Several definitions can be used to describe handwriting, which can be defined in terms of a motoric ability or as the production of written language.1 Other authors define handwriting as a process of forming letters, figures, or other significant symbols, predominantly on paper. Handwriting is a functional task used for communication and recoding of thoughts and experiences.2 Handwriting is a complex activity in which perceptual–motor processes and cognitive processes continuously interact.3–5 Perceptual–motor processes in handwriting consist of perception of either visual or auditory information, fine motor coordination, and visual-motor integration. Cognitive processes involved in handwriting can be divided into more generic processes such as cognitive planning or working memory processes6 and more specific language processes, such as phonological and orthographic coding.3 Legibility and writing speed are the cornerstones of functional handwriting.7 Handwriting legibility generally includes the quality of letter formation, alignment and spacing of letters and words, and sizing of letters.8,9 Handwriting speed refers to the rate of letter production.10 Studies in which the correlation between the legibility and speed of handwriting was examined either found only weak positive correlations,11,12 or no significant relationship between the variables.4,13,14
Estimates of the prevalence of handwriting dysfunction in children range from 10% to 30%.11 Handwriting dysfunction is present when the written output is illegible, handwriting speed is too slow, arm movement during writing cannot be taught, or pain is reported during writing without an intellectual or somatic pathology.1 Handwriting difficulties have negative effects on a child's academic performance and self-esteem.15,16 Problems with handwriting have been identified as one of the most common reasons for referring school-age children to occupational or physiotherapy services.17,18 Furthermore, poor handwriting has been mentioned as one of the diagnostic criteria for developmental coordination disorder (DSM-IV-TR19; ICD-10, WHO20). In this study, we focus on 1 aspect of handwriting, namely handwriting speed.
Handwriting speed is likely influenced by gender, culture, and methodological factors. No consensus about the effect of gender on handwriting speed is found in the literature. In previous research, girls have generally been found to write faster than boys,4,21–23 but others indicate that boys write faster.24 Results of another study indicate that between the ages of 7 and 10 years, girls write faster, and at the age of 11 years, boys write faster.25 Despite these inconsistencies, gender appears to have an effect on writing speed.26 Clearly, handwriting speed studies done in different countries yield inconsistent results. This disparity may be attributed to the different approaches to teach writing internationally and differences in letterforms. The inconsistency in writing speed measurements may also reflect variation in methodological factors (eg, differences in the type and duration of writing assignments, the time needed to perform the writing task, the writing accessories used, and the evaluator or the instruction given to children on how to perform).26 Results from a Belgian study of 438 boys and 422 girls between 7 and 12 years old showed handwriting speeds ranging from 122.4 (33.9) letters per 5 minutes for those 7 years old to 259.6 (51.0) letters per 5 minutes for those 12 years old.27 In several other studies, handwriting speed has been shown to increase with age.9,12,23,28 Researchers found that this change was not linear and differed in boys and girls. Between the ages of 6 and 9 years, the pace of handwriting is relatively constant for boys and girls. Girls’ handwriting speed tends to level off between 9 and 10 years and increase again at the age of 11 years. Boys’ handwriting speed increases between the age of 9 and 11 years, but at a slower pace than at earlier ages. For both genders, there is a leveling off at the age of 14 to 15 years.4
Over the years many methods have been developed for the evaluation of handwriting performance. Most are based on analyzing the handwritten product and speed.26 The evaluation tools developed worldwide to measure handwriting speed count the number of written letters or words in 1 or more minutes.1 These evaluation tools can be different in type of assessment, task type, paper type, pencil type, timing, instruction given, validation group, scoring system, script assessed, and psychometric properties.26,29,30 To measure handwriting speed, several Dutch tests such as the “Beknopte beoordelingsmethode voor kinderhandschriften,”31 the “Systematische Opsporing van Schrijfproblemen,27,32 and the “Vlaamse snelheidstest”33; and numerous English tests such as the Minnesota Handwriting Assessment,34 the Evaluation Tool of Children's Handwriting,7 the Test of Handwriting Skills,35 the Handwriting Speed Test,23 and the Detailed Assessment of Speed of Handwriting (DASH)36 are available. In the DASH, a set of 5 tasks is included, which cover a fairly broad range of the component skills involved in the process of handwriting. When using this test for Dutch-speaking children, difficulties for the younger children might be expected. In Belgium, primarily Dutch-speaking children are exposed to several foreign languages. French and English are formally taught in school beginning at ages 10 and 12 years, respectively. But whether the DASH can be used to evaluate handwriting in school-age children in Flanders is still unknown.
This study evaluates the reliability of the DASH scale to investigate whether this test can be used to evaluate handwriting speed in Dutch-speaking children. The goals for this study were (1) to evaluate the reliability of the DASH in Dutch-speaking children using interrater agreement, test-retest reliability, and internal consistency; (2) to compare handwriting speed in boys and girls; and (3) to compare handwriting speed using the DASH in Dutch-speaking children with the norms of native English-speaking children published in the DASH manual.
A total of 1163 children (650 boys and 513 girls) aged 9 to 16 years were recruited to take part in the study. Inclusion criteria were age 9 to 16 years, regular school attendance, proficiency in cursive writing, and Dutch as their primary language. Socioeconomic and ethnic information was not collected. Children were recruited from 11 different primary and secondary schools in Flanders. The sample was divided into 3 subgroups—a total group (n = 1163); out of this group, a test-retest group (n = 266) and an interrater group (n = 61) were obtained (Table 1).
Parents and school directors agreed to the testing and children signed a written consent form before participation in accordance with the Declaration of Helsinki.
The children's handwriting speed was evaluated using a modified form of the DASH.36 The DASH consists of 5 tasks: copy best, alphabet writing, copy fast, graphic speed, and free writing. These are a series of tasks that encompass as many different aspects of the skill of handwriting speed as possible. The DASH has been standardized on a stratified sample in the UK.36 Of the 5 component tasks in the test copy best, alphabet writing, copy fast, and free writing are writing tasks. Graphic speed is a “pure” measure of perceptual–motor competence. The graphic speed test is uncontaminated by anything related to language. Both copying tasks involve copying the same sentence: “The quick brown fox jumps over a lazy dog.” First, in the student's best handwriting for 2 minutes, then as quickly as possible, but legibly, again for 2 minutes. The first provides a baseline estimate of what the child is capable of doing when trying his or her best. The second judges the extent to which the child can follow a “speed” instruction. Both copying tasks (copy best and copy fast) are written on lined paper during the 2 minutes; the number of words per minute is counted. In the alphabet-writing task, the child has to write the alphabet in lower case continuously for 1 minute. The number of letters in the right order is counted. The graphic speed task requires the child to make a series of crosses within circles, focusing more on the fine motor/precision aspects of making a mark. It represents a measure of perceptual–motor competence uncontaminated by anything related to language. The number of correct crosses drawn in a minute is counted. The last task in the DASH is a free writing task. In the free writing task, the students have to write about their life during a 10-minute period. The number of words per minute is taken into account. The topic chosen, “My life,” has been carefully selected as one that enables students to generate material easily without too much thought or effort. According to the authors of the DASH, a task like this is close to what a student must do in an examination setting. Total administration time for the test is 30 minutes. The children can be evaluated individually or in a group. For the English norm group, the interrater agreement estimated by the intraclass correlation (ICC) for absolute agreement varies between 0.85 and 0.99. The test-retest reliability for the total test is 0.85. The values for Cronbach's alpha for the total scores were between 0.83 and 0.89. The reliability estimate for the total score was 0.87.36 A significant age effect is found in the data, meaning that older children can produce more in a given time than their younger peers. According to the manual, the test can discriminate between students with and without dyslexia. Students with dyslexia wrote on average 5 words per minute less on the free writing task than the nondyslexic group. As the scores on 4 of the 5 tasks are very highly correlated, they can be summed and converted into a meaningful total standard score, which can be seen as a global measure of handwriting speed. The norms of the test were based on 546 students (boys = 254; girls = 292) between 9 years 0 months and 16 years 11 months.36 The stratified sample closely approximated the UK 2001 census data.
The testing procedure described in the DASH manual was followed to evaluate the Flemish children; only the free writing task was written in Dutch instead of English. This was done because it could not be expected that the children were able to write about their life fluently in a foreign language for 10 minutes. Alphabet writing and graphic speed were performed as in the DASH manual. All children were tested during a group test setting. Raw scores were calculated for copy best, alphabet writing, copy fast, and the free writing task. The raw score for copy best is the number of words per minute; for alphabet writing, the number of lower case letters in the right sequence; for copy fast, the number of words per minute; and for free writing, the number of legible words per minute. For each age group and subtest, different norms are used to convert the raw scores into standard scores (mean = 10; SD = 3). The sum of these 4 standard scores is again converted into the total standard score (mean = 100; SD = 15).
All data were tested for normality of distribution using the Kolmogorov-Smirnov test. The present study used an alpha level of 0.05 for all statistical tests. Descriptive statistics were used to calculate the means and standard deviations for all assessment scores.
Furthermore, this study used the same statistics as reported for the DASH to test reliability for a sample of Dutch-speaking, school-age children. Interrater agreement was calculated from data of 61 children using an ICC for absolute agreement for average measures ICC = (3, k). Interrater agreement for each of the 5 DASH tasks and total agreement were calculated using the students’ raw scores for each task and their total score. Test-retest reliability was evaluated in 266 children with a Pearson correlation coefficient for the test-retest group for the standard scores of each writing task and the total score. The internal consistency of the total standard score for all 1163 children was evaluated using Cronbach's alpha. Cronbach's alpha was calculated for different age groups and for the total sample using the normalized standard scores. A multivariate analysis of variance was used to estimate interaction effects between age and gender on handwriting speed and main effects of age and gender on handwriting speed. Further analysis of variance was performed to examine the source of the significant differences between groups. Significant analysis of variance results were followed up using Tukey's post hoc comparisons. Unpaired t tests were used to investigate the differences between boys and girls and to compare the results of the Flemish children with the UK norms. The Bonferroni correction was used for multiple comparison testing. The size effect was calculated, whereby 0.2 represents a small effect; 0.5, a medium size effect; and 0.8, a large effect.
Significant differences in age were found between boys (mean = 13.17 years; SD = 2.46 years) and girls (mean = 13.64 years; SD = 2.39 years) (t = −3.14; P = .001). One thousand children performed the test with their right hands (86%) and 163 children with their left hands (14%).
Two raters scored the performance of the children independently. The extent to which their scores correlate with one another provides a coefficient of reliability. The reliability coefficients were calculated using the English pangram in both copying tasks, whereas the free writing task was written in Dutch. Great accuracy of the scoring rules for each task of the DASH was found, with all ICCs above 0.94 (95% confidence interval: 0.903-0.965; P < .001) and an ICC for the total score of 0.93 (95% confidence interval: 0.887-0.959; P < .001) (Table 2).
Test-retest reliability refers to the consistency with which a given score is assigned to an individual's test performance on different occasions. This study evaluated the test-retest reliability with a Pearson correlation coefficient. The correlation coefficients ranged from 0.65 for the free writing task to 0.77 for the copy fast task. The Pearson correlation coefficient for the total score was r = 0.81, indicating that under standardized test conditions, performance remains stable (Table 3).
Internal consistency measures whether several items that are supposed to measure the same general construct produce similar scores. The values for Cronbach's alpha for the total standard scores ranged from 0.88 to 0.94 (Table 4), indicating good internal consistency of the results of the different writing tasks for each age group, meaning that scores on various tasks are highly intercorrelated and calculation of a total score is justified. Cronbach's alpha for the total sample was 0.91.
Differences Between Boys and Girls
A multivariate analysis of variance was conducted to examine the effect of age and gender on handwriting speed. The interaction effect between age and gender was not significant (F [7. 1144] = 0.34; P = .93), although the main effects were significant for age (F [7.1144] = 21.85; P < .001; Cohen's f = 0.37, a medium effect) and gender (F [1.1144] = 17.43; P < .001; Cohen's f = 0.14, a small effect). Results of follow-up analysis with Tukey's post hoc comparisons showed that girls (mean = 40.70; SD = 9.15) wrote significantly faster than boys (mean = 38.01; SD = 8.82) (P < .001). When looking more in detail, differences between boys and girls were found, with a tendency for the girls to write faster than the boys in every age group. However, differences were significant only at the age of 14 years (t (149) = −2.93; P < .01; effect size = −0.42, a small effect) (Table 5).
Differences Between Flemish and UK Norms
Results of an unpaired t test comparing the total sample of the Flemish children to the UK norms showed no significant differences between the 2 samples (t (1159) = −1.53). However, the results indicate that the Flemish children at 9, 10, and 11 years wrote significantly slower than reported for the English norms at those ages (all P < .001). For the 14- and 16-year-old Flemish children, significantly higher handwriting speed scores were found compared with the English norms (P < .001). At the ages of 12, 13, and 15 years, no statistical differences were found between the Flemish children's handwriting speeds and English norms.
Interrater agreement calculated with an ICC showed similar results to those reported in the original DASH manual for the copying tasks, alphabet writing, and free writing. Only the ICC for graphic speed was higher in our study (0.97) than that reported in the DASH manual (0.85).36 As each of the coefficients was at least 0.94 in our study, sufficiently high interagreement for all subtasks was found, because most researchers agree on a correlation coefficient of at least 0.90 as high agreement.37
The test-retest reliability coefficients for the total score are similar in our study and in the DASH manual. A Pearson correlation coefficient of r = 0.81 was found in comparison to a Spearman correlation coefficient of r = 0.85 reported in the DASH manual. In the DASH manual, the test-retest reliability is calculated with a Spearman correlation coefficient because of the small sample size. The correlation coefficients of the copy best and copy fast tasks were also comparable with those reported in the DASH manual, with r = 0.61 and r = 0.50, respectively, for the 9- 10-year-old age group. For the age group of 14 to 15 years, correlation coefficients were found for the copy best and copy fast tasks of r = 0.72 and r = 0.75, respectively.36 In our sample of participants (9-16 years), the correlation coefficient was r = 0.67 for copy fast and r = 0.77 for copy best. The Pearson correlation coefficients for alphabet and the free writing task were r = 0.76 and r = 0.65, respectively. Those reported in the DASH manual ranged between 0.84 and 0.91. The test-retest reliabilities reported in the DASH manual are better, but were calculated for very small sample sizes (n = 16, age 9-10 years; n = 12, age 14-15 years).
Cronbach's alphas calculated in the present study ranged from 0.88 to 0.94. Very similar results were reported in the DASH manual, with Cronbach's alphas ranging from 0.84 to 0.89. Cronbach's alpha for the total sample was similar as well—0.91 for the Flemish children (n = 1164) and 0.87 reported for the English norms. A high degree of homogeneity was observed across the 4 DASH tasks and excellent reliability for each age group. The high Cronbach's alpha for the total score indicates that the calculation of a total score is reliable. The results of the interrater agreement, test-retest reliability, and internal consistency showed us that the reliability of the DASH remained sufficiently high although the Dutch-speaking students had to write the copying task in English.
Differences Between Boys and Girls
The results of this study indicate that girls wrote faster than boys. The older age of the girls in this sample could be the cause of this finding. When looking at each age group, girls obtained higher total scores on the DASH at all ages, but these differences were only significant at age 14 years. Our finding is in agreement with some previous researchers who found that girls wrote faster.4,21–23 However, the results of another study showed no differences between boys and girls,8 whereas a study by Ziviani12 indicated that boys wrote faster than girls. In the DASH manual, no handwriting speed difference between boys and girls is mentioned and the norms are identical for both genders. Gender-related differences are likely the result of both biological and environmental factors.38 There is literature documenting the more advanced development of fine-motor coordination in girls relative to boys.39 Cultural stereotypes are also likely to influence handwriting development; usually girls are assumed to be better hand writers than boys.40 More research is needed about this topic to specifically address the effects of nature and nurture in the development of differences in the handwriting of male and female students.4
Differences Between Flemish Children and the UK Norms
One of the major outcomes of this study is that it seems that the UK norms of the DASH can be used for Flemish children, because no significant differences were found between the Flemish children and English norms. For the younger children (9-11 years), these norms should be used with some caution, because their scores were significantly lower than the UK norms. A possible reason may be the use of the English sentence for copying, where the Flemish children have to write in a foreign language. English training in Flanders starts at 12 years of age. It is unclear why the Flemish children at ages 14 and 16 years performed better than the English norms. Maybe it is because they know multiple languages, and this makes it easier for them to memorize such a short sentence. Another possibility is that the sentence used in the DASH is known from the children's computer use. This sentence is used widely to provide an example of the different letters types.
The free writing task was performed in Dutch in our study, whereas in the English version this task is written in English. Children between 9 and 16 years are unable to perform the free writing test in a foreign language. So the present study assumed that their output is comparable to the written output of the English norms. Both copying tasks were performed in English. For these tests, slower handwriting speeds for the Dutch-speaking students compared with the English norms might be expected because of the foreign language. This disadvantage would probably be more obvious in the younger children than in the older children. Further research should be done using a Dutch pangram instead of the English one to determine whether reliability could be higher for the younger children.
This study evaluated the reliability of the DASH scale to investigate whether this test can be used to evaluate handwriting speed in Dutch-speaking children. Therefore, this study analyzed the 3 separate aspects of reliability reported in the DASH manual: interrater agreement, test-retest reliability, and internal consistency. Although all reliability coefficients should meet a minimum standard, acceptable values depend on the specific use or function of the test. Our results indicate very good reliability for interrater agreement and the internal consistency of the total score. However, the test-retest reliability for the total score was lower, but still sufficiently high. With these results, other researchers can further analyze the types of validity most relevant to the DASH, that is, discriminating between children with dyslexia and those who are not dyslexic.
1. Overvelde A, van Bommel I, Bosga I, et al. KNGF Evidence Statement: Motorische schrijfproblemen bij kinderen. K Ned Genootschap voor Fysiotherapie. 2011;121(2):1–65.
2. Ziviani J, Wallen M. The development of graphomotor skills. In: Henderson A, Pehoski C. eds. Hand Function in the Child. Foundations for Remediation. 2nd ed. St Louis, MO: Mosby Elsevier; 2006:217–238.
3. Berninger VW, Swanson HL. Modifying Hayes and Flower's model of skilled writing to explain beginning and developing writers. In: Carlson JS (Series Ed.), Butterfield EC (Vol. Ed.). Advances in Cognition and Educational Practice: Vol. 2 Children's Writing: Toward a Process Theory of the Development of Skilled Writing. Greenwichn CT: JAI; 1994:55–57.
4. Graham S, Berninger VW, Weintraub N, Schafer W. Development of handwriting speed and legibility in Grades 1-9. J Educ Res. 1998;92:42–52.
5. Van Galen GP. Handwriting: issues for a psychomotor theory. Hum Movement Sci. 1991;10:165–191.
6. McCutchen D. Knowledge acquisition, processing efficiency and working memory: implications for a theory of writing. Educ Psychol. 2000;35:13–23.
7. Amundson SJ. Evaluation tool of children's handwriting. Homer. AK: OT kids. APA. (2000). Diagnostic and Statistical Manual of Mental Disorders (4th rev. ed.). Washington, DC: American Psychiatric Association; 1995.
8. Feder KP, Majnemer A, Bournonnais D, Blayney M, Morin I. Handwriting performance on the ETCH-M in Grade One Regular Education Program. Phys Occup Ther Pediatr. 2007;2(2):43–62.
9. Tseng MH, Chow S. Perceptual-motor function of school-age children with slow handwriting speed. Am J Occup Ther. 2000;54:83–88.
10. Kushki A, Schwellnus H, Faizah I, Chou T. Changes in kinetics and kinematics of handwriting during a prolonged writing task in children with and without dysgraphia. Res Dev Disabilites. 2011;32:1058–1064.
11. Karlsdottir R, Stefansson T. Problems in developing functional handwriting. Perceptual Mot Skills. 2002;94:623–662.
12. Ziviani J, Elkins J. An evaluation of handwriting performance. Educ Rev. 1984;36:249–261.
13. Sovik N, Arntzen O. A developmental study of the relation between the movement patterns in letter combinations (words) and writing. In: Wann J, Wing A, Sovik N, eds. Development of Graphic Skills: Research, Perspectives and Educational Implications. London: Academic Press; 1991:77–89.
14. Weintraub N, Graham S, Writing legibly and quickly: a study of children's ability to adjust their handwriting to meet common classroom demands. Learn Disabil Res Pract. 1998;13:146–152.
15. Margalit M. Loneliness and coherence among preschool children with learning disabilities. J Learn Disabil. 1998;31:173–180.
16. Pavri S, Monda-Amaya L. Loneliness and students with learning disabilities in inclusive classrooms: self-perceptions, coping strategies, and preferred interventions. Learn Disabil Res Pract. 2000;15:22–33.
17. Bosga-Stork I, Overvelde A, Van Bommel-Rutgers I, et al. Inventarisatie van verwijzingspatroon. onderzoek en behandeling van kinderen met schrijfproblemen. Een digitale enquête. [Inventarisation of the way of referring. assessment and therapy in children with writing problems] Ned Tijdschr Voor Kinderfysiotherapie. 2009;2:14–18.
18. Hammerschmidt SL, Sudsawad P. Teachers’ survey on problems with handwriting: referral, evaluation, and outcomes. Am J Occup Ther. 2004;58(2):185–192.
19. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed [text revision (DSM-IV-TR)]. Washington, DC: American Psychiatric Association; 2000.
20. WHO. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva: World Health Organization; 1992.
21. Cohen MR. Individual and sex differences in speed of handwriting among high school students. Percept Mot Skills. 1997;84:1428–1430.
22. Karlsdottir R. Development of cursive handwriting. Percep Mot Skills. 1996;82:659–673.
23. Wallen M, Bonney M, Lennox L. The Handwriting Speed Test. Adelaide, Australia: Helois; 1996.
24. Ziviani J. Use of modern cursive handwriting and handwriting speed for children aged 7 to 14 years. Percept Mot Skills. 1996;82:282.
25. Ziviani J, Watson-Will A. Writing speed and legibility of 7–14 year old school students using modern cursive script. Aust Occup Ther J. 1998;45:59–64.
26. Rosenblum S, Parush S, Weiss PL. Computerized temporal handwriting characteristics of proficient and non-proficient handwriters. Am J Occup Ther. 2003;57:129–138.
27. Van Waelvelde H, De Mey B. Smits-Engelsman BCM. Handleiding SOS. Systematische opsporing van schrijfmotorische problemen. Versie november [Manual: systematic inventarisation of writing problems] 2008. Gent: Revalidatiewetenschappen en kinesitherapie Gent; 2008.
28. Dutton KP. Writing under examination conditions: establishing a baseline. Professional Development Initiatives. 1989-90, 1990; 189–210.
29. Feder KP, Majnemer A. Children's handwriting evaluation tools and their psychometric properties. Phys Occup Ther Pediatr. 2003;2:65–84.
30. Tomchek S, Shneck C. (2006). Evaluation of handwriting. In: Henderson A, Pehoski C, eds. Hand Function in the Child, Foundations for Remediation. 2nd ed. St Louis, MO: Mosby Elsevier; 291–309.
31. Hamstra-Bletz L, De Bie J, Den Brinker BPLM. Beknopte Beoordelingsmethode Voor Kinderhandschriften - BHK. [Short assessment method for childrens’ handwriting]. Lisse: Swetz & Zeitlinger; 1987.
32. Smits-Engelsman B, Vrenken I, Stevens M, Van Hagen A, Systematische opsporing Schrijfproblemen (SOS): Een hulpmiddel voor leerkrachten bij het signaleren van motorische schrijfproblemen van leerlingen in primair onderwijs. [Systematic inventarisation of writing problems] Kinderfysiotherapie. 1999;17(47):16–20.
33. Vanderheyden V. Vlaamse Schrijfsnelheidstest: Evaluatie van de Schrijfsnelheid. [Flemish test for writing speed] Herent: OSPOO; 2003.
34. Reisman JE. Minnesota Handwriting Assessment. Los Angeles: Psychological Corporation; 1999.
35. Gardner M. The Test of Handwriting Skills: Manual. Hydesville. CA: Psychological and Educational Publications; 1998.
36. Barnett A, Henderson S, Scheib B, Schultz J. Detailed Assessment of Speed of Handwriting (DASH). London: Pearson; 2007.
37. Downing SM. Reliability: on the reproducibility of assessment data. Med Educ. 2004;38:1006–1012.
38. Graham S, Weintraub N. A review of handwriting research: progress and prospects from 1980 to 1994. Educ Psychol Rev. 1996;8:7–88.
39. Hartley J. Sex differences in handwriting: a comment on spear. Br Educ Res J. 1991;17:141–145.
40. Spear M. Differences between the written work of boys and girls. Br Educ J. 1989;15:271–277.