Generalized Estimating Equations (GEE) Model of the Association between the Proportion of Qualitative Research and Factors Pertaining to Journals' Characteristics and Policies.

b- Unstandardized coefficient; SE- Standard Error; CI- Confidence Interval.*Statistically significant at the 0.05 level.1The proportion of qualitative research was calculated by dividing original qualitative studies published in journals of general and internal medicine from 1998 to 2007 (numerator) by all original research articles published in journals from 1998 to2007 (denominator). The proportion was calculated for each journal (unit of analysis) and for each year.2Country of publication- as derived from journals' website- was categorized into 3 categories: US, UK, and all other countries (defined as ‘other’). The ‘other’ category was not divided into sub-categories to facilitate statistical analysis-due to the relatively small number of journals (n = 67). The UK and US category were each compared to the referent category- ‘other’.3Editorial/Methodological papers- defined as appearance of editorials and/or methodological papers on qualitative research (excluding original research) during the 10 year period (1998–2007). The presence of editorial/methodological was compared to the referent category- absence of editorial/methodological papers.4Aims/guidelines- defined as the presence of specific guidelines for authors pertaining to qualitative research and/or specific mention of qualitative research in the journals' aims or scope. The presence of aims/guidelines was compared to the referent category -absence of aims/guidelines.5Impact factor – impact factor was derived for each year of the 10 year period (1998–2007) from the ISI web of science Journal Citation Reports.10 Since the 2007 impact factor was highly and significantly correlated with the 10-year mean (r = 09.37; P = 0.01), the 2007 impact factor was utilized.