210 likes | 320 Views
How good is our research? New approaches to research indicators. Average is a metric; distribution is a picture. ‘Average impact’ is a good bibliometric index but not sufficient A tool for reporting but not for action Data are skewed, so average is not central
E N D
How good is our research?New approaches toresearch indicators
Average is a metric; distribution is a picture • ‘Average impact’ is a good bibliometric index but not sufficient • A tool for reporting but not for action • Data are skewed, so average is not central • Many papers are uncited and a few papers are very highly cited • New approach looks at where the spread of performance falls • Activity is located within distribution by more than a single metric • Thresholds help in describing peak of performance. • This improves descriptive power, information content and management value
Distribution of “research performance” A good ‘indicator’ should capture and reflect this in some meaningful way Do current metrics do this?
Distribution of data values - income Minimum Maximum
Distribution of data values - impact The variables for which we have metrics are skewed and therefore difficult to picture in a simple way
Simplifying the data picture • Scale data relative to a benchmark, then categorise • Could do this for any data set • All journal articles • Uncited articles (take out the zeroes) • Cited articles • Cited less often than benchmark • Cited more often than benchmark • Cited more often but less than twice as often • Cited more than twice as often • Cited less than four times as often • Cited more than four times as often
Categorising the impact data This grouping is the equivalent of a log 2 transformation. There is no place for zero values on a log scale.
UK ten-year profile 680,000 papers MODE (cited) AVERAGE RBI = 1.24 MODE MEDIAN THRESHOLD OF EXCELLENCE?
Implications • Is the UK as good as we thought? • YES - the average is unchanged • What lies beneath just became apparent • The effective peak is very concentrated • Other countries would probably look similar • New metrics are needed • Average impact not indicative of distribution • Need to add median, mode • Proportion of activity at thresholds of excellence • Above world average, More than 4 x world average, etc • Evaluate methodology • Does it work by year and by subject? • How can we apply it?
HEIs – 10 year totals – 4.1 Smoothing the lines would reveal the shape of the profile
HEIs – 10 year totals – 4.2 Absolute volume would add a further element for comparisons
What next? • Profiles • Create a view of the distribution of performance • Provide more information useful to management • Require a change in metrics • Applications • Disaggregate the components of the research base • Track institutional profiles against benchmark • Evaluate the link between platform and peak • Track papers through time: e.g. leaders vs. climbers
How good is our research?New approaches toresearch indicators