home index by author index by subject

Poster P3

Friday stream 1-4 Session 16.10 - 17.25 Length 25 minutes

Writing Speed and Extra Time in Examinations

Erica Bishop
Anthony Esgate

Ampthill, Bedford Andrew.Bishop@waii.com

Abstract

A study was carried out with 220 Year 9 students to replicate writing rates that would be expected in an extended writing test carried out under examination conditions. The results were analysed to shed light on three issues:

1. What is slow writing? Writing speed alone may not always correctly predict the total number of words written down. Actual writing speeds are compared with writing speeds predicted by published writing speed tests, widely used when considering examination allowances.

2. Who are the slow handwriters? This investigates the variation in the writing rates by gender and by writing style.

3. Is extra time beneficial to slow handwriters? This compares the marks gained by a group with and without extra time with a group allowed 25% extra time.

The results of the experiment could be interpreted to show that the actual writing speed is significantly slower than is generally accepted. Girls at 13 years 9 months on the whole write more quickly than boys do. The implication is that when a judgement is made on selection of slow handwriters using mixed sex data, some girls are falsely excluded whereas some boys are falsely included.

Print and cursive styles of handwriting appear to be written at a similar rate. However, print writers tend to write fewer words and use less time.

The data could be said to demonstrate that extra time does result in extra marks. There is a sliding scale whereby slower writers gained a greater proportionate increase in marks compared to faster writers. An implication could be that slower writers would justifiably benefit from extra examination time.

1. INTRODUCTION

The use of fixed-time national examinations to assess ability and award qualifications has been employed in schools for many years. The strengths of this system are that it is seen to be fair, impartial, and unvarying, thereby providing a result that can be used both as a direct measure of an individual's ability and as a comparative measure against his peers. The weakness of the system is that it consists of two components - the test of the student's knowledge of the subject, and the assumption that the student has the ability to express this knowledge adequately in a timed examination.

For those students who have no difficulty in the non-subject related component, the results of examinations usually accurately reflect their ability in that subject. However, for students with difficulties associated with the techniques of producing written output (e.g. those who are dyslexic, and others with slow processing rates) the true assessment of subject knowledge may be masked by their inability to express themselves adequately on paper in a set time.

Students with particular learning difficulties have historically been given extra time in public examinations. The rationale for this practice is that such students have particular difficulties that prevent them from doing themselves justice within the normal time constraints. Sawyer (1993) surveyed the practice of Examination Boards and concluded that extra time is the most readily and frequently offered concession to students with specific difficulties. The commonest justification for allowing extra time is slow handwriting.

Each year Examination Boards are overwhelmed with requests for extra time. Practice in addressing this problem was to award an extra time provision to those students who were identified as needing extra time based on the recommendations of an Educational Psychologist report. Assessment has been extended to include recommendations made by selected specialist teachers (e.g. those who hold a RSA / SpLD diploma).

A slow writing student may be allowed 25% extra time in SAT or GCSE (and other) examinations. However the definition of "slow writer" is ambiguous, and the value of extra time is contentious. Classroom teachers are being asked to make judgements on these issues.

2. LITERATURE REVIEW

What is slow handwriting?

It seems that the type of assessment affects the rate quoted. Published articles show that writing speed varies with the task set. References are divided into purely mechanical and those measures which portray 'real-life situations'.

• Sawyer (1991) said that, handwriting speed might be seen as a particular motor co-ordination task. Sawyer views the 0+0+0+… test as providing a reasonably pure measure of motor speed.
• Hedderly (1992) reinforces the idea that an accurate test of writing speed needs to consider access skills that include cognitive variables, but without making a judgement upon knowledge of the exam subject. The situation should be a physically tiring experience of writing at length, in combination with a real life setting.

What is a mean writing speed for an extended writing test? A review of these rates is presented, with the faster ones presented first.

• Dutton, (1992) reported that the average writing speed of young workers who left school at 16 was around 20-21 wpm.
• The SCEEB (The Scottish Certificate of Education Examination Board) suggest a 'rule of thumb' of 16 wpm as a minimum writing speed for exam candidates.
• The SCEEB (The Scottish Certificate of Education Examination Board) suggest a 'rule of thumb' of 16 wpm as a minimum writing speed for exam candidates.
• Hedderly gives a range of writing speeds between 10 and 20 wpm as normal at age 15. With an average speed of around 15-wpm he suggests that "those pupils writing at a speed of 8 wpm will almost certainly be handicapped."
• Ashton (1997) reported that the average speed in an English paper for a secondary modern pupil is about 10-12 wpm.

Who are the slow handwriters?

Gender and writing rates

Authors agree that girls generally write more quickly than boys do.

• Ames (1984) work reflects that writing speed improves rapidly in the teenage years, linked to physical maturity. Girls on average achieve a maximum writing speed earlier than boys do.
• Alston (1995) found the average rate was 14.7 wpm for girls and 13.8 wpm for boys.
• Dutton (1992) found that although a reasonable benchmark as an average score, the 16 wpm 'rule of thumb'. This criterion would be failed by more than 50% of boys, although a reasonable figure for girls.

Handwriting style.

• Thomas (1998) suggests that a joined-up handwriting style is quicker to produce than printing.
• Turner (1997) says it is not just an issue of speed, but there are even more important aspects to a joined writing style for dyslexic students.
• Connor, (1995) focuses attention upon senior pupils, especially boys, with a view to encouraging and developing fluent and rapid handwriting.
• Thomas (1998) reports that, France, by contrast to the UK teach handwriting, particularly a flowing joined up writing - le graphisme. It is considered fundamental, a physical skill that, once mastered, which unlocks the mind.

Is extra time beneficial to slow handwriters?

• Turner (1997) says that constrained by time, a student may drop marks, other sources dismiss the advantage of extra time.
• Knight (1992) report that, there is no evidence to indicate whether additional time for GCSE is an advantage or not.
• Sawyer (1991) infers that giving extra time to some is unfair to the other candidates on the grounds that they all could use and gain marks from it.
• Hedderly (1996) followed up several 'A' level and GCSE candidates who had been granted extra time. It became apparent that not one of the 'A' level students received an improved grade as result of the extra time.
• Ashton (1997) suggests that any one that requests more time should receive it. Schools could request extra time on the basis that: it has been seen to benefit the pupil in SATs; the parent and child wants extra time; and the use of extra time would be noted on the certificate.

3. METHODOLOGY

The Sample

• Approximately 220 Year 9 students took part in this study, 113 boys and 106 girls. The average age was 13 years and 9 months. A large sample of students was used. Robson (1993) points out that the larger the sample the lower the likely error in generalising.
• Researchers have advocated Year 9 SAT exams being used as a model for concessions needed in 2 years time at the GCSE exams. Year 9 SAT exams are the last public exams at which all students sit the same English essay question.
• 8 groups of Year 9 students each containing about 25 students. Two parallel sets of 4 ability sets took part in this study.

The Essay Task

• A 40-minute essay question taken from a past test paper was set. 25% (ten minutes) extra time was given to half of the sample of students.
• The essay was taken from Key Stage 3 English SAT. The paper was from 1997, level 4-7, and question 4.
• The use of a specific test was recommended by Dutton (1992) who recommended that the task be set as close to examination conditions as possible, including a formal presentation, a standard writing paper and a precise time limit.
• The students marked their scripts every three minutes. This enabled accurate measurement of the actual time spent writing and accurate calculation of the writing rate for that period.
• As the students were writing, the teacher called out 'Time Mark' at three-minute intervals. When the students heard this, they marked the point in the line of writing with the mark // and then carried on writing (this was demonstrated on blackboard by writing a 'line of // writing'). No matter how much they had written since the last mark, even no words at all. This method allowed accurate targeting of slow writers.
• Critical thresholds for low writing students were set at 1 standard below the mean.
• A different coloured pen was used to delineate Extra Time.

What is slow handwriting?

Calculation of writing rates

• The following formula is used to calculate writing speed, and to predict how many words are written in a set time. Writing rate (words per minute or WPM) is one member of a tripartite relationship. The other 2 members are Total Words Produced (TWP) and Time. If two parts are known, the third can be determined.

Formula for writing speed WPM = TWP / Time TWP = WPM x Time Time = TWP / WPM

Fig 1. Writing Speed (WPM), the total of words produced (TWP) and time used

Who are the slow handwriters?

Gender and writing rates

The gender of each student was recorded. The data was analysed on the basis of gender for writing speed, TWP and time spent writing.

Handwriting style

• Each script was subjectively analysed for the amount of joins in the writing.
• Scheme for ranking handwriting (HW) style (Table 1):

• Computer analysis was used to evaluate the hypothesis was that a print style of handwriting is slower to produce than a cursive style. The alternative hypothesis was that they take the same time to produce.

Is extra time beneficial to slow handwriters?

Calculating Marks

• The validity of extra time was investigated by dividing the cohort was into two. The Control group was allocated 40 minutes ("Time") and the Experiment group was allocated 50 minutes ("Extra Time").
• Class teachers marked the scripts in accordance with the Government's School Curriculum and Assessment Authority (QCA, 1997) who published the Mark Scheme for this exam.
• Slow writers were identified in the control and experimental groups. By comparing marks gained in 40 mins with marks gained in 50 mins, the validity of extra time was assessed.

4. RESULTS

What is slow handwriting?

• Writing Speed (WPM) - whole cohort.

To establish the writing speed, the total number of words written is divided by the time the student actually wrote for (to the nearest 3 mins).

68% write between 9-15 wpm, with a mean value of 13 wpm. 95% of the group write in the range of 5-20 wpm.

• Total Word Production (TWP) - whole cohort.

The mean TWP was 419; 68% of students produced between 273 and 565 words.

• Time used by whole cohort.

• Establishing Base Lines for slow writers

The base lines were calculated from the Control data, that is all students writing for 40 minutes without extra time. One standard deviation either side of the mean, contains 68% of the values of the population. A student scoring more than 1 SD away from the mean would be considered as unrepresentative of that population. For each parameter those students performing below 1 standard deviation were taken as the "low" group.

• Summary: Low/critical level of writing production (control group)

Table 2: Group Characteristics

Who are the slow writers?

• Gender and Writing Speed

The whole cohort was used to search the data for any significant difference between boys' and girls' rates, as there was no significant difference between boys' rate in the control or experimental groups. This was also true for girls.

The mean writing speed for boys is 11.9 wpm; the slow score is 8 wpm.
The mean writing speed for girls is 13.3wpm; the slow score is 10.0 wpm.

Writing speed for boys and girls were compared using Analysis of Variance (ANOVA). The ANOVA shows that there is a significant difference between the writing rates for boys and girls (F=8.62; P=0.004, d.f. =1, 119). The high F-statistic and low P-value indicate there is a statistically significant difference between boys and girls.

The 95% confidence interval for average WPM of boys is between about 11.2 and 12.6 wpm. For girls, it is between 12.7 and 13.86 wpm.

• Gender and Total Word Production (TWP) - Table 3:

Girls at 40mins produced 448 words, whilst boys produced 397. Slow girls produced 306 words, whilst slow boys produced 250.Girls at 50 mins produced 593 words, whilst boys produced 465. Slow writing girls produced 434 words; slow writing boys produced 284 words.

A one-way Analysis of Variance examined the difference between TWP for boys and girls (Experiment-50 mins). The discrepancy is shown to be significant. Using a single factor sample F test, the independent variable was TWP. The difference between the groups is significant (f=13.9, p=0,0 with 1, 96 and 97 df).

• Handwriting style

Data was analysed from the whole cohort. The TWP was calculated from the words written at the 40-minute mark.

The division into 5 types was subjective and therefore open to different interpretation. The data was analysed on the basis of 3 types, entirely printed (1); mixed print and joined (3) and cursive (5).

Table 4: Results for the 3 styles as writing speed, TWP and time used.

The hypothesis for WPM and then TWP is that the values for print style are lower than for fully cursive handwriting. The null hypothesis is that there is no significant difference between print or cursive HW styles.

Analysis of Variance for Total No. of Words between Print and Cursive styles

A print style results in a significantly lower TWP than a fully cursive style. (f=5.22, p=0.033 with 1, 191 df).The p-value is below 0.05 however the f-value is small indicating the significance is low. About 7% of the cohort had either fully print or fully cursive handwriting. The sample numbers are low.

Analysis of Variance for writing speed between print and cursive HW styles.

The null hypothesis is accepted, there is no significant difference between the rate of producing print or cursive HW. (f=1.84, p=0.190 with 1, 29 df).

It seems that although both print and cursive are written down at the same rate, cursive writers write down more words. In general the cursive writers must keep writing when the print writers have stopped.

Is extra time beneficial to slow handwriters?

• A significant extra amount of words were written in extra time

One-way Analysis of Variance - Between TWP for Control and Experiment (f=22.33, p=0.00 with 1, 191 df)
The analysis output shows that Experiment and Control TWP are significantly different. There is a high F-statistic and a corresponding low P-value, showing that More time appears to result in more words being written.

• A significant extra amount of marks were gained in extra time.

The mean for 40 mins was 11.5 marks, SD=6.5. The mean for 50 mins was 15 marks, SD=7.5. This suggests that extra time result in extra marks.

One-way Analysis of Variance - Comparing Marks for all students in Control and Experiment (f=7.11, p=0.009 with 1, 191 df). When an F test is performed to test the null hypothesis that the marks are the same for control and experiment, because the p-value is 0.009 (a value smaller than 0.05), the null hypothesis can be rejected. The marks increase significantly in extra time.

• Slow writers were shown to benefit from extra time - Table 5.

Slow writing boys and girls were identified using single gender data groups. Their marks were compared with and without extra time.

Fig. 5 (Left): Comparing marks in Control (1) vs. Experiment (2) for slow writing Boys Surface validity suggest that slow writing boys gain more marks in extra time.	Fig. 6 (Right): Comparing marks in Control (1) vs. Experiment (2) for slow writing Girls Surface validity suggests a smaller gain in marks for slow writing girl

The median and range of the boys group is much greater with extra time. The median of the girls' data is little different, although the range is larger. The small number of students involved makes ANOVA unreliable.

5. CONCLUSIONS

What is slow writing?

• Definition of slow writing rate

The slow group mean writing speed is 9 wpm, TWP of 263, averaging 26 mins.

• Writing speed may not predict how many words are actually written.

Although individual students have predictable writing rates, the amount of time spent on the task before stopping is not. If a student continues to write throughout the examination until the end, then the formula for speed maybe valid. The number of words written would predict writing speed. If a student stops writing before the end of an exam, the formulas are invalid. Writing speed does not indicate the total of words produced.

• Observed writing rates may be slower than predicted by published rates.

Comparing observed writing speed in this experiment with a published 10 min writing test, the observed rate is slower.

Table 6: Robin Hedderly Sentence Completion test

The writing rate quoted in Hedderley's test for year 9 students has a mean of 15 wpm, and a slow level of 12 wpm. The "average" student in this study wrote at the rate quoted for a year 5-7 (mean was 12.8 wpm, with a low level of 9.1 wpm). The Hedderley test, calculated over 10 minutes, is shown to be inaccurate as a realistic predictor of speed in a longer answer question in an exam.

Who are the slow writers?

• There is a difference between boys and girls writing speed.

Gender was seen to be significant, with boys in general, writing more slowly than girls were (boys mean was 11.9 wpm: slow was 8.0 wpm / girls mean was 13.3 wpm: slow was 10.0 wpm). The critical level for slow writers will be different if calculated for single-sex or a whole cohort group. Year 9 boys' achieve a Hedderly Y 6, and girls' achieve Y 7/8.

• Total Word Production was generally different for boys and girls.

An average of 11% extra words were written in extra time. For both boys and girls groups there were a significant amount of extra words written in extra time.

The total number of words written is the key to success in acquiring marks. In allowing extra time, examiners are implicitly not concerned with writing speed itself, but are indirectly addressing the need for extra words.

There is a gender discrepancy, similar to that discussed for writing speed, with girls producing significantly more words than boys (boys mean was 465 words: girls mean 593 words in 50 mins). Girls persisted for a longer time than boys did.

This deficit widens with an increase in writing task length.

Each group (all, boys, or girls) will have different means and standard deviations, and these means do not coincide. The mean for mixed group is between those for boys (lower) and girls (higher). Therefore there will be some students who fall outside one category but inside another, or vice versa.

The significance is that, some girls are more than one SD below the mean for the girl's group, but less than one SD below the mean for the mixed group.

Similarly, some boys may be more than one SD below the mean for mixed group, yet less than one SD below the mean for boys.

If extra time were granted solely on the basis of being more than one SD below the average for mixed group students, some girls would miss out, and some boys would be wrongly included.

• Writing style

Cursive handwriting resulted in a significantly greater amount of words written down than a printed handwriting style.

Is extra time beneficial to slow writers?

• Extra Time resulted in extra marks

Extra time may be said to be valid if students produce extra words in extra time and if extra words result in extra marks, i.e. extra time must equal extra marks.

There is a gender effect in the distribution of marks. Girls wrote more words in extra time, yet boys increased their marks by a higher proportion. This effect tends to be more noticeable at the slower, lower end of the spectrum.

• Slow Writers

There is a significant increase in words written between the 40-minute control group and the 50-minute experiment group. There is a positive correlation between marks and total word production. This was not directly proportional. The relationship was linear at low values of TWP and curvilinear at higher values of TWP. This means that a small increase in words at the lower end generates proportionally more marks than a similar increase at the higher end. Students who have a low total word production will therefore benefit proportionally more from extra time than normal or higher total word producers.

The results of the study found that not only is there an increase in the number of words written in extra time, but that marks increased as well.

In general over the whole cohort, the increase in marks was more significant for boys than for girls.

This effect is a limiting factor on the trend of more words / more marks. "At risk" poorer students would benefit from extra time proportionally more than the more able. If access to extra time is limited to those students with a demonstrable need, it would seem a fair concession that would not disadvantage other students.

Extra time results in extra marks especially for students with a low word production. For such a student prepared to keep on writing, extra time is judged to be valid.

6. RECOMMENDATIONS

Handwriting speed

• WPM may not predict how many words are actually written.

Writing speed does not per se predict the number of words written. The assessment model used to predict writing speed is of crucial importance. A short timed test will give a writing speed but may not predict how many words will be written in an exam. An assessor might be well advised to be highly sceptical of published tests of writing speed. The assessor may be better off asking: "Does a particular student have a history of writing slowly? Is there evidence to show that the student can sustain writing over time? Can it be demonstrated that the student has benefited from extra time before?"

The slow Handwriters

• There is a difference between boys and girls writing speed.

The significance of this is that if selection takes place on any fixed-figure criteria calculated from a mixed-sex group, some girls will be unfairly excluded while some boys may be unfairly included. Using 40 min results; 1 extra girl was identified as a slow writer and 7 boys are excluded when the data is analysed in separate gender groups compared to the mixed gender group. Writing speed should be assessed in single-sex groups. The alternative view is that separation on the basis of gender could be construed as unfair, and could be seen as positive discrimination in favour of girls. This issue may be a topic for further research.

• Handwriting style

School can influence handwriting style. Perhaps a cursive handwriting should be introduced as the initial style at Infant school. Subsequent levels of education should encourage cursive writing.

Extra time is beneficial to slow writers.

• Extra Time resulted in extra marks

Giving extra time will help those who need it most but will not significantly benefit a student who is already writing a high number of words. Conversely, withholding extra time may seriously disadvantage the slow writers. Thus there is a limited "fail-safe" effect whereby giving the benefit of any doubt to a borderline, possibly slow writer will not confer an unfair advantage on them.

7. REFERENCES

1. Alston, J. (1995) Assessing and Promoting Writing Skills. Nasen Enterprises Ltd.

2. Ashton, C. (1997) "The Assessment of Handwriting Speed", Dyslexia Review, Vol 9, p8-11.

3. D'Arcy, J. (1997) Provision of Additional Time for Candidates. Joint Forum for the GCSE and GCE, Standing Committee on Research.

4. Department for Education and Employment (1994), Code of Practice on Identification and Assessment of Special Educational Needs. HMSO, London.

5. Department for Education and Employment (1997), Assessment and Reporting Arrangements - Key Stage Three. HMSO London.

6. Dutton, K.P. (1992) Writing Under Exam Conditions Establishing a Base Line. Hand Writing Review, Vol 2, pp 80-101.

7. Hedderly, R. (1992) Psychologists Assessments of Specific Learning Difficulty (Dyslexia) and Examination Boards: Policies and Practices. Educational Psychology in Practice Vol 6 (1) pp 32-42.

8. Hedderly, R. (1995) The Assessment of SpLD Pupils for Examination Special Arrangements. Dyslexia review, Vol.7, pp 12-19.

9. Hedderly, R. (1996) Assessing Pupils with Specific Learning Difficulties for Examination Special Arrangements at GCSE, 'A'-level and Degree-level. Educational Psychology in Practice, Vol. 12, pp 36-44.

10. Joint Forum for GCSE and GCE (1997), Candidates with Special Assessment Needs - Special Arrangements and Special Considerations. London; HMSO.

11. Qualifications and Curriculum Authority (1997), Assessment and Reporting Arrangements. London, HMSO.

12. Robson, C. (1994) Real World Research. Blackwell, Oxford.

13. Rowntree, D. (1981) Statistics Without Tears. (pp 96). Penguin Group.

14. Sawyer, C. E. (1991), On reading and the GCSE. The Psychologist, 4,5,221-222.

15. Sawyer, C. Gray, F. and Champness, M. (1996) Measuring speed of Handwriting for the GCSE candidates. Educational Psychology in Practice, Vol. 12, pp. 9-23.

16. Sawyer, C. Francis, M. Knight, E. (1992) Handwriting Speed, Specific Learning Difficulties and the GCSE. Educational Psychology in Practice, Vol 8, pp. 77-81.

17. Turner, M. (1997) Psychological Assessment of Dyslexia. Whurr Publications.