Good Plotting Practices

In the pharmacokinetic literature, three types of plots are most commonly seen scatter plots, histograms, and bar charts. Good plotting practices for each of these graph types will be presented. Beyond these plots are others, like QQ plots, dot plots, and three-dimensional surface plots. It is beyond the scope of this chapter to cover every conceivable plot, but many of the good plotting practices that will be put forth should carry over into other plot types as well. 

A common plot in pharmacokinetics is the scatter plot, e.g., Fig. 1.5. If x and Y are vectors of two continuous measurements, each observation pair is represented by a plotting symbol, such as a solid circle ( ) or open circle (o), in an x-Y coordinate system. Sometimes the observations are categorized by strata, i.e., a third variable, such as treatment, in which case a unique symbol is used for each grouping variable. Vision appears to consist of two different processes. Rapid visual processing is the mechanical analysis of pictures to surfaces, 

Tremmel (1995) later extended this study focusing on black/white symbol contrasts. The best symbols to use were those that were readily distinguishable. 

Many of Cleveland s guidelines are common sense, but when has that ever stopped someone from making a graph. Cleveland used extensive scientific research to determine what graph properties lead to efficient transfer of information to the reader. 

In one study, Cleveland and McGill (1984a 1985) described 10 elementary perceptual tasks viewers use to 

---

Plot maintenance B Expertise must be available to maintain the test site and, if cropped, to take care of the crop For bare-soil studies, the soil surface must be carefully prepared prior to test substance application and kept weed-free without disturbing the test areas. If the test is cropped, the crop should be treated according to Good Agricultural Practice. In case of a soil accumulation study, the field may be cultivated and cropped each season for up to 6 years... 

Also shown in Figure 22.8 is the pop-plot run distance versus pressure. Note that the minimum diameter for which detonation could be achieved at any given pressure is approximately equal to the ideal run distance at that pressure. Good design practice is to midce sure that detonation is always achieved within the constant pressure cone. Therefore, flyer diameter should always be equal to or greater than twice the run distance. 

In theory, optimmn accmacy and precision values approach zero. Note that the smaller the specified volume chosen for evaluation, the greater the effect of volmne variation on accuraey and precision. Therefore, it is good laboratory practice to plot the results of accuracy and precision for a pipet s specified volmne on a data chart over time. 

Another topic to establish a good TMDSC practice deals with Lissajous figures (or Bowditch curves), plots of the time-dependent heat-flow rate HF(t) or AT(t) versus... 

The laboratory staff are relieved from non-productive routine activities such as reading meters and plotting of results the data evaluation is assisted by automated analysis routines. For validated processes or Good Laboratory Practice (GLP), a digital laboratory journal is evident. 

The sources of noise (defined as any type of additive unwanted signal) in a seismic record include inherent instrument noise (self-noise), environmental effects, excessive instrument noise, installation-driven noise, and ancillary equipment noise. A seismometer s self-noise is usually specified by the manufacturer, typically as a power spectral density (PSD) plot, and this establishes a performance baseUne for an installation. Good installation practice will include mitigating sources of environmental noise as much as is practical. Excessive instrument noise is defined as noise originating within the seismometer in addition to its published self-noise characteristic and may represent variability in the... 

While it is, in principle, desirable to take the limiting slope of the log CLD plot, in practice the limiting slopes are very susceptible to experimental noise and baseline choice issues. Moad and Moad24 have shown that very little error is introduced by systematically taking the slope over the top 10% or the top 20% of the chain length distribution. The values for the slopes will overestimate ln(0). However, because the discrepancy is systematic, the Mayo" analysis still provides a good estimate for C ( 6% error for the example in Figure 6.1). 

In this instance, the (5)-enantiomer data have been negated prior to plotting. From previous discussion of the antisymmetry of the parameters under enantiomer exchange (e.g., Section III.A) it is recognized that it is then to be expected that the (R)- and (5)-enantiomer data should fall on the same experimental trend line. That they do indeed do so shows, as was argued in the Section IV.A for fenchone, that the behavior is at least qualitatively in accord with a pure electric dipole model. Furthermore, combining two distinct data sets [(/ )- and (5)-enantiomers] in this manner provides a consistency check on the reproducibility of the PECD data. It seems good practice to include measurement of both enantiomers, where this is feasible, in an experimental study. 

---