You are on page 1of 13

Identifying an Indicator of Growth in Early Writing Proficiency

for Elementary School Students


Erica Lembke and Stanley L. Deno
University of Minnesota
Kim Hall
Robbinsdale Area Public Schools
New Hope, MN
In this study we compared alternative measures of early writing performance to identify an indicator of
writing proficiency that teachers could use to monitor student growth and evaluate instructional effec-
tiveness. The participants were 15 second-grade students in a summer school program for students at risk.
The performance sampled was word copying, sentence copying, word dictation, and sentence dictation.
Two- to three-minute samples of performance were obtained and several different scores were examined.
The criterion validity of the early writing measures was examined using both global judgments of compe-
tence and quantitative scoring of story writing as criteria. Writing from word and sentence dictation

entered into very high correlations with both criteria, showing promise as indicators of early writing pro-
ficiency. The results are discussed in terms of desirable features of assessment procedures including tech-
nical adequacy, administrative efficiency, and consumer acceptability.

Monitoring student progress in academic reliable measures for evaluating written


areas is criticalstudent success, especially
to expression interventions include the number
for students who are at risk for academic fail- of words written, number of words spelled cor-
ure (Deno, 1992). For these students at risk, rectly, or number of correct word sequences in
progress monitoring is particularly important a 3-minute writing sample (Deno, Marston, &

during the early elementary grades (K-2), as Mirkin, 1982; Deno, Mirkin, & Marston,
this is the time when students’ progress might 1980; Videen, Deno, & Marston, 1982).
best be accelerated through improved instruc- These indicators derived from students’
are
tional interventions. To monitor student writing in response to a story prompt. The

progress in academic subjects, teachers need empirical validity of these measures as overall
to use technically adequate measures and sub- indicators of written expression has been rel-
sequently respond to student needs based on atively well established for elementary stu-
the data from such measures ( Deno, 1985). dents in grades 2-6 (Marston, 1989) and, to a
One technically adequate method of lesser extent, for secondary students (Espin,
progress monitoring is curriculum-based mea- Scierka, Skare, & Halverson, 1999; Espin,
surement (CBM), which provides teachers Shin, Deno, Skare, Robinson, & Benner,
with reliable, time-efficient, easy-to-adminis- 2000). However, for students in the early
ter measures that are overall indicators of aca- stages of writing (e.g., K-2 ), or struggling
demic progress (Deno, 1985). CBM indicators learners, the CBM measure of written expres-
have been identified in reading, math, sion is too difficult to capture progress in
spelling, and written expression. Current learning to encode oral language into writing.
CBM indicators that are viewed as valid and In grades K-2, writing is typically not empha-

23

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


24

sized as much as reading or math ( Bradley- dents different levels, criterion validity,
at
Johnson Lesiak, 1989; Parker, Tindal,
& & sensitivity to growth, and face validity.
Hasbrouck, 1991; Roit & McKenzie, 1985). Identifying these indicators is viewed as
However, developing early writing skills is a important for teachers assessing students who
critical component of early literacy, and mas- are just learning how to write. The target stu-

tering these skills is necessary for later success dents include those in grades 1 and 2 and
in written expression. For students in the those who have identified leaming needs or
early elementary grades, teachers may focus are at risk for academic failure. Whereas the
on legible handwriting, letter and word for- existing CBM writing measure is relatively
mation, and basic sentence composition unstructured, the proposed measures for early
when assessing students’ writing. Criterion- writing assessment ask students to encode
referenced assessment of specific skills can controlled vocabulary. If measures can be
provide information about mastery. However, identified for early elementary students that
mastery of these skills does not provide infor- are predictive of later writing, such measures
mation about students’ general proficiency in could serve as indicators, providing data to
writing over time. The existing CBM measure the teacher in evaluating and planning
of written expression correlates well with instruction. Ideally, an early form of ongoing
other criterion variables in writing, but young monitoring will result in implementation of
or struggling students may have difficulty with
specific interventions for individual students
the advanced requirements of this task with- designed to increase writing performance and
out aid from the teacher. As a result, measures
eventually lead to better overall writing. The
of writing proficiency need to be identified alternative measures to be examined included
that allow teachers to assess student progress writing from word or sentence dictation and
of initial acquisition of proficiency at encod- word or sentence copying, each administered
ing oral language into writing. for 2-3 minutes.
Tindal and Parker (1991) identified four
criteria that can be used to guide the search Method
for useful assessment procedures: (a) consis- and
tent administration and reliable scoring, (b)
Partic:ip3nts Siting
ability to discriminate among students at dif This study took place in an elementary
ferent skill levels, (c) criterion validity (to school in a first-ring suburb of a large, mid-
other accepted assessment methods), and (d) western metropolitan area. Approximately
sensitivity to growth. These criteria consti- 50% of the students were from minority back-
tute a condensed version of the criteria used grounds, and 52% of students received free
in the initial development of CBM (Deno, and reduced cost lunch. Participants were 15
1985). Another important characteristic for students in a second-grade summer school
the task is &dquo;face validity.&dquo; In this context, face program for students at risk, (8 females and 7
validity refers to the degree to which the males). General education teachers had
assessment task is perceived as valuable by the referred these students for the summer school
teacher and whether the task is authentic, program because of their low academic
indicating that a student could complete it in achievement. Of the 15 subjects, 6 were
the classroom. receiving special education services: 1 student
The purpose of this study was to identify with leaming disabilities, 3 with mild/moder-
indicators of early writing performance with ate mental impairments, and 2 with speech

respect to the criteria identified by Tindal and and language difficulties. All students and
Parker: consistent administration, reliable their parents signed letters of consent for par-
scoring, effective discrimination among stu- ticipation.

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


25

Pirooecbare Pisdktor Variables


Students completed four writing tasks at the The writing tasks were scored using stan-
beginning and the end of the summer school dardized methods (Deno et al., 1980).
session: word copying, sentence copying, word Methods used to score the writing tasks were
dictation, and sentence dictation. Materials used as the predictor variables in the study,
used in these tasks included controlled vocabu- with scoring method varying by task. For
lary, with words taken from Basic Elementary example, the word copying and word dicta-
Reading Vocabulary (Harris & Jacobson, tion tasks were scored according to number of
1972). Each task was individually administered words written, number of words spelled cor-
for 2-3 minutes by the classroom teacher using rectly, and correct letter sequences. Sentence
standardized administration directions. For copying and sentence dictation were scored
each task, the student completed a sample item according to number of words written, num-
before beginning. Fidelity of implementation ber of words spelled correctly, correct word
checks were performed by a graduate student. A sequences, and correct minus incorrect word
description of each task follows: sequences.
1. Word copying. Students copied printed Words written included all words that the
words onto lines below each word. The student wrote, a word being defined as a series
task was composed of 30 words, and was of letters with spaces on both sides. A word
administered for 2 minutes. was counted even if it was spelled incorrectly.
2. Sentence copying. Students copied printed Words spelled correctly included all correctly
sentences onto lines below each sentence. spelled English words, with no regard to con-
The task was composed of 12 sentences, text. Correct letter sequences were defined as
each with five to seven words, and was two letters adjacent to each other that are
administered for 3 minutes. correctly placed in the word (Deno et al.,
3. Word dictation. Students wrote dictated 1980). Correct word sequences were defined as
words. Each word was dictated twice by the any two adjacent correctly spelled words that
teacher and then written by the student. were semantically and syntactically accept-

Students were instructed to do their best able within the context of the sentence
when attempting to spell words, and teach- according to a native speaker of the English
ers moved on to the next word after stu- language (Videen et al., 1982). Word
dents had finished the current word, or sequences were counted at the beginning and
when the students had paused for 5 sec- the end of sentences, giving students credit or
onds. The task was composed of 30 words, penalizing them for capitals on the first letters
and was administered for 3 minutes. of sentences, and punctuation at the end of
4. Sentence dictation. Students wrote dictated sentences. In all scoring methods, students
sentences. Each sentence was dictated were penalized for letter reversals within

twice by the teacher and then written by words.


the student. Students were instructed to do The typical interrater reliabilities reported
their best when attempting to spell words, when scoring words written, words spelled
and teachers moved on to the next sen- correctly, correct letter sequences, correct
tence after students had finished the cur- word sequences, and correct minus incorrect
rent sentence, or when the students had word sequences range from .87 for correct
paused for 10 seconds. The task was com- word sequences to .99 for words written
posed of 12 sentences, each five to seven (Marston & Deno, 1981; Parker et al., 1991;
words in length, and was administered for 3 Tindal & Parker, 1989; Watkinson & Lee,
minutes. 1992).

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


26

Ctiterim Variables ity, based on all of her experience with the


students’ writing. Independent ratings from ini-
Scores obtained on each writing task using
tial and final writing samples consisted of two
the various scoring methods were correlated
with seven criterion variables, which includ- independent raters ranking three of the stu-
dents’ journal writings from the beginning of
ed two qualitatively different sets. Criterion
the summer and three from the end of the
variables were grouped into atomistic and the
summer on a bipolar rating scale from 1 to 7,
holistic sets (Isaacson, 1988), with the atom-
with 1 being the worst and 7 being the best.
istic variables utilizing discrete, countable fea-
tures of student writing, and the holistic vari-
This scale did not include rubrics. Raters were
instructed to examine all sets of initial and
ables utilizing teacher ranking or writing sam-
final journal writings, and identify samples
ple rating. The four atomistic variables that that were the best out of all that they had
enumerated student performance included
the average number of words written, the seen, samples that were average, and samples
that were poorest. They were instructed to
average number of words spelled correctly, the
look at punctuation, grammar, spelling, hand-
average number of correct word sequences,
and the average number of correct minus writing, spacing, and indentation of the first
sentence and to try to rate them holistically,
incorrect word sequences on writing samples
that the students completed at the end of that is, they were not to pick out all the punc-
tuation and grammar mistakes in each writing
summer school. The three holistic variables,

which provided a more global judgment of sample, but to look at each child’s overall
writing performance, included the classroom writing ability based on these samples. After
teacher’s ranking of students’ general writing examining the students’ samples, they rated
each sample on a scale of one to seven. The
ability, the average of two holistic ratings of
students’ initial writing samples, and the aver- interscorer agreement between the two inde-

age of two holistic ratings of students’ final pendent raters was 84%.
writing samples. Results
The atomistic variables included scoring
methods that were the same as those for the To identify writing measures that would be
writing tasks, with scoring completed on stu- the best indicators of early writing proficien-
dents’ writing samples taken at the end of the cy, data analyses focused on the relation
summer program. The holistic variables between students’ performance on the poten-
included rankings or ratings of students’ writ- tial writing measures and performance on the
ing. Writing samples completed at the end of the criterion scores representing general writing
summer program involved students writing for proficiency. In conducting our analyses, we

5 minutes on a picture story starter where key were interested in measures that embodied
words were listed. To obtain scores on this several different characteristics. Ability to
task, students completed three writing sam- discriminate among students at different skill
ples, which were scored according to number levels and sensitivity to growth (Tindal &
of words written, number of words spelled cor- Parker, 1991) were two of the criteria that we
rectly, correct word sequences, and correct hoped our writing measures would possess.
minus incorrect word sequences. This criteri- The potential of a measure to discriminate
on variable is similar to the current written among students at different skill levels
expression measure in CBM. To obtain the increases if it produces a wide dispersion of
teacher’s ranking, the summer school teacher student scores (a large standard deviation).
was asked to rank her students from most Sensitivity to growth would be established if
skilled to least skilled on overall writing abil- there was not a floor or ceiling effect for the

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


27

measure and if students had room to grow on atomistic and holistic measures. Atomistic
the measure, without reaching the maximum criterion variables included the average num-
score. ber of words written, words spelled correctly,
correct word sequences, and correct minus
Performance en Vkitin _g 7&#a incorrect word sequences. These scores were
In our first analysis we examined the taken from students’ writing samples given at
descriptive data for each writing measure to the end of the summer program. Holistic cri-
determine whether sensitivity to growth and terion variables included the classroom
discrimination criteria were met. The means teacher’s holistic ranking of students’ writing
and standard deviations for each of the writ- ability, the average of two holistic ratings
ing measures are reported in Table 1. from students’ initial writing samples, and the
Measures given at the beginning of summer average of two holistic ratings from students’
school are referred to as Initial, whereas mea- final writing samples. The correlations
sures given at the end of summer school are between the atomistic and holistic criterion
referred to as Final. As seen in Table 1, none scores are presented in Table 2. Inspection of

of the mean scores for the measures approach- this table reveals that the coefficients ranged
es the total possible, so students had room to from .53 to .98, indicating that the two sets of
improve on all measures. In other words, they criterion measures thought to be broadly asso-
were not approaching a ceiling. In addition, ciated with overall writing performance were,
variability among participants on each mea- in fact, related.
sure was considerable, with standard devia-

tions ranging from 6.92 to 23.27. Thus, there Qimfm Validity


was a large dispersion of student scores for The key characteristic to be demonstrated
each measure. Measures with the largest stan- by the early writing indicators is criterion
dard deviations included Word Copying- validity with other accepted assessment meth-
Correct Letter Sequences, Word Dictation- ods (Tindal & Parker, 1991). The criterion
Correct Letter Sequences, Sentence validity of our measures was evaluated
Dictation-Correct minus Incorrect Word through a series of correlational analyses.
Sequences, and Sentence Copying-Correct Results of the analysis between predictor vari-
Word Sequences. Thus, all measures were ables and the atomistic criterion variables for
able to discriminate among students at differ- the initial and final data collection are pre-
ent skill levels and were sensitive to growth, sented in Table 3. Predictor variables are list-
with several measures showing particular ed horizontally, whereas atomistic criterion
strength on these criteria. It should be noted variables are listed vertically. Due to the large
that one student was able to complete the number of correlation coefficients, we set a
word copying task at both initial and final test stringent alpha level of p < .001. When exam-
sessions, although the total possible for this ining the columns for each measure, it is evi-
task was still not met. dent that most of the strong, statistically sig-
nificant correlations with the atomistic crite-
Cbrrelaticns Between the Criterion
ria are obtained for the dictation measures.
variables
Dictation measures included both word dicta-
The next step in our analyses was to deter- tion (14 out of 24 correlations were signifi-
mine whether the criterion variables were surf cant, p < .001, column two) and sentence dic-
ficiently related to be appropriate for use in tation (17 out of 32 correlations were signifi-
subsequent analyses. For that purpose, the cri- cant, p < .001, column four). Sentence
terion variables were grouped into sets of Copying, in the third column, also showed

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


28

- ç: C’CI o0 z

<c s

o 6ê s
N
~
&Scaron; x
G
-£ &copy; .- <
i !
_#f 0
.y. 2 x~ s
J
°~ 3
i
?
C’CI _£
.:;::

.~M
>1
M
~ .s

3
S %m
o
Uh
--
S? ~0
w
rA
-’sf - fl
i
n. .~
E
ed
-
en
.~CU

ç:
w
u
c
m N
c~4 I g’§fi< ’&dquo;

E O U
L ~~]0
a
n
.~ ~
.9 ~ C’CI
P

a~ S ~
’O t0 G
z

fi
E t ~ fi
0 ~=! ~ ’~ E
L
(A
~-
±
&dquo; __-~ oj3
p S* :E
0
u .~ ,_ £
u .,¡:::
L

10- o .-

H
e .a 0v. ~ ~^v. c:
0
~5
>< ~ . S ~ : I
~#~ ä08~
Gw.
°§
a .. S
I’f&dquo;B
&6scaron; ~~~ E
16.
? ..9~~
% b
-T3
c
]..gä3
m ~ o
(on -S~E
s 8 ~ ¡;
§§ °~%f,&cop£y;n~8(+
TS
m
(A
c
m
_~~) K m # #~~ &copy;~< o < .5- o £ õ
0)
~ ~ ~ g ~
!-~!!
1 0P3
j
’p < _ Kti
-
VJ
.il -o
m
9 C Z &scaron;~..g

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


29

(A
QJ
:3
e_~0
.¡:

C
o
L.
QJ
°11
U
U
°$g
.4 1
E
0
Q
&dquo;&dquo;C
C
~
. -L,
’$g
.~
O
:::J:
c
cu
QJ
A
QJ
m
CII
C
o
°fl
. tv
GJ
-

0
U 0
9
N vi
QJ 1-1.
» *

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


30

significant correlations with some of the Compared holistic criteria, once again,
to
atomistic criterion variables for the initial and word and sentence dictation had the strongest
final data collections (7 out of 32 correlations criterion validity.
were significant, p < .001 ).

Examining the columns again, but now Discussion


noting differences in individual scoring meth- The purpose of this study was to examine
ods (i.e., WW, WSC, CLS, CWS, C-IVVS), alternative approaches to sampling early writ-
when the word dictation, sentence copying,
ing performance to identify measures that
and sentence dictation tasks were scored
might be used to monitor individual student
using words written, words spelled correctly, growth and to formatively evaluate the effec-
correct letter sequences or correct word
tiveness of writing instruction. Scores from
sequences, correlations were higher than four writing tasks were examined to deter-
when the writing measures were scored using mine their criterion validity and evaluated
correct minus incorrect word sequences, with
with respect to criteria that are important
many correlations ranging from .80 to .92. In when selecting assessment procedures: consis-
the case of the word copying task, however,
tent administration, reliable scoring, effec-
correlations were in the low to low-moderate
range for all scoring methods. Additionally,
tively discriminating among students at differ-
ent levels, sensitivity to growth, and face
the majority of correlations increased from
the initial to the final data collection (as can validity (Tindal & Parker, 1991 ). The alter-
native performance scores examined were
be seen when examining the horizontal rows
taken from writing from word or sentence dic,
in Table 3). Thus, it appears that word dicta-
tation and word or sentence copying.
tion, sentence dictation, and sentence copy-
An examination of the descriptive data for
ing were the most promising measures com- all of the performance scores reveals that all
pared to the atomistic criteria, particularly tasks result in a wide dispersion of student
when scored using words written, words
scores. The largest standard deviations occur
spelled correctly, correct letter sequences, or
correct word sequences.
for sentence copying. The large standard
Correlations between predictors and holis- deviations were apparent when the sentence
tic criterion variables are presented in Table copying task was scored using any of the scor-
4. Few significant correlations were obtained ing methods (i.e., words written or words
between the writing measures and the holistic spelled correctly). In addition, standard devi-
criteria. However, as can be seen in columns ations for all measures apparently are larger
two and four, correlations between the writ- when the measures are scored using correct
ing measures and the holistic criteria were letter sequences (for word dictation and word
strongest for word dictation (8 out of 18 cor- copying), correct word sequences (for sen-
relations ranging from .62 to .83, with one sig- tence copying), or correct minus incorrect
nificant correlation, p < .001 ) and sentence word sequences (for sentence dictation).
dictation ( 17 out of 24 correlations ranging When considering the range of possible
from .61 to .84, with three significant correla- scores, it appears that students do not score
tions, p < .001 ). No particular scoring method zero or reach a ceiling on any of the tasks. In

appeared to produce higher correlations for particular, student performance for sentence
the holistic variable set. The rows in Table 4 copying and sentence dictation measures is
also show correlations increasing from the ini- low compared to the total possible.
tial to final data collection for most measures, Consideration of both the dispersion and
with the exception of sentence copying, for range of scores is important because these
which many correlations decreased. technical characteristics limit the ability of a

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


31

1% £ enen u
&dquo; g
’ C63
~
N

! s
en ;:I ~ V en
VY
£ O 3 &copy;
O
UD %
e <u
’E 6
.8~OJ) £
v
v
...c:: v
9
» &dquo;t7

N
9 .9

1
c
G
~
~ V s
::. v ~ .- >- gf 0- 5 ;:I
E
tv
<u~ 0 0. ~ UJ 7
E
cu
fl n n~ ’:&dquo;;0>~.
v
’&dquo;0

w
. L-1
-_J~i ~
<S
-~
S’-’ ~
ot
~5
M)
~£ Y ~

Q
0 iN S~ ~Y ~.w-6 S!
.- ¡::
S
v ¡:
fl

T3
c4 fl X S
c
m
V)
JJ
’&dquo;0 ~ ~ O c1 U
S
Ot
~
v~
F
cu t
.9 ~
m

Q- ~ ~

’&scaron; c: s rfJ
’&dquo;0 ~ ¡:: OJ)
U s
</) tv w ’§~
w
u
c
s
0 s :
3 ’&dquo;0 t) (f)~ ~vs
jQ
E
E ~ i ~
~) o
~
M)
_

C;;
- y’i
~.0- G
°>. °2 %
v
i Gw
s ~.~
~
E ’&dqOuo;0 . . . <.) . .c: ’-’.£¡: u N
ll
0 3 i M
i
i G h
M) ~-! U

p
0
£ £ / ’ q~
y OJ) 1- x
i*
i8~
~
ul ê [0
c ’&scaron; t ll~
n
8
%
°
w
40
3 ?
’n ~ ~
fl j 0-
UJ
8 /J fl , %
’S q~ §£ £ 9 r11
en
m
c
O
~~~ ~
~~ .

c
0 ~~vll,..~n
~ ~~ ~ &5
m a~ Y
~
i
0
3e~~&dquo;
Y~ ~G5 ’~~~ §
n.au .~2 U’EvI 5

<J ~ ~ .s ~ ~
M i ri, 3
K < # ) , fl §£ w ~ £ m G * I ~ m # I l 8 £ Vl
.R
.

ja <ü $: vi
¡¡¡ &dquo;

9 o~.’ain,
z s..s.&scaron;~:*

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


32

*g 55o
%
g
~- s s 6
~ N
S
~
4.
~ °i
¡::

3S j~ ?
.G
~
N
O
# j
-5 6
M) 6 U
...c;¡ ’&dquo;0
GJ
9
:&scaron;
.-

i
~ w

G ~ 3 &c U

_~#sjfl n f.§oOy£i~~w ~c55fgl§8.


SS
.f
~

’&dquo;0 c; U ’S

% £ &copy; 3 ) _ W
fi = $ S
tv
S! .5i~ ~ -

Q,/
h..
1 ~j .S ~
~! 1S 6g
u
u
M)

0
°4 g
% <g ~
I ~a~r -~s c1) l
c
(V i & u

~
ri)
M > e
]/.-P@fW#_
Q.
E ~ # c
m
CA
£
.~ ’b cJ
n
Q,/
w
c
i ~ &dquo;&dquo;8
M _J )
tv
E ’I # Dc b£ st
0~
’&dquo;0 . &dquo; 8(f) I
I’ll o3 s
E , Z£ Qu)
9~e- ~£~ fi
0
It: s! ~0 30 u% W
s ê ~~ ~
0 ’&scaron; ~.r; 8
u
U) ~!&dquo;
s§~ K
~
c
Q,/
&dquo;’&dquo; rJ). &
’b£ w°i~ y ~11I I ·t (Cn
i .~ ’õ ~ c;
’s
OQ ’s’s’<~ ~
’6 5 &dquo;ti’ ~ ~
c ~~~~&S
0
~0
4m, £ _ , ,
-tv
S
3 ‘~ GII .S ~
0
~~¡õê
n ,h , <
a
’r

U
.4
3 N-ow 3 -oo ’~,~g
a
J3 :<~~~~8
~o ~~.~~’Qj ~~ II.a’Vlvi
,

9 2 ~t

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


33

measure to discriminate change in individual beginning of the summer program. The


student scores across time. It appears then, stronger relationships between the scores
that the measures of sentence copying and obtained from the performance samples and
sentence dictation showed the greatest atomistic criteria are not surprising, and are

promise as indicators of early writing profi- most likely associated with the technical
ciency when considering their potential for characteristics of the criterion scores. The
discriminating performance within and CBM procedures used to score student’s writ-
between individuals. ten expression produce a wide dispersion of
The criterion validity of a measure is one of scores. The holistic ratings, in contrast, have

its most important characteristics. To forma- a restricted range. Differences in the distribu-

tively evaluate instruction and to make pre- tion of scores necessarily result in differences
dictions about the acceptability of student in the strength of correlations based on those
growth, it is essential to establish an empirical distributions. A related point is that the valid-
basis for using data to inform those decisions. ity of teacher judgment of student writing
In the present study we examined both atom- proficiency, while widely accepted, is not
istic and holistic criteria in an effort to broad- always certain.
en the base for identifying the measures that With few exceptions the correlations
might be most informative. The results of the increased from the initial to the final data col-
analyses related to criterion validity provide a lection. From this we conclude that the writ-
basis for concluding that word and sentence ing measures provide a stable measure of per-
dictation can function as measures of early formance and that the criterion validity is
writing proficiency. That is, scores from both likely to be maintained in future explorations.
approaches to sampling performance enter One concern is that student scores on the per-
into high correlations with the criterion vari- formance samples did not increase from the
ables. Performance on the word and sentence beginning to the end of the summer program.
dictation tasks shows particularly close associ- We are not certain whether this means that
ation with the number of words spelled cor- the students did not improve in their writing
rectly, and the number of correct letter or skills during that relatively short period of
word sequences. Results indicate that sen- instruction, or whether the prospective indi-
tence copying might also be used to measure cators are not sufficiently sensitive to growth
writing performance, since it entered into in the short term. Since the summer program
many significant relationships with atomistic in which the students were enrolled was only
criteria; however, the strength of the sentence 5 weeks believe that there was not
long, we

copying relationship is not as strong as those enough time for these students to significant-
for the dictation measures. Additionally, the ly grow. Nevertheless, future studies are nec-
range of reliable relationships between sen- essary to examine whether student growth on
tence copying and the holistic criteria seems the early writing measures will be revealed
narrower, which makes it less certain as an when measures are utilized on a frequent basis
indicator of broad writing proficiency. (e.g., once per week).
Generally, the relationship between the Face validity is often a concern when rec-
potential indicators of writing proficiency and ommending procedures for teachers to use in
atomistic criteria was higher than that formatively evaluating their instruction. In
between the writing indicators and the holis- selecting measures we were mindful of this
tic criteria. One notable exception is that the issue and tried to examine measures that
word dictation scores seem to be strongly involve straightforward scoring of actual sam-
related to independent ratings, where moder- ples of student writing. We thought that the
ate to strong correlations were found at the teachers would view the early writing tasks as

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


34

appropriate to use with their students and References


that the tasks might provide clear data on Bradley-Johnson, S., & Lesiak, J. L. ( 1989). Problems in
student performance in writing. The measures written expression: Assessment and remediation. New
are easy to create, as teachers can construct York: Guilford.
probes with words and sentences taken from Deno, S. L. (1985). Curriculum-based measurement:
The emerging alternative. Exceptional Children,
high-frequency word lists. In addition, the 52(3), 219-232.
graphed results of these short measures of stu- Deno, S. L. (1992). The nature and development of cur-
dent performance in writing can be valuable riculum-based measurement. Preventing School
to teachers as they examine instructional Failure, 36(2), 5-10.
effectiveness. Another positive feature of the Deno, S. L., Marston, D., & Mirkin, P L. (1982). Valid
measurement procedures for continuous evaluation
measures is that they can be completed in 3
of written expression. Exceptional Children, 48, 368-
minutes, thus taking up very little instruc- 371.
tional time. Further research must be con- Deno, S. L., Mirkin, P. K., & Marston, D. (1980).
ducted to determine whether differences in Relationships among simple measures of written expres-
sion and performance on standardized achievement tests
sample duration affect the ability of the mea- (Research Report No. 22). Minneapolis: University
sures to predict writing proficiency. Perhaps of Minnesota Institute for Research on Learning
even shorter samples (1 or 2 minutes) of per- Disabilities.
formance will demonstrate potential as indi- Espin, C. A., Scierka, B. J., Skare, S., & Halverson, N.
cators of writing proficiency. (1999). Curriculum-based measures in writing for
Based on the results of this study, it appears secondary students. Reading and Writing Quarterly,
15, 5-27.
that indicators of early writing proficiency Espin, C. A., Shin, J., Deno,S. L., Skare, S., Robinson,
can be identified that will allow us to monitor S., & Benner, B. (2000). Identifying indicators of
growth in early writing proficiency and evalu- written expression proficiency for middle school stu-
ate instruction. This is important because the
dents. The Journal of Special Education, 34, 140-153.
Harris, A. P., & Jacobson, M.D. (1972). Basic elementary
current CBMs of written expression function
reading vocabularies. New York: Macmillan.
well after students become fluent in encoding Isaacson, S. (1988). Assessing the writing product:
oral language into written, but they are less Qualitative and quantitative measures. Exceptional
serviceable for beginning writing instruction. Children, 54, 528-534.
The results provide a basis for optimism that Marston, D. ( 1989). A curriculum-based measurement
these early writing measures will serve teach- approach to assessing academic performance: What
it is and why to do it. In M. Shinn (Ed.), Curriculum-
ers and researchers as well as CBM has done based measurement: Assessing special children (pp. 18-
in early reading instruction. We hope that the 78). New York: Guilford.
findings presented here will encourage others Marston, D., & Deno, S. L. (1981). The reliability of sim-
to conduct research examining the effective- ple, direct measures of written expression (Research
Report No. 50). Minneapolis: University of
ness of these early writing measures with larg-
Minnesota, Institute for Research on Learning
er groups of students in grades 1 and 2, and Disabilities.
over longer periods of time. Eventually, per- Parker, R., Tindal, G., & Hasbrouck, J. (1991).
Countable indices of writing quality: Their suitabili-
haps, normative development and benchmark
ty for screening-eligibility decisions. Exceptionality,
standards will be established that enable edu-
2, 1-17.
cators to not only describe, but also improve
Roit, M., & McKenzie, R. ( 1985 ). Disorders of written
the writing instruction of beginning writers. communication: An instructional priority for LD

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015


35

students. Journal of Learning Disabilities, 19, 258-260. Watkinson, J. T, & Lee, S. W. (1992). Curriculum-
Tindal, G., & Parker, R. (1989). Assessment of written based measures of written expression for learning-
expression for students in compensatory and special
disabled and nondisabled students. Psychology in the
education programs. The Journal of Special Education,
23 , 169-183. Schools, 29, 184-191.
Tindal, G., & Parker, R. (1991). Identifying measures
for evaluating written expression. Learning Disabilities
Research and Practice, 6, 211-218. Author Note
Videen, J., Deno, S. L., & Marston, D. (1982). Correct
word sequences: A valid indicator of proficiency in writ- We thank Andrea Thompson for her assis-
ten expression (Research Report No. 84). tance with rating the students’ writing sam-
Minneapolis: University of Minnesota, Institute for
Research on Learning Disabilities. ples.

Downloaded from aei.sagepub.com at DALHOUSIE UNIV on June 21, 2015

You might also like