Findings of the present study indicate that asking questions about cognitions regarding physical activity positively influence subsequent self-reported participation in this behavior over a three-month period. Moreover, these results compare favorably to a recent meta-analysis that observed short- and mid-term (i.e., six to twelve months of follow-up) small-to-medium effect sizes on self-reported physical activity (SMD = .28, CI95% = 0.15 to 0.41) in more intensive interventions among adults . Thus, results from the present study suggest that having only overweight/obese individuals complete a TPB questionnaire that also included other theoretical constructs such as anticipated regret, moral norm, and positive feelings is an easy and inexpensive way to promote exercise, especially in clinical settings where time is scarce (e.g., in the physician's office).
In spite of the growing scientific evidence of the effect of the mere-measurement of cognitions on behavior, the review by French and Sutton  highlights the fact that this phenomenon is expected to be small for health-related behaviors and to be mostly observed for non-complex behaviors such as those requiring a single action (e.g., giving blood). However, our results do not support this view given that physical activity is considered to be a complex behavior defined along dimensions of frequency, intensity of practice and duration of exercise sessions. According to French and Sutton , this mere-measurement effect is also more likely to be detected when health-related behaviors are subjectively assessed. As such, it is noteworthy that our measure of physical activity was self-reported. However, mere-measurement effects were observed for physical activity assessed either by self-reported or accelerometer (in the corrected model) in a study by van sluijs et al. . There are also few studies that showed this effect for objectively assessed behavior [8, 9, 34]. According to these latter authors, the question now is not about the existence of mere-measurement, but about what is causing this effect and under which conditions this phenomenon could be observed.
In the present study, the full constructs of the TPB as well as additional variables from other theories were evaluated in the psychosocial questionnaire on LTPA. This questionnaire measured ten theoretical constructs and three sets of beliefs that required about fifteen to twenty minutes to complete. Consequently, individuals were guided towards deliberation about their behavior throughout the questionnaire (i.e., weighting the desirability and feasibility of taking action) that could lead to greater introjections and behavioral resolutions to become more active. Although the design of the present study could not allow any conclusion about the mechanisms underlying this mere-measurement effect, one could surmise that such an effect is less likely to emerge after completing shorter questionnaires. However, previous studies have reported that assessing only intention by means of a few items changes subsequent performance of studied behaviors at follow-up [10, 12, 35], including physical activity . Others might also infer that this effect might be the consequence of assessing specific cognitions such as those included in the questionnaire of the present study. In this line of thought, Sandberg and Conner  observed that measuring anticipated regret (i.e., the anticipation of regret not to perform the behavior) was an important variable responsible for mere-measurement effect on cervical screening attendance. Since anticipated regret was one of the variables assessed in our extended TPB questionnaire, it can be hypothesized that participants in the present study anticipated feelings of regret about not being physically active in response to the questionnaire and, consequently, increased their level of physical activity.
The precise mechanism behind the mere-measurement effect is still a matter of debate [6, 19]. At this time, the dominant explanation of this effect is that asking behavioral intention questions heightens the accessibility of the person attitude towards the behavior, which in turns increases the likelihood that the behavior will be performed . Morwitz and Fitzsimons  showed that responding to a query about one's purchase intention increases the activation level of one's pre-existing brand attitude. When the respective brand attitude was both highly accessible and positively valenced then participants were likely to choose that brand. On the other hand, when the activated attitude was negatively valenced this led to a decrease in the choice of this brand. However, the mere-measurement effect may only operate among those whose thoughts and feelings are favorably disposed towards the behavior. In a laboratory setting Morwitz and Fitzsimons  showed that positive attitudes increased brand choice, while negative attitudes decreased brand choice. Obviously, future research in this direction is needed to elucidate the potential mechanisms of action of the mere-measurement effect on changes in physical activity behavior.
The relatively modest costs and simplicity of the intervention may add to the appeal of mere measurement as an important additional strategy for improving public health via health behavior change. Nevertheless the use of the mere-measurement effect as a mean to promote public health is likely to be limited by the compliance of individuals to complete adequately the questionnaire. Indeed, questionnaire completion appears to be a prerequisite for a mere-measurement effect to occur [8, 9, 34]. Consequently, an important implication for using the mere-measurement effect to promote health behaviors is that studies will need to maximize completion rates of the questionnaire in order to produce the greatest impacts on the targeted behavior. Further research could also examine the added value of combining the mere measurement effect with interventions that promote questionnaire completion and/or positive attitudes toward the target behavior. For example, Dillman  provides guidance on the content and type of cover letter, questionnaire format, token incentive and return envelop that could be used to improve response rates to mailed questionnaire.
Notwithstanding this positive and significant effect of mere-measurement of cognitions, it must be acknowledged that the effect remains small and might benefit from being used in combination with other techniques for behavior change (see Abraham and Michie  for a description of a set of theoretical techniques). Moreover, from a public health perspective, if asking questions regarding a given behavior reinforces or induces its adoption, caution should be exercised before asking questions about sedentary behavior. Indeed, Williams, Block and Fitzsimons  observed an increase in illegal drug use following the completion of one item about intention to use illegal drugs. Thus, researchers investigating cognitions regarding sedentary behaviors among overweight or obese individuals instead of those related to physical activity participation must be aware that they may cause more harm than good in asking such questions.
Some limitations of this study must be acknowledged. Firstly, the present study was completed among a group of volunteers representing individuals interested in the topic of the study. Thus, this effect might be more important in this kind of sample, given that the cognition scores were relatively high (data not shown). Sprott, Spangenberg and Fischer  also investigated the effect of mere-measurement (described as self-prophecy) when the behavior under investigation was considered by the participant as right or wrong. Their results indicated that this effect was more likely to occur among participants with stronger normative beliefs regarding low-fat snack consumption and health and fitness assessment. Secondly, although we used an experimental design, it was not possible to apply an intention-to-treat approach given that different measures of physical activity were used at baseline and follow-up. Thirdly, subjective measures of physical activity were used in the present study, although both tools have been validated. Also, given that in the present study the measure of physical activity at follow-up differed from the measure used at baseline, it is less likely that the size of the observed effect might be attributed to measurement habituation. Nonetheless, it will remain important in future studies to conduct randomized controlled trials with objective measures of physical activity. Finally, additional research should be conducted among other segments of the overweight/obese population, especially among sub-groups with low socio-economic status.