Joachim Kunert

Systematic variation improves reproducibility of animal experiments

measures to compare between-experiment variation for the standardized and heterogenized design. Whereas strain differences were relatively consistent among heterogenized experiments, they varied considerably between standardized experiments (Fig. 1a–c). In 33 of 36 measures, between-experiment variation was lower in the heterogenized design, indicating better reproducibility. We also analyzed each experiment separately as if conducted independently in different laboratories and assessed the effect of ‘strain’ on each of the 36 measures using a general linear model (GLM). Based on the 2 × 2 factorial nature of the heterogenized design and cage position in the rack, we divided each replicate experiment into four ‘blocks’, each comprising one cage per strain (Supplementary Fig. 1), and included ‘block’ nested within experiment as blocking factor in the GLM (Supplementary Methods). Whereas the effect of ‘strain’ was stable in the four heterogenized experiments, outcomes of the four standardized experiments were highly variable (Supplementary Fig. 2), suggesting that withinSystematic variation improves reproducibility of animal experiments

Biopersistence of synthetic mineral fibers as a predictor of chronic inhalation toxicity in rats.

In December 1997 the European Commission (EC) adopted Directive 97/69/EC (O.J. L 343/19 of 13 December 1997) in which criteria were established for the classification and labeling of synthetic mineral fibers. This directive was derived based upon an extensive program evaluating current scientific knowledge on fiber pathogenicity and its relationship to the biopersistence of long fibers. Within this context, the biopersistence of fibers longer than 20 microm was found to be a good predictor of the lung burden and early pathological changes in chronic inhalation studies with fibers as well as of the tumor response in chronic intraperitoneal studies with fibers. The analysis that provided the scientific basis for the relationship of biopersistence to the chronic inhalation results is presented in detail. Proportional odds regression techniques were used to determine the relationship between both inhalation and intratracheal instillation biopersistence clearance half-times and the collagen deposition at the broncho-alveolar junction as determined following 24 mo in chronic inhalation toxicity studies. The results indicate all the indicators of biopersistence considered are equally good predictors of the early long-term change that occurs in the lung in response to more durable fibers. This change, the collagen deposition at the broncho-alveolar junction, is a precursor of interstitial fibrosis, which has been shown to be associated with tumor response in fiber-exposed animals. The results show that the clearance half-times set in the EC directive are within the baseline for this parameter.In December 1997 the European Commission (EC) adopted Directive 97/69/EC (O.J. L 343/19 of 13 December 1997) in which criteria were established for the classification and labeling of synthetic mineral fibers. This directive was derived based upon an extensive program evaluating current scientific knowledge on fiber pathogenicity and its relationship to the biopersistence of long fibers. Within this context, the biopersistence of fibers longer than 20 µm was found to be a good predictor of the lung burden and early pathological changes in chronic inhalation studies with fibers as well as of the tumor response in chronic intraperitoneal studies with fibers. The analysis that provided the scientific basis for the relationship of biopersistence to the chronic inhalation results is presented in detail. Proportional odds regression techniques were used to determine the relationship between both inhalation and intratracheal instillation biopersistence clearance half-times and the collagen deposition at the broncho-alveolar junction as determined following 24 mo in chronic inhalation toxicity studies. The results indicate all the indicators of biopersistence considered are equally good predictors of the early long-term change that occurs in the lung in response to more durable fibers. This change, the collagen deposition at the broncho-alveolar junction, is a precursor of interstitial fibrosis, which has been shown to be associated with tumor response in fiber-exposed animals. The results show that the clearance half-times set in the EC directive are within the baseline for this parameter.

Biopersistence of synthetic mineral fibers as a predictor of chronic intraperitoneal injection tumor response in rats.

In December 1997 the European Commission (EC) adopted Directive 97/69/EC (O.J. L 343/19 of 13 December 1997), in which criteria were established for the classification and labeling of synthetic mineral fibers. This directive was derived based upon an extensive program evaluating current scientific knowledge on fiber pathogenicity and its relationship to the biopersistence of long fibers. Within this context, the biopersistence of fibers longer than 20 µm was found to be a good predictor of the lung burden and early pathological changes in chronic inhalation studies with fibers as well as of the tumor response in chronic intraperitoneal studies with fibers. The analysis that provided the scientific basis for the relationship of biopersistence to the chronic intraperitoneal (ip) results is presented in detail. Analysis of the relationship of biopersistence clearance half-times to ip tumor response shows a statistically significant relationship of ip tumor response to not only the number of fibers injected, but also the median length of the fibers injected and their solubility (clearance half-time). The results show that the biopersistence half-times as determined by intratracheal instillation (T 1/2 of WHO fibers or weighted T 1/2 of fibers with L > 20 µm) and as determined by inhalation (weighted T 1/2 of fibers with L > 20 µm) are equivalent predictors of the ip results. From these ip studies, fibers that can be exonerated from classification as carcinogens in Europe have a relative tumorigenic potency in the ip cavity of between 66 and 2500 times less than fibers that have been shown to produce a significant increase in tumors following chronic inhalation exposure. In addition, based upon the ip results, there is no statistical difference between the EC and the other fiber exoneration criteria, such as the German Gefahrstoffverordnung of 1999.In December 1997 the European Commission (EC) adopted Directive 97/69/EC (O.J. L 343/19 of 13 December 1997), in which criteria were established for the classification and labeling of synthetic mineral fibers. This directive was derived based upon an extensive program evaluating current scientific knowledge on fiber pathogenicity and its relationship to the biopersistence of long fibers. Within this context, the biopersistence of fibers longer than 20 microm was found to be a good predictor of the lung burden and early pathological changes in chronic inhalation studies with fibers as well as of the tumor response in chronic intraperitoneal studies with fibers. The analysis that provided the scientific basis for the relationship of biopersistence to the chronic intraperitoneal (ip) results is presented in detail. Analysis of the relationship of biopersistence clearance half-times to ip tumor response shows a statistically significant relationship of ip tumor response to not only the number of fibers injected, but also the median length of the fibers injected and their solubility (clearance half-time). The results show that the biopersistence half-times as determined by intratracheal instillation (T(1/2) of WHO fibers or weighted T(1/2) of fibers with L > 20 microm) and as determined by inhalation (weighted T(1/2) of fibers with L > 20 microm) are equivalent predictors of the ip results. From these ip studies, fibers that can be exonerated from classification as carcinogens in Europe have a relative tumorigenic potency in the ip cavity of between 66 and 2500 times less than fibers that have been shown to produce a significant increase in tumors following chronic inhalation exposure. In addition, based upon the ip results, there is no statistical difference between the EC and the other fiber exoneration criteria, such as the German Gefahrstoffverordnung of 1999.

Optimal Crossover Designs in a Model With Self and Mixed Carryover Effects

We consider a variant of the usual model for crossover designs with carryover effects. Instead of assuming that the carryover effect of a treatment is the same regardless of the treatment in the next period, the model assumes that the carryover effect of a treatment on itself is different from the carryover effect on other treatments. For the traditional model, optimal designs tend to have pairs of consecutive identical treatments; for the model considered here, they tend to avoid such pairs. Practitioners have long expressed reservations about designs that exhibit such pairs and about the traditional model. The new model provides an attractive alternative that leads to appealing optimal designs.

On the Triangle Test with Replications

Abstract We consider the triangle test with replications. A commonly used test statistic for this situation is the sum of all correct assessments, summed over all assessors. Several authors argue that the binomial distribution cannot be used to analyse this kind of data. Brockhoff and Schlich [ Brockhoff, P.B., & Schlich, P. (1998). Handling replications in discriminations tests. Food Quality and Preference, 9, 303–312. ] propose an alternative model for the triangular test with replicates, where the assessors have different probabilities to correctly identify the odd sample even if the products are identical. Although we agree that assessors will have different probabilities of correct assessment if there are true differences, we do not think that Brockhoff and Schlichs model makes sense under the null hypothesis of equality of treatments. We show that all assessments are independent and have success probability 1 3 , if the null hypothesis is true and the experiment is properly randomized and properly carried out. This implies that the sum of all correct assessments is binomial with parameter p= 1 3 . Therefore the usual test based on this sum and the critical values of the binomial distribution is a level α test for the null hypothesis of equality of the products, even if there are replications.

Comparing generalized procrustes analysis and statis

We consider a model for sensory profiling data including translation, rotation and scaling. We compare two methods to calculate an overall consensus from several data matrices: GPA and STATIS. These methods are briefly illustrated and explained under our model. A series of simulations to compare their performance has been carried out. We found significant differences in performance depending on the variance of random errors and on the dimensionality of the true underlying consensus. Therefore we investigated on the dimensionality of the calculated group averages. We found both methods to give too many dimensions compared to the true consensus. This finding is supported by some theoretical considerations. Finally we propose a combined approach which takes advantage of both methods and which gave better results in the simulations.

Exact optimal designs for weighted least squares analysis with correlated errors

In the common linear and quadratic regression model with an autoregressive error structure exact D-optimal designs for weighted least squares analysis are determined. It is demonstrated that for highly correlated observations the D-optimal design is close to the equally spaced design. Moreover, the equally spaced design is usually very efficient, even for moderate sizes of the correlation, while the D-optimal design obtained under the assumptions of independent observations yields a substantial loss in efficiency. We also consider the problem of designing experiments for weighted least squares estimation of the slope in a linear regression and compare the exact D-optimal designs for weighted and ordinary least squares analysis.

Suitability of free choice profiling for assessment of orange-based carbonated soft-drinks

A Free Choice Profiling test was carried out to find out whether assessors with a basic sensory knowledge are able to characterise flavour and odour of orange-based lemonades. The experiment was run with eight different samples of orange-based soft drinks in four replicates. Generalized Procrustes Analysis combined with Principal Component Analysis were used to analyse and plot the data. The analysis of the four separate replicates showed, that the assessors were not able to reproduce their description of the sensory characteristics suitably, there were relatively large differences between the assessments of the same products by the same assessors in the different replicates. Even for this data, however, the permutation test showed a highly significant group average configuration. This was mainly due to differences between the replicates. But a large residual variance remained. Only about 30% of the variability could be explained by the group average configuration, and only about 12% of the total variance could be explained by the first two dimensions of it. When the four assessments of each product were averaged to stabilize the data, 60% of the total variance could be explained by the group average configuration and the first two dimensions of it could explain 30% of the total variability. Therefore, at least a coarse interpretation of the consensus space could be made. Our general finding was that profiling of orange-based lemonades by nearly untrained assessors appears to be useful but should be considered with some reserve. It seems that the differences between the products were too subtle for a reproducible profiling by these methods. However, some interpretable results on the structure of the products and on the variables could be gained.

Designs balanced for circular residual effects

Magda (1980) introduced a model for repeated measurements designs with a circular structure of the residual effects. He proved the universal optimality of circular balanced uniform designs over a subclass of the possible designs. We strengthen his result to optimality over the set of all designs with the same number of experimental units, periods and treatments.

On the use of the factor-sparsity assumption to get an estimate of the variance in saturated designs

In unreplicated multifactor designs, it is customary to nominate some factor interactions as not active—that is, to assume that they have no influence on the response—and to use the corresponding sums of squares to calculate an estimate of the variance. This estimate usually is based on very few degrees of freedom. An estimate of the variance is needed if one wants to find out which of the factors are really important. In many cases, especially in screening designs, it can be assumed that most factors and interactions are 0. When the design is orthogonal, then this assumption, called factor sparsity, is often used to get alternative estimates that are based on more degrees of freedom. There are several proposals for such estimators. This article uses approximations to discuss the respective advantages and disadvantages of four of them. I also show that such estimates of variance proposed for orthogonal designs can be extended to the nonorthogonal case. A method of analysis for nonorthogonal data is propos...