ERIC EJ1153576: Double Time? Examining Extended Testing Time Accommodations (ETTA) in Postsecondary Settings

2017·0.4 MB·English

by  ERIC

Journal of Postsecondary Education and Disability, 30(2), 185-200 185
Double Time? Examining Extended Testing
Time Accommodations (ETTA) in Postsecondary Settings
Laura Sokal1
Laurie Anne Vermette2
Abstract
Over eight thousand test administrations across two universities were examined to determine whether students
with disabilities were being given the necessary extended testing time accommodations and whether their
use of extended time decreased over the course of their programs. Findings revealed that commonly accepted
recommendations about appropriate durations of accommodations were not suitable in meeting individual stu-
dents’ needs and that students used more time on these types of accommodations as they moved through their
first three years of their postsecondary programs. Recommendations are provided.
Keywords: Disability, accommodation, testing
The increasing number of postsecondary students and four-year colleges in the United States in 2008-
with disabilities who request extended testing time ac- 2009 reported having learning disabilities. Further-
commodations (ETTA) places a spotlight on issues of more, the number of students registering with their
fairness and validity in testing. As postsecondary insti- university’s accessibility services in order to access
tutions re-allocate resources in order to meet their legal accommodations is increasing (Cairns, Massfeller, &
duty to accommodate the needs of students with disabil- Deeth, 2010). Although many of these students qual-
ities (Kettmann et al., 2007; Wolgast, Rader, Roche, & ify for and use multiple accommodations (Brincker-
Thompson, 2005), questions arise about the fairness of hoff & Banerjee, 2007), ETTA is one of most common
ETTA to students with disabilities, other students, and accommodations (Sokal & Desjardins, 2016; Kim &
professors, and additionally provoke questions about Lee, 2015; Lindstrom, 2010; Lovett, 2010; Stretch &
the suitability of timed testing per se. Although there Osborne, 2005), if not the most common (Sireci, et
is a high level of controversy about providing students al., 2005), and is usually accompanied by an accom-
with ETTA (Lindstrom, 2010; Sireci, Scarpati, & Li, modation where the students write the tests in quiet,
2005), little is known about how much time students separate settings (Sokal, 2016). Test accommodations
with disabilities actually use when provided with the are defined as altering the processes of test administra-
recommended increases of 50-100% time allotments in tion in such a way that the test can accurately measure
testing situations. The purpose of the current study was how well the student has learned the materials taught
to examine the duration of time used by students pro- (validity) without altering or “watering down” the
vided with ETTA in course-based tests administered in construct being tested (Sireci, Li, & Scarpati, 2006).
postsecondary settings. Their goal is to “level the playing field” so that the
learning (knowledge, skills and abilities) of students
Increasing Requests for ETTA with disabilities can be accurately measured (Sireci,
Rothstein (2006) demonstrated that the generally et al., 2005, p. 457).
accepted prevalence of learning disabilities in college While there has been recent research conducted
populations is one in every eleven students in Austra- on university students regarding the use of test ac-
lian schools, and Raue and Lewis (2011) showed that commodations, most research on this topic pertains to
almost one third of students who attended two-year children (Runyan, 1991). Thompson, Blout, and Thur-
1 University of Winnipeg; 2 University of Manitoba
186 Sokal & Vermette; Double Time?
low (2002) published a review of 46 empirical studies proposed processes to determine appropriateness of
related to test accommodations, and only three of the accommodations (Brinckerhoff, et al., 1992), as the
studies were conducted with postsecondary samples. same accommodation may affect students within the
Runyan (1991) presented evidence that, insofar as it same categories of disability differently (Lindstrom,
pertains to meeting the testing needs of students with 2010; Medina, 2000), and while “certain testing ac-
learning disabilities, separate research about children commodations may benefit some students with learn-
and adults can be generalized between these groups. ing disabilities, no single accommodation has been
shown to benefit all students with learning disabili-
How is ETTA Used? ties” (Brinckerhoff & Banerjee, 2007, p. 247). This
When ETTA is used, it is common practice to al- statement takes on even greater meaning when the di-
low students either 50% or 100% more time than the versity of disabilities addressed through universities’
standard time allocated to students without disabilities student accessibility services is considered.
who are writing the same test (Lewandowski, Cohen, So, why then is the use of ETTA so prevalent? Re-
& Lovett, 2013; Lovett, 2011). Although this practice search has shown that many students with disabilities,
is the most common way to enact ETTA, other proce- both in grade school and university, view ETTA as
dures have also been used alone or in addition to this an effective way to meet their learning needs (Sokal,
time allocation enhancement, including rest breaks 2016; Sokal & Desjardins, 2016; Elliott & Marquart,
between sessions and having students write portions 2004). University accessibility services are finding
of the test on consecutive days. These practices have that their students’ needs exceed their offices’ resourc-
been used effectively with younger students (Elliott & es and that providing ETTA satisfies both their stu-
Marquart, 2004). dents’ wishes and their own duty to accommodate in
The intuitive appeal of ETTA is not easily denied a time-effective way (Sokal, 2016; Lovett, 2011). The
when it comes to students with disabilities. Given that most cost-effective accommodation options that meet
specific disabilities, such as learning disabilities or students’ immediate needs may be selected ahead of
anxiety disorders—now the most common category one-on-one counseling on test-writing skills, and oth-
of disabilities being served by university accessibil- er accommodations that are more expensive or inten-
ity services (AUCCCD, 2014)—are often character- sive for accessibility service providers (Brinckerhoff,
ized by slower processing speeds, it is common sense et al., 1992). As such, ETTA has now become the de-
that a more accurate, valid picture of student learning fault accommodation in many cases.
would result from allowing these students additional
test-writing time (Lovett, 2011, Stretch & Osborne, Theoretical Basis for ETTA
2005; Weiler, et al., 2000). That is, ensuring that stu- The intuitive appeal of using ETTA is bolstered by
dents with slower response times have the opportuni- strong theoretical support. For the purposes of clarity,
ty to access and process all of the test questions would we will make reference to the Interaction Hypothesis
seem to allow for better measurement of their actual (see Sireci, et al., 2005), also called the Accommoda-
learning as opposed to their speediness at demonstrat- tion-Disability Interaction Paradigm (Elliott & Mar-
ing it. However, given that most students with disabil- quart, 2004) or the Maximum Potential Thesis (Zu-
ities use multiple accommodations concurrently, little riff, 2000), and contrast it with the Differential Boost
research has been conducted about the effects of any Theory (Fuchs & Fuchs, 2001). In essence, the inter-
given accommodation in isolation, making it difficult action hypothesis proposed that providing additional
to determine how effective ETTA is at addressing the time should result in higher performance in students
needs of specific students (Lindstrom, 2010). with disabilities but should not result in higher test
Despite its widespread use, researchers have scores in students without disabilities. The hypothesis
questioned the appropriateness of ETTA as a “blan- is based on the premise that students without disabil-
ket” accommodation for students with disabilities and ities are able to complete the test when working to
have instead proposed that accommodations are more their maximum potential under timed testing situa-
appropriate when they are tailored to the needs of tions and that well-designed tests therefore provide a
both the student with a disability as well as the intents reliable, valid measure of their learning of the content
and design of the specific test (Brinckerhoff, Shaw, & materials. Students with disabilities, however, are at
McGuire, 1992; Lovett, 2011). Various scholars have a disadvantage, as slower processing times result in
Journal of Postsecondary Education and Disability, 30(2) 187
them running out of time before they complete the alternative psychological student response to being
test (Cahalan-Laitusis, Morgan, Bridgeman, Zanna, provided with ETTA is that the students may perceive
& Stone, 2007). In these instances, the well-designed the accommodation as a validation of their lower skill
test is not an accurate and valid measure of their level, and ETTA may therefore inhibit self-percep-
learning but instead a measure of their speediness tions of efficacy in testing situations (Elliott, Yssel-
in accessing knowledge. Without having the time to dyke, Thurlow, & Erickson, 1998).
access and attempt all of the test questions, students
are denied the opportunity to fully demonstrate their Literature Supporting and Refuting the Theoretical
learning. Therefore, according to the interaction hy- Basis of ETTA
pothesis, the provision of extra time to students with As the use of high-stakes testing has grown, so has
disabilities allows a more valid and accurate measure the study of how ETTA effects both the testing process-
of their learning, but should not affect the scores of es and the use of the results. Copious research has ex-
students without disabilities in the same way. amined the interaction hypothesis to determine wheth-
The differential boost theory is similar in many er it stands up in various situations. Sireci et al. (2005)
aspects, yet it differs in one important way. The differ- completed the most influential and recent examination.
ential boost theory (Fuchs & Fuchs, 2001) also pro- These researchers found that, based on a review of
posed that extended time will enhance performance on over 40 empirical studies, the hypothesis was partial-
tests for students with disabilities, but does not require ly supported. In most studies they examined, students
that there are no similar positive effects on students with disabilities performed better when they were al-
without disabilities. The important distinction in this lotted more time than when they were not. In many
theory is that the gains made by the students with dis- of the studies they reviewed, however, it was shown
abilities must be significantly greater than the gains that students without disabilities also performed better
made by the students without disabilities, hence the on tests with extended time. Thus, Sireci, et al. (2005)
“differential boost” to the students with disabilities. proposed a modification to the interaction hypotheses
In this way, the differential boost to the achievement that, in effect, validated the differential boost theory: in
of students with disabilities when ETTA is provided ETTA situations, the scores of students with disabili-
suggests that this accommodation is an appropriate ties should be significantly greater when the students
response to the student’s specific disability. are provided with ETTA than when they are not, and
There are other theories from the field of psychol- gains made by students with disabilities should be sig-
ogy that inform understanding of why ETTA may or nificantly greater than those made by students without
may not result in higher test scores in students with disabilities in ETTA settings.
disabilities. Social Learning theory (Bandura, 1991)
proposed that people’s perceived efficacy in a given Interpretation of Increased Gains for All Students
situation will affect their functioning. Accordingly, Using ETTA
Elliott and Marquart (2004) proposed that students In framing this modification to the interaction the-
may process being provided with ETTA in one of ory, Sireci, et al. (2005) argued that the findings indi-
two ways. First, the students may be motivated by cating that all students benefited from extended test
increased self-efficacy, as they perceive that the test time did not mean that ETTA is unfair. That is, it is
is now achievable with the ETTA provided (Sokal & not the case that ETTA is necessarily unfair when all
Desjardins, 2016). In addition, the extra testing time students make gains, rather ETTA is viewed as unfair
may result in lower levels of anxiety and therefore when all the students make similar gains. Fuchs and
allow the students to focus more effectively on com- Fuchs (2001) clarified this position:
pleting the test (Perlman, Borger, Collins, Elenbogen,
& Wood, 1996). This possibility is noteworthy, as When accommodations increase scores for stu-
anxiety disorders have now surpassed depression as dents with learning disabilities no more than is
the most common disability in the general population, expected for non-disabled students, then we might
as well as in postsecondary populations (AUCCCD, conclude that the test accommodation does not
2014), and are often comorbid with other disabili- speak to the nature of the student’s disabilities in
ties, suggesting that processes that address anxiety in any essential way. On that basis, we may also in-
testing situations would have broad application. An fer that the accommodation is not fair. (p. 176)
188 Sokal & Vermette; Double Time?
Other research has refuted that the intent of the mod- than typical students or students “at risk” when all
ified interaction hypothesis and differential boost the- three groups were given ETTA. This finding is in-
ory are accomplished through the use of ETTA. Re- teresting in that the students with disabilities did not
cent research by Lewandowski, Lovett, and Rogers demonstrate a differential boost in their achievement
(2008), and Lewandowski, Lovett, Parolin, Gordon, when compared with the other groups in this study,
and Codding (2007) showed that extended time pro- but still experienced better affective outcomes (re-
vided even greater advantage to students without dis- laxation, motivation, less stress) in the ETTA con-
abilities than it did to students with attention deficit dition. Research with university students with anxi-
hyperactivity disorder (ADHD) or reading disabili- ety disorders (Sokal & Desjardins, 2016) found that
ties. Likewise, Fuchs and Fuchs (2001) demonstrated students who were provided with ETTA felt calmer,
that ETTA sometimes fails to result in better perfor- and during testing time experienced fewer of the
mance in students either with or without disabilities. gastrointestinal problems that are often associated
Of the seven studies about ETTA that Thompson et with high-stress events.
al. (2002) reviewed, four showed positive effects of
ETTA and three showed no effects of ETTA, again Criticisms Related to Fairness
suggesting that the appropriateness of using ETTA to Given the lack of consensus on whether provid-
address students’ testing needs is as yet inconclusive. ing ETTA results in a differential boost for students
However, even in studies that showed no overall dif- with disabilities, and also controversy on how to in-
ferential boost favoring either students with disabil- terpret the finding that other students sometimes also
ities or those without, there were still a minority of perform better when extra time is given to them in
students with disabilities who demonstrated differen- testing situations, it is not surprising that the use of
tial boosts: Fuchs and Fuchs (2001) showed that 23% ETTA as a common testing accommodation has been
of students with disabilities benefitted substantially widely questioned and criticized. The arguments tend
more from ETTA than students without disabilities to fall into four categories: (1) unfairness to students
when extended time was provided on math and read- without disabilities; (2) unfairness to students with
ing tests where no overall differential boost between disabilities; (3) unfairness to professors; and, (4) un-
groups was demonstrated. Medina (2000) had similar fairness to pedagogical development.
findings with the university students she studied us- Unfairness to students without disabilities.
ing both course-based and standardized assessments. The finding that all students benefit from additional
These finding suggest that attention must be paid to test time has been used to argue that providing ETTA
individual learning needs and accommodations, as only to students with disabilities gives them an unfair
opposed to group or standard practices of accommo- advantage over students in the same testing situation
dation (Fuchs, Fuchs, & Capizzi, 2005). Considered who are not given extra time (Lovett, 2011; Sireci, et
together, the current research literature demonstrates al., 2006; Sireci, et al., 2005). Indeed, the burgeon-
that there is no consensus on the accuracy of the inter- ing requests for ETTA on the Scholastic Aptitude Test
action hypothesis (Sireci, et al., 2005), nor on the dif- (SAT) administration have resulted in more stringent
ferential boost theory (Fuchs & Fuchs, 2001), as they documentation requirements and more students being
relate to ETTA and its effects on learners in groups or denied accommodation (Moore, 2010), suggesting
on individuals. that this question of providing unfair advantage to
Other research gives some support to Bandura’s students with disabilities is especially salient in high-
social learning and self-efficacy theories as a means stakes tests (Brinckerhoff & Banerjee, 2007).
of understanding how ETTA may affect student Unfairness to students with disabilities. Un-
performance. In addition to the cognitive benefits fairness to students with disabilities is claimed in two
of providing opportunities for students with slower different ways. First, when ETTA is not provided, stu-
processing speeds to access more of the test content dents perceive that they are being disadvantaged by
through ETTA, research with middle school children their institution’s reluctance to meet its legal obliga-
demonstrates that there are also potential psycholog- tions to accommodate:
ical processes at work. Elliott and Marquart (2004)
found that grade eight students with disabilities felt When students with learning disabilities tell us that
less frustrated, more relaxed, and more motivated some instructors engage in non-accommodation,
Journal of Postsecondary Education and Disability, 30(2) 189
there is an obvious challenge…. Denial of the ex- struct under examination is compromised. Although
istence of the learning disabilities or denying ac- a primary role of the professoriate is to work with
commodation likely contributes to instructional accessibility services to ensure that both the fidelity
and institutional environments that delay students’ of the test and the fairness to the student are main-
graduation, limit their academic success, and ulti- tained (Brinckerhoff, et al., 1992), professors are not
mately undermine their ability to use higher edu- always given this opportunity to collaborate but rather
cation as a stepping stone towards meaningful life are simply instructed to provide ETTA (Sokal, 2016).
goals. (Quinlan, Bates, & Angell, 2012, p. 230) Author showed that many professors question the use
of ETTA privately, but simply comply when asked to
The corollary is that these needs are met when ETTA provide it. Even when professors agree with the use
is provided. Sireci, et al. (2006) indicated that, “stu- of ETTA, Stretch and Osborne (2005) suggest that
dents with…disabilities demonstrate their true abil- instructors who are not familiar with how validity is
ities more clearly when they are allowed accom- affected by ETTA may introduce error into the test
modations” (p. 3). Lin (2010) therefore argued that process that in turn limits the use of the testing results.
accommodations increase test validity in that they The second way that ETTA is perceived to create
allow students to demonstrate their skills and abilities unfairness to professors relates to their impression of
in situations where standard testing procedures would the perceptions of students without disabilities. Re-
prevent it. search (Sokal, 2016; Bruder & Magro-Wilson, 2010;
Second, some critics view accommodations such Izzo, Hertzfeld, Simmons-Reed, & Aaron, 2001) has
as ETTA as a failure to meet individual student’s needs shown that professors are very concerned, especial-
in meaningful ways. ETTA, in particular, is viewed as ly in competitive programs, that other students will
a cost-effective, blanket accommodation that releases perceive the accommodations provided to some stu-
institutions from investigating student-specific, more dents and not to others as unfair. While confidentiality
suitable accommodations. According to Brinckerhoff, prohibits professors from explaining why specific stu-
et al. (1992), it is essential that accommodations are dents are provided with longer test times, these pro-
developed in order to meet the learning needs of a fessors, nonetheless, remain concerned that the other
specific student, rather than providing common ac- students will see them as giving some students an un-
commodations to all students who identify with dis- fair advantage (Sokal, 2016).
abilities. Lovett (2011) argued that “easy fixes” such Unfairness to pedagogical development. The
as ETTA take attention away from interventions that last way that ETTA is perceived as unfair is that by
have been shown to increase student agency and de- accepting that some students need accommodations
crease their dependence on accommodations, such as in order to provide valid test results, the professori-
test-taking strategies and stress-reductions practices. ate is diverted from an examination of timed testing
“Accommodations are overly tempting,” he said, “be- procedures per se. That is, in many tests, speediness
cause they are easier than interventions” (p.2). Critics is a factor that affects the students’ performance. Yet,
of ETTA suggest that students who are provided with in very few situations is the intended construct be-
these sorts of accommodations can become overly ing measured actually speediness. Considering the
dependent on them and find that similar accommoda- many ways to assess student learning as well as the
tions are not as readily available in work situations. limitations of testing, it seems counter intuitive that
Thus, these students enter the workforce less prepared timed tests have become the default method in mea-
than other students (Brinckerhoff, et al., 1992). suring student learning. Stretch and Osborne (2005)
Unfairness to professors. Likewise, arguments therefore suggested that timed tests should be a rare
that ETTA is unfair to professors follow two path- exception when choosing assessment practices. These
ways. First, research shows that there are times when authors, as well as Elliott, Braden, and White (2001),
ETTA is inappropriate, such as in situations where posited that using untimed testing is a more logical
speediness is a construct being evaluated (Brincker- and less problematic choice, while at the same time
hoff & Banerjee, 2007; Lovett, 2010; Phillips, 2002). recognizing that giving a prescribed time for a test
When professors are told that they must provide provides a logistical scheduling benefit to schools.
ETTA to students, it can sometimes create a situation
where the capability of the test to measure the con-
190 Sokal & Vermette; Double Time?
How Much Time is “Reasonable?” It is therefore a challenge to determine a reason-
If the advice of Stretch and Osborne (2005) and able amount of time to allow when ETTA is used, be-
Elliott, et al. (2001) were taken, the practical ques- ing as there are no established processes to determine
tion would arise regarding how long an untimed test this answer. The Cahalan-Laitusis, et al. (2006) study
would typically take. As seen previously, it is com- showed that students with disabilities needed no ex-
mon practice to allow students either 50% more time tra time on some types of test questions but needed
or 100% more time than the standard time allocated to a small amount (4-18% more per section) on other
students without disabilities who are writing the same types of questions, suggesting that the test design
test, a practice recommended by Ofiesh and Hughes may also affect the appropriateness of time allotted
(2002). It should be noted that Ofiesh and Hughes’ (Ofiesh & Hughes, 2002). Research studies about the
recommendation was generated from an analysis of use of accommodations have further suffered from
only seven quasi-experimental studies where the du- small sample sizes (Thompson et al., 2002). More-
ration of time used by students with disabilities writ- over, research designs investigating this question
ing almost exclusively standardized tests under ETTA have mainly been restricted to the study of the effects
conditions were reported. Furthermore, these studies of ETTA on high-stakes tests such as SAT (Elliott &
did not focus on the question of how much time was Marquart, 2004; Ofiesh & Hughes, 2002) in experi-
appropriate, but serendipitously reported these times mental settings, thus limiting what we know about the
as part of the data collected while examining other ETTA time used in day-to-day situations of students
factors. In six of the studies, the students were given with disabilities writing real, course-based tests. Ca-
unlimited time to finish the tests and told that their halan-Laitusis, et al. (2006) therefore have recom-
time would be recorded, a design feature that Ofiesh mended “Future research may wish to examine time
and Hughes posited may have inflated the total time used by students with… disabilities during an opera-
used. It is noteworthy that no empirical evidence ex- tional administration” (p. 12). The purpose of the cur-
ists to support these practices as recommendations for rent study was to examine the duration of time used
effective ETTA durations (Lewandowski, et al., 2013; by students provided with ETTA Spin course-based
Lovett, 2011), and the research literature, with a few tests administered in postsecondary settings. Specific
exceptions such as Ofiesh and Hughes’ work, is “si- research questions included:
lent on this issue” of what appropriate time allowanc-
es should be (Lewandowski, et al., 2013; Stretch & 1. Do students with disabilities who write their
Osborne, 2005). tests with ETTA use their full allotted time;
Some research has suggested that Ofiesh and and
Hughes’ (2002) recommended allowances are far too 2. Do students use longer durations of addition-
generous. Cahalan-Laitusis, King, Cline, and Bridge- al testing time relative to the maximum time
man (2006) posited that 25% more time is a more provided to students without ETTA in lower
suitable allotment. These authors found that students level than in higher level courses?
with disabilities who were writing their SATs in un-
timed conditions needed only 8% to 14% more time Methods
in order to access the same number of questions as
their peers without disabilities. Furthermore, Brooks, Population
Case, and Young (2003) found that giving students The data used in this study were taken from sec-
with disabilities excessive time allotments did not ondary data sets accessed through the student acces-
result in higher test scores. Cahalan-Laitusis, et al.’s sibility service offices at two mid-western, Canadian
(2006) recommendations are bolstered by dated, yet universities. Given that student records related to dis-
relevant, research by Perlman et al. (1996) who found abilities are protected under the Freedom of Informa-
that most students provided with ETTA did not use all tion and Protection of Privacy Act (FIPPA), all identi-
the additional time they were allocated. This finding fying information was removed from the data before
was further supported by the perception of university it was accessed by the researchers. Thus, it is impos-
students in a recent study by Author (in press a), and sible to report on the demographic information of
was also demonstrated by high school students in a the specific students whose testing data we accessed.
study by Cahalan-Laitusis, et al. (2006). However, general information about each university
Journal of Postsecondary Education and Disability, 30(2) 191
can be used as a proxy to likely describe the repre- Comparisons Between Universities
sentative participants. The smaller university offered Three time durations were reported for each case
mainly undergraduate programs, while the larger uni- by SAS. Table 3 reports the descriptive statistics of
versity provided undergraduate, graduate, and pro- these data from each university: (1) The Standard test
fessional programs. While both universities provided duration is the maximum duration of time in minutes
data regarding the diversity of their students, the larg- provided to all members of the class who did not have
er university published reports about the use of its ser- ETTA provided; (2) The ETTA test duration is the
vices, and therefore those data regarding the students maximum duration of time in minutes provided to a
registered with Student Accessibility Services (SAS) specific student writing that same test with ETTA; (3)
during the study years were available to the research- The Used test duration is the actual duration of time in
ers and are provided in Table 1. minutes used by a specific student writing that same
test with ETTA.
Design Three additional variables were computed and
The current research study entailed a post-hoc reported in Table 4. The ETTA/Standard score (E/S
analysis of data provided on exams written with ETTA score) represents the ETTA test duration divided by
by students registered with accessibility services. The the Standard test duration, indicating the relationship
exams were written during the years 2013-2014 and between the time provided to the specific students un-
2014-2015. After removing cases where no ETTA was der ETTA and to the other students without ETTA.
provided and where students failed to report to the test- For example, if the E/S score was 1.5, a particular
ing location, data from 2,414 exams were used from the student in the sample was given 1.5 times the duration
smaller university and 6,443 exams were used from the of time provided to the students who did not qualify
larger university. After consulting with and obtaining for ETTA on that particular test. The second comput-
the agreement of the coordinator of student accessibil- ed variable was the Used/Standard score (U/S score)
ity services at each university, and obtaining approval and represents the actual test duration used divided by
of the representative research ethics boards, data were the Standard test duration, indicating the relationship
provided by both universities including: (1) year in the between the time used by the specific student under
program to which each exam applied; (2) the standard ETTA and the time maximum provided to other stu-
exam time; (3) accommodated exam time under ETTA; dents without ETTA. For example, if the U/S score
(4) the time actually used to write the exam. were 2, the student used double the duration of time
provided to the students who did not qualify for ETTA
Findings on that particular test. The third variable computed
was the Used/ETTA score (U/E score), and was de-
Given Aud et al.’s (2013) caution that contextu- rived from dividing the students’ actual time used to
al factors specific to individual settings recommend complete the test by the maximum durations that were
against collapsing data sets across settings, we began allowed under ETTA. Thus, if a student’s U/E score
our analysis by examining each data set separately. was .5, it would indicate that the student completed
the test in 50% of the time allotted under ETTA.
Descriptive Statistics Although the values presented in Tables 3 and 4
At the smaller university, 1,235 tests were ana- appeared strikingly similar between the universities, a
lyzed from the 2013-2014 school year, and 1,179 tests MANOVA was conducted with each of the six scores
were analyzed from the following year, for a total of as dependent variables and the university as the in-
2,414 tests. At the larger university, 2,989 tests were dependent variable. Results indicated that there were
analyzed from 2013-2014 and 3,454 tests were ana- no significant differences between the universities in
lyzed from the subsequent school year, comprising terms on the standard test durations provided, (f(1,
6,443 tests in total from the larger university, and 8,857 8,856)=1.25, p =.26) and the Used/Standard score
tests in all. Given the differences in programming and [f(1, 8,856)=2.95, p=.09]. However, significant dif-
graduate level courses offered at each university, the ferences emerged between the two universities’ sam-
distribution of the level of each exam--corresponding ples in terms of ETTA provided [ f(1, 8,856)=8.38,
to the year of the course in the program are--presented p=.01], the actual Used minutes [f(1, 8,856)=7.89,
separately by university in Table 2. p=.01], the ETTA/Standard score [f(1, 8,856)=3.00, p
192 Sokal & Vermette; Double Time?
≤ .001], and the Used/ETTA score [f(1, 8,856)=29.68, As a result, we chose to conduct a Kruskal-Wallis
p ≤ .001]. Examination of the means previously pre- test, because it is recommended as an alternative to
sented indicated that the students at the smaller uni- ANOVA procedures in cases with non-parametric
versity were provided with an average of five extra variances (Lund Research, 2013a). Furthermore, the
minutes on exams under ETTA. In contrast, students data satisfied the four assumptions of using this test:
at the larger university used an average 5 more min- (1) the dependent variable was continuous; (2) the
utes to complete their tests. In terms of the relation- distribution of the data was not normal; (3) each case
ships between the ETTA durations compared to the (i.e. exam) was represented in only one group; and
standard test times at each university, the smaller uni- (4) the groups of two or more were categorical and
versity allowed on average an additional 62% of the independent (Lund Research, 2013a). We maintained
standard test time while the larger university offered the collapsed category five, which represented grad-
an additional 58% of the standard time. Finally, while uate level courses, and therefore we examined exams
the ETTA accommodations were slightly more gener- across five categories (first year, second year, third
ous at the smaller university, these students used only year, fourth year, and fifth the ninth year). The Used/
72% of the ETTA allowance on average, compared Standard score was used as the independent variable
with the students at the larger university who used an because this variable was not significantly different
average of 75% of their ETTA. between the universities. The results indicated that
Once it was established that the mean duration of there were significant differences between the Use/
Used/ETTA was .75 of the maximum for the larger Standard scores of exams written at different course
university and .72 of the maximum for the smaller levels, H(4)=128.25, p. ≤ 001, with mean ranks of
university, it was clear that the first research question 4,058.32 for first year exams, 4,386.81 for second
was answered: Many students with disabilities who year exams, 4,741.41 for third year exams, 4,885.45
write their tests with ETTA do not use the full allotted for fourth year exams, and 5,658.38 for exams from
time. We conducted follow-up analyses to tease out years five to nine. Follow-up Mann-Whitney tests
intricacies within this finding. First, through analysis were chosen to determine where the significant differ-
of the Used/Standard data frequencies, we determined ences could be found, being as they are recommended
how much, if any, of the standard time was used by as alternatives to t-tests when non-parametric groups
students who were provided with ETTA and reported are present and also because our data met the four
the findings in Table 5. We chose to use the Used/ assumption of using this test (Lund Research, 2013b).
Standard variable because it was not significantly dif- Results indicated that the mean rank was significant-
ferent between the two universities and therefore al- ly higher in second-year tests than in first-year tests
lowed us to examine the large data set as one. (U= 4,940,345.50, p ≤ .001) and was also significant-
In order to investigate whether the ETTA times ly higher in third-year tests than in second-year tests
provided followed these trends, we used similar fre- (U= 2,028,565.00, p ≤ .001). The mean ranks were
quency analysis on the ETTA/Standard scores and not significantly different between tests from year 3
reported the findings in Table 6. However, being as and 4 (U= 425,397.00, p = .21) or between years 4
the ETTA/Standard scores were significantly differ- and 5 (U= 304, p = .23). Our second question was
ent between the universities, we examined and have therefore answered, as students actually used increas-
presented each university’s frequencies separately. ingly more testing time relative to the maximum stan-
We then turned our attention to the second re- dard test time in third than first level courses. While
search question: Do students use longer durations of this escalation stopped between year three and year
testing time relative to the maximum time provided to four testing levels, it did not decrease.
students without ETTA in lower level than in higher
level courses? In order to investigate this question, Discussion and Implications
we had originally planned to use an ANOVA. How-
ever, Levene’s Statistic indicated that the variances The findings of the current research contribute to
within the exam levels were not homogeneous [F(5, the understanding of how ETTA is enacted in two uni-
8,774)=3.3, p= .01], even when we collapsed the versities of different sizes. Taking the advice of Aud
graduate level courses from years five to nine where et al. (2013), the two schools were initially examined
cell sizes were comparatively smaller (see Table 2). separately, despite the impression that the data from
Journal of Postsecondary Education and Disability, 30(2) 193
both universities appeared quite similar. Indeed, the Conclusions and Recommendations
average standard testing time, the average ETTA du- Based on these three trends, it would seem logi-
ration provided, and the average actual testing time cal to recommend that students be provided with both
used differed only five minutes or less across univer- shorter ETTA as well as more supports for develop-
sities. However, when analyses were conducted on ing other strategies to use less ETTA as they progress
these variables as well as those derived from them throughout their studies. However, a blanket recom-
(ETTA/Standard score, Used/Standard Score, and mendation such as that suffers from the same limita-
Used/ETTA score), significant differences emerged. tion as those of Ofiesh and Hughes (2002) in that it
Three of the findings of the current study will be ignores individual differences.
examined in greater depth in order to inform process- While increasingly tight university budgets make
es for supporting students who request ETTA. First, a “one-size-fits-all” approach to meeting students
it is noteworthy that many students complete testing testing accommodation needs more attractive and
with only a small proportion of additional time com- while the current findings show that students rarely
pared to the standard. Of the sample of 8,857 exams, use more than 25% additional time when compared
it was found that in 3,059 (35.5%) of exams, students to the standard test times, therefore creating a tempta-
did not use any ETTA, while in 5,798 (64.5%) ex- tion to endorse blanket decreases in ETTA, abruptly
ams, students used at least some of it. These findings reducing ETTA would be a mistake. ETTA in many
support those of Cahalan-Laitusis et al. (2006), who ways can be compared to home insurance or health
conducted their research with high school students insurance: Just because one does not use it does not
and showed that 8-14% additional time is usually suf- mean that one does need it. Other research has shown
ficient for ETTA. They therefore recommended that that just the presence of the extra time is enough to
the standard ETTA be set at 125% of the standard test decrease student stress so that they do not use the ex-
time. Likewise, our data showed that while over 55% tra time allowed on a test (Sokal & Desjardins, 2016).
of the students complete their tests with an addition For this reason, caution against drastically reducing
of 25% of the standard time of less (see Table 5), only ETTA even in the situations where it is not used is
1.8% of students at the smaller university and 8.7% advisable. Students should be given the opportunity
of students at the larger university were limited to this to gradually decrease their ETTA use in situations
duration (see Table 6). Furthermore, 85% of students where that is possible and should be active agents in
completed their tests with an additional 50% of the the goal-setting and discussions that lead to decisions
standard time or less (see Table 5), yet only 58% at about ETTA durations.
the smaller university and 70.5% at the larger univer- Furthermore, individual differences must be con-
sity were limited to this ETTA duration (see Table 6). sidered when setting goals around reduced ETTA. For
The second trend to be highlighted is that students example, it is possible that a student who experiences
used more ETTA as they moved through their first test anxiety may learn additional coping strategies and
three years of university. It may be that these findings therefore use less ETTA over time. However, it is un-
speak to the commensurate challenge level of increas- likely that a student who has permanent language pro-
ing course levels, or it may be that students are failing duction difficulties and uses a scribe will show the same
to develop other strategies that allow them to either trends. Indeed, although the frequency of students re-
maintain or decrease their ETTA used. quiring more than twice the standard test time are rare,
The third trend is that the ETTA provided at both these students are present in our university populations
universities clusters around the time points recom- and have equal rights to appropriate accommodations.
mended by Ofiesh and Hughes (2002). It is notewor- Individual needs and capacities must be considered
thy that there is a clustering of scores at both uni- both in setting ETTA and considering whether reduc-
versities within the range that includes 1.5 times the ing ETTA is a reasonable accommodation.
standard test time and a second cluster at the 2 times How then can the approach to ETTA honor indi-
the standard testing time range at the larger university vidual differences and capacity at the same time as
only. This finding suggests that some ETTA provid- it fosters learner growth? Similar to school systems
ers may have accepted the recommendations of Of- that require students with disabilities to have annual
iesh and Hughes (2002) without examining their very individual educational plans, universities are advised
weakly supported research origins. to meet annually with students to review progress in-
cluding the students’ Used/Standard scores and their
194 Sokal & Vermette; Double Time?
ETTA allowances. If students are given information by research evidence and is insensitive to individu-
about their own trends in use of ETTA, they will al learning needs. Rather than focusing on the short-
be able to set more accurate goals for their futures. term, time-efficient means of providing accommoda-
This approach not only recognizes the efficacy needs tion through ETTA alone, we suggest that universities
of adult students (Bandura, 1997) but also fosters pay more attention to both the individual needs of
self-determination (Getzel, 2008), a main predictor of students in testing situations as well as to goal set-
student success. Self-determination was described by ting and supports that foster greater learning indepen-
Getzel as “acceptance of a disability and how it af- dence whenever possible.
fects learning; understanding which support services
are needed; knowing how to describe one’s disability References
and the need for certain supports to service providers;
and having the determination to overcome obstacles Association for University and College Counseling
that may be presented” (p. 210). In addition, Getzel Centre Directors Annual Survey [AUCCCD].
showed that student skills such as organizational (2014). AUCCCD monograph public 2013.
skills, time management, goal setting skills, and an Aud, S., Wilkinson-Flicker, S., Kristapovich, P.,
awareness of how technology can support their learn- Rathbun, A., Wang, X., & Zhang, J. (2013). The
ing were predictive of success in students with dis- condition of education 2013 (NCES 2013-037).
abilities. Of course, the goals set by students would U.S. Department of Education, National Center
need to be accompanied by supports from the univer- for Education Statistics. Washington, DC.
sities such as those mentioned by Getzel as well as Bandura, A. (1997). Self-efficacy: The exercise of
self-regulation strategies, stress reduction, test-taking control. New York: W.H. Freeman.
strategies, and the like. In this way, students are sup- Bandura, A. (1991). Social cognitive theory of
ported to become self-determined, active agents in self-regulation. Organizational Behavior and Hu-
planning and enacting their growth and independence man Decision Processes, 50, 248-287.
as students. Brinckerhoff, L. C., & Banerjee, M. (2007). Mis-
All research presents limitations, and the current conceptions regarding accommodations on high-
project is no exception. The first limitation relates stakes tests: Recommendations for preparing dis-
to the data set. Due to privacy laws, the data were ability documentation for test takers with learning
cleaned of information about specific students and disabilities. Learning Disabilities Research &
their disabilities before being provided to the re- Practice, 22(4), 246-255.
searchers. If provided with this information and also Brinckerhoff, L. C., Shaw, S. F., & McGuire, J. M.
provided with information about the same students’ (1992). Promoting access, accommodations, and
Used/Standard scores over time, it would be possi- independence for college students with learning
ble to make more specific recommendations about disabilities. Journal of Learning Disabilities, 25,
which students would benefit most from gradual de- 417-429.
creases in ETTA accompanied by other skill training Brooks, T. E., Case, B. J., & Young, M. J. (2003).
and which students would not. Kim and Lee (2015) Timed versus untimed testing conditions and stu-
showed that the influence of testing accommodations dent performance. Pearson Assessment Report.
varies by disability, and having access to these data Pearson Education, Inc.
about the specific disabilities in the current study’s Bruder, M., & Magro-Wilson, C. (2010). Student and
sample would have allowed a more nuanced analysis. faculty awareness and attitudes about students
The second limitation is that the recommendations with disabilities. Review of Disability Studies: An
generated here infer allocation of funding in order to International Journal, 6, 3-13.
create individualized plans for each student. The lim- Cairns, S., Massfeller, H., & Deeth, S. (2009). Why
itation to this recommendation is the will of the poli- do postsecondary students seek counselling? Ca-
cy-makers and university administrators who allocate nadian Journal of Counselling and Psychothera-
budget dollars. py / Revue Canadienne de Counseling et de Psy-
Overall, the current research findings suggest chothéRapie, 44(1), 34-50.
that adhering to allotments of 50-100% ETTA, while
cost-effective and easily administered, is unsupported