Data Analysis and Interpretation

¶ … Tobacco control and stress reduction

Data interpretation is a critical component of the research process, one which is necessary to ensure that the data has program significance -- i.e., can provide useful guidelines for changes in policy or to add to the existing research -- and also to ensure that the data simply meets basic statistical criteria for validity. In the case of the smoking cessation program profiled in the text by McKenzie, Neiger, & Thackeray, R. (2013), there are several obvious problems with the data -- only a limited number of the pooled workplaces (67%) actually completed the survey, 251 in total. 198 of the profiled workplaces (79%) had fewer than 51 employees. This means that the data may be incomplete regarding the full extent that such programs have been implemented (or the lack of extent to which they have been implemented) and most of the programs have a relatively limited number of involved persons. According to Fitzpatrick (2004 et al., cited by McKenzie, Neiger & Thackeray 2013) one method of data analysis is "comparing assessed performance levels on critical variables to expectations of performance or standards." Using such a limited pool of workplaces for a multifactorial policy as anti-smoking is inherently problematic, given the variability which exists within such programs and in terms of enforcement policy from workplace to workplace.

This variability, it should be noted, is obvious in the available data. For example, 52% of organizations did not allow smoking within the buildings at all while 37% allowed it only within designated areas (10% had no such policies). There was also an overrepresentation of large businesses within the sample, further complicating the results. Thus, the data presented is problematic for a variety of potential uses -- in terms of evaluating the utility of such a program in discouraging tobacco use, the surveyed entities are extremely varied in terms of the specifics of the programs they are using. The sampling is relatively incomplete and skewed in favor of particular types of businesses at the expense of others. Variance in program types for an already small sampling further indicates possibly inaccurate results, given that different anti-smoking approaches may be missing from the data, particularly of the kinds of businesses that did not respond to the survey. For a survey intended to be relatively encompassing of an entire locality, this is particularly problematic. While it may be true that the data is characteristic of the programs overall, there is a clear bias in favor of a certain 'type' of company and therefore a certain 'type' of employee. Another consideration in evaluating data according to Fitzpatrick (2004) is the value of the accomplishments chronicled in the data. While assessing the existence of anti-smoking programs in the workplace may be useful, there was no data gathered regarding efficacy or the degree to which such guidelines were enforced.

The stress reduction survey offers more complete information -- it contains both quantitative and qualitative measures. The subjects were asked to record both pre and post-intervention scores and also submitted to semi-structured interviews about their experience. This mixed methods approach enhances the validity of the data. The concept of 'stress' can be particularly subjective, and merely because subjects recorded changes in their self-perceptions of how stressed they were before and after the program, one person's definition of stress is not necessarily the same as someone else's. Many other variables can affect perceptions that someone is more or less stressed. There is also no notation that the instrument used is widely-accepted in the psychological community.