Promoting scientific thinking with information handling programs

Alistair Ross
Institute for Policy Studies in Education, University of North London
a.ross@unl.ac.uk
This article first appeared in MAPE Focus on Science Autumn 2000

Scientific Thinking is a phrase in which the stress should be on the word 'thinking' rather than 'scientific'. This article is about encouraging primary aged children ­ years 3 to 6 ­ to think about meaning, to play with ideas, to test hypotheses.

It is not about 'scientific facts' or about 'data'. Data, or information, is intrinsically boring to adults and to children: what is interesting and often exciting is knowledge and meaning.

Many people think that facts are necessary to further understanding, and must be learnt first.

Margaret Thatcher put this case well in her autobiography, writing about history:

'History is an account of what happened in the past. Learning history, therefore, requires knowledge of events. It is impossible to make sense of such events without absorbing sufficient factual information and without being able to place matters in their chronological framework ­ which means knowing dates.

No amount of imaginative sympathy for historical characters or situations can be a substitute for the initially tedious but ultimately rewarding business of memorising.'

Margaret Thatcher, The Downing Street Years,

1993 p 595

Each sentence here ­ no, each clause ­ is utter nonsense. The fact that knowledge and meaning is constructed around facts does not make the facts themselves any more interesting, nor mean that 'the facts' must be learned first. This is as true in science as in history, and is as true in the primary school as it is in postgraduate research.

This article will first consider how children learn, and then look at three specific examples of the development of scientific thinking. The case studies on which these are based were carried out in the early 1980s, on computers that are now positively palaeolithic in character, but the principles of teaching and learning remain as true. There seems to have been relatively little progress since work such as this was undertaken: soon after the last of the case studies described, the National Curriculum was devised, based very much on the sort of thinking typified by the quotation from Thatcher above. Perhaps this explains the break in progress.

Children's learning

Using the real world

Much of children's learning is based on direct experience.

They discuss their families, the weather, the local street and its shops, litter and traffic, and so on. British primary schools have a particularly good reputation for harnessing these experiences. They often encourage children to observe and describe their social and physical environment carefully, to collect both objects and information, and to use many of the so-called 'basic skills' in their responses and explorations. The emphasis is twofold: first, to build on the direct and concrete experiences of children, using their findings as a starting point for further discovery; and second, to stress learning through performing a process rather than through acquiring facts about a process. Thus, in primary school, children learn a good deal about the world by observing, describing and analysing that which is near to them. In the process of doing this, they acquire skills that will help them better observe, describe and analyse other aspects of the world which they may meet up with later.

Analysing and categorising

Children learn not just by looking and describing what is seen. They learn by thinking about their observations ­ by arranging them into mental categories of alike and unlike, by constructing abstract generalisations about events that will in future help them to learn quickly, classify and predict. We expect even very young children to learn in this way:

think of the enormous range of physical appearance in various breeds of dog, and of how soon young children can categorise all of them as 'dog'. Or how children learn to predict that hot stoves, and then hot things in general, are potentially hurtful. Primary school children learn to classify and sort an extraordinary variety of items, events and people. This process continues through life in the way that we all consider the social and scientific world about us.

Making hypotheses

The way in which we sort out what we observe is by the process of making and testing hypotheses.

Hypotheses are generally based on casually acquired personal experiences. If, for example, in a young child's experience of his own family and relatives, the father and other male adults go out to work, while mothers and other married females work at home, the child may make the very reasonable hypothesis that in all families adult females do housework, and adult males are in paid employment. The specific information has been generalised into a universal rule. Only later may fresh information challenge this rule.

Testing hypotheses

Hypotheses are constantly being tested and found wanting. Fresh hypotheses are then framed to meet the new evidence. Some hypotheses are very persistent, and it may take more than mere casual observation to find them wanting. Galileo and the testing of the acceleration of falling bodies of different weights from the tower of Pisa is an example of the disproving of a long and widely held hypothesis.

The philosopher Karl Popper has advanced the notion that all human understanding of the physical and social world is based on the building and testing of hypotheses and that hypotheses can only be proved false. No hypothesis, he argues, can be proved true. Hypothesis succeeds hypothesis, each new stage being more precise and refined than the previous stage. Even should the 'final' hypothesis be in fact true, we would have no way of knowing that we had reached this stage.

Problems: a range of results

Making hypotheses can be difficult. One reason is that there is often a range of observations, scattered about. For example, the various properties of a dog are not found in all dogs: we recognise an animal as a dog if it has a sufficient number of doggy features. We all recognise a family, but it can be very hard to define exactly what is meant by a family if we are to cover the full range of families. Most children, for example, generalise from their own experiences and those of their peers, and imagine that all families have children as members. If children measure the length and areas of horse-chestnut leaves, they find that the majority of leaves fall within a narrow band of lengths and areas but that a minority are larger, smaller, longer or shorter. The range of results makes it difficult to come to a generalised conclusion about dogs, families or horse-chestnut leaves.

Problems: making generalisations

In order to generalise, it is useful to have more than simply one's own experiences and observations.

We need an exchange of information, so that one person's false inference can be countered by the experiences of others. Learning, therefore, involves listening to and evaluating the observations of other people, and being prepared to modify or even abandon hypotheses as new information becomes available. This open-mindedness must be matched by an ability to investigate and share in a social manner, in discussion and argument with one's peers.

Problems: large numbers

The more experiences one tries to bring together, and the greater the scatter or range of observations, the more difficult it is for primary school children to generalise. Sharing information and making hypotheses from the evidence of three or four people is fairly easy; when it comes to a classful of experiences, problems of handling numbers arise.

When the teacher tries to bring in even more observations ­ which is necessary in order to continue hypothesising and testing ­ it can become very difficult. Sometimes statistics help ­ fractions, percentages, pie-charts and histograms; at other times probability language is used ­ 'most', 'many', 'a lot of', 'sometimes', 'usually', etc.

Information retrieval ­ banking information

Information retrieval with a computer can begin to meet some of these problems. Collecting together a body of information, for example, means that larger numbers of observations can be collated, and these can be quickly and accurately searched through.

Large numbers will not be such a problem if the information is held electronically, in a form that can rapidly be made accessible ­ unlike the casual sampling of the memories of a class (sometimes unreliable, often volatile, and on occasions distorted by absences), or lists on paper (dog-eared, mislaid and misread); the data file on disc can be read easily, quickly and reliably.

Much discussion and sharing of ideas is essential to agree on the nature of what data is to be collected and the standard of observation that is required, and this social interaction of a class or group engaged in building or interrogating a database should not be overlooked. It is in the discussion about how to proceed, and the debate about the significance of what has been found, that learning occurs.

Information retrieval ­ testing hypotheses

Children using an information retrieval system should always be talking and discussing together what they are doing. Sometimes discussions will be about advancing and testing hypotheses: could it be that there is some association between one finding and another? Sometimes the discussion will be to do with logic: how can we collect or structure the data to find the information we need? Some discussions will be between the two: what exactly do we mean when we categorise groups in this way? At a later stage, talk will be focused more upon significance: what have we found? How does this match our expectations: what we thought we would find?

Information retrieval ­ an interactive process

Information retrieval can thus greatly enhance the learning process. In particular, it should be seen as interactive inquiry, encouraging a dialogue not just between child and machine, but also between child and child, child and teacher. By liberating the class from the drudgery of number-crunching, and enabling children to have a sufficiently large body of data from which to generalise with confidence, it can enormously enlarge the range and depth of learning activities in the primary curriculum.

The case studies

In scientific investigations, whether in the primary classroom or the research laboratory, experimental results rarely offer the clean-cut and precise proofs that fiction portrays. In 'real' science, experiments are repeated and the results are rarely identical.

Variables are measured and considered, and experiments undertaken which try to hold these constant. In primary science the same sometimes happens, but the variations in measurement ­ experimental error ­ are greater, and if there are more than one or two variables things can become too complex to handle.

Some kinds of scientific investigations can be made more rewarding ­ and better science ­ through using information handling programs. The three case studies that follow describe how children between 8 and 10 years old made and tested scientific hypotheses. These two processes ­ putting forward possible explanations and then testing to see if the explanation fits the facts ­ lie at the heart of the scientific process. They are not usual activities: they can occur with young children every day, but school science is not always the forum in which they are developed. The case studies show children developing the skills of putting forward and testing their ideas about scientific information that interests them, and doing so in a situation where there is not necessarily a 'right' answer to be discovered.

Fossils

The second year junior class went fossil hunting.

We made a day trip to a disused chalk quarry on the North Downs, taking with us as many parents as we could muster. The quarry had many examples of fossil shells, bellemnites and sea urchins, and I was very confident that every child would come away with at least one example. The quarry was arranged in two steps, so that children were fairly easily able to search for fossils from two different strata of chalk, one some 5 m deeper than the other.

We had discussed how fossils were formed and how, over vast lengths of time, strata of rocks were formed. It seemed possible that we might be able to compare the types of fossils found in the two layers. Both layers would have been part of the same Cretaceous sea-bed, but the sea-life present at the two different periods of time might have been different.

To find out if this were so, we went armed with lots of plastic bags and copies of a sketch of the quarry. As each child found a fossil (and they all found several) they popped the fossil in a bag with a sketch of the quarry marking the spot where it was found and their name.

We had to climb the chalk cliff to reach the chalk face. It was very steep. We had to scramble. It was difficult. Lots of us slipped down. It wasn't very safe. We got scratched from brambles and stung by nettles. But most of us got to the top of the slope and could start looking for fossils.

Hardly anyone really hurt themselves. We dug for nearly a whole day, and brought back 97 rocks to display. There are probably millions more just waiting to be found.

We only had to tap the rock lightly, then it would crack open.

Back in school, the 97 rocks revealed 148 different fossils. Some were just fragments, others were complete. The children began the process of trying to identify each type. Naturally, the children rapidly acquired (and revelled in) the specialist vocabulary of palaeontology. This became easier as the principal varieties were recognised, and there was much interesting work to be done on recognising different kinds of symmetry.

One of the most common fossils we found was Inoceramus. An Inoceramus is a shellfish, and the shell is not symmetrical. This means that the shell is not balanced not the same on the left as on the right. But the Inoceramus had two shells, and these are mirrors of each other. The Inoceramus was connected to the bottom of the sea by a muscle foot that held the shells together. This rotted away very quickly, so you wouldn't usually find an Inoceramus with two shells sticking together, because when the Inoceramus died the hinge broke.

Rhynchonella fossils were very different from other shells because Rhynchonella is in the shape of a triangle. The ribs stick out at the side. They lived on the seabed.

The two shells of the Rhynchonella are not the same size. One is bigger than the other and has a different shape. But each shell is symmetrical, which is one way to tell the difference between it and Inoceramus and Entolium.

Entolium is in the same family as Inoceramus but is smaller, and the curves are smoother and closer together. There are two shells with the animal in between.

The shells are not symmetrical. They are different on each side. But the shells are mirror images of each other. The animal could swim in the water and was not fixed to the bottom like Inoceramus. They clapped their shells together and the water squirted out so they moved.

These descriptions also show how they managed successfully to synthesise their own observations with the information that they were able to get from reference books. They had to use adult books for this purpose, but were pretty determined to make sense of the rather difficult text. As a learning project, it was already a success. Skills of observation, communication, measurement and using reference materials were tackled with enthusiasm.

But the material that they had collected had so much more potential. There was enough of it to determine patterns of distribution, to make generalisations about types, to hypothesise about changes.

However, making generalisations about 148 different fossils was difficult. It was not possible for children to keep track of all the information in their heads. For each fossil they now knew the name of its finder, the name of the fossil, where it was found in the quarry, and whether it was complete or not. They added more information as they began to measure the dimensions of each fossil, and to sort them into actual fossils and impressions. Each fossil was also found in one of the 97 pieces of rock . . . and by now they had eight pieces of information on 148 fossils ­ almost 1200 items.

Sorting them physically would have been difficult ­ and have led to damage of the fragile shells.

This is the point where the computer came into its own. We put all the data we had collected into an information handling programme. All the numbers, names and records were entered at the keyboard into a file of data, which we called FOSSIL. The information handling program allowed the children to create the FOSSIL file, to type in the information (and correct it where errors were made) and then to interrogate the file to locate particular kinds of information (Table 1).

The easiest way to understand this is to imagine all the results displayed on a vast table like that in the diagram but going on for another 143 lines.

Each column is called a field, and contains similar kinds of data. In this example, there are 10 fields.

Each horizontal row is called a record, and contains all the information about each fossil. The fields called Fossil and Whole refer to whether the fossil was an actual fossil (Y) or impression (N) or to whether the fossil was whole (Y) or not (N). MapE and MapN are two co-ordinates that locate where the fossil was found on the sketch of the quarry.

Now imagine using this large table to sort out information. What was the average length of the Inoceramus, for example? Running a finger down the Name field one searches for all the instances of Inoceramus, noting, each time, the value in the Length field. Then one calculates an average. Having done this one realises that many of these examples would have been incomplete fossils, so one searches again with two fingers ­ one on the Name field (for Inoceramus), the other on the Whole field (for Y). Only when one finds these in both columns does one record the value in Length.

The computer does all this very quickly and very accurately. All the child has to do is to type a command to sort through the file searching the two fields for examples where Name = Inoceramus AND Whole = Y. Another simple command will then give the average of all entries in the Length field.

The children set up the data file by using the program to create the ten fields they needed. They specified how wide the fields needed to be (Rhynchonella has 12 letters) and if the information in each field was alphabetical or numerical. The program then prompted the children to feed in all the records, listing each field for each record. With a class full of children, entering all the records was relatively simple (six records each).

Groups of children came up with different ideas of what to look for. Some searched for the distribution of dimensions and averages of each particular species, discovering the typical size of each. Others searched for the different kinds found, showing them as histograms or pie charts (Fig. 1). Others looked at the distributions on the sketch of the quarry, seeing if there was any pattern in where species were found. They found that most of the Inoceramus were in the lower of the two strata, and that the Entolium and Rhynchonella were found in both layers.

Bellemnites were only in the lower level.

As each discovery was made, the children came up with new ideas to try out. Their hypothesising was firmly rooted in the data, with which they were quite familiar, but they nevertheless were encouraged to speculate freely because there was very little penalty in testing a 'wrong' hypothesis ­ the computer did the boring searches, and did them accurately and without wasting time.

The project ended with an exhibition of all the fossils and their findings, an assembly to explain everything to the rest of the school, and a booklet for each member of the class.

Conkers

The third year class became interested in the strength of conkers.

It began with a conversation one September morning on how to prepare a strong conker, with recipes of vinegar, slow baking, varnish and the like being paraded. The discussion spread to the whole class, and then concentrated on the natural qualities that gave strength. Various theories were put forward:

 The bigger the better.
Strong conkers are large and heavier.
The oldest conkers are strongest.
The heavier the better but size is not important

 I encouraged them to write down their theories, and then we discussed how we might test them.

Various methods of testing strength were discussed and tried. Dropping them from the first floor window to see if they cracked was abandoned when it was realised, first, that this method didn't allow for dropping from different heights, and second, that the strength of the impact would be partly dependent on the weight of the conker.

Dropping a standard weight onto the conker from gradually increasing heights seemed to offer a better method. The strength of a conker thus became the height at which a 1 kg weight cracked it. Simply dropping the weight often meant that the edge of the weight caught and split the conker: what was needed was some way of accurately hitting the conker with the flat surface of the weight. The children started by taking a set of cardboard tubes, of gradually increasing height.

The conker was put under the smallest, the weight dropped from the top, and the conker inspected.

The class soon realised that the tubes were interfering with the free fall of the weight and a group devised the Destructor Mark 2 ­ a cardboard aiming device, fixed to a rule for measuring children's height. A cardboard base with a cross marked the precise position for the conker to be placed.

Conkers were collected. In the end we had over 300, including a precious store of 1-year-old conkers sacrificed by one child in the cause of increased scientific knowledge. Each conker was numbered with a spirit pen, weighed, its age recorded (in days from falling from the tree), and its volume recorded by measuring the displacement in water. Then it was despatched to the Destructor.

We tested most of the 300 conkers, and each conker had, on average, ten drops of the kilogramme weight before it cracked. Each drop gave a dull reverberating thud that echoed around the school: the other classes and teachers were most forbearing!

Finding the volume of the conker
We got a measuring cylinder and filled it with
water up to 10 cubic cm. We dropped in the
conker and looked to see how much the
water had risen. Some of the little ones
floated so we had to push it down with the
end of a pencil. The amount of water risen
was the volume.

After the testing was over, the class were in a similar situation to the fossil hunters. They had a table of results ­ 1 m wide and 4 m long. The information handling program was essential if any sense was to be extracted from all these figures. A data file was constructed:

Number reference number

Age in days

Weight in grammes

Volume in cubic centimetres

Strength in centimetres × 1 kilogramme

Skinthick in millimetres

The skin thickness was measured with a calliper gauge on a number of conkers after they had been cracked, but we realised that this was taking a great deal of time and that each skin seemed to vary in thickness. These measurements were therefore abandoned after a while.

The children typed in the data as before with small groups working together typing and checking.

Before any analysis could take place, however, the children needed to decide what they meant by categories such as small and large, strong and weak. The class was divided into five groups, each of which had to determine the exact categories for weights, volumes, ages, strengths and skin thicknesses.

Each group made a simple enquiry to arrange the contents of a single field in order. For example, the volume field produced a string of numbers from 2 cc (of which there were 3) to 22 cc (of which there were 7). This data was presented in the form of a bar chart (Fig. 2). There was a bell-shaped distribution curve and from this the children decided that 'small' meant with a volume of less than 10 cc and 'large' was with a volume greater than 15 cc. This gave about 30% as large, 30% as small and the rest in between.

Similar categories were arrived at by the other groups of children. Thus the 73 'strong' conkers were the ones with a cracking point of over 32 cm, while the 'weak' ones were the 67 which were cracked at 20 cm or less.

Hypotheses were now tested. Scatter graphs were drawn showing the range of results. This demonstrated that there were no absolute rules, merely tendencies that were being investigated. The results of the scatter graphs were difficult to interpret.

A better way was to construct three block graphs, showing, for example, the strengths of the large, medium and small sized conkers. This produced observations such as:

Most of the big conkers are the strong ones.
The medium sized conkers aren't as good as the big sized conkers, but the small sized ones are definitely the worst.

A third tactic was to make smaller groups of one variable (say all the 1­2 day old conkers, then all the 3­4 day conkers and so on), and to find the average strength of each group using the computer to do this calculation.

They were all astonished to find that there was a negative correlation between age and strength.

Fresh conkers, less than 3 days old, were, on average, twice as strong as conkers that were 13­14 days old. There was a slight exception in that the year-old conkers proved to have regained some (but not all) of their strength.

There was a great deal of discussion about this discovery, and about whether the results should be communicated. There was a strong temptation to keep the information private, in order to have a maximum advantage in the next season's battles!

However, in the end, the need for proper scientific publication of results prevailed, in the form of an assembly. The conclusion was:

The strongest conker in the world is . . .
With a weight of over 17 grammes
With a volume of over 14 cubic centimetres
It should be 1­4 days old (or over 365 days old)
Big conkers are better than small conkers and
New conkers are better than old ones.

Parachutes

This investigation was undertaken by a group of third year juniors in early July ­ they were nearly a year older than the conker group.

'What makes a good parachute?' I asked.

There was, at first, a little confusion. Some of the children thought a 'good' parachute was one that made a dramatically fast fall and hit the ground with a satisfying thud. On reflection, they modified their view to fit in with the majority: 'the more gently it goes down, the better,' as one said.

We started making toy parachutes, using sheets of polythene, thread and yoghurt pots (Fig. 3). The children rapidly decided on a standard shape with a number of 'things we can change' (i.e. variables).

The chute was always circular with a variable radius. The number of strings could vary, as could their length. The empty yoghurt pot slung underneath could have its weight varied by adding plastic weights. These parachutes sailed down from our classroom's first floor window without much difficulty. Three more design problems remained.

First, the release of the parachute had to be standardised in some way. Hand-held releases were too close to the wall, and too much subject to human variability (such as a slight toss, or a jerk that opened the chute giving an unfair advantage).

After some design experimentation, a releasing device was constructed from a clothes peg, a bamboo cane and a piece of string. One arm of the clothes peg was fixed to the cane and the other to the piece of string. The chute was clamped in the jaws of the peg and then the pole thrust out of the window as far as possible. Tugging the end of the string remaining in the classroom released the parachute.

Second, they wanted to measure how softly the parachute landed. A good idea seemed to be to place a tray of carefully levelled sand under where the parachute would drop. But a few trials showed that it was impractical: even the slightest breeze caused sufficient drift to miss the target. They then tried sticking a lump of Plasticine to the base of the yoghurt pot, with a paperclip protruding exactly 1cm from this. The harder the impact with the ground, the further the clip was pushed into the Plasticine.

The final problem was to measure the exact time taken for the descent. Several children had electronic wrist watches with stopwatch facilities. The most accurate method of timing, they discovered, was for the person with the stopwatch to wait in the playground below and to watch carefully for the moment that the release string was pulled. By my reckoning, they were probably usually accurate to within about a fifth of a second, but several of them recorded their results to three decimal places!

Having found solutions to these technical problems, the class were ready to start making and testing their parachutes. But first we had an extended discussion about what we predicted would affect the speed of descent. The first proposal was that the larger the area of the chute, the slower would be the fall. This was agreed likely by the majority. A girl then suggested that the heavier the weight carried, the faster the fall. The class split on this and it was pointed out that if the weight was too light, the parachute would sail off in the wind. It was agreed that there would have to be a minimum weight sufficient to prevent this.

'What about the length of the strings?' asked another child. When pressed he re-framed this into a proposition ­ 'The longer the strings, the worse the parachute'. This was not agreed by most of the class, but neither was the majority sure that the converse was true. The thickness of the chute material was held to be important, speeding up the fall. One child then suggested that colourful materials would perform badly, but he declined to explain why he thought this. Another child suggested that the greater the number of strings, the better, but he was alone in this. Finally it was suggested that making a vent in the chute would cause a faster descent.

Table 2 summarises the various hypotheses.

At this point, one child pointed out that it might be a combination of these various factors working together that influences the speed of descent.

Instead of testing just one item at a time, we should look at combinations of variables. This caused a great deal of discussion. It was eventually agreed to test only one variable at a time initially.

Five groups each tested one variable. They agreed a common 'standard' parachute ­ with a 20 cm radius, six strings of 25 cm length, a 25 g weight and no vents. (We never got round to testing different materials.) Each group experimented with just one of the variables: thus group 'A' tried chutes of 10, 15, 20, 25 and 30 cm radius ­ all with six 25 cm strings and a 25 g weight. In all, 27 parachutes were made and tested over the week.

My group was C. We tested the length of the strings. Me, Felix and Berta made C1 (35 cm) and Alison and Cady made C2 (15 cm). Me and Felix made the parachute and Berta went into the playground and timed the parachute and Felix hung it out of the window. C2 was Cady and Alison. They made one but the plastic burst so they made another one but they haven't tested it yet.

 

There were still some problems with the investigation. It was agreed to stop testing if there was any wind (we lost two early examples to neighbouring rooftops). The strings got tangled and I gained a new reputation.

Our parachute strings kept on getting mixed up. It was quite hard to untangle it at first but I eventually did it. Then Mr Ross told us a new way of untangling the strings.

 

But nevertheless, results were beginning to emerge.

Our group is testing the weight. We used 1 gm cubes and put them in the cup of the parachute. Me, Tania and Nicola put 5 cubes in one parachute and Boris and Jan put 25 in theirs. What we wanted to find out was if more or less gm cubes made it go slower, which is what we were aiming for. Our first time was 4.19 secs and our second 3.18.
Boris and Jan's were 2.18 and 2.41. So it is the less the better so far.

Note the awareness of the provisional nature of the results 'so far'. The same child continued the next day:

 We have found out so far by testing a 10 gm weight and a 30 gm weight that the one with the 10 gm in it went down slower. This is how I thought it would be. We are going to carry on testing different weights to see if the heaviest are faster ones and if the lighter are slower and better. I definitely think the heavier are faster and the lighter ones slower.

It was at this point that the class created a data file of all the results they had collected so far. Each record consisted of the details of one experimental flight, so that there might be 7 or 8 records for each parachute. The file contained ten fields:

Number          reference number of that parachute

Radius            radius of the chute (in cm)

Strings            number of strings

Length            length of strings (in cm)

Weight            weight of load (in gm)

Ventno            number of vents in the chute

Ventsize         area of vent (in sq cm)

Ventshape     shape of vent(s)

Time                time taken to fall (in seconds)

Impact            force of impact (in mm of paper clip pushed in)

Entering all the data was a quick job ­ the fields were all short. By the end of the week the class has 127 records and began to interrogate the file. The first step was to decide what was meant by a 'slow' or a 'fast' parachute. An initial enquiry sorted all the times into order and they turned this into a histogram. From this they were able to decide what was fast (most agreed on less than 2.5 seconds) and what was slow (most agreed on more than 4 seconds). The rest were 'average'.

Groups now began to investigate the various characteristics of the three groups of parachute descents (fast, slow, average). One group of children, for example, examined the basic premise that slow falling parachutes landed with a softer impact than faster ones. The hypothesis was confirmed, but it was not a direct and absolute correlation.

 Our group was testing weight. By using lighter and heavier weights we found that the lighter ones went slower and the heavier ones faster. We worked out the average time for each parachute by adding all the times for that parachute and dividing by the number of tests for that parachute.* We put the averages on a graph and by that we can tell by looking at it.

 *(in fact the computer did this for them)

One group of children did not follow the simple three-fold division into fast, average and slow.

They created six categories.

less than 2 seconds       very fast
2­2.49 seconds               fast
2.5­2.99 seconds            fairly fast
3­3.99 seconds               fairly slow
4­4.99 seconds               slow
more than 5 seconds     very slow.

This group then found the average value for each variable, and created a series of graphs with titles such as 'Does the radius of the Chute Change the Speed of the Parachute?' (Figs 4 and 5).

These charts show us that one of the things that changed the speed of the parachute the most is chute radius. A 26 cm radius chute descends much slower than an 18 cm chute, which descends in about 1 second. If you wanted to make a very fast parachute you would need less strings, shorter strings and a heavier weight. If you wanted to make a slower chute you would need exactly the opposite: a larger radius, more strings, longer strings and a lighter weight.

The children were now beginning to think in terms of interacting variables. With more time it could have been taken further.

But our final experiment was to test the power of our predictions. The class used the computer to calculate the average dimensions of the parachutes that took 2.5 seconds or less to fall: an 18.5 cm radius, 5.4 strings each 23.5 cm long, with a 19 g weight and a 0.3 sq cm vent. This parachute was then made up ­ with 5 strings! A similar exercise undertaken for the parachutes that took longer than 4 seconds to descend gave a 24.5 cm radius, 6 strings of length 28 cm, a 16 g weight and a tiny 0.1 sq cm vent. This too was made up.

The two parachutes were released side by side.

The larger parachute took 4 seconds to fall ­ exactly 1.5 seconds more than the smaller parachute.

Conclusions

These three projects happen to demonstrate three different kinds of scientific observation, undertaken by children at average age 8.7 (fossils), 9.6 (conkers) and 10.4 (parachutes).

The quarry showed the problems of tracking the physical distribution of natural phenomena. Many biological distribution patterns are like this: how does the density of a particular plant vary according to soil type, rainfall and acidity? why are different species of mini-beasts found in different parts of the pond? The opportunities for sampling the distribution of plants and animals are enormous.

The information handling program records each item.

The conker experiment was rather different.

Here we were again measuring individual items, but each was an experiment ­ and an experiment to destruction. No experiment could be repeated, and we were faced with a problem of the distribution of results. Experiments on a single conker, or even a few conkers, would not reveal the rule. Mass testing was necessary to measure any tendency that was present. The information handling program records each experimental item.

The parachute experiments were repeatable. The problem of these experiments was not one of distribution of results, but of uncontrollable experimental error. It was not possible to ensure that the parachute was always released in precisely the same manner, or to control the variations in the breeze. The measurement of the time of descent was fairly crude with much observer error. All that one could do was to repeat the experiment and arrive at an average result. The information handling program recorded each experiment.

The use of data handling programs led to new possibilities in the case of each of the investigations.

I want to conclude this article by suggesting that these new possibilities were of a kind that makes it possible for primary aged children to take on hitherto unattainable levels of abstraction and hypothesis, and to develop skills of scientific thinking in an enjoyable ­ in a playful ­ manner. In each project, the investigation was possible without the use of a computer. It would have been successful without the computer. The three projects would have been broadly scientific, based on experiment and observation, bringing in a range of cross-curricular skills in language and mathematics, social skills of co-operative learning and often also skills in aesthetic areas such as art and drama.

However, the use of an information handling program, I suggest, did more than simply enhance these skills, making them better and more developed. It changed the qualitative nature of each project, because it positively encouraged children to formulate and test hypotheses in a way that was previously virtually impossible. Trying out ideas may sound straightforward, but when testing those ideas involves repetitious counting and sorting, it becomes an activity that children avoid. But if the sifting through the data, the mechanical task, is made quick, efficient and simple, then the intellectual task, of having the ideas and seeing if they fit, becomes much more attractive.

The children using the computer to sort the data were, in a very real sense, playing with information. They were trying out ideas in an almost casual way: casual, because it didn't really matter if the idea was wrong. It was easy to test, and nothing much was lost if nothing was proved ­ not much more than a little time.

But if the idea had something in it, then there was an immediate reward. And there was an incentive to explore how the data could be presented to make the pattern more explicit. Would changing the parameters of the categories make the relationship clearer? It was easy to try out ideas and see what happened. Children played with hypotheses and categorisation in an intellectual, a scientific, fashion.

Further, each successful hypothesis proved, invariably led to three or four more ideas to be followed up. I think that one could have charted some kind of exponential growth of the number of investigations going on, as the data became more familiar and the patterns and co-relationship became clearer. Forming and testing hypotheses may be at the heart of scientific experimental method, but it can be laborious and frustrating.

Taking the labour and frustration out of the process allows room for enthusiasm for discovery. This unshackling of the fetters is what I see as making a qualitative difference to children's scientific investigations.

There was also a degree of progression in the sophistication of the hypotheses offered by children of different ages. The youngest children (average age 8.7 years), working on fossils, made only implicit hypotheses. 'Let's see if the two types of fossils are different' was a typical remark. They were able to explore difference, and indeed to anticipate its presence, but not to attempt any prediction. This may in part have been due to the nature of the data on which they were working.

The next group of children (average age 9.6 years), working on the conkers, were more able to frame distinct hypotheses, and often to do so using two or more variables at the same time. In one sense, many of their predictions were intuitive ­ but this is also true of predictions made by many adults. In their testing and analysis, however, they tended to isolate each variable and to assume that it acted independently, so that even multivariate hypotheses were disaggregated into a series of mono-causal explanations.

Some of the oldest group of children (average age 10.4 years), working on the parachutes, were able to progress beyond this. This investigation had the largest number of variables, and some of the children were not simply advancing hypotheses involving several variables but were also able to see that there might well be an interaction between them. With some more help and more time, some of them might have been able to explore multivariate analysis a little, trying to detect if the variables did interact.

The progression in the development of hypothesising skills seemed linked to age although the three investigations, fortuitously, also offered increasing numbers of variables. The three groups of children were largely different (a small number of the parachute group had earlier been involved with the conker investigation) so one might speculate that, had there been a progressive exposure to situations in which hypotheses could be advanced and tested, there might have been an even more pronounced progression.

In all this work the facts were not important ­ they were, for the most part, trivial and did not need to be 'known' in any sense. The answers arrived at were relatively insignificant ­ children don't really need to know what makes a parachute slow, or what makes a conker strong. What made the learning experiences, described here, powerful was the process of learning: they were concerned with identifying and describing data, with the creation of useful definitions, with the suggestion of ideas, with playing with data to see how it fitted with different explanations. It was scientific thinking in action. But it's doubtful whether Mrs T would approve.

The major part of this article has been taken from Primary Teaching Studies, Vol. 4, No. 2, Polytechnic of North London, 1989.

[top of page]