assessing the educational consequences of teacher deficits ... · basic math . distribution of...
TRANSCRIPT
Assessing the Educational Consequences of Teacher Deficits in Content Matter and
Language of Instruction (L2)
Steve Walter - GIAL
Some conventional wisdom: Schools are only as good as their teachers.
Research Context
Location (Cameroon, Philippines) Experimental program testing L1 instruction Early exit program 12 experimental schools; 12 comparison schools Testing began in 2008 Testing in Reading, Math, English All teachers speak L1
Key observation about Kom schools
Standard Schools Mean Mean Experimental Schools
GS Mentang 10.3 31.7 CBC Fujua CS Kindoh 14.1 31.8 GS Ngwah GS Yuwi 14.5 33.9 GS Laikom GS Meli 15 42.1 GS Kitchu CBC Fundeng 20.4 46.9 GS Bolem GS Baichu 23.2 47.9 GS Muteff GS Atondum 24.6 50.5 GS Ameng GS Fundong Village 24.8 50.5 GS Mboh CBC Abuh/Wainchia 26.6 54.6 CBC Kikfuini GS Njinikijem 28.3 64.7 GS Wombong PS Ngwah 36.3 65.6 GS Ilung CS Wombong 37.6 72.6 CBC Belo
Sources of Variation – some obvious and some less so
Some “obvious” hypotheses Motivation (avocation) Commitment to teaching Training and Education Ability and creativity Teacher-centered instruction Low compensation
Some “less obvious” hypotheses Low content mastery on the part of teachers Low mastery of the language of instruction by teachers Low time on task Crowded curriculum requirements Poorly constructed curricula Erratic attendance; automatic promotion Clumsy screening and placement practices
The Investigation Venue:
Government Teacher Training Center in Fundong
Participants: 185 students tested (139 female, 44 male) 150 in Form 3 (the last year of training) 29 in Form 2 6 in Form 1
Content of Tests: English (including reading comprehension) Basic math
Distribution of Participants by Age
3230282624222018
20
15
10
5
0
age
Pe
rce
nt
Histogram of age
The Test Instruments
The English/reading Assessment
Two texts for reading comprehension “Tenali and the Thieves” Grade 3 level 668 words 13.36 words per sentence 10 comprehension questions “Living an Adventurous Life” Grade 9 level 468 words 19.5 words per sentence 10 comprehension questions
English knowledge and vocabulary Word meanings and associations – 10 items ex. 6. crush a. sweet b. cheap c. change d. smash
Synonyms and antonyms – 10 items ex. 4. Which of the following words is closest in meaning to the word resolve? a. faint b. require c. weird d. determination
Words in context (naturalness) – 10 items ex. 5. The teacher praised the girl because her ideas were so __________. a. unique b. odd c. purple d. foolish
Nature of the Comprehension Questions
General comprehension (vocabulary, textual content, etc.) – 7 items 4 items from Tenali 3 items from Adventure Understanding the plot structure – 2 items 2 items from Tenali Simple factual comprehension – 4 items 1 items from Tenali 3 items from Adventure Drawing inferences and conclusions – 7 items 3 items from Tenali 4 items from Adventure
Nature of the Vocabulary Section
30 vocabulary items (taken from the Cameroon textbook series for teaching English) Level 4 – 10 items ex. oil, respect, good, open Level 6 – 9 items ex. ambition, estimate, acquire, pursuit Level 8 – 6 items ex. cunning, resolve, prior, endorse Levels 10 and 12 – 5 items ex. mutual, crush, unique, append Categories of vocabulary items General use vocabulary – 23 items ex. shelf, trick, chance, dead Academic vocabulary – 7 items ex. structure, resolve, mutual, append
The Math Test
Classification by response method Free response (supply a math fact or do an actual computation) 12 items Multiple choice 30 items Classification by manner of presentation Direct (simple computation or directed solution) 22 items
Story problems (an abstract scenario is created from which information must be gleaned in order to solve a problem)
20 items
The Math Test – Classification by Content
Basic operations – 4 items Math facts (basic and advanced) – 4 items Operations with Fractions – 3 items Basic algebra – 6 items Conversions and equivalences – 2 items Probability and statistics – 2 items Logical and applied analysis – 7 items Symbolic representation – 6 items Problem solving with percentages – 2 items Area and perimeter – 3 items (two of which were discarded) Graphical representation – 1 item Ratios – 2 items
A Look at the Results
Distribution of Overall Performance
70605040302010
25
20
15
10
5
0
OVERALL_PER
Pe
rce
nt
Histogram of OVERALL_PER
Distribution of Performance on the English Assessment
907560453015
20
15
10
5
0
eng_PER
Pe
rce
nt
Histogram of eng_PER
Distribution of Performance in Math
70605040302010
25
20
15
10
5
0
Math_PER
Pe
rce
nt
Histogram of Math_PER
Summary Statistics of Performance
Mean Stan. Deviation
Math 37.77 10.51
English/reading 65.18 13.79
Overall 52.41 9.42
The Results for English and Reading
Suggested Standards for Reading Comprehension
Barr, Blachowitz, Katz, and Kaufman (2002) Proficiency Level Level of Reading Comprehension
Independent 90 – 100 percent Instructional 75 – 79 percent Borderline 50 – 74 percent Frustration below 50 percent
Reading Comprehension – a closer look
Comprehension vs. Textual Difficulty
Vocabulary Knowledge vs. Textual Difficulty
Predictive probabilities .92 for Tenali .37 for Adventure
Vocabulary – a closer look
Relationship between Academic and General Vocabulary among GTTC trainees
100806040200
100
80
60
40
20
0
Vocab_no_ac_PER
Vo
ca
b_
ac_
PER
S 18.2455R-Sq 29.3%R-Sq(adj) 28.8%
Fitted Line PlotVocab_ac_PER = 9.333 + 0.5839 Vocab_no_ac_PER
Mean AWL = 48.0 Mean GWL = 66.2
Best Predictors of Reading Comprehension
For the Level 3 Text Level 4 vocabulary (F = 18.68; p = 0.000; r-sq. = 10.61%) For the Level 9 Text Academic vocabulary (F = 23.33; p = 0.000) Form (years in GTTC) (F = 8.05; p = 0.000; r-sq. = 17.35%)
Vocabulary and Text Comprehension
Comparison to Grade 5 Readers Level 3 Text
Distribution of Reading Comprehension among GTTC trainees relative to the Barr et al. Standards
Barr, Blachowitz, Katz, and Kaufman (2002) When reading a… Proficiency Level Percent Comprehension Level 3 Text Level 9 Text
Independent 90 – 100 percent 77.6% 2.4% Instructional 75 – 79 percent 9.1% 2.4% Borderline 50 – 74 percent 11.5% 61.2% Frustration below 50 percent 1.8% 33.9%
Implications and Questions raised by the Data
Levels of general proficiency in English are obviously low. The level of reading comprehension is quite limited. How effectively can such teachers teach via English? Is there a self-perpetuating negative spiral in place in terms of
learning and teaching English? What is the probability of educational success for non English-
speaking children when their teachers have low proficiency in the language of instruction?
How much improvement in teachers’ proficiency is needed to raise students’ performance?
What steps could be taken to improve teachers’ proficiency in English?
The Math Results
Distribution of Performance in Math
1009080706050403020100
25
20
15
10
5
0
Math_PER
Pe
rce
nt
Histogram of Math_PER
Mean = 37.77 SD = 10.51
Performance by Content Area
No. of items
Mean Score
Standard Dev.
Number who answered zero items correctly in
this category
Basic operations 4 49.4 22.2 3
Math facts 4 46.8 24.6 11
Fractions 3 33.8 26.9 44
Basic algebra 6 25.3 23.1 47
Percentages 2 23.4 28.6 90
Conversions 2 60.4 34.4 24
Probability and statistics 2 32.9 32.3 69
Logical analysis 7 42.0 16.5 3
Symbolic representation 6 30.9 20.2 17
Geometry (area and perimeter) 3* 24.7 43.3 119
Graphical representation 1 9.5 29.4 143
Ratios 2 62.7 35.2 24
GTTC trainees vs. Class 5 students
Class 5 Students GTTC
Skill or knowledge domain Mean SD Mean SD
Basic operations 10.9 17.7 49.4 22.2 The number system 19.9 14.7 52.1 21.6 Fractions 12.6 13.3 22.8 34.1 Algebra -- -- 24.3 20.7 Word Problems Basic operations 19.5 19.8 31.7 28.4 Analysis -- -- 44.8 14.9 Geometric figures 23.8 23.1 22.2 30.1 Graphs and tables 23.3 21.7 22.2 27.4 Percent, interest, commissions -- -- 30.1 34.3
Basic statistics -- -- 32.9 32.3 Math Overall 16.6 7.7 37.8 10.5
Some Inferences for Math Learning and Instruction
Mastery of basic operations is not strong. It is reported that many have taken no math courses
beyond primary (Grade 6) and the data appear to support this claim.
Complex numbers including decimals are troublesome. Mathematical abstractions (models, variables,
symbols) are especially difficult. If teachers lack knowledge and skill, we must assume
that students will as well.
Data from the Maguindanaon MLE Experience in the Philippines:
Comparing L1 vs. L2 speakers of the
Language of Instruction
At the Kindergarten Level
At the Grade 1 Level
At the Grade 2 Level
(Negative) Impact on Instructional Efficiency
A Simple Probabilistic Model of Learning
S = student T = teacher Prob. X Prob. X Prob. = Prob. T knows T can explain S will understand S learns
A Simple Probabilistic Model of Learning
Suggested probabilities Teachers’ performance on primary-level math items = 42.24% Teachers’ performance on English assessment (Grade 3 text plus
vocabulary) = 75.9% Class 5 students’ performance on English assessment (easiest story plus
vocabulary) = 38.6% * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Prob. X Prob. X Prob. = Prob. T knows T can explain S will understand S learns .4224 X .759 X .386 = .124 (actual mean math score = 17.0%)
Conclusions and Challenges
Teacher mastery of curricular content, while frequently considered suspect, has not been widely measured.
Limitations in linguistic competence are widely recognized in fact, but little understood in terms of instructional impact.
The weakness of educational systems in linguistically diverse developing countries involves teachers as well as students.
How much competence in a language is required to enable quality instruction?
What kinds of language learning/teaching interventions are needed to developed the needed competence?
Thank you.