Nurses' perceptions of shortage effects Nursing stats

Statistical methods were not always explained or even mentioned in the methods section of the articles, but were often buried in the text of the results section or listed only as footnotes to tables. In several instances, no statistical procedure was specified, but the presence of a p value indicated that a test had been performed.

Hellems, Gurka and Hayden (2007),

Buerhaus, DesRoches, Donelan, Dittus and Ulrich (2007) report a six-year survey of U.S. nurses performed over 2002-2006 and compares mean responses for a sample they allege is large enough to represent the current U.S. nurse population. They report and discuss changes in nurse perceptions of conditions and health care outcomes derived from national, regional and local nurse shortages over the survey years and indicate where changes are statistically significant, presenting 'p-values' to indicate effect strength of the change over various years. As Hellems, Gurka and Hayden (2007) indicate above, however, peer-reviewed academic studies increasingly fail to fully explain the statistics they base their assertions on (1085). The 2007 Buerhaus, DesRoches, Donelan, Dittus and Ulrich study is one such paper, presenting conclusions supported by p-values for change over time, supporting the random survey methodology and generalizeable sample size, but nowhere explaining how they achieved statistical significance for these claims.

Nonetheless the study presents interesting findings, using nonparametric, descriptive statistics to compare respondent perceptions of effects deriving from nurse shortages. Smith (1985) explains nonparametric statistics are appropriate whether distributions are normal or not (596). These results come from methodically controlled random samples that the authors allege are large enough to generalize to the wider population (Buerhaus, DesRoches, Donelan, Dittus and Ulrich, 2007, p. 70), although they do not show the calculations used to estimate that precise cutoff...

The authors qualify their presentation by asserting that complete methodologies were published after the 2002 and 2004 rounds, which may have contained more extensive description of the methods used to support statistical significance of change over time widely reported throughout the study results (Buerhaus, DesRoches, Donelan, Dittus and Ulrich, 2007, p. 70).
Results indicate extremely high dissatisfaction with many conditions derived from workplace stress that would be alleviated with increased supply of nurses. A number of these negative perceptions fell over the course of the three rounds of surveys, but many still display alarmingly high levels of workplace dissatisfaction, improvement notwithstanding. Perceptions that supply was "much less than demand" have fallen overall over the four-year period prior to 2007, but increased from 2004-2006 by four percent (Buerhaus, DesRoches, Donelan, Dittus and Ulrich, 2007, p. 73). At the same time, respondents considering supply of nurses "less than demand" remained the same over 2002-2006, but include an increase of 11% between those years (Buerhaus, DesRoches, Donelan, Dittus and Ulrich, 2007, p. 73). Four percent of this difference represents a shift to the higher category, since no other category decreased, with the rest (7%) shifting to perceptions supply had increased from 2004-2006. These results were similar or worse for wider geographic and national labor pools (Buerhaus, DesRoches, Donelan, Dittus and Ulrich, 2007, p. 72). When changes are statistically significant, the authors indicate this with p-values, which reveal the nonparametric character of the underlying means tests (Smith, 1985, p. 596) although those procedures are never described.

Other results include nurses' perceptions of quality of and change in quality of patient care over the two survey rounds 2004-2006. These impacts include patient complaints and increased wait time, delayed response to calls, conflict between workers, increased physician workloads and reduced access…