Chris Badger, David Stephenson, and Dr. Tim Smith, Counseling Psychology
Introduction
Relationships have already been shown to have clear psychological benefits for those involved. (Meyers, 1999) Research has also shown an inverse relationships between the number and quality of relationships and health status. (House 1988, Knox 1998) In addition to health status, an incredible amount of research has examined the relationship between relationships and mortality. Other researchers have attempted to summarize this data in metaanalyses. (HoltLunstad, Smith & Layton 2010, Manzoli 2007) However, past metaanalyses of mortality research have either focused on relationships in general or on a specific subgroup of married people, like the elderly. The current project looked specifically at marriage, and will include all age groups, not just the elderly. We were also able to look at subgroups of marriage data, like married, never married and divorced. This data will be particularly valuable to those looking to account for this effect as a possible covariate in other studies.
Methodology
First students searched multiple scholarly databases for articles that might have information linking marriage with mortality. The students developed search strings for individual databases, and strings that were used across databases. After possible articles were identified, experienced coders went through the articles and circled the specific effect sizes that linked marriage and mortality. If no effect size was present, the article was rejected. Then, pairs of coders went through each article with a circled effect size and gathered more information, like how many people were included in each effect size, how old the sample was, gender, and the statistical controls applied to in each effect size. The most problematic things we encountered were keeping the definitions that we used for each category consistent. We often needed to convert from different ways of measuring effects into our chosen method, hazard ratios. We found lots of articles that used OR, p values, and just straight numbers of people in each category. Each conversion had a specific procedure that we needed to follow to accurately represent each effect. Authors sometimes didn’t report the exact data that we needed to do these conversions in the article, so we had to devise together with Tim ways of estimating these numbers.
Results
The results of our data support the widely held assumption that marriage has a strong negative effect on mortality. In a comparison between married and single populations it we found that the protective effect of marriage to be slightly greater than the effect experienced between alcohol abstinence and heavy alcohol consumption.
Discussion
We are excited for the impact our research will have on the current body of knowledge. In confirming the generally held hypothesis we enable future researchers to control for the effects of marriage in survival analysis. This has significant implications beyond relationship science. Researchers are now able to better isolate the effect of other variables through our metaanalysis of the existing 258 articles with applicable data. As we pursue future publication in (insert likely journal names here) we anticipate this study to become a work of reference in the field of epidemiology.
Conclusion
We’ve learned a lot about how metaanalyses work, and feel that the experiences we’ve had working with Tim and Julianne on this project will help us in our future careers as doctors. We’ve learned a lot about how different types of statistical analyses affect the way data is interpreted, and have improved our ability to evaluate academic literature.