High Percentage Of Missing Data Research Articles

Data‐driven Buck converter model identification method with missing outputs

AbstractA data‐driven Buck converter model identification method is proposed to deal with missing (incomplete) outputs, which is robust to the data length and percentage of missing data. A nuclear norm based convex optimization problem instead of linear interpolation, to guarantee the recovered missing data satisfying the potential model structured low‐rank character, is constructed to estimate missing outputs. The alternating direction method of multiplier strategy is used to solve the nuclear norm based convex optimization problem. In this way, the high‐quality missing data can be estimated, even for short data length and high percentage of missing data. Based on the recovered data, the subspace identification method provides accurate estimates of the structure and parameter of the Buck converter synchronously. By applying the proposed method to a Buck converter, experimental results demonstrate its effectiveness.

Solving the missing value problem in PCA by Orthogonalized-Alternating Least Squares (O-ALS)

Dealing with missing data poses a challenge in Principal Component Analysis (PCA) since the most common algorithms are not designed to handle them. Several approaches have been proposed to solve the missing value problem in PCA, such as Imputation based on SVD (I-SVD), where missing entries are filled by imputation and updated in every iteration until convergence of the PCA model, and the adaptation of the Nonlinear Iterative Partial Least Squares (NIPALS) algorithm, able to work skipping the missing entries during the least-squares estimation of scores and loadings. However, some limitations have been reported for both approaches. On the one hand, convergence of the I-SVD algorithm can be very slow for data sets with a high percentage of missing data. On the other hand, the orthogonality properties among scores and loadings might be lost when using NIPALS.To solve these issues and perform PCA of data sets with missing values without the need of imputation steps, a novel algorithm called Orthogonalized-Alternating Least Squares (O-ALS) is proposed. The O-ALS algorithm is an alternating least-squares algorithm that estimates the scores and loadings subject to the Gram-Schmidt orthogonalization constraint. The way to estimate scores and loadings is adapted to work only with the available information.In this study, the performance of O-ALS is tested and compared with NIPALS and I-SVD in simulated data sets and in a real case study. The results show that O-ALS is an accurate and fast algorithm to analyze data with any percentage and distribution pattern of missing entries, being able to provide correct scores and loadings in cases where I-SVD and NIPALS do not perform satisfactorily.

Performance of Different Imputation Methods in Logistic Regression with Multicollinearity

The existence of missing values in the data, which must be taken into account, is an important issue in research. In particular, in the case of medical data, missing data can lead to a loss of trust in patient evaluation and make it impossible to classify people according to their level of health or disease. Furthermore, multicollinearity between independent variables can lead to misleading results. Therefore, the present research aims to study the efficiency of missing data imputation techniques for logistic regression with complete multicollinearity. The missing data imputation methods considered in this research were MEAN (mean imputation), MI (multiple imputation), KNN (k-nearest neighbor imputation), RF (random forest imputation), SRI (stochastic regression imputation), and BRI (Bayesian linear regression imputation). In this study, the simulation conducted was done with sample sizes of 20, 50, 100, 150, 200, 500, and 1000, as well as percentages of missing data at 10, 20, 30, and 40%. To compare efficiency, the EMSE (estimated mean square error) was used. The results showed that the RF method was most effective when the sample size was large and with a high percentage of missing data, whereas the MEAN method had the worst performance in all cases. When the sample size is small with a high proportion of missing data, KNN might be a preferable alternative for imputation. The inquiry into expanding the number of independent variables and different patterns of multicollinearity may be essential for future work.

Equivalence of variance components between standard and recursive genetic models using LDL′ transformations

BackgroundRecursive models are a category of structural equation models that propose a causal relationship between traits. These models are more parameterized than multiple trait models, and they require imposing restrictions on the parameter space to ensure statistical identification. Nevertheless, in certain situations, the likelihood of recursive models and multiple trait models are equivalent. Consequently, the estimates of variance components derived from the multiple trait mixed model can be converted into estimates under several recursive models through LDL′ or block-LDL′ transformations.ResultsThe procedure was employed on a dataset comprising five traits (birth weight—BW, weight at 90 days—W90, weight at 210 days—W210, cold carcass weight—CCW and conformation—CON) from the Pirenaica beef cattle breed. These phenotypic records were unequally distributed among 149,029 individuals and had a high percentage of missing data. The pedigree used consisted of 343,753 individuals. A Bayesian approach involving a multiple-trait mixed model was applied using a Gibbs sampler. The variance components obtained at each iteration of the Gibbs sampler were subsequently used to estimate the variance components within three distinct recursive models.ConclusionsThe LDL′ or block-LDL′ transformations applied to the variance component estimates achieved from a multiple trait mixed model enabled inference across multiple sets of recursive models, with the sole prerequisite of being likelihood equivalent. Furthermore, the aforementioned transformations simplify the handling of missing data when conducting inference within the realm of recursive models.

Enhancing environmental data imputation: A physically-constrained machine learning framework

In water resources management, new computational capabilities have made it possible to develop integrated models to jointly analyze climatic conditions and water quantity/quality of the entire watershed system. Although the value of this integrated approach has been demonstrated so far, the limited availability of field data may hinder its applicability by causing high uncertainty in the model response. In this context, before collecting additional data, it is recommended first to recognize what improvement in model performance would occur if all available records could be well exploited. This work proposes a novel machine learning framework with physical constraints capable of successfully imputing a high percentage of missing data belonging to several environmental domains (meteorology, water quantity, water quality), yielding satisfactory results. In particular, the minimum NSE computed for meteorologic variables is 0.72. For hydrometric variables, NSE is always >0.97. More than 78 % of the physical-water-quality variables is characterized by NSE > 0.45, and >66 % of the chemical-water quality variables reaches NSE > 0.35. This work's results demonstrate the proposed framework's effectiveness as a data augmentation tool to improve the performance of integrated environmental modeling.

Is it time to rethink disability assessment in low back pain? Reliability, internal consistency, and construct validity of the Brazilian WHODAS 2.0 for chronic low back pain.

The World Health Disability Assessment Schedule 2.0 (WHODAS 2.0) was developed to assess health and disability based on the biopsychosocial model. The WHODAS 2.0 has not been validated for Brazilians with chronic non-specific low back pain (LBP). We aimed to evaluate the reliability, internal consistency, and construct validity of the Brazilian version of the WHODAS 2.0 in patients with chronic LBP. Methodological study. The Brazilian version of the WHODAS 2.0 was applied to 100 volunteers with chronic nonspecific LBP. Test-retest reliability, internal consistency, and construct validity were assessed using the Spearman correlation test, Cronbach's alpha (α) coefficient, and Spearman's correlation test between WHODAS 2.0, the Oswestry Disability Index (ODI), Roland-Morris Disability Questionnaire (RMDQ), and Fear Avoidance Beliefs Questionnaire (FABQ), respectively. WHODAS 2.0 showed satisfactory test-retest reliability with a moderate correlation for total WHODAS 2.0 (r=0.75, p<0.05). Internal consistency was adequate for all domains and total score (α=0.82-0.96). Regarding construct validity, WHODAS 2.0, ODI (r=0.70, p<0.05), and WHODAS 2.0 and RMDQ (r=0.71, p<0.05) had significant correlations. Total WHODAS 2.0 and FABQ-Phys subscale scores correlated moderately (r=0.66, p<0.05). The Brazilian WHODAS 2.0 was proved to be a valid and reliable tool for patients with chronic LBP. The item referring to sexual intercourse had 27% and 30% of the missing values during the test and retest stage, respectively and had a high percentage of missing data for work-related questions (41% missing data) in the life activities domain; therefore, the data must be interpreted with caution. WHODAS 2.0can be used as a disability assessment strategy from a biopsychosocial perspective in this population.

New computations for RMSEA and CFI following FIML and TS estimation with missing data.

The full-information maximum likelihood (FIML) is a popular estimation method for missing data in structural equation modeling (SEM). However, previous research has shown that SEM approximate fit indices (AFIs) such as the root mean square error of approximation (RMSEA) and the comparative fit index (CFI) can be distorted relative to their complete data counterparts when they are computed following the FIML estimation. The main goal of the current paper is to propose and examine an alternative approach for computing AFIs following the FIML estimation, which we refer to as the FIML-corrected or FIML-C approach. The secondary goal of the article is to examine another existing estimation method, the two-stage (TS) approach, for computing AFIs in the presence of missing data. Both FIML-C and TS approaches remove the bias due to missing data, so that the resulting incomplete data AFIs estimate the same population values as their complete data counterparts. For both approaches, we also propose a series of small sample corrections to improve the estimates of AFIs. In two simulation studies, we found that the FIML-C and TS approaches, when implemented with small sample corrections, estimated the population-complete-data AFIs with little bias across a variety of conditions, although the FIML-C approach can fail in a small number of conditions with a high percentage of missing data and a high degree of model misspecification. In contrast, the FIML AFIs as currently computed often performed poorly. We recommend FIML-C and TS approaches for computing AFIs in SEM. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Descriptive Assessment of Race, Ethnicity, Comorbidities, and SARS-CoV-2 Infection- Fatality in North Carolina

Background SARS-CoV-2 infection has caused variable clinical outcomes including hospitalization and death. We analyzed state-level data from the North Carolina COVID-19 Surveillance System (NC COVID) to describe demographics of those infected with SARS-CoV-2 and to describe factors associated with infection-fatality in North Carolina. Methods This was a retrospective cohort study using surveillance data on positive SARS-CoV-2-infected individuals (N = 214,179) identified between March 1, 2020, and September 30, 2020. We present descriptive statistics and associations among demographics, medical comorbidities, and SARS-CoV-2 infection-fatality. Results Median age for residents with reported SARS-CoV-2 was 38 (IQR 23–54). Age was strongly correlated with SARS-CoV-2 infection-fatality. Greater infection-fatality was noted among those who identified as Black across all comorbidities. Coexisting chronic disease was associated with greater infection-fatality, with kidney disease demonstrating the strongest association. Limitations A high percentage of missing data for race/ethnicity and comorbidities limits the interpretation of our findings. Data were not available for socioeconomic measures that could aid in better understanding inequities associated with SARS-CoV-2 infection-fatality. Conclusions Among North Carolinians identified with SARS-CoV-2 via surveillance efforts, age, race, and comorbidities were associated with infection-fatality; these findings are similar to those of studies using different source populations in the United States. In addition to age and other nonmodifiable variables, systematic differences in social conditions and opportunity may increase the risk of SARS-CoV-2 infection-fatality among Black Americans compared to other races/ethnicities.

Phylogeny of sea spiders (Arthropoda: Pycnogonida) inferred from mitochondrial genome and 18S ribosomal RNA gene sequences

The phylogeny of sea spiders has been debated for more than a century. Despite several molecular studies in the last twenty years, interfamilial relationships remain uncertain. In the present study, relationships within Pycnogonida are examined in the light of a new dataset composed of 160 mitochondrial genomes (including 152 new sequences) and 130 18S rRNA gene sequences (including 120 new sequences), from 141 sea spider morphospecies representing 26 genera and 9 families. Node congruence between mitochondrial and nuclear markers was analysed to identify the most reliable relationships. We also reanalysed a multilocus dataset previously published and showed that the high percentages of missing data make phylogenetic conclusions difficult and uncertain.Our results support the monophyly of most families currently accepted, except Callipallenidae and Nymphonidae, the monophyly of the superfamilies Ammotheoidea (Ammotheidae + Pallenopsidae), Nymphonoidea (Nymphonidae + Callipallenidae), Phoxichilidioidea (Phoxichilidiidae + Endeidae) and Colossendeoidea (Colossendeidae + Pycnogonidae + Rhynchothoracidae), and the sister-group relationship between Ammotheoidea and Phoxichilidioidea. We discuss the morphological evolution of sea spiders, identifying homoplastic characters and possible synapomorphies. We also discuss the palaeontological and phylogenetic arguments supporting either a radiation of sea spiders prior to Jurassic or a progressive diversification from Ordovician or Cambrian.

Perbandingan Performa Metode Pohon Model Logistik dan Random Forest pada Pengklasifikasian Data

Multicollinearity and missing data are two common problems in big data. Missing data could decrease the prediction accuracy. Logistic model tree (LMT) is used to handle multicollinearity because multicollinearity does not affect the decision tree. Random forest can be used to decrease variance in prediction case. This study aimed to study the comparison of two methods, LMT and random forest, in multicollinearity and missing data in various cases using simulation study and real data as dataset. Evaluation model is based on classification accuracy and AUC measurement. The result stated that random forest had better performance if the multicollinearity level is moderate. LMT with omitted missing data is proven to have better performance for big data and when a high percentage of missing data occurred, and the multicollinearity level is severe. The next step is analysed real data with different sample size. The result stated that random forest have better performance. Omitted missing data have better performance in classification “breast cancer” data which consist 0,3 % missing data.

Estimativa de dados faltantes de precipitação mensal no sudoeste da Colômbia: comparação de diferentes métodos

ABSTRACT Historical rainfall records are relevant in hydrometeorological studies because they provide information on the spatial features, frequency, and amount of precipitated water in a specific place, therefore, it is essential to make an adequate estimation of missing data. This study evaluated four methods for estimating missing monthly rainfall data at 46-gauge stations in southwestern Colombia covering 1983-2019. The performance of the Normal Ratio (NR), Principal Components Regression (PCR), Principal Least Square Regression (PLSR), and Artificial Neural Networks (ANN) methods were compared using three standardized error metrics: Root Mean Square Error (RMSE), Percent BIAS (PBIAS), and Mean Absolute Error (MAE). The results generally showed a better performance of the nonlinear ANN method. Regarding the linear methods, the best performance was registered by the PLSR, followed by the PCR. The results suggest the applicability of the ANN method in regions with a low density of stations and a high percentage of missing data, such as southwestern Colombia.

121 (PB111) - Systematic evaluation of label-free protein quantification pipelines in 12 mouse syngeneic models

Background: The widespread application of omics technologies in the past two decades has significantly expanded our knowledge of cancer biology. Proteins act as the link between genetic code and phenotype and can better reflect the dynamic state of a cell. With rapid advances in the field of proteomics, especially the label-free proteomics quantification technology powered by LC-MS/MS, it is now a routine to characterize and quantify thousands of proteins in tumor samples in a timely manner. Data-dependent acquisition (DDA) has been the workhorse for bottom-up proteomics within the past few years, however, it suffers from a high percentage of missing data originating from its stochastic nature.

Study protocol for UNICEF and WHO estimates of global, regional, and national low birthweight prevalence for 2000 to 2020.

Background Reducing low birthweight (LBW, weight at birth less than 2,500g) prevalence by at least 30% between 2012 and 2025 is a target endorsed by the World Health Assembly that can contribute to achieving Sustainable Development Goal 2 (Zero Hunger) by 2030. The 2019 LBW estimates indicated a global prevalence of 14.6% (20.5 million newborns) in 2015. We aim to develop updated LBW estimates at global, regional, and national levels for up to 202 countries for the period of 2000 to 2020. Methods Two types of sources for LBW data will be sought: national administrative data and population-based surveys. Administrative data will be searched for countries with a facility birth rate ≥80% and included when birthweight data account for ≥80% of UN estimated live births for that country and year. Surveys with birthweight data published since release of the 2019 edition of the LBW estimates will be adjusted using the standard methodology applied for the previous estimates. Risk of bias assessments will be undertaken. Covariates will be selected based on a conceptual framework of plausible associations with LBW, covariate time-series data quality, collinearity between covariates and correlations with LBW. National LBW prevalence will be estimated using a Bayesian multilevel-mixed regression model, then aggregated to derive regional and global estimates through population-weighted averages. Conclusion Whilst availability of LBW data has increased, especially with more facility births, gaps remain in the quantity and quality of data, particularly in low-and middle-income countries. Challenges include high percentages of missing data, lack of adherence to reporting standards, inaccurate measurement, and data heaping. Updated LBW estimates are important to highlight the global burden of LBW, track progress towards nutrition targets, and inform investments in programmes. Reliable, nationally representative data are key, alongside investments to improve the measurement and recording of an accurate birthweight for every baby.

Preprocessing Methods for Ambulatory HRV Analysis Based on HRV Distribution, Variability and Characteristics (DVC).

Thanks to wearable devices joint with AI algorithms, it is possible to record and analyse physiological parameters such as heart rate variability (HRV) in ambulatory environments. The main downside to such setups is the bad quality of recorded data due to movement, noises, and data losses. These errors may considerably alter HRV analysis and should therefore be addressed beforehand, especially if used for medical diagnosis. One widely used method to handle such problems is interpolation, but this approach does not preserve the time dependence of the signal. In this study, we propose a new method for HRV processing including filtering and iterative data imputation using a Gaussian distribution. The particularity of the method is that many physiological aspects are taken into consideration, such as HRV distribution, RR variability, and normal boundaries, as well as time series characteristics. We study the effect of this method on classification using a random forest classifier (RF) and compare it to other data imputation methods including linear, shape-preserving piecewise cubic Hermite (pchip), and spline interpolation in a case study on stress. Features from reconstructed HRV signals of 67 healthy subjects using all four methods were analysed and separately classified by a random forest algorithm to detect stress against relaxation. The proposed method reached a stable F1 score of 61% even with a high percentage of missing data, whereas other interpolation methods reached approximately 54% F1 score for a low percentage of missing data, and the performance drops to about 44% when the percentage is increased. This suggests that our method gives better results for stress classification, especially on signals with a high percentage of missing data.

Exploration of Rain Gauge Quality Issues in Northern England

The rain gauge, which is the preferred device for rain rate quantification, is susceptible to various types of errors. The problems of inconsistency and incompletion of rainfall time-series, in addition to data quality issues obstruct hydrometeorological analysis. Daily rainfall data from 30 gauges in Northern England were quality controlled by a double quality check procedure for the period 2006–2014 to solve any specific data quality issues. It emerged that fault readings happened only over a few days compared with the high percentage of missing data which could reach years. Thus, factors that impact gauges’ missing data were inspected. It was found that snowfalls, land cover-land use and bird distribution are the dominant factors that malfunctioning the rain gauge for long time periods, often producing successive missing days that could be extended months in total. However, when the missing data happened on separate days, then it seemed that high winds related with light rainfall resulted in drifting away the rainfall from the rain gauge and thus caused missing data. Moreover, the rainfall itself was not one of the major factors to malfunction the gauge, because the values of missed rainfall were not quite extreme to do this.

Water-Quality Data Imputation with a High Percentage of Missing Values: A Machine Learning Approach

The monitoring of surface-water quality followed by water-quality modeling and analysis are essential for generating effective strategies in surface-water-resource management. However, worldwide, particularly in developing countries, water-quality studies are limited due to the lack of a complete and reliable dataset of surface-water-quality variables. In this context, several statistical and machine-learning models were assessed for imputing water-quality data at six monitoring stations located in the Santa Lucía Chico river (Uruguay), a mixed lotic and lentic river system. The challenge of this study is represented by the high percentage of missing data (between 50% and 70%) and the high temporal and spatial variability that characterizes the water-quality variables. The competing algorithms implement univariate and multivariate imputation methods (inverse distance weighting (IDW), Random Forest Regressor (RFR), Ridge (R), Bayesian Ridge (BR), AdaBoost (AB), Hubber Regressor (HR), Support Vector Regressor (SVR) and K-nearest neighbors Regressor (KNNR)). According to the results, more than 76% of the imputation outcomes are considered “satisfactory” (NSE &gt; 0.45). The imputation performance shows better results at the monitoring stations located inside the reservoir than those positioned along the mainstream. IDW was the model with the best imputation results, followed by RFR, HR and SVR. The approach proposed in this study is expected to aid water-resource researchers and managers in augmenting water-quality datasets and overcoming the missing data issue to increase the number of future studies related to the water-quality matter.

The potential of genome-wide RAD sequences for resolving rapid radiations: a case study in Cactaceae

The reconstruction of relationships within recently radiated groups is challenging even when massive amounts of sequencing data are available. The use of restriction site-associated DNA sequencing (RAD-Seq) to this end is promising. Here, we assessed the performance of RAD-Seq to infer the species-level phylogeny of the rapidly radiating genus Cereus (Cactaceae). To examine how the amount of genomic data affects resolution in this group, we used datasets and implemented different analyses. We sampled 52 individuals of Cereus, representing 18 of the 25 species currently recognized, plus members of the closely allied genera Cipocereus and Praecereus, and other 11 Cactaceae genera as outgroups. Three scenarios of permissiveness to missing data were carried out in iPyRAD, assembling datasets with 30% (333 loci), 45% (1440 loci), and 70% (6141 loci) of missing data. For each dataset, Maximum Likelihood (ML) trees were generated using two supermatrices, i.e., only SNPs and SNPs plus invariant sites. Accuracy and resolution were improved when the dataset with the highest number of loci was used (6141 loci), despite the high percentage of missing data included (70%). Coalescent trees estimated using SVDQuartets and ASTRAL are similar to those obtained by the ML reconstructions. Overall, we reconstruct a well-supported phylogeny of Cereus, which is resolved as monophyletic and composed of four main clades with high support in their internal relationships. Our findings also provide insights into the impact of missing data for phylogeny reconstruction using RAD loci.

Ultra-Conserved Elements and morphology reciprocally illuminate conflicting phylogenetic hypotheses in Chalcididae (Hymenoptera, Chalcidoidea).

Recent technical advances combined with novel computational approaches have promised the acceleration of our understanding of the tree of life. However, when it comes to hyperdiverse and poorly known groups of invertebrates, studies are still scarce. As published phylogenies will be rarely challenged by future taxonomists, careful attention must be paid to potential analytical bias. We present the first molecular phylogenetic hypothesis for the family Chalcididae, a group of parasitoid wasps, with a representative sampling (144 ingroups and seven outgroups) that covers all described subfamilies and tribes, and 82% of the known genera. Analyses of 538 Ultra-Conserved Elements (UCEs) with supermatrix (RAxML and IQTREE) and gene tree reconciliation approaches (ASTRAL, ASTRID) resulted in highly supported topologies in overall agreement with morphology but reveal conflicting topologies for some of the deepest nodes. To resolve these conflicts, we explored the phylogenetic tree space with clustering and gene genealogy interrogation methods, analyzed marker and taxon properties that could bias inferences and performed a thorough morphological analysis (130 characters encoded for 40 taxa representative of the diversity). This joint analysis reveals that UCEs enable attainment of resolution between ancestry and convergent/divergent evolution when morphology is not informative enough, but also shows that a systematic exploration of bias with different analytical methods and a careful analysis of morphological features is required to prevent publication of artifactual results. We highlight a GC content bias for maximum-likelihood approaches, an artifactual mid-point rooting of the ASTRAL tree and a deleterious effect of high percentage of missing data (>85% missing UCEs) on gene tree reconciliation methods. Based on the results we propose a new classification of the family into eight subfamilies and ten tribes that lay the foundation for future studies on the evolutionary history of Chalcididae.

Genetic structure of Italian population of the grapevine downy mildew agent, Plasmopara viticola

AbstractDowny mildew, caused by the Oomycete Plasmopara viticola, is one of the most important diseases affecting the Eurasian grapevine, Vitis vinifera. The pathogen originated in Northern America and its presence was signalled for the first time into Europe in 1878. In this study, the genetic variability and structure of Italian P. viticola population was investigated by genotyping 106 P. viticola strains belonging to 12 different regions, at 31 microsatellite markers. As a result of the high percentage of missing data, 96 strains and 19 loci were retained for the data analysis. The overall Italian population presents absence of clones, evidence of sexual and asexual reproduction and a low genetic diversity, as expected for an introduced pathogen, but a slightly higher genetic diversity than that found in other European populations, based on allelic diversity at the investigated microsatellite loci. Out of 19 loci, half shows deviation from Hardy–Weinberg equilibrium and, indeed, structure analysis indicates the presence of two separate genetic clusters, with little but significantly different distribution according to geography (west–east gradient) and climatic conditions. Overall, the analysis of the P. viticola population, 140 years after its appearance in Italy, provides indication on the pathogen adaptability. This should be taken into consideration in the future for preserving the durability of disease resistant varieties in open field. In this view, all the disease control methods available should be integrated in order to reduce the selection of pathogen strains able to overcome plant resistance.

Biases with the Generalized Euclidean Distance measure in disparity analyses with high levels of missing data

AbstractThe Generalized Euclidean Distance (GED) measure has been extensively used to conduct morphological disparity analyses based on palaeontological matrices of discrete characters. This is in part because some implementations allow the use of morphological matrices with high percentages of missing data without needing to prune taxa for a subsequent ordination of the data set. Previous studies have suggested that this way of using the GED may generate a bias in the resulting morphospace, but a detailed study of this possible effect has been lacking. Here, we test whether the percentage of missing data for a taxon artificially influences its position in the morphospace, and if missing data affects pre‐ and post‐ordination disparity measures. We find that this use of the GED creates a systematic bias, whereby taxa with higher percentages of missing data are placed closer to the centre of the morphospace than those with more complete scorings. This bias extends into pre‐ and post‐ordination calculations of disparity measures and can lead to erroneous interpretations of disparity patterns, especially if specimens present in a particular time interval or clade have distinct proportions of missing information. We suggest that this implementation of the GED should be used with caution, especially in cases with high percentages of missing data. Results recovered using an alternative distance measure, Maximum Observed Rescaled Distance (MORD), are more robust to missing data. As a consequence, we suggest that MORD is a more appropriate distance measure than GED when analysing data sets with high amounts of missing data.