CHAID Research Articles

408 - Muscle strength is an important predictor for limitations in basic activities of daily living among people with severe disability

Objective:This study aimed to identify predictors of limitations in basic activities of daily living (BADL) among people with severe disabilities.Methods:4075 long-term care beneficiaries with severe disabilities in Guangzhou, China, were included during July 2018 and March 2019. BADL was assessed using the Barthel index (BI). Muscle strength was measured by using the Lovett Rating Scale. Age, gender, comorbidities, and muscle strengths were collected as independent variables. Chi-square Automatic Interaction Detector (CHAID) method was used to examine associations between independent variables and item scores of the BI.Results:Muscle strength and history of stroke were parent node and child node for most of BADL limitations, respectively. Upper limb muscle strength (≤ 3) was a major predictor for dependence in feeding, grooming, toileting, dressing, and transfer, while lower limb muscle strength (≤ 3) was a major predictor for limitation in mobility.Conclusions:Muscle strength was the strongest predictor of BADLs among people with severe disability. Muscle strength grading may be optimal for designing supporting strategies for people with severe disabilities.

Synergistic Effect of Static Compliance and D-dimers to Predict Outcome of Patients with COVID-19-ARDS: A Prospective Multicenter Study.

The synergic combination of D-dimer (as proxy of thrombotic/vascular injury) and static compliance (as proxy of parenchymal injury) in predicting mortality in COVID-19-ARDS has not been systematically evaluated. The objective is to determine whether the combination of elevated D-dimer and low static compliance can predict mortality in patients with COVID-19-ARDS. A “training sample” (March–June 2020) and a “testing sample” (September 2020–January 2021) of adult patients invasively ventilated for COVID-19-ARDS were collected in nine hospitals. D-dimer and compliance in the first 24 h were recorded. Study outcome was all-cause mortality at 28-days. Cut-offs for D-dimer and compliance were identified by receiver operating characteristic curve analysis. Mutually exclusive groups were selected using classification tree analysis with chi-square automatic interaction detection. Time to death in the resulting groups was estimated with Cox regression adjusted for SOFA, sex, age, PaO2/FiO2 ratio, and sample (training/testing). “Training” and “testing” samples amounted to 347 and 296 patients, respectively. Three groups were identified: D-dimer ≤ 1880 ng/mL (LD); D-dimer > 1880 ng/mL and compliance > 41 mL/cmH2O (LD-HC); D-dimer > 1880 ng/mL and compliance ≤ 41 mL/cmH2O (HD-LC). 28-days mortality progressively increased in the three groups (from 24% to 35% and 57% (training) and from 27% to 39% and 60% (testing), respectively; p < 0.01). Adjusted mortality was significantly higher in HD-LC group compared with LD (HR = 0.479, p < 0.001) and HD-HC (HR = 0.542, p < 0.01); no difference was found between LD and HD-HC. In conclusion, combination of high D-dimer and low static compliance identifies a clinical phenotype with high mortality in COVID-19-ARDS.

Consumer Profile Segmentation in Online Lottery Gambling Utilizing Behavioral Tracking Data from the Portuguese National Lottery.

The present study is the first to examine account-based tracking data of Portuguese online lottery players comprising the gambling activity of all active players over a one-year period (N = 154,585). The main research goal was the identification of groups or segments of players by their engagement levels (high, neutral, low) and to assess preferences in product category with the use of CHAID (Chi-Square Automatic Interaction Detection) segmentation models, based on expenditure and sociodemographic variables. Findings showed that (1) age was found to be the most influential differentiating variable in player segmentation and had a positive correlation with expenditures and wagers, (2) gender was the second most influential variable (males represented 78.7% of players), (3) education the third most influential variable and had a negative correlation with expenditure, and (4) region was the least relevant variable. The models generated several players segments that engaged in different games. Older males (54–64 years; ≥ 65 years) were the most engaged overall. Younger males (18–34 years) were the least engaged but showed preferences for lotto as did females (35–49 years). Lower educated males and older males (49 years+) with a high school education were the most engaged in instant lottery games. These findings show that Portuguese lottery players can be grouped into several segments with distinct demographic characteristics and corresponding engagement levels. These findings help support more effective marketing segmentation and will help in the targeting of responsible gambling approaches.

Effects of proximal priority and distal priority robotic priming techniques with impairment-oriented training of upper limb functions in patients with chronic stroke: study protocol for a single-blind, randomized controlled trial

BackgroundThe sequence of establishing a proximal stability or function before facilitation of the distal body part has long been recognized in stroke rehabilitation practice but lacks scientific evidence. This study plans to examine the effects of proximal priority robotic priming and impairment-oriented training (PRI) and distal priority robotic priming and impairment-oriented training (DRI).MethodsThis single-blind, randomized, comparative efficacy study will involve 40 participants with chronic stroke. Participants will be randomized into the PRI or DRI groups and receive 18 intervention sessions (90 min/day, 3 days/weeks for 6 weeks). The Fugl-Meyer Assessment Upper Extremity subscale, Medical Research Council Scale, Revised Nottingham Sensory Assessment, and Wolf Motor Function Test will be administered at baseline, after treatment, and at the 3-month follow-up. Two-way repeated-measures analysis of variance and the chi-square automatic interaction detector method will be used to examine the comparative efficacy and predictors of outcome, respectively, after PRI and DRI.DiscussionThrough manipulating the sequence of applying wrist and forearm robots in therapy, this study will attempt to examine empirically the priming effect of proximal or distal priority robotic therapy in upper extremity impairment-oriented training for people with stroke. The findings will provide directions for further studies and empirical implications for clinical practice in upper extremity rehabilitation after stroke.Trial RegistrationClinicalTrials.gov NCT04446273. Registered on June 23, 2020.

Body weight prediction using different data mining algorithms in Thalli sheep: A comparative study.

Background and Aim:The Thalli sheep are the main breed of sheep in Pakistan, and an effective method to predict their body weight (BW) using linear body measurements has not yet been determined. Therefore, this study aims to establish an algorithm with the best predictive capability, among the Chi-square automatic interaction detector (CHAID), exhaustive CHAID, artificial neural network, and classification and regression tree (CART) algorithms, in live BW prediction using selected body measurements in female Pakistani Thalli sheep.Materials and Methods:A total of 152 BW records, including nine continuous predictors (wither height, body length [BL], head length, rump length, tail length, head width, rump width, heart girth [HG], and barrel depth), were utilized. The coefficient of determination (R2), standard deviation ratio, root-mean-square error (RMSE), etc., were calculated for each algorithm.Results:The R2 (%) values ranged from 49.28 (CART) to 64.48 (CHAID). The lowest RMSE was found for CHAID (2.61), and the highest one for CART (3.12). The most significant predictors were the HG of live BW for all algorithms. The heaviest average BW (41.12 kg) was observed in the subgroup of those having a BL of >73.91 cm (Adjusted p=0.045).Conclusion:Among the algorithms, CHAID provided the most appropriate predictive capability in the prediction of live BW for female Thalli sheep. In general, the applied algorithms accurately predicted the BW of Thalli sheep, which can be very helpful in deciding on the standards, available drug doses, and required feed amount for animals.

Employment of CHAID and CRT decision tree algorithms to develop bid/no-bid decision-making models for contractors

PurposeBid/no-bid decision is a significant and strategic decision, which must be finalized at an early stage of the bidding process. Such decision-making may have significant impact on the performance of the contractors. Using Chi-square Automatic Interaction Detector (CHAID) and Classification and Regression (CRT) decision tree algorithms, this paper aims to develop bid/no-bid models for design-bid-build projects for contractors.Design/methodology/approachThe models in this study have been developed using CHAID and CRT algorithms. Thirty-four bid/no-bid key factors were collected via extensive research. The bid/no-bid factors were listed based on their importance index as a result of a questionnaire distributed among the construction professionals. These factors were divided into five main risk categories – owner, project, bidding situation, contract and contractor – which were taken as inputs for the models. Split-sample validation was applied for testing and measuring the accuracy of the CHAID and CRT models. Moreover, Spearman's rank correlation and Analysis of Variance (ANOVA) tests were employed to identify the statistical features of the received 169 responses.FindingsThe key bid/no-bid factors in construction industry were categorized in five related groups and ranked based on the relative importance index. It was found that the top 6 ranked bid/no-bid factors were (1) current workload, (2) need for work, (3) previous experience with employer; (4) timely payment by the employer; (5) availability of other projects for bidding (6) reputation of employer in the industry. Matrix comparison between all bid/no-bid groups was performed using Spearman's correlation to measure the relationship between each of the two paired groups. It was concluded that all the relationships were positive.Originality/valueExisting bidding models require many inputs and advanced understanding of mathematics and software to run the model. Contractors tend to use easy, fast and available support methods. Excluding a great number of the bid/no-bid factors may affect the final decision. This paper proposes a bid/no-bid decision tree models for contractors of different sizes. It is the first study in the literature, to the best of authors' knowledge, to study bid/no-bid decision with the proposed decision tree algorithm. The proposed models in this study overcome the shortfalls of most previous models such as avoiding the complexity and difficulties of applying the concept. The proposed model will provide the contractors with a bid/no-bid decision based on the input for the defined bid factor groups. The proposed models display the soft spots and hot spots between the independent and dependent variables, which leads to a better decision. The proposed models display the result effectively in visual terms, easy to understand and easy to apply. The proposed models are a form of multiple effect (or variable) analysis which allows the companies to explain, describe, predict or classify an outcome.

Dose Adjustment Model of Paliperidone in Patients With Acute Schizophrenia: A post hoc Analysis of an Open-Label, Single-Arm Multicenter Study.

This study aimed to investigate the factors that influenced the clinicians to adjust the paliperidone dose in the acute phase of schizophrenia. This was a post hoc study of an 8-week, open-label, single-arm multicenter trial which evaluated the efficacy, safety, and tolerability of flexible doses of paliperidone ER (3–12 mg/day) in patients with acutely exacerbated schizophrenia. Patients were divided into groups according to the dose at week 8 (3, 6, and 9–12 mg). The responder was defined as the reduction percentage in the Positive and Negative Syndrome Scale (PANSS) total score of ≥30%. According to the chi-squared automatic interaction detection algorithm, decision tree models predicting an increase in the dose of paliperidone ER were established. A decision tree, based on 4-week Marder positive factor, Clinical Global Impression (CGI), and BMI, was established to guide the dose adjustments of paliperidone ER in the acute phase of schizophrenia. The multivariable logistic regression analysis showed that lower age at onset, higher baseline PANSS positive subscale score, and lower baseline Personal and Social Performance Scale (PSP) score were significant predictors of increased dose in responders. Patients with young-onset age, severe baseline symptoms, and poor function are more likely to benefit from high dosage.

Antibiotic Discontinuation 24 h After Neonatal Late-Onset Sepsis Work-Up—A Validated Decision Tree Model

Objectives: Neonatal late-onset sepsis work-up is a frequent occurrence in every neonatal department. Blood cultures are the diagnostic gold standard, however, a negative culture prior to 48–72 h is often considered insufficient to exclude sepsis. We aimed to develop a decision tree which would enable exclusion of late-onset sepsis within 24 h using clinical and laboratory variables.Study Design: Infants evaluated for late-onset sepsis during the years 2016–2019, without major malformations, in a tertiary neonatal center were eligible for inclusion. Blood cultures and clinical and laboratory data were extracted at 0 and 24 h after sepsis work-up. Infants with bacteriologically confirmed late-onset sepsis were compared to matched control infants. Univariate logistic regression identified potential risk factors. A decision tree based on Chi-square automatic interaction detection methodology was developed and validated.Results: The study cohort was divided to a development cohort (105 patients) and a validation cohort (60 patients). At 24 h after initial evaluation, the best variables to identify sepsis were C-reactive protein > 0.75 mg/dl, neutrophil-to-lymphocyte ratio > 1.5 and sick-appearance at 24 h. Use of these 3 variables together with blood culture status at 24 h, enabled identification of all infants that eventually developed sepsis through the decision tree model. Our decision tree has an area under the receiver operating characteristic curve of 0.94 (95% CI: 0.90–0.98).Conclusions: In non-sick appearing infants with a negative blood culture at 24 h and normal laboratory values, sepsis is highly unlikely and discontinuing antibiotics after 24 h is a viable option.

Beyond Groceries: An Analysis of Referral Needs to Address Underlying Causes of Child Hunger among Households Accessing Food Pantries

ABSTRACT Very low food security among children (VLFS-C), often referred to as child hunger, can profoundly hinder child development, family well-being, and community health. Food pantries are important community resources that routinely serve at-risk families. This study investigated the influence of various candidate risk factors for VLFS-C within a food pantry population to inform the development of the “Pantry Assessment Tool against Child Hunger (PATCH).” We collected standardized surveys among a representative sample of households with children accessing food pantry services in Oklahoma (n = 188). Weighted analyses revealed a large majority of households experienced child-level food insecurity (70.6%), with nearly half reporting low food security and nearly one-quarter reporting VLFS-C. We then used logistic regression to identify factors associated with VLFS-C, followed by chi-square automatic interaction detection (CHAID) to assess if, and in what progression, significant risk factors predicted VLFS-C. In unadjusted models, annual household income <$15,000, non-urban residence, lack of health insurance, unstable housing, heavier food pantry reliance, fair or poor adult health, adult anxiety, and adult smoking to reduce hunger pangs were all positively associated with VLFS-C. Receipt of Special Supplemental Nutrition Program for Women, Infants and Children (WIC) and higher social support were protective against VLFS-C. However, in adjusted models, only receipt of WIC remained significant. CHAID analysis revealed that access to insurance best differentiated groups with and without VLFS-C. Informed by these analyses, the PATCH tool may be useful for the development of screening programs to identify and address potential root causes of VLFS-C in pantry settings.

Systemic Lupus Erythematosus Disease Activity Score (SLE-DAS) enables accurate and user-friendly definitions of clinical remission and categories of disease activity

ObjectivesThere is an unmet need for accurate and user-friendly definitions of systemic lupus erythematosus (SLE) disease activity and remission. We aimed to derive and validate the SLE Disease Activity Score...

Exploring Quality of Life Reported by Norwegian Older Adults Using Classification Tree Approach on Group Profiles

This study aims to explore the variation between- and within subgroups of older adults with regard to low, medium, and high levels of self-reported quality of life (QoL) measured by the WHOQOL-BREF scale. The contribution of interacting personal and contextual life conditions to QoL was examined in a sample of 1,910 (sample frame 6,000) Norwegian men and women aged 62 to 99 years. The data collected by a postal questionnaire were analyzed using the Chi-square Automatic Interaction Detection (CHAID) classification method in order to detect unique profiles of groups who shared common characteristics. The CHAID model revealed 15 relatively homogenous groups, but distinct from one another, whose profiles were defined by unique constellations of several interacting variables significantly related to a given QoL level. Mental functioning was predominantly linked to perception of life meaning along with health status, and/or in some cases living arrangement, loneliness, neighborhood quality, and satisfaction with income, and placed an individual at different likelihood levels of reporting low, medium, or high QoL. Socio-demographics had no statistically significant impact on QoL for any subgroup. Through this individual-oriented approach, a periodically ongoing assessment of subjective quality of life (QoL) may be sufficiently powerful to allow detecting and addressing personal concerns and specific needs that detract from quality of life in advancing age.

A hierarchy of correlates impacting adults’ sensor-based physical activity and sedentary time

ABSTRACT Investigation into correlates across all levels of the socio-ecological model predictive of objectively measured physical activity has rarely been assessed in adults. While considering a diversity of correlates, we determined which correlates best predict sensor-based moderate-to-vigorous physical activity (MVPA) and sedentary-time (ST) in adults. A Chi-squared Automatic Interaction Detection algorithm was used to hierarchize the correlates associated with high ST (≥66.6thpercentile) and sufficient MVPA (≥150 min/week) in 865 adults. The main correlate predictive of being active was currently partaking in sport/exercise. The following relevant correlates were being male for the exercisers and having trees in the neighbourhood for the non-exercisers. The final correlate to boost male exercisers’ MVPA was having lots of shops in the neighbourhood and not having television in the bedroom for women. The primary correlate for high ST was job activity level, with individuals having highly active jobs being less likely to exhibit high levels of ST; being single, male, and a former athlete also increased the chances of being highly sedentary. To increase adults’ MVPA, promotion of sport participation, neighbourhood landscape planning, shop availability, as well as limiting television in the bedroom must be prioritized. For counteracting ST, increasing workplace activity level is warranted.

Characteristics of child welfare investigations reported by healthcare professionals in Ontario: secondary analysis of a regional database

ObjectivesThis study examines the characteristics and outcomes of child welfare investigations reported by hospital-based and community-based healthcare professionals.MethodsA sample of 7590 child maltreatment-related investigations from the Ontario Incidence Study of...

How Does the Facilitation Effort of Clinical Educators Interact With Aspects of Organizational Context to Affect Research Use in Long-term Care? Evidence From CHAID Analysis.

Organizational context influences the effect of facilitation efforts on research use in care settings. The interactions of these factors are complex. Therefore, the use of traditional statistical methods to examine their interrelationships is often impractical. Big Data analytics can automatically detect patterns within the data. We applied the chi-squared automatic interaction detection (CHAID) algorithm and classification tree technique to explore the dynamic and interdependent relationships between the implementation science concepts-context, facilitation, and research use. Observational, cross-sectional study based on survey data collected from a representative sample of nursing homes in western Canada. We assessed three major constructs: (a) Conceptual research utilization (CRU) using the CRU scale; (b) facilitation of research use measured by the frequency of contacts between the frontline staff and a clinical educator, or person who brings new ideas to the care unit; and (c) organizational context at the unit level using the Alberta Context Tool (ACT). CHAID analysis was performed to detect the interactions between facilitation and context variables. Results were illustrated in a classification tree to provide a straightforward visualization. Data from 312 care units in three provinces were included in the final analysis. Results indicate significant multiway interactions between facilitation and various aspects of the organizational context, including leadership, culture, evaluation, structural resources, and organizational slack (staffing). Findings suggested the preconditions of the care settings where research use can be maximized. CHAID analysis helped transform data into usable knowledge. Our findings provide insight into the dynamic relationships of facilitators' efforts and organizational context, and how these factors' interplay and their interdependence together may influence research use. Knowledge of the combined effects of facilitators' efforts and various aspects of organizational context on research use can contribute to effective strategies to narrow the evidence-practice gap in care settings.

Investigations involving urgent protection concerns vs. chronic needs: are there differences in post-investigation service decisions?

ABSTRACT Previous research has proposed an urgent-chronic taxonomy of maltreatment investigations to better understand how child welfare services across Canada operationalized the dual mandate of child safety and well-being. Data from the Ontario Incidence Study of Reported Child Abuse and Neglect (OIS-2013a) were used to examine post investigation service decisions using this taxonomy. Logistic regression and chi-square automatic interaction detection (CHAID, a non-parametric classification tree) assessed the characteristics associated with the decision to provide ongoing child welfare services for investigations involving urgent vs. chronic concerns. Of an estimated 123,162 child protection investigations, approximately 11% were considered urgent and the other 89% involved chronic need. A quarter of investigations were transferred to ongoing services for both groups, and an assessment that there was future risk of maltreatment and caregiver mental health issues were significant and substantial factors, along with economic hardship and housing problems. Uniformity in the investigative response suggests that the decision to provide more intensive child welfare services is based more on the circumstances of the family than the nature of the primary protection concern. Since child protection investigations are oriented toward addressing urgent maltreatment, greater precision about the purpose of child welfare involvement in Ontario is needed.

A Stacking-Based Data Mining Solution to Customer Churn Prediction

In today’s competitive world, organizations are in a constant struggle to retain their current customers while attracting new customers through various methods. Customer churn is a major challenge in different industries and companies. Despite their initial successful attempts at attracting customers, organizations soon face the fact that their current customers may turn away toward their rivals. By identifying churn candidates, organizations will be able to guarantee their future success by revising their customer relationship management policy. Analyzing the data of the telecommunications industries, this study provided an effective early-churn-detection solution using modern techniques by stacking data mining algorithms. Research findings indicate that integrating support vector machines (SVMs) with the chi-square automatic interaction detection (CHAID) decision tree can yield the best outcome. The results show the proper accuracy of the proposed churn prediction solution. In addition, stacking contributed to improved customer churn detection results.

Identifying large freight traffic generators and investigating the impacts on travel pattern: A decision tree approach for last-mile delivery management

Large urban freight traffic generators (LTGs) are large specialized buildings or landmarks housing multiple establishments and generate a significant truck trips at both disaggregate and aggregate levels. Identification of LTGs and quantifying their relationship with freight travel characteristics helps policymakers formulate necessary logistical interventions and reduce externalities from freight activity. Hence, this study proposes a methodology for identifying LTGs and exploring their interactions on freight travel, expenditure pattern, shipment pattern, and other establishment characteristics. A decision-tree approach called chi-squared automatic interaction detector (CHAID) algorithm is used to map these interactions. Results suggest that LTGs are distinctly associated with multiple variables such as shipment size, shipper expenditure, commodity classification, and business age characteristics. Business age is the best predictor across all models. These associations vary based on LTG definitions. Implications of this study would augment the efforts on interlinking LTGs with urban freight demand modeling systems and enable sustainable city logistics initiatives and last mile delivery management.

Factors contributing to successful employment outcomes for individuals who are hard-of-hearing

BACKGROUND: The State-Federal Vocational Rehabilitation (VR) Program provides rehabilitation services to people with disabilities with the intention of assisting them in securing competitive employment. The VR services offer substantial resources to help individuals who are hard-of-hearing to enhance their quality of life and employment opportunities. OBJECTIVE: The current study investigated the impact of demographic variables and the use of VR services on employment outcomes among hard-of-hearing consumers. Specific VR services that lead to successful employment among hard-of-hearing consumers were thoroughly examined. METHODS: Binary logistic regression, Chi-square, and Chi-square Automatic Interaction Detector analyses were used to analyze the data extracted from the 2014 fiscal year US. Department of Education Rehabilitation Service Administration Case Service Report (RSA-911). RESULTS: Logistic regression reveals that VR services such as diagnosis and treatment of impairments (p-value 0.000), counseling, and guidance (p-value 0.000), and rehabilitation technology (p-value 0.000) were influential factors in determining the successful employment outcome among the consumers. The relative importance of the factors based on the mean decrease in accuracy in CHAID identifies rehabilitation technology (0.264), diagnosis and treatment of impairments (0.090), job placement assistance (0.016), transportation (0.016), and secondary disability (0.010) to be among the most contributing factors. CONCLUSION: Overall, rehabilitation technology services were especially beneficial, particularly for minority consumers, in achieving a successful employment outcome.

A characterisation of social media users within the primary care system in Colombia and predictors of their social media use to understand their health

IntroductionSocial media use is growing in Latin America and is increasingly being used in innovative ways. This study sought to characterise the profile of social media users, among primary care patients in Colombia, and to assess predictors of their use of social media to search for health and mental health information (searching behaviour). MethodsAs part of a larger scale-up study, we surveyed 1580 patients across six primary care sites in Colombia about their social media use. We used chi-square and Student’s t-tests to assess associations between demographic variables, social media use and searching behaviour, and a Chi-square Automatic Interaction Detector (CHAID) analysis to determine predictors of searching behaviour. ResultsIn total, 44.4% of respondents reported that they were social media users. Of these, 35.7% used social media to search for health-related information and 6.6% used it to search for mental health-related information. While the profile of individuals who used social media to search for health-related information was similar to that of general social media users (the highest use was among women living in urban areas), the presence of mental health symptoms was a more important predictor of using social media to search for mental health-related information than demographic variables. Individuals with moderate-severe symptoms of anxiety reported a significantly higher percentage of searching than individuals without symptoms (12.5% vs. 5.2%). ConclusionsGiven that some individuals with mental health disorders turn to social media to understand their illness, social media could be a successful medium for delivering mental health interventions in Colombia.

Predictors of the Level of Knowledge about Sexual Harassment and Assault among College Students: A Chi-Squared Automatic Interaction Detection Analysis

This study aims to determine college students’ levels of knowledge about sexual harassment and assault, and to identify the predictors of this level of knowledge. Participants have been selected using a multi-stage sampling method and consist of 7,302 college students from a state college in Turkey. The age of participants ranged from 17 years to 29 years with a mean age of 21.33 years (SD = 2.04). The Sexual Harassment and Assault Knowledge Test, the Survey of College Students’ Exposure to Sexual Harassment and Assault, and a socio-demographic form were used to obtain data from the sample. The two-steps cluster analysis revealed that 70.2 % of the all participants had moderate level of knowledge about sexual harassment and assault, while 13.6 % had a low level of knowledge and 16.2 % had high level of knowledge. The CHAID analysis indicated that sex was the main predictor of level of knowledge about sexual harassment and assault. Also, exposure to sexual harassment or assault, dating experience, and having gender-equality education were found as other predictor variables. Finally, classification accuracy of the CHAID model was found 79.5 % within the sample.