JOURNAL OF RESEARCH IN SCIENCE TEACHING
VOL. 22, NO. 5, PP. 421-436(1985)
HIGH SCHOOL STUDENTS’ UNDERSTANDING OF FOOD WEBS:
IDENTIFICATION OF A LEARNING HIERARCHY
AND RELATED MISCONCEPTIONS
ALAN K. GRIFFITHS and BETTE A. C. GRANT
Department of Curriculum and Instruction, Memorial UniversiQ of Newfoundland,
Newfoundland, Canada AIB 3x8
Abstract
Developing an understanding of the nature of food webs is an important topic in today’s
biology curricula. The relationships represented in a food web are rule-like in nature. Hence,
it should be possible to construct a learning hierarchy for this concept. A hierarchy leading to
the ability to determine how a change in the size of one population can affect another population
in the same web but not on the same chain was hypothesized. Data from 200 subjects were
extremely consistent with the hierarchy. A second major focus related to the identification of
specific misconceptions held by subjects for food webs. The need to identify students’ misconceptions of important concepts has been expressed widely in the recent science education
literature. In the present article, an argument is presented for the usefulness of learning hierarchies in this work. Specific misconceptions and the frequencies of their Occurrence are reported.
Introduction
This article has two foci. First, it is concerned with the development and validation
of a learning hierarchy relating to the concept of “food web,” an important concept
in biology cumcula. Second, it is concerned with the use of the learning hierarchy
model as a tool to aid identification of students’ misconceptions in science. Particular
misconceptions relating to the concept of food web are reported.
Gagne’s learning hierarchy model is well known and will not be discussed in
detail here. However, it is noted that the model refers to learning in one of five
domains posited by Gagne (1972), namely the domain referred to as “intellectual
skills.” An intellectual skill relates to the ability to apply knowledge in familiar
situations. Hence, the ability to determine the effect of a change in one population in
a food web on another population in the web represents an intellectual skill, whereas
knowing that “populations in a given food web interact,” does not. The first case
requires that an individual apply his knowledge of food webs, while the latter requires
the recall of verbal information,
0 1985 by the National Association for Research and Science Teaching
Published by John Wiley & Sons, Inc.
CCC oO22-4308/85/05O421-16$04.00
GRIFF’ITHSAND GRANT
Biology teachers rate the concept of food web as a very important topic in school
biology (Finley, Stewart, & Yarroch, 1982). It appears to be central to an understanding of more complex ecological principles (Alexander, 1982). These, in turn, form
the basis for an understanding of crucial environmental issues including conservation,
pollution, and population management. In light of a leading professional ecologist’s
view that “we are abysmally ignorant of the ecosystems of which we are dependent
parts” (Odum, 1977, p. 1289), the importance of the food web concept becomes
apparent.
Interestingly, Finley et al. (1982) suggest that teachers do not perceive food webs
and related concepts to be particularly difficult, but Johnstone and Mahmoud (1980)
note that examiners’ reports on school-leaving examinations in Scotland over an eightyear period indicated that students experienced substantial difficulty. The rule-like
nature of the relationships in food webs suggested to us that a potentially profitable
approach through which to promote more effective learning would be the learning
hierarchy model. The model was attractive to us for another reason, which we will
now elaborate.
An emerging area of substantial interest in science education research is the
identification of students’ misconceptions and alternative conceptions of scientific
phenomena. The importance of this research is cogently expressed by Simpson and
Arnold (1982):
There is however a growing recognition that it is not merely the absence of
appropriate mental operations, concepts or skills which inhibits learning; the presence of previously acquired theories, informationand skills which may be incorrect
or inappropriately applied may actively interfere with the acquisition of new material. (p. 174)
There is evidence that teachers are typically unaware of their students’ particular
misconceptions (Johnstone, Macdonald, & Webb, 1977; Hart, 1979; Simpson & Arnold, 1982). Clearly there is merit in the position taken by Johnstone et al. (1977)
and Nussbaum (1981) that teachers need to be made more aware of students’ particular
misconceptions and how to identify them. We concur with the position advocated by
Strike and Posner (1982) that “accommodation is more likely to occur if instruction
can be organized so that teachers can spend a substantial portion of their time in
diagnosing and correcting errors in student thinking” (p. 239).
The literature reveals a number of interesting reports of students’ particular misconceptions and alternative conceptions of scientific phenomena. Most of the concepts
studied are broad in nature and relate to the physical sciences. For example, among
the concepts studied are the particulate nature of matter (Doran, 1972; Novick &
Nussbaum, 1978); dynamic equilibrium (Johnstone, Macdonald, & Webb, 1977); heat
(Erickson, 1979, 1980); chemical equilibrium (Wheeler & Kass, 1978); elementary
dynamics (Viennot, 1979); light (Stead & Osborne, 1980); gravity (Gunstone & White,
1981).
Very few misconceptions studies have related to biological concepts, although
Brumby (1982) reports data relating to students’ misconceptions of “life,” and Deadman and Kelly (1978) report on misconceptions about “evolution.” It is particularly
timely, therefore, to add to the literature on students’ misconceptions about biological
phenomena.
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
422
423
Finally, before turning to the details of our study, we comment on the potential
usefulness of the learning hierarchy model as a tool for the identification of students’
misconceptions. Typically, researchers have investigated students’ misconceptions and
alternative conceptions for broad concepts of the kind listed above. Concepts which
may be represented as intellectual skills are less exotic but nevertheless important.
They are more narrowly defined in terms of specific operations, and are typically
encountered for the first time in school learning. They are amenable to task analysis
and representation in learning hierarchies. For such concepts the underlying source of
misconceptions causing inadequate representation of the overall concept may be traced
to inadequate learning of subordinate skills. Unless these subordinate skills are identified and specifically tested, the underlying cause of a particular misconception may
not be readily identified.
The Hypothesized Hierarchy
The initial step in any learning hierarchy study is to attempt to prescribe the level
to which the concept will be taken, and to state the target skill in behavioral terms.
Consideration of the concept of food web led to some restrictions, which we will now
explain:
A food web is a model, a simplified representation of feeding relationships in a
given community. The more realistically the model mirrors nature, the more complex
it becomes. The process of determining how an unusual increase or decrease in one
population will affect another population in a real life situation is an onerous, if not
impossible, task. In much the same way, complex food web diagrams present an
arduous task to anyone attempting to determine the net effect of a change in one part
of a food web or another. The exercise becomes one of frustration in attempting to
untangle endless numbers of pathways. Surprisingly, students are occasionally presented with such tasks. A typical example may be found in a popular biology text
(BSCS, 1970, p. 79). In the present study, subjects were not required to consider
food webs involving more than four routes between populations in question. It was
felt that increasing the complexity of food webs beyond this level would only increase
the tediousness of the task, without demanding greater understanding of the basic
skill.
The terminal skill of the hypothesized hierarchy was defined as, “Given a good
web diagram, determine the effect of a sudden size change in one population on a
second population which is not on the same food chain, when the effect may be
transmitted along more than one route.” The hypothesized hierarchy was derived by
asking of this skill, and then each successive skill, “What should the learner already
be able to do in order to learn this skill?” Test items for each skill in the hierarchy
were developed at the same time. This procedure involved a number of revisions, but
eventually field testing was conducted. Following this the hierarchy was substantially
revised. The component skills and a specimen test item for each are represented in
Table I. For convenience, in the present article all specimen items refer to the food
web illustrated in Figure 1, but in the actual tests different food webs were used for
different items. Figure 2 shows the arrangement of these skills in the hypothesized
hierarchy.
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
TABLE I
Descriptions of Skills Comprising the Food Web Hierarchy"
Skill
Description
9
Given a food web diagram, determine the effect o f a sudden size change i n one
population on a second population which i s not on the Sam food chain, when the
e f f e c t m y be transmitted along nure than one route,
For example, "determine the e f f e c t of a sudden increase i n population F on the
s i z e o f population R" (Figure 1 ) .
8
Given a food web diagram, determine the e f f e c t o f a sudden s i z e change i n one
population on a second. non-adjacent, population located lower on the same food
chain, when the e f f e c t i s transmitted along more than one mute.
For e x q l e , "determine the e f f e c t Of a sudden decrease i n population P on the
s i z e o f population L" (Figure 1 ) .
7
Given a food web diagram, determine the e f f e c t o f a sudden s i z e changC i n one
population on a second, non-adjacent. population located higher on the same food
chain, when the e f f e c t i s transmitted along more than one mute.
For example. "drtarmine the effect of a sudden increase i n population L on the
s i z e o f population P" (Flgure 1).
6
Given a food web diagram, recognize a l l possible pathways thmugh which the
e f f e c t of a change i n one population i s transmltted t o a second population.
For example. " c i r c l e the l e t t e r s i n d i c a t i n g populations thmugh which the e f f e c t
o f a'chpnge i n population 0 i s passed on t o population P" (Figure 1).
5
Given a food web diagram, determine the e f f e c t o f a sudden s i z e change i n one
population on a second, non-adjacent', population located lowcr on the sam f w d
chain. when the e f f e c t i s transmitted along only one route.
For example, "determine the e f f e c t o f a sudden decrease i n population H on the
slze o f population N" (Figure 1).
4
GiJen a food m b diagram, determine the e f f e c t o f a sudden s i z e changc i n one
population on a second population. n o t located on the same food chain, when the
e f f e c t i s transmitted along only one mute.
For example, "determine the e f f e c t o f a sudden increase i n population G on the
s i z e o f population A" ( F i g u n 1).
3
Given a food web diagram, determine the e f f e c t o f a sudden s i z e change i n one
population on a second, non-adjacent, population located higher on the same food
chaln. whrn the e f f e c t i s transmitted along only one route.
For r x u p l e , "datermine the e f f e c t o f a suddan decrease i n population N on the
s i z e o f population H" (Figure 1).
2
Given a food web diagram, detarmlne the e f f e c t o f a sudden s i z e change i n a
predator population on i t s prey population.
For example. "deternine the e f f e c t o f a sudden increase i n population G on the
s i z e o f population N" (Figure 1).
1
Given a food web diagram, determine the e f f e c t o f a sudden s i z e change i n a P m y
population on I t s predator population.
For exungle, "deternine the e f f e c t o f a sudden decrease i n population N on tha
s i z e o f population G" (Figure 1).
ONote: Each skill is illustrated by reference to the food web shown in Figure 1 .
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFITHS AND GRANT
424
425
Fig. 1. A sample food web (each letter denotes a component population of the
web).
Hierarchy Validation
General Principles
Ideally, a learning hierarchy consists of a series of superordinate-subordinate
relationships between pairs of intellectual skills which are related according to two
criteria. First, there should be no instances of individuals who can exhibit the superordinate skill in a pair without being able to exhibit the subordinate skill. This is the
inclusion hypothesis. Second, learning of the subordinate skill should enhance learning
of the immediate superordinate skill. This is the transfer hypothesis.
It has been argued (Bergan, 1980; Griffiths, Kass, & Cornish, 1983; White &
Gagne, 1974) that hierarchies should be validated according to both of these criteria.
However, with a few exceptions (Bergan & Jeska, 1980; Griffiths, Kass, & Cornish,
1983) recent studies have considered only the inclusion hypothesis,
In practice, some exceptions to the inclusion hypothesis are usually found and it
is necessary to estimate the level of statistical confidence at which these can be ascribed
to chance. The White and Clark (1973) test of inclusion has been used for this purpose
in a number of recent studies (Beeson, 1977; Linke, 1975; White, 1974). A more
recent method (Dayton & Macready, 1976) subsumes the White and Clark test and,
because it considers skills in groups or for small hierarchies even the complete hierarchy, is more robust.
It is not possible to fully describe, in the space available, the development of
Dayton and Macready’s model. Briefly stated, the probability of a subject producing
a specific response pattern u, if the hierarchy is valid, is given by
. .
SKILL
1
. .
SKALL
Fig. 2. The arrangement of component skills in the food web hierarchy.
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
GRIFFITHS AND GRANT
where
vI represents the set of j consistent (“true”) response patterns, and k is the number
of skills in the postulated hierarchy. a,and p, represent the probability of observing
1-for-a-0 and 0-for-a- 1 errors, respectively, and are labeled guessing and forgetting
parameters by Dayton and Macready .
Essentially what is implied by the use of Eqs. (1) and (2) is that the values of
the misclassification parameters a, and p, are, respectively, raised to a power representing the number of guessing and forgetting corrections necessary to fit all true
pattern vectors to an observed data vector. Similarly, (1 - a,)and (1 - p,) are
raised to the number of “correct” responses in each case. The product of these overall
possible response patterns is Eq. (2). Multiplying this by the probability (0,) that the
jth true pattern vector occurs, and summing this product for all true pattern vectors,
yields Eq. (l), which represents Dayton and Macready’s general probabilistic model.
The goodness of fit between data and the hierarchy model is then calculated by both
a Pearson chi-square test and a likelihood ratio expressed in the form of a chi-square.
The latter appears to be more useful as it is less severely distorted by small frequencies.
The Dayton and Macready procedure was used in the present study.
With respect to transfer, a situation in which all subjects exhibited all skills would
be consistent with perfect transfer. Similarly, a situation in which all subjects failed
all except the bottom skill(s) would be consistent with lack of transfer. However, in
each situation it would be impossible to demonstrate transfer. Hence, as in the case
of the inclusion hypothesis, moderate discrimination between successive skills is the
most desirable response configuration for the purpose of establishing the transfer
validity of a hierarchy. No generally accepted procedure to establish transfer validity
has yet been established. Early studies applied Gagne’s (1962) proportion positive
transfer, now discredited by White (1974). White and Gagne (1974) suggest a direct
method involving teaching some subjects in the hypothesized hierarchial direction and
others in the opposite direction. However, the number of subjects needed for other
than very small hierarchies is large, and the deliberate teaching of some subjects in
the wrong order raises ethical questions. More recently, Bergan (1980) suggests the
use of part coefficients between skills as measures of transfer, while Griffiths (1979)
applies a chi-square test to determine directly whether gain of a skill between two
testing situations is related to gain of its directly subordinate skill(s). This method
was applied in the present study.
Procedure
The transfer design incorporated a test-retest format, with a specially designed
individualized remedial treatment interposed between the two tests. The purpose of
the remediation was to enhance learning of skills failed in the first test, so that the
pattern of gain of skills from one testing situation to another, and hence the degree
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
426
427
of transfer, could be determined. The first test was designated pretest, even though it
was administered after normal classroom instruction. The second test was administered
after remediation and was designated posttest. Each test was composed of nine subtests, one for each skill, with two equivalent items per skill. Identical items were used
in the two tests, but in different random arrangements. Each item was free response,
and involved a food web that was different but geometrically similar to those used in
other items. Populations were represented by letters rather than names of real species,
and the letters had no common meaning from item to item. Hence, recall of specific
items from one test to the next was discounted, and was certainly not evident in
subjects’ response patterns. In contrast to the test items, in the remedial unit food
webs were expressed in terms of familiar species. Both tests and remedial unit were
field tested and modified as necessary prior to the main data collection phase.
In the main data collection phase, each instrument was administered to 200 grade
ten biology students in three St. John’s schools. Eight coeducational classes and three
teachers were involved. The pretest was marked and returned to subjects two days
after testing. At that time each subject received the remedial unit together with an
individual prescription to study those skills failed in the pretest. Three days later,
subjects were administered the posttest.
When the data collection phase was complete all papers were re-evaluated, and
each item was analyzed to identify the particular errors made by subjects. For each
subject the same error was counted only once for each pair of items testing a given
skill, regardless of the number of times the error was exhibited.
Test Validity and Reliability
Construct and content validity were established through the expert opinion of a
group of five science teachers and educators. Reliability for each subtest was determined by consideration of the level of agreement in subjects’ responses to the two
items involved.
Ideally, in a two-item test of the same skill, subjects should get both items right
or both items wrong. Agreement was never less than 90% for any pair of items.
Whenever possible a tetrachoric coefficient was determined. However, it is to be
anticipated that the lower skills in a hierarchy will be exhibited by most subjects,
while the upper skills may be failed by many subjects. Inevitably, data will be more
skewed for skills at the extremities of the hierarchy than in the middle. In the present
case, the double testing design incorporating the same items allows some relief from
this problem. Except for Skill 2, as indicated in Table I1 a tetrachoric coefficient
incorporating a correction for skewness (Jenkins, 1955) could be obtained from at
least one of the tests. All of these coefficients indicated good agreement between the
pair of items involved. In the case of Skill 2, 97.5% of the sample were correct on
both items in the pretest and 98.5% were correct on both items in the posttest. Hence,
for Skill 2 the data were too skewed to allow determination of the coefficient. In
general, however, the reliability of each subtest is well supported.
Analysis
As preliminary evidence and to illustrate the general character of the data, the
level of difficulty of each of the nine skills is presented in Table 111. The order of
difficulty is consistent with that expected from the hierarchy except for the relation-
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
TABLE I1
Tetrachoric Correlation Coefficients (r& F’re- and Posttests
Sklll
1
2
3
4
5
6
7
8
9
Pretest
Sample S i t e
(n)
rtet
P
2.13
5;ls
6.15
10.16
497
12,17
1.14
9.11
3.18
149
161
135
118
163
78
169
182
116
.85
<.05
4,11
2.17
5;l8
6.10
1,16
7,14
3.12
9.13
8.15
164
135
123
164
143
1 30
161
157
149
Corresponding
Test Items
Posttest
t
.62
.79
.72
.87
c.01
c.00
e.00
c.00
t
*
.75
.89
.98
.95
.95
.75
c.00
c.01
<.oo
c.00
<.OO
c.00
*rtctcould not be determined for items for which zero frequencies occurred in
at least one cell of the data matrix.
TABLE I11
Proportion of Sample Mastering Skills of the Hierarchy
Ski 11
1
2
3
Pretest
Posttest
1
2
3
4
Sample S l r e
(n)
Proportion o f Sample
Masterln t h i s S k i l l
149
161
135
118
163
78
169
182
116
91.3
97.5
81.5
83.1
62.0
17.9
164
135
123
164
143
130
161
157
149
99.4
98.5
91.9
84.1
91.6
74.6
56.5
54.1
18.1
9%)
0.0
0.0
0.0
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFITHS AND GRANT
428
429
TABLE IV
Dayton and Macready Analysis for Goodness of Fit of Subhierarchies
A, B, and C
Guessing
Parameter
Forgetti ng
Parameter
Maximum
Likelihood
Degrees
o f Freedom
A
0.00
0.00
0.00
8
8
0.00
0.00
0.00
8
<o.oo
<o.oo
C
0.00
0.00
0.00
8
co
A
0.00
0.02
0.94
8
<
6
0.00
0.00
0.00
8
c
C
0.00
0.00
0.00
8
<
SubHierarchy
Pretest
.00
Posttest
.oo
.oo
.oo
ships between Skill 9 and Skills 7 and 8, respectively, for the pretest where no subjects
were successful for these three skdls. However, for the posttest many more subjects
were successful on the items representing these skills, and the order of difficulty was
completely consistent with that predicted by the hypothesized hierarchy. However,
while increasing difficulty between successive skills is consistent with a valid hierarchy, it is not sufficient to establish the validity of the hierarchy.
To establish the inclusion validity of the hierarchy the Dayton and Macready
procedure was applied to pretest and posttest data. Because the number of possible
patterns in the complete hierarchy was too great to be accommodated by the computer
programme available, three subhierarchies involving Skills 9, 8, 7, 6 (subhierarchy
A), Skills 8, 5 , 4, 2 (subhierarchy B) and Skills 7, 4, 3, 1 (subhierarchy C), respectively, were considered. Table IV indicates the results of these analyses.
Examination of Table IV indicates an extremely good fit between hypothesized
hierarchy and data (< 0.00).
Generally, it is to be expected that few subjects will fail the lower skills in a
hierarchy. Hence, for these skills it is difficult to establish transfer. This was true in
the present case for Skills 1 to 5 . Consequently, for each connection involving these
skills the subsample derived from the pretest was too small to allow a meaningful
test. Hence, the test was applied to Skills 6 , 7 , 8, and 9 only. The results are presented
in Table V.
In each case, significant transfer (< 0.00) was found. The connections from Skill
6 to Skill 7, Skill 6 to Skill 8, and Skill 7 to Skill 9 are strongly supported in terms
of their transfer characteristics. It would clearly be preferable to have been able to
test for transfer between the lower skills and from the lower skills to the upper skills.
However, given the relatively low level of difficulty of the lower skills it is more
important to have established good transfer for the upper skills. In particular, the
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
TABLE V
Transfer of Learning Test, x2 Valuesa
Connection
6
7
Subordinate Not Gained
SuperSuperordinate
ordinate
Failed
Passed
(n)
(n)
10
0
Subordinate Gained
SuperSuperordinate ordinate
Failed
Passed
(n)
(n)
2
X2
18
18.91
p
c.00
6
8
10
0
2
18
18.91
..OO
7
9
26
0
15
15
15.30
<.OO
8
9
24
0
14
16
15.72
c.00
“Note: All subsamples ( n ) are taken from those subjects failing both skills of a
particular connection in the pretest.
transfer data highlight the crucial importance of Skill 6 , namely the need for the
learner to be able to recognize all possible pathways through which the effect of a
change in one population is transmitted to a second population.
Misconceptions
The test items developed for the present study followed a free response format.
Upon responding to a given item, subjects were asked to indicate the reasoning used
in arriving at their answer. This information proved to be useful in evaluating subjects’
mastery status with respect to a given skill. Just as importantly, it was useful in
providing information which gave clues to subjects’ misconceptions. In some cases,
it was possible only to make inferences regarding a possible underlying misconception,
upon considering the errors made by subjects; in other cases, subjects stated misconceptions openly.
In order to identify misconceptions, all items which were answered incorrectly
on the pretest were scrutinized. Taken one item at a time, subjects’ explanations of
the rationale used in answering these items were analyzed for key ideas or common
criteria, and were then transcribed on field cards. A series of categories of misconceptions were developed and coded numerically. In this way, the frequency of each
particular misconception could be determined for each skill and for groups of skills.
It should be noted that categories of misconceptions were developed after the major
common ideas had been identified in subjects’ explanations; they were not predetermined categories of misconceptions. Where a subject exhibited more than one misconception in responding to a given item, each individual misconception was reported.
Misconception 1: The interpretation of food web dynamics in terms of a food
chain.
Almost the entire sample (95.5%) made a common error when asked to determine
the effect of a sudden size change in one population on a second population which is
part of the same food web. These subjects failed to consider that the effects of a
change in one population could be passed along several different pathways as it
approached the population in question. Rather, they tended to select one of several
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
GRIFFITHS AND GRANT
430
43 1
pathways and considered successive predator-prey relationships until they reached the
population in question. Such students appear to have dealt with food web relationships
by applying a strategy better suited to food chains. In a food chain, effects of a size
change in one population travel to a second population from one direction, along one
pathway of predator-prey relationships. In contrast, the nature of the food web model
permits the effects of a change in one population to spread through a “web” of
pathways, prior to reaching a population in another part of the web. Each pathway
transmits an effect on the population in question, and each should be considered in
determining the net effect of the initial change. Unfortunately, when subjects were
asked to disclose their strategy, they indicated that they utilized one pathway in the
web, without explaining why they chose a particular pathway, or why they based their
answer upon consideration of only one pathway. Although the underlying misconception was not revealed directly, a possible explanation for this incorrect strategy can
be inferred from student responses.
Typically, students encounter the food chain model prior to the food web model.
This is quite reasonable, considering that the former represents a less complex model
on which students may begin their understanding of feeding relations between different
populations in a community. Food webs are often introduced as a more realistic model
for the representation of feeding relationships than the simple food chain model.
Structurally, food webs resemble, and may be described as, a network of food chains.
However, misconceptions may occur if students consider a food web to be functionally
like a network of individual food chains. Subjects in the present study did not appear
to perceive the dynamics of food web relationships in terms of an interrelated network
of populations. Rather, they persisted in apparently arbitrarily singling out an individual food chain or single pathway of populations in the food web structure and
considering only those populations along this chain. The influence of the food chain
strategy is apparent, for example, in the following subjects’ responses to food web
problems presented in the pretest:
If population K undergoes a sudden increase, it will not affect population G because
both are on different food paths.
The effect varies because of different routes a food chain could take.
Brumby (1982) encountered a similar problem in a study involving 52 first-year
biology students at a British university. She found that over one-half of the sample
interpreted statements involving the food web concept in terms of a food chain.
Although Brumby does not address the question of why students tend to interpret the
notion of a food web in terms of a food chain, she does note that this problem persisted
despite years of secondary school biology.
In the present study, this misconception was encountered for items representing
all skills in which effects spread from one population to another along more than one
pathway (Skills 6, 7, 8, and 9).
Misconception 2: In a food web, a change in one population will only affect
another population if two populations are directly related as predator and prey.
Surprisingly, following classroom instruction on the food web model, 16%of the
sample proposed that if one population in a food web undergoes a sudden size change
it will have no effect on a second population which is not directly related as its predator
or prey. The occurrence of this misconception is especially surprising as it indicated
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
GRIFFITHS AND GRANT
that a considerable proportion of the sample did not appear to understand a basic
premise underlying the food web concept: that populations which are not directly
related as predator and prey can still influence each other because they are part of a
common food web. This misconception was identified for all skills in which subjects
were asked to determine the effect of a change in one population on a second, nonadjacent population (Skills 3, 4, 5, 7, 8, and 9).
A small group of subjects (2.5% of the sample) exhibited a related misconception.
When asked to determine the effect of a change in one population on a second population in another part of the web, they suggested the two populations were “too far
apart” to affect each other. These subjects appear to have set an arbitrary distance
with regard to how far removed two populations may be and still affect each other,
as this misconception occurred in response to Skill 9 only. The populations referred
to in items testing Skill 9 tended to be placed on opposite sides of the food web,
whereas in other items, the populations under consideration were not so far removed
in the web. The following responses are indicative of this misconception:
There will be no effect. The two populations are not really closely linked.
No effect. The populations are too far away on the food chain.
Misconception 3: A population located higher on a given food chain within a
food web is a predator of all populations located below it in the chain.
Almost one-fifth of the sample (17.5%) incorrectly assumed that a population
located higher in a given food chain within a food web is a predator of any population
located lower in the chain. As a result, these subjects were prevented from correctly
solving problems involving populations which were not directly related as predator
and prey. This misconception was observed in response to items testing skills which
require that subjects determine the effect of a change in one population on a second,
nonadjacent population in the food web (Skills 5, 7, 8, and 9).
Misconception 4: A change in the size of a prey population has no effect on its
predator population.
Six percent of the sample mistakenly believed that a change in the size of a prey
population would have no effect on its predator population. These subjects appear to
be unaware that food supply, in the form of a prey population, is an important factor
influencing the size of a predator population. Although this misconception which
involves a simple predator-prey relationship did not occur frequently, that such a
misconception is found at all at this level of study is a cause for concern. In a study
of the predation concept involving elementary students, Powell and Powell (1982)
suggest that a realistic understanding of this concept is important. They advise that
children who learn to understand predation will have a broader perspective upon which
to base important societal decisions concerning conservation of wildlife and population
management. It is therefore disturbing to note the persistence of this misconception
at the grade ten level.
Misconception 5: If the size of one population in a food web is altered, all other
populations in the web will be altered in the same way.
Four percent of the sample appeared to believe that “whatever happens to one
population in a food web, will happen to all populations in the web.” The following
responses are representative of this misconception:
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
432
433
TABLE VI
A Summary of Misconceptions Related to the Food Web Concept
Misconception
S k i l l s i n which the Misconception Occurred
1
1.
2.
3.
2
3
4
5
6
7
0
9
x
x
x
x
x
x
x
x
X
x
x
x
x
x
The interpretation o f food
web dynamics i n t e r n o f a
food chain.
I n a food web. a change i n
one population w i l l only
a f f e c t another population
i f they are d i r e c t l y r e l a t ed as predator and prey.
x
x
A population located higher
on a given food chain w i t h i n a food web i s a predator
o f a l l populations located
below i t i n the chain.
4.
A change i n the size o f a
prey population has no
e f f e c t on i t s predator
population.
5.
I f the size o f one populat i o n i n a food web i s
altered. a l l other opulations i n the web w l f l be
altered i n the same way.
X
x
x
X
X
x
x
Population L will decrease because if one part of the food web decreases, this wilI
cause populations throughout the food web to decrease because there is less food
(in response to pretest item 1).
If there is an increase, then all other populations increase throughout the web
(in response to pretest item 3).
Population L would decrease in size also because they are each interacted,
and so if one decreases, everything else will be affected also (in response to pretest
item 1).
In utilizing these strategies in determining the effect of a change in one part of a food
web on another part of the web, these subjects appear to ignore basic predator-prey
relationships. They seem to prefer to adopt a strategy which may result from some
misconception of the concept of food web stability, in which all populations interact
and react to changes in the web. For some reason, these subjects suggest they all
react in the same way.
A summary of the five major misconceptions and the skills in which they were
encountered is presented in Table VI. Examination of Table VI indicates that misconceptions 1, 2, 3, and 5 were identified from the responses of some subjects to the
item testing the terminal skill. Further, these misconceptions ran through responses
to items testing a number of skills. Conversely, misconception 4, although identified
for a number of skills, was not identified in items testing the terminal skill, emphasizing the usefulness of the hierarchical model as a means to diagnose students’ precise
learning difficulties.
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
GRIFFITHS AND GRANT
Implications
Identification of a learning hierarchy for the food web concept clearly has implications for teaching and design of instructional materials relating to this concept.
Instruction which follows the order of skills in the hierarchy, while not guaranteeing
success, enhances the prospect of it. Failure to learn skills which are subordinate to
one or more other skills in the hierarchy, while not guaranteeing failure of these other
skills, seriously diminishes the prospect of success. One skill in the hierarchy, Skill
6, is worthy of special mention in this regard. This skill focuses on the need to
recognize all possible pathways through which the effect of a change in one population
is transmitted to a second population. The data in the present study brought the
importance of this skill into sharp focus. Without it, the learner is essentially confined
to food chain problems. Mastery of it is the key to expansion to food web problems.
While this may seem obvious, it was apparently not to the teachers involved in our
study, nor to many writers of textbooks. Moreover, the problem persists to the tertiary
level (Brumby, 1982). Teachers would do well to pay attention to this problem, as
well as to the other misconceptions revealed by the present study, whether or not they
wish to apply the learning hierarchy model to their resolution. For those who favor
this model the prospect of remediation through combination of a generalizable learning
structure with specific identifiable individual misconceptions is likely to be enhanced
by the study reported here.
References
Airasian, P. W., & Bart, W. M. (1975). Validating a prior: instructional hierarchies. Journal of Educational Measurement, 12(3), 163-173.
Alexander, S . K. (1982). Food web analysis: An ecosystem approach. The American Bio!ogy Teacher, 44(3), 186, 189-190.
Beeson, E. W. (1977). Hierarchical learning in electrical science. Journal of
Research in Science Teaching, 14(2), 117-127.
Bergan, J. R. (1980). The structural analysis of behavior: An alternative to the
learning-hierarchy model. Review of Educational Research, 50(4), 625-646.
Bergan, J. R., & Jeska, P. (1980). An examination of prerequisite relations,
positive transfer among learning tasks, and variations in instruction for a seriation
hierarchy. Contemporary Educational Psychology, 5(3), 203-2 15.
Biological Sciences Curriculum Study. (1970). Biological science: Patterns and
processes (Teacher's edition). Toronto: Holt, Rinehart and Winston.
Brumby, M. (1982). Students' perceptions of the concept of life. Science Education, 66(4), 61 3-622.
Dayton, C. M., & Macready, G. B. (1976). A probabilistic model for validation
of behavioral hierarchies. Psychometriku, 41(2), 189-204.
Deadman, J. A . , & Kelly, P. J. (1978). What do secondary school boys understand about evolution and heredity before they are taught the topics? Journal of Biological Education, 12(l), 7-15.
Doran, R. L. (1972). Misconceptions of selected science concepts held by elementary school students. Journal of Research in Science Teaching, 9, 127-137.
Driver, R., & Easley, J. (1978). Pupils and paradigms: A review of literature
related to concept development in adolescent science students. Studies in Science
Education, 5 , 61-84.
John Wiley-Journal of Research in Science Teaching-WG: C2994
Pack 2 (MV8000/1)-Bin 6&-Pg. Dsn. 123-03-05-85 09-1 1-23 DJM
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
434
435
Erickson, G. L. (1979). Children’s conceptions of heat and temperature. Science
Education, 63(2), 221-230.
Erickson, G. L. (1980). Children’s viewpoints of heat: A second look. Science
Education, 64(3), 323-326.
Finley, F. N., Stewart, J., & Yarroch, W. L. (1982). Teachers’ perceptions of
important difficult science content. Science Education, 66(4), 531-538.
Gagne, R. M. (1962). The acquisition of knowledge. Psychological Review,
69(4), 355-365.
Gagne, R. M. (1972). Domains of learning. Interchange, 3, 1-8.
Griffiths, A. K. (1979). The mole concept: Investigation of an hierarchical model.
Unpublished doctoral dissertatjon, University of Alberta.
Griffiths, A. K. (1980). Reproducibility and learning hierarchies: A further clarification. Journal of Research in Science Teaching, 17, 83-85.
Griffiths, A. K., & Cornish, A. G. (1979). An analysis and comparison of three
recent methods of validating learning hierarchies. St. John’s, Newfoundland, Canada:
Memorial University of Newfoundland (ERIC document, ED 167625).
Griffiths, A. K., Kass, H., & Cornish, A. G. (1983). Validation of a learning
hierarchy for the mole concept. Journal of Research in Science Teaching, 20, 639-654.
Gunstone, R. F., & White, R. T. (1981). Understanding of gravity. Science
Education, 65(3), 291-299.
Hart, K. (1979). A hierarchy of understanding in mathematics. In W. F. Archenhold, R. H. Driver, A. Orton, & C. Wood-Robinson, (Eds.) Cognitive development
research in science and mathematics. Leeds: University of Leeds.
Jenkins, W. L. (1955). An improved method for tetrachoric. Psychometriku,
20(3), 253-258.
Johnstone, A. H., & Mahmoud, N. A. (1980). Isolating topics of high perceived
difficulty in school biology. Journal of Biological Education, 14(2), 163-166.
Johnstone, A. H., Macdonald, J. J., & Webb, G. (1977). Misconceptions in
school thermodynamics. Physics Education, 12(4), 248-25 1.
Linke, R. D. (1975). Replicative studies in hierarchical learning of graphical
interpretation skills. British Journal of Educational Psychology, 45, 39-46.
Novick, S . , & Nussbaum, J. (1978). Junior high school pupils understanding of
particulate nature of matter: An interview study. Science Education, 62(3), 273-28 1 .
Nussbaum, J. (1981). Towards the diagnosis by science teachers of pupils’ misconceptions: An exercise with student teachers. European Journal of Science Education, 3(2), 159-169.
Odum, E. P. (1977). The emergence of ecology as a new integrative discipline.
Science, 195(2484), 1289-1 293.
Owston, R. D. (1979). A maximum likelihood approach to the test of inclusion.
Psychometrika, 44(4), 421-425.
Powell, C. B., & Powell, R. A. (1982). The predator-prey concept in elementary
education. Wildlife Society Bulletin, 10(3), 238-244.
Simpson, M., & Arnold, B. (1982). The inappropriate use of subsumers in biology learning. European Journal of Science Education, 4(2), 173-182.
Stead, B. E., & Osborne, R. J. (1980). Exploring science students’ concepts of
light. Australian Science Teachers Journal, 26(3), 84-90.
Strike, K. A., & Posner, G. J. (1982). Conceptual change and science teaching.
European Journal of Science Education, 4(3), 23 1-240.
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
UNDERSTANDING OF FOOD WEBS
GRIFFITHS AND GRANT
Viennot, L. (1979). Spontaneous reasoning in elementary dynamics. European
Journal of Science Education, 1(2), 205-221.
Wheeler, A. E., & Kass, H. (1978). Student misconceptions in chemical equilibrium. Science Education, 62(2), 223-232.
White, R. T. (1974). Indexes used in testing the validity of learning hierarchies.
Journal of Research in Science Teaching. 11(1), 61-66.
White, R. T. (1974). The validation of a learning hierarchy. American Educational Research Journal, 11, 121-136.
White, R. T., & Clark, R. C. (1973). A test of inclusion which allows errors of
measurement. Psychomefrika, 38, 77-86.
White, R. T., & Gagne, R. M. (1974). Past and future research on learning
hierarchies. Educational Psychologist, 11( l), 19-28.
Manuscript accepted September 17, 1984
10982736, 1985, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/tea.3660220505 by Charles Darwin University Library, Wiley Online Library on [27/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
436