click here to return to main Research Page

Statistically significant improvements in Reading and Maths achieved by Special Needs Students as a result of Brain Gym ® intervention - New re-evaluation of data published in Smart Moves by Carla Hannaford (1995)

Introduction:

In this paper, a statistical analysis is carried out on data from a research experiment which was described in the book, Smart Moves by Carla Hannaford, published in 1995. This new statistical evaluation of the published data demonstrates that nineteen children aged 10-11 years old (USA 5th Grade students), with Special Educational Needs, made statistically significant improvements in reading, reading comprehension and mathematics as a result of a year-long Brain Gym ® programme.

The study was done in the USA where school years are called Grades, Grade 1 = age 6-7 yrs; Grade 2 = 7-8 yrs etc. Reading performance is usually measured as the equivalent Grade level at which the student is reading. The students' reading age was on average more than 3 years behind calendar age at the start of the experiment, and after the intervention, their reading age had improved to be on average less than 1.5 years behind. In contrast, without costly special individual intervention and only using the typical interventions used in schools for Special Needs, such students would normally have fallen further behind (typically by at least a further half year). Thus, the total improvement is close to two extra years in reading age, which is demonstrated to be highly statistically significant using Student's T-tests. The improvement in mathematical ability is somewhat smaller, and also statistically significant.

The Experiment:

The experiment involved nineteen "fifth grade" Special Education students (Age 10-11; equivalent to UK Year 6). Ability levels in reading, reading comprehension and mathematics were monitored at the beginning of the grade year and again at the end of the school year using the Brigance Inventory of Basic Skills (Albert H Brigance, 1977, Curriculum Associates).

The intervention consisted of a Dennison Laterality Repatterning procedure (Personalised Whole Brain Integration, Paul and Gail Dennison, 1985) which was carried out at the start of the experiment on each student, and thereafter each child did 5 to 10 minutes of  Brain Gym movements daily (Brain Gym, simple activities for whole brain learning, Paul and Gail Dennison, 1986). The specific movements which were used in the experiment were not indicated in the book Smart Moves.

click here to return to top of page

Results:

Raw data: Graphs of the results of the Brigance Skills assessment are provided in the book Smart Moves on page 114. For copyright reasons, the graphs are not reproduced here. The data values have been extracted from the graphs and are tabulated in the tables below. Table 1 shows the raw data extracted directly from the graphs. The values are of the Grade level for reading or maths (i.e. the normal age group with this level of reading or mathematical ability).  Note that the US Grade levels start at Grade 1 for age range 6-7 years, so a Grade level of 0 is below that of an average 6 year old, while a Grade level of 3 is equivalent to the reading ability of an average 8-9 year old, for example.

Note that the test is calibrated for student age and date of test, so an average student should score at their Grade level throughout the school year. I.e. NO change in Grade Score is expected between pre- and post-tests. The arithmetic means (averages), standard deviations and standard error values have  been calculated in this review.

Table 1: Raw data on Reading from Figure 7.2, p 114 Smart Moves, C Hannaford, 1995.

Reading: At the start of the experiment, the average reading grade of 1.95 years was more than 3 years behind calendar age (which should give a grade score of 5, as the students are in Grade 5).

After the year-long intervention, the average value of reading grade of 3.53 demonstrated an improvement of an average of 1.58 grade points in reading, during the 10 months of the experiment. During this time the student's ability should have improved because of their increased maturity. We assume that the underlying trend is of 0.5 months improvement per month of maturity (typical for special needs students). This means that the total improvement for these students is 1.58 Grade points plus an extra 5 months for maturity. I.e. converts to an improvement of 24 months over a 10 month period, or 2.4 months per month of intervention. The maximum improvement seen for any one student was an improvement of two grade years in one academic year, and 10 students improved by this amount.

Figure 1. Reading Grade data from Hannaford 1995. Purple bars indicate the number of students with Reading Grade in the indicated ranges at the start of the Brain Gym intervention, Green bars indicate the number of students with the indicated Reading Grades at the end of the experiment.

Table 2: Raw data on Mathematics from Figure 7.2, p 114 Smart Moves, C Hannaford, 1995

Maths:  Initially, the mathematical ability levels of these children was better than their reading levels. At the start of the experiment, the average value of mathematical grade level was 3.25, in other words one and three quarters of a year behind the average.

After the year-long intervention, the average value of mathematical grade had increased to 4.31, now only 0.7 grade points behind average, an improvement of 1.06 years, during the 10 months of the experiment. Since these student started at a higher level in maths than reading, we can assume that their maths ability improved at approximately 0.8 months per month of extra maturity (typical for the higher performing special needs student). This converts to an improvement of 20.7 months over a 10 month period, or 2.1 months per month of intervention.

Figure 2. Mathematics Grade data from Hannaford 1995. Purple bars indicate the number of students with Maths Grades in the indicated ranges at the start of the Brain Gym intervention, Green bars indicate the number of students with the indicated Maths Grades at the end of the experiment.

Are these changes significant?

Firstly, without costly individual remedial methods, students who are already failing are found to fall further behind their peers as they grow up (Outcome of treatment of visual problems in children with reading difficulties, Clisby C, Fowler MS, Hebb GS, Walters J, Southcott P & Stein JF, (Professional Association of Teachers in Special Situations (PATOSS) Bulletin Nov. 2000, 9-14.). Changes of between 0.56 to 0.88 (average 0.72) months improvement per month of intervention are typical using methods which are used regularly in schools for Special Needs children. If a child of 7 started with a reading age of 6, one year behind the average, then by age 11, their reading age would only have improved to 8 years 11 months, and they would now be more than two years behind the average, if their reading improved at 0.72 months per month of maturity.

The changes reported by Hannaford which convert to 2.4 month per month of intervention for reading, and 2.1 months per month of intervention for mathematics, are clearly greater than the average value of 0.72 months per month quoted above. The true significance of the data can only be determined by a full statistical analysis.

click here to return to top of page

Statistical analysis of data.

Why should we want to do statistical analysis of data?

The main aim of this experiment was to find out what benefits the Brain Gym movement programme provided for the learning abilities of the children who took part. All of the children's abilities either improved or stayed the same, none got worse.

There are two key pieces of information that are required to make this type of study useful in the classroom. One is to assess the overall benefit to a group of children, as a kind of "cost-benefit" exercise, to see what is the "value-added" for the whole group, so that this can be compared to the whole group benefits of other interventions. The other key piece of information needed is what is the probability that an individual child will gain personal benefit from the intervention, compared to other types of intervention.

With these two pieces of information, the educator can make a rational decision about whether taking on a new intervention would bring sufficient benefits to balance the extra efforts required to introduce and maintain a new intervention programme.

Basic principles of statistical analysis in brief:

The data has been analysed by using the standard Student's T-test for paired samples. This test is applicable when a group of measurements has been taken from a group of children, and each child has been assessed twice. In this case each child has a "before" and and "after" score, and there are 19 children in the experiment.

Gosset published the T-test in Biometrika in 1908, under the anonymous name "Student", hence the name of the test. The test assesses mathematically whether the average score of a group of measurements "before" is statistically distinguishable from the average score "after".

Figure 1: synthetic data set to illustrate

the principles of statistical analysis

This is very important, because if there is a lot of variability in the ability levels of the children, for example, it is difficult to tell if a small change in the average score is significant and reflects a real change, or if the change is only an illusion, because of natural errors in making measurements.

In the synthetic example in the figure shown to the right, the purple data indicates the scores in an imaginary test with a possible maximum score of 12 points. The bars show the number of students with each score, and the purple colours indicate the imaginary range of scores before an intervention, and the green bars indicate after an intervention. In both examples the average score before (purple) is 5 and the average score after (green) is 6.

Is the improvement brought about by the intervention statistically significant?

It is NOT possible to guess the answer by looking at the graphs. The range of scores for "before" and "after" overlap by a large amount, for both examples.

Running the Student t-tests for the top set of data gives the result that these two sets of data (purple and green, before and after) are significantly different at the 95% confidence level. However, the two sets of data in the bottom graph are NOT significantly different, and cannot be distinguished statistically.

There are two possible ways of analysing pairs of data using the T-test. One is the "one-tail" test, which is valid for use only when you are testing whether the mean of the data has increased, and don't care if it actually decreased between the two measurements. The alternative "two-tail" test is more rigorous, and should be used where ever possible.

click here to return to top of page

Applying statistics to the data from Hannaford, 1995:

The Student's T-test has been applied to both the Reading Grade data set and the Mathematical Grade data set.

reading grade T-test results:

The first stage of the T-test is to decide what is the change in reading grade which would be expected to be observed between the first and second measurements of reading grade, if the intervention had had no effect. Since the test is designed for there to be NO change between tests, then we are looking for changes in reading grade that are significantly greater than 0. This of course is a conservative estimate, as we know that these children have consistently performed below average, and poor readers typically drop behind further as time elapses, which would provide an estimated negative change in grade. We do NOT use this estimate, because we want the analysis to be as strict as possible.

Mathematical programmes are available which calculate a number called the "T statistic" from the two lists of values of measurements before and after the intervention. For Hannaford's data, the T Statistic value is calculated to be 14.36 no expected change. We should use the two-tail T-test for proper statistical rigour. So, we compare the critical 2-tail value with the T-statistic value. The calculated T-statistic (= 14.36) is much greater than the critical 1-tail value of 2.10, so we can say that this indicates that the reading grades have increased by a value that is significantly greater than the 10 elapsed months. The probability level for this change to be due to random chance is P = 0.000 or less than one chance in a million.

The reading grade data therefore pass the T-Test, and the improvements in reading brought about by this Brain Gym programme are highly statistically significant with 99.999% confidence.

maths grade T-test results:  

Similarly, since the test is designed for there to be NO change between tests, then we are looking for changes in maths grade that are significantly greater than zero.

The T statistic value is calculated to be 2.40 for no expected change. If we compare the T statistic value (2.4) with the critical 2-tail value (2.10), then the improvement in maths is significant. The probability level for this change to be due to random chance (p = 0.027) is less than 1 chance in 22 (97% confidence).

The maths grade data therefore pass the T-Test, and the improvements in mathematics brought about by this Brain Gym programme are just significant statistically with 97% confidence. The students already started at higher levels in maths than reading, so it is reasonable that the improvements are not as great.

Conclusions:

This new statistical analysis indicates that in the experiment described in Hannaford (1995), regular 10 - 15 minute sessions using a selection of Brain Gym ® movements throughout the academic year produced highly significant improvements in reading ability and significant improvements in mathematical abilities of nineteen 10-11 year-old children with Special Educational Needs.

Improvements in basic skills often are accompanied by improvements in behaviour and other emotional indicators, and Hannaford comments that "the most surprising results (of this study) were the remarkable improvements in self-esteem and in their ability to focus on task".

click here to return to top of page

References:

Clisby C, Fowler MS, Hebb GS, Walters J, Southcott P & Stein JF, 2000. Outcome of treatment of visual problems in children with reading difficulties,  Professional Association of Teachers in Special Situations (PATOSS) Bulletin Nov, 9-14.

Dennison, P.E. & Dennison, G.E., 1985. Personalised Whole Brain Integration. Ventura, CA: Edu-Kinesthetics, Inc.

Dennison, P.E. & Dennison, G.E., 1986.  Brain Gym, simple activities for whole brain learning. Ventura, CA: Edu-Kinesthetics, Inc.

Hannaford, C., 1995. Smart Moves, why learning is not all in your head. Great Ocean Publishers.

click here to return to top of page

Brain Gym ^®  is a registered trademark of the Educational Kinesiology Foundation/ Brain Gym International.

This site was last updated 11/25/09