You are on page 1of 6

The Learning Styles Inventory (LSI) and Instrument Bias

Steven W. Lamb, Indiana State University


Samuel C. Certo, Indiana State University

ABSTRACT Although some research investigating this model


dwells on the model itself (Certo, 1977), perhaps
most of it concerns a related instrument called
This study Investigated The Learning Styles Inven- the Leaming Styles Inventory (LSI). The LSI
tory (LSI) and instrument bias. A comparison of (Kolb, 1971) presented in Figure 1, is purported
results based upon the LSI and a modified Likert- to measure an individual's relative emphasis on
scale version of the LSI suggests that it is high- the above two learning dimensions and corresponding
ly likely that the LSI contains significant in- learning abilities in training situations which he
strument bias. Eliminating this bias and the encounters. The LSI has been used as the basis for
development of other related instruments are en- such activities as: 1) comparing perceived learning
couraged as worthwhile future work in this area. in pedagogically different training situations
(Wolfe and Byrne, 1976); 2) investigating the rela-
tionship between learning style, perceived learn-
INTRODUCTION ing, and performance (Fritzsche, 1976), and; 3)
recommending experimental designs for experiential
studies (Brenenstuhl and Catalanello, 1976).
A close examination of the use of experiential
learning techniques in management education re- Despite the repeated use of the LSI, little re-
veals that an increasing number of related empiri- search has been conducted which investigates the
cal investigations has been reported over the last instrument itself. One such investigation conduct-
four years. The predominance of more subjective ed by Freedman and Stumpf (1977) focused on the
evaluations of experiential pedagogy (Dimock, validity of the LSI. These writers concluded that
1961; Mood and Spect, 1954; Cohn and Rhenman although data gathered and analyzed using the LSI
1961) is slowly giving away to a stream of empiri- seemed to yield results which are somewhat consis-
cal experiential studies. Overall, these experi- tent with Kolb's bi-polar experiential learning
ential studies are distributed from such diverse construct, it is uncertain if this support is at-
areas as the comparison of experiential training tributable to LSI instrument bias.
methods with other training methods (Wolfe, 1975;
Chesser and Martin, 1976) and the identification
The present study is a logical extension of the
of the primary learning dimensions within experi-
Freedman and Stvnnpf study. This study will inves-
ential learning situations (Certo and Peter, 1976;
tigate the relationship between the LSI's ability
and Randolph, 1977).
to generage two bi-polar learning dimensions
through instrument bias. The issue of instrument
One experiential learning area which seems to have bias is important since such a bias can artifically
generated a significant amount of scientific in- support the validity of empirical test instruments
vestigation involves Kolb's (1974) four-stage ex- (Lamb, 1970). Given the increasing use of the LSI
periential learning model. According to this as a research tool, the timeliness and worth of
model: such an investigation seems obvious.
"Learning is conceived to be a four-
stage cycle. . . The learner, if he is
to be effective, needs four different METHOD
kinds of abilities - concrete experience
abilities (CE), reflection observation
abilities (RO), abstract conceptuali- The instrument bias referred to by Freedman and
zation abilities (AC), and active Stumpf is related to LSI completion and scoring
experimentation abilities (AE). procedures. When completing the LSI, subjects are
forced to rank nine sets of four words. LSI scor-
Indeed a closer examination of the ing procedures involve totaling six of these rank-
four-stage learning model reveals that ings over the nine sets to determine a total score
learning requires abilities that are for each of the four columns. Since each column
polar opposites . . . More specifi- represents one of Kolb's four main learning abili-
cally, there are two primary dimen- ties, and since a high rank on one column necessi-
sions to the learning process. The tates a low rank on another column, data analysis
first dimension represents the con- could yield results which are forced into support-
crete experiencing of events at one ing the bipolar learning dimensions and correspond-
end and abstract conceptualization ing abilities suggested by Kolb's experiential
at the other (AC-CE dimension). The learning construct.
other dimension has active experimen-
tation at one end and reflections ob-
To investigate this potential bias, results using
servations at the other (AE-RO dimen-
the LSI were compared to results using a modified
sion)." (Kolb, 1974, p. 28).
LSI. The modified LSI simply contained the same

28
FIGURE 1

LEARNING-STYLE INVENTORY*

This inventory is designed to assess your method of learning. As you take the
inventory, give a high rank to those words which best characterize the way you
learn and a low rank to the words which are least characteristic of your learn-
ing style.

You may find it hard to choose the words that best describe your learning style
because there are no right or wrong answers. Different characteristics described
in the inventory are equally good. The aim of the inventory is to describe how
you learn, not to evaluate your learning ability.

Instructions

There are nine sets of four words listed below. Rank each set of four words
assigning a 4 to the word which best characterizes your learning style, a 3
to the word which next best characterizes your learning style, and 2 to the
next most characteristic word, and a 1 to the word which is least character-
istic of you as a learner. Be sure to assign a different rank number to each
of the four words in each set. Do not make ties.

[#]** [GE#] [R0#] [AC//] [AE//]


1, ^discriminating _tentative _involved practical
2, preceptive _relevant _analytical impartial
3, feeling jwatching _thinking ^doing
4, accepting _risk-taker _evaluative aware
5, intuitive _productive _logical ques t ioning
6, ^abstract _observing _concrete ^active
7, present-oriented _reflecting _future-oriented pragmatic
8, experience _observation _conceptualization ^experimentation
reserved rational ^responsible
9, intense

FOR SCORING ONLY


RO AC AE
GE 234578 136789 234589 136789

*In order to score each of the four abilities, six of the nine items under each
classification are totaled. Then, to determine the extent to which an indivi-
dual emphasizes abstractness over concreteness (AC-CE), the total for concrete
experience is subtracted from the total for abstract conceptualization. Next,
to determine the extent to which an individual emphasizes active experimenta-
tion over reflection (AE-RO), the total for reflective observation is subtracted
from the total for active experimentation, (Kolb, 1974),

**The bracketed labels are included so that the reader may identify the items
used in each table. They were not included in the test instrument.

word items as on the original LSI, Each modified Data were analyzed by following a three-step pro-
LSI item, however, was placed on a seven-point cedure. First, overall correlation matrices were
Likert scale and rated individually as opposed to developed for the AC, CE, AE, and RO learning abi-
the four-column, dependent ranking of the original lity column totals using both the original and
LSI, In effect, the modified LSI eliminated the modified LSI, Second, correlation matrices were
completion and scoring dependency mentioned by also developed for individual items within the AC-
Freedman and Stumpf, CE learning dimensions using both the original and
modified LSI, Third, step two was repeated for
Subjects were 450 undergraduate School of Business the items within the AE-RO learning dimension (Nie,
students. Subjects were first asked to complete et, al,, 1975),
the modified LSI, Immediately after all subjects
had completed the modified LSI they were asked to
complete the original LSI, Both instrument forms RESULTS
contained similar directions. Of the original
sample, 383 subjects completed both instruments
in a usable fashion. Table 1 contains the overall correlation matrices
generated for the AC, CE, AE, and RO learning

29
TABLE 1
Comparative Correlation Matricies of the AC, CE, AE, and RO Column Totals for the Original LSI,
The Theoretical Experiential Learning Construct, and the Modified LSI.

ORIGINAL LSI CORRELATIONS REFLECTED MODIFIED LSI


(Forced-Rank) IN THEORY CONSTRUCT (Likert Scale)

AC CE AE RO AC CE AE RO AC CE AE RO
AC
1 - :hk* :88r :88i AC 1 0 0 AC 1
:88i :35i :88i
CE 1 - . CE 1 0 0 CE 1
i5i :?§4 :e8i :&8i
AE 1 - AE 1 - AE 1
:55i :88i
RO 1 RO 1 RO 1

*Values below correlations are significance levels.

ability column totals for both the original and The two dimensional, bi-polar experiential learn-
modified LSI. Note that this table contains re- ing theory states that items measuring the AC
sults using the original LSI, results which would construct should correlate negatively with items
reflect the two dimensional bi-polar learning con- measuring the CE construct. Using the original
struct developed by Kolb, and results using the forced rank instrument, 28 of the 36 pairs do cor-
modified LSI. Results based on the original LSI relate negatively; 17 of these are significant at
approximate the correlation structure in the the .05 level. Note, however, that the largest
theory matrix. The modified LSI, however, de- five negatively correlated values circled are form-
viates significantly from both the original LSI ed from pairs appearing in the same row of the
and the theory matrix by exhibiting positive sig- orced-rank LSI. Also, if all values of l-rl<.25
nificant, and meaningful correlations among all were discounted only those items circled would re-
summative scales. main. Results based upon the modified LSI, how-
ever, are again much different. Here there exist
Table 2 contains the comparative correlation mat- only positive correlation?, 32 of which are sig-
rices generated for individual items that are con- nificant at the .05 level indicating that these
sidered in determining the AC-CE learning dimen- items do not reflect a bi-polar AC-CE construct.
sion using both the original and modified LSI.

TABLE 2
Comparative Correlation Matricies for Individual AC, CE Items Using
Original and Modified LSI.

ORIGINAL LSI MODIFIED LSI


(Forced Rank) (Likert Scale)

CE2* CE3 CE4 CE8 CE2 CE3 CE4 CE5 CE7 CE8
AC 2 AC2 •. 001 •.hh tool :88i •.hh
AC 3 AC3
: 68i :85i lotji •.hh Ml
AC4 AC4
-:88i ^ :83i :66i :hhi :fi8i .19
AC5
-.hh -.^s AC5
:88i
AC8 AC8
;o44 :o39 :83i :8§
AC9
-:8§i ~:io4 ~:hh AC 9
:88i

*Items used to generate this table are contained in Figure 1.


**Values below correlations are significance levels.

30
Table 3 contains the comparative correlation mat- strument bias. Although shortcomings of the
ricles generated for Individual Items of the AE-RO Likert scales used in the modified LSI must also
learning dimension using both the original and be considered (Kerlinger, 1973), the obvious lack
modified LSI. The two dimension bi-polar learning of supportive correlations based upon the modified
theory states that items measuring the AE con- LSI renders the validity of the original LSI high-
struct should correlate negatively with the items ly suspect.
measuring the RO construct. Using the original
forced-rank LSI, 27 of the 36 correlations are Overall, the LSI is an instrument which attempts
negative, 21 of these are significant at the 05 to empirically delineate dimensions within the ex-
level. Note that of the largest eight negatively periential learning situation. The worth and
correlated values, six (circled beloy)are formed timeliness of such an instrument is evidenced by
from pairs appearing in the same row of the its repeated use in more current experiential
forced-rank LSI. Note also, if all values of learning literature. Based upon the results of
l-rl<.25 were discounted^ only 3 negative cor- this study, however, significant instrument bias
relations, besides those items circled would re- within the LSI seems likely. Modification of the
main. LSI to eliminate this potential bias and the devel-
opment of other similar instruments seem like
Using the modified LSI, however, there are only appropriate directions for future work in this
four negative correlations; one of which is sig- area.
nificant at the .05 level. Of the 32 positive
correlations, 26 are significant, at the .05 level.
In essence, results based upon the modified LSI
items do not represent a bi-polar AE-RO construct.

TABLE 3
Comparative Correlation Matricies for Individual AE, RO Items Using
Original and Modified LSI.

ORIGINAL LSI MODIFIED LSI


(Forced Rank) (Likert Scale)

ROl* R03 R06 R07 R08 R09 ROl R03 R06 R07 R08 R09
AEl AEl
'.hil '.hh
'.%e :824 -:9?7 :9^3 '.hh ihk :8!4 -:i!)94
AE3 -:88i (:33J) • •:?$5 -.hk
-:38i AE3
:884 :83i :83i :hh :85i -:2f3
AE6 -:8ii -:38i { -Ms -:88i -.hh AE6
-:Sl6 :h^i :88i Ml :i5i :Sl6
AE7
:hh -:9^8 ••:T8i ( IQOl '.%S AE7
:9^3 '.ih '.hh:8ii ,:8§9 :83i
AE8 -.%s AE8
:?9i ••:8ii ••:i94 riooy :i88 '.m :85i :3ii :18i :33i :i93.
AE9
-.Us -:i96 :i88 ••:8§3 AE9
: J52 :l^i :88i :hk :88i :hh

*Items used to generate this table are contained in Figure 1.


** Values below correlations are significance levels.

DISCUSSION AND IMPLICATIONS REFERENCES

The preceding results seem clear. When the ori- Brenenstuhl, Daniel C. and Catalenello, Ralph
ginal forced-rank LSI was used, the correlations F., An Analysis of the Impact Upon Learning
matrices derived from LSI Column totals as well Effectiveness of Traditional Instruction, Sim-
as from individual AC, CE items and the indivi- ulation Gaming, Computer Simulation and Learn-
dual AE, RO items support the AC-CE and AE-RO bi- ing Theory, Association for Business Simulation
polar learning dimensions. When the modified and Experiential Learning Proceedings (1976),
Likert-scale LSI is used, however, there does not pp. 463-473.
appear to be any substantial support for the
claim of the two bi-polar learning dimensions. Certo, Samuel C. and Peter, Paul J., An'Explo-
ratory Analysis of Perceived Leaming Dimen-
These results seem consistent with the notion that sions in an Experiential Learning Situation,
the appearance of two bi-polar learning dimensions Academy of Management Proceedings (1976),
based upon the original LSI is largely due to in- pp. 25-28.
31
Certo, Samuel C., Stages of the Kolb-Rubln- Business Policy Course, Association for Busi-
Mclntyre Experiential Learning Model and Per- ness Simulation and Experiential Learning
ceived Trainee Learning: A Preliminary In- Proceedings (1976), pp. 474-482.
vestigation, Academy of Management Proceedings
(1977), pp. 21-24. 17. Wolfe, Joseph, A Comparative Evaluation of
the Experiential Approach as a Business Policy
Chesser, R. J. and Richard C. Martin, An In- Learning Environment, Academy of Management
vestigation into the Relative Effects of the Journal (Vol. 18, No. 3 ) , 1975, pp. 242-252.
Experiential and Case Teaching Methods,
Academy of Management Proceedings (1976),
pp. 29-33.

5. Cohn, Kalman J. and Eric Rhenman, The Role of


Management Games in Education Research,
Management Science (January, 1961), pp. 131-"
161.

6. Dimock, Hedley G., Improving Communication


Skills Through Training, The Journal of
Communication (1961), pp. 149-156.

7. Freedman, Richard D. and Stephen A. Stumpf,


Validity Issues in Experiential Learning,
Academy of Management Meeting (1977).

8. Fritzsche, David J., On the Relationships 6f


Learning Style, Perceived Learning, and Per-
formance in an Experiential Learning Environ-
ment, Computer Simulation and Learning Theory.
Association for Business Simulation and Ex-
periential Learning Proceedings (1976).
pp. 455-462.

9. Kerlinger, Fred N., Foundation of Behavioral


Research. Holt, Rinhart and Winston, Chicago,
1973, 2nd Ed. p. 549.

10. Kolb. David A., Individual Learning Styles and


the Learning Process, Working Paper, #535-71,
M.I.T., Sloan School (1971).

11. Kolb, David A., On Management and the Learning


Process, in Organizational Psychology: A Book
of Readings. David A. Kolb, Irwin M. Rubin,
and James M. Mclntyre eds.. Prentice Hall,
New Jersey, 1974, pp. 27-42.

12. Lamb, Steven W., The Determination of the


Benefits of Economies of Scale as Applied to
Rural Road Systems in Kansas, unpublished Ph.D.
Dissertation, Kansas State University, (1970),
pp. 39-51.

13. Mood, A. M. and R. D. Specht, Gaming As A Tech-


nique of Analysis, the Rand Corporation,
Santa Monica, 1954.

14. Nie, Norman H., C. Hadlai Hull, Jean G.


Jenkins, Karin Steinbrenner, and Dale H. Bent,
Statistical Package for the Social Sciences,
McGraw Hill, New York, 1975.

15. Randolph, Alan W., Perceived Learning Dimen-


sions of an Experiential Training Unit in
Effective Group Functioning, Southern Manage-
ment Proceedings (1977), pp. 118-120.

16. Wolfe, Douglas E. and Eugene T. Byrne, A


Comparison of Perceived Learning in Three
Pedogogically Different Sections of a Required

32

You might also like