Professional Documents
Culture Documents
Introduction
The first time sixth grade students learn the concept of acid-base balance. Two-tier multiple-choice questions (knowwhat then know-how) are effective to diagnose misconceptions. Persons response patterns prevalently were used to analyze students learning misconceptions.
Purposes
To examine the effect of classifying response patterns on an acid-base test by a Guttmanbased person-fit index set
Questions
Q1 : How do the classified clusters based on the BW indices be verified? Q2 : What characteristics of the classified clusters will reveal? Q3 : What characteristics will change for individuals between the know-what test and know-how test?
BW family indices
Person-facet
W (Carelessness) B (Guessing) C (Capability) M (Misconception)
Wi =
Ti j =1
( 1 uij ) ( q* qij ) iT [( K 1 ) / 2 ]
Bi
j = Ti + 1
u ij ( q i j q * ) iT
[( K 1 ) / 2 ]
Ti
Ci
j =1
u ij ( q * q i j ) iT
[( K 1 ) / 2 ]
K
Mi
j =Ti +1
(1 - uij ) ( qij q* ) iT [( K 1 ) / 2 ]
Method
Subject 67 valid 6th grade students in Taiwan. 41 boys (61.19%) and 26 girls (38.81%).
Instrument
20 two-tier multiple-choice acid-base questions Item difficulty
Easy : .65 <P<1.0 (7 items) Moderate: .4 <P <.64 (7 items) Hard: .0 <P <.39 (6 items)
Item discrimination
High discrimination: D >.50 (9 items ) Middle discrimination: D >.30 (8 items ) Low discrimination: D <.20 (3 items )
Example 1
Know-what question (Answer) ( ) According to the weather report, rain water from most parts of the world is acidic rain. What do you think the nature of acidic rain is? (1) Acidic (2) Neutral (3) Basic (4) Uncertain Know-how question (Reason) ( ) What causes it to possess such property? (1) Because of openings in the ozone layer. (2) Because factories, vehicles and scooters release acidic substances into the air, therefore rain water is acidic. (3) Because factories release basic substances into the air, therefore rain water is basic. (4) Whether the nature of rain water is acidic, neutral or basic, it depends on the substances in the air at the moment.
Example 2
Know-what question (Answer) ( )There are many types of aqueous solutions in our daily lives and one of them has a slippery feeling when touched. What kind of property might it posses? (1) Acidic (2) Basic (3) Neutral Know-how question (Reason) ( ) What do you think the best reason for this is? (1) Soap water feels slippery (2) Tap water feels slippery (3) Peeling pears feel slippery
Scoring system
S1. Dichotomous scoring for 1st tier items as kk1 S2. Dichotomous scoring for 2nd tier items as kk2 S3. Coding cognitive styles truly known: kcogn = 1, if (kk1 = 1 & kk2 = 1) guessing: kcogn = 2, if (kk1 = 1 & kk2 = 0) careless: kcogn = 3, if (kk1 = 0 & kk2 = 1) unknown: kcogn = 4, if (kk1 = 0 & kk2 = 0) S4: Summing the number of each cognitive type across 20 items within subject, and create four variables as TK (truly known), GS (guessing), CR (careless), and (UK) unknown to serve as the predictors of discriminant analysis and multinominal logistic regression procedures on the criteria of group classifications from clustering.
Classification by BW indices
Analysis
BW indices estimation The BW indices (W,B,C,M) were calculated through the first and second tier items (i.e., right or wrong). Changes from the two tiers of test can be compared through standardized indices. Classification: Three cluster analysis methods based on would classify samples into 2-4 groups based on the BW indices by
Between method (Average linkage between groups) Ward method K-mean method
Verification: The four variables (TK, GS, CR, UK) would serve as the predictors on the criteria of group classifications from clustering by two identifying approaches
Discriminant Analysis (DA)~ linearly prediction Multinominal Logistic Regression (MLR) ~ nonlinearly prediction
RESULTS
Classification
Both identifying approaches confirmed the BW-based classifications were valid with very high hit rates of predictions, especially the nonlinear multinominal logistic regression. Since the small number of cases (n=2) clustered by the Between method in the first two types of groups (Group = 2 and 3) and that (n=4) by the K-mean and Ward methods in the final group (Group =4) for both DA and MLR procedures, we will discuss the properties of clusters by the K-mean and Ward methods in the first two types of groups.
Properties of 2 clusters
.PHDQ
:DUG
* UR X S
0 & %
* UR X S
0 & %
Properties of 3 clusters
.PHDQ
:DUG
0 * UR X S & %
0 * UR X S & %
:
'% $ (
H1 H1
H1 L1
)1 HP H 6
"! ) #$ @A
76 98
)2
HQ
S 4" 3 4 Y b cd d Y a Y b a cd ba c ef !5 " 30 ! R 97 8T 9S 7 RI
L1 H1
eg VU YXW ` eU eh
H1 A3
p XV Vi iWq Xp
fW
4. Examining the changes of standardized personfit indices between the know-what test and the know-how test can provide much diagnostic information.