we standardized for gender, our
male-to-female ratio might not be completely comparable to that of the UK as a whole. We broadened the data collection at the participating sites in order get student samples that are representative of their universities; however, our sample remains a convenience sample. Such convenience samples are common in student surveys
across the world (1, 44, 47, 63). Even when response rates are relatively high, possibly due to data collection being undertaken in the class/ lecture room, students not interested in healthful practices could be under-represented in our samples leading to an underestimation of unfavourable such practices. The data that was collected did not include variables related to the background of the University as well as the background of the students and the situational context. Such information would have enabled a detailed comparison of potential explanations of the differences
found between the participating universities. Likewise, the degree of observed clustering of a more or less favourable levels of the variables at some universities might be subject to many factors (usually not measured) that would confound such complex and intricately associated constellations of relationships that are usually challenging to unpack, let alone attribute to certain aspects of the university, region, country or participating individuals (58). Future research should attempt to address these limitations.