Quantifying performance using multiple measurements

<p>In order to overcome the problems of skewing the incentives that <a class="wikilink" href="/measuring_performance_using_single_criteria/">measuring performance using single criteria</a> generates, we could think about using many criteria at once. For example, we could see the number of products a factory makes, and the number of accidents in the plant. The problem here is that there'll need to be some normalization. </p> <p>The biggest challenge is that the way in which we normalize and add together the metrics can change over time and from place to place[<a class="litnote tooltip" href="/literature/@ridgway1956">@ridgway1956<span class="tooltiptext">Dysfunctional Consequences of Performance Measurements</span></a>]. This is exactly what happens when one thinks about the <a class="wikilink" href="/exchange_rate_of_scientific_output/">exchange rate of scientific output</a>, since they are many metrics normalized somehow, especially the case that <a class="wikilink" href="/citation-based_metrics_incentivize_competition_over_collaboration/">citation-based metrics incentivize competition over collaboration</a>. </p> <blockquote> <p>Without a single over-all composite measure of performance, the individual is forced to rely upon his judgment as to whether increased effort on one criterion improves over-all performance, or whether there may be a reduction in performance on some other criterion which will outweigh the increase in the first. [<a class="litnote tooltip" href="/literature/@ridgway1956">@ridgway1956<span class="tooltiptext">Dysfunctional Consequences of Performance Measurements</span></a>]</p> </blockquote>