TREND: WRITING REPORTS FOR QUASI-

EXPERIMENTS
by Simon Moss

Introduction

The TREND statement—an acronym that represents the Transparent

Reporting of Evaluations with Non-randomized Designs—is a checklist writers
should utilize when they report a quasi-experiment. To clarify, researchers often
compare two or more conditions, such as two teaching methods. In many of these
studies, the participants or units are not randomly assigned to these conditions.
This design is called a quasi-experiment or non-randomized design. TREND is a set
of principles that researchers should observe whenever they report quasi-
experiments or non-randomized designs.

Illustration

Suppose a researcher wants to explore whether lecturers who wear casual

attire enhances student engagement more than do lectures who wear formal attire.
In this study, students are not randomly allocated to these lecturers. Instead,
whether they observe lecturers in casual attire or formal attire depends on whether
they attended Tuesday classes or Wednesday classes.

Purpose of this document

This document lists every TREND principle and then includes examples to
illustrate each principle. These principles do not encompass every paragraph you
should include. Instead, these principles stipulate the most important details you
should include.

TREND checklist

Title and abstract

Item Description of the item—as Example
specified by Des Jarlais et al
(2004)
1 Title Information on how unit were  Participants who were assigned to

Page 1 of 11
 and allocated to interventions the Tuesday classes observed the
abstract lecturer wearing casual attire;
participants who were assigned to
the Wednesday class observed
the lecturer wearing formal
attire…
Structured abstract NB. Some journals prefer writers
recommended subdivide the abstract into distinct
sections such as
 Background
 Method
 Findings
 Implications
Information on target  The participants were psychology
population or study sample students at Charles Darwin
University, aged between 18 and
65.

Introduction
Item Description of the item Example
2 Scientific background and  According to the deviant
Background explanation of rationale hypothesis, when individuals are
exposed to cues that typify
convention, such as formal attire,
they tend to behave more
conventionally. Their creativity
diminishes.
 Because creativity has been shown
to promote positive emotions, this
decrease in creativity is likely to
limit student engagement.
Theories used in designing  Not applicable to this study
behavioural interventions

Method
Item Description of the item Example
3 Eligibility criteria for  All students enrolled in a specific
Participants participants, including first year psychology unit were
criteria at different levels invited to participate.
in recruitment/sampling
plan (e.g., cities, clinics,

Page 2 of 11
 subjects)
Method of recruitment  The students received an email
(e.g., referral, self- from the researcher immediate
selection), including the after the first class. This email
sampling method if a asked students whether they would
systematic sampling plan like to complete a survey about the
was implemented first class.
Recruitment setting  “

Settings and locations  Students received online surveys

where the data were during the first week and last week
collected of the semester, immediately after
the class. The students were
prompted to complete these
surveys within 5 days of receiving
the survey at a time and location of
their choice. They were
encouraged to complete the
surveys in a location in which they
will not be disturbed.
4 Details of the interventions 
Intervention intended for each study
s condition and how and
when they were actually
administered, specifically
including
Content: what was  The students attended 12 lectures
presented? on psychology, each lasting
approximately 2 hours, on either a
Delivery method: how was Tuesday or Wednesday at 10.00 in
the content presented? the morning. The same individual
presented all lectures, but wore
Unit of delivery: how were casual attire on the Tuesdays and
the subjects grouped formal attire on the Wednesdays
during delivery  In each condition, all the students
attended the same lectures
Setting: where was the  The lectures were presented in an
intervention delivered auditorium that seats 100
individuals, although only about 50
Deliverer: who delivered students, on average, representing
the intervention? 80% of the cohort, attended each
lecture
 These lectures were presented at
Charles Darwin University,
Casuarina campus, by a male
senior lecturer who was 45 years of

Page 3 of 11
 age.
Exposure quantity and  On average, the students who were
duration: how many assigned to the Tuesday classes
sessions or episodes or attended 10.4 of the 12 lectures
events were intended to be (sd = 1.42). The students who
delivered? How long were were assigned to the Wednesday
they intended to last? classes attended 10.2 of the 12
lectures (sd = 1.39). An
Time span to deliver the independent t-test uncovered no
intervention to each unit? significant difference in attendance
rate, t(98) = .947, p > .05
Activities to increase  Whether or not students attended
compliance or adherence was recorded at the start and end
(e.g., incentives) of each lecture. Furthermore, to
promote attendance, students
received five clues to the
forthcoming exam during each
class.
 Students also received a text
reminder before each class.
5 Objectives Specific objectives and  The aim of this study was to
hypotheses explore whether clothing can affect
student engagement. The
hypothesis was that casual attire
should enhance student
engagement more than formal
attire—but this effect might
diminish over time as individuals
become habituated to the clothes
of this lecturer
6 Outcomes Clearly defined primary  To measure student engagement,
and secondary outcome participants completed the student
measures engagement scale—a scale that
comprises 15 items, such as “I felt
absorbed during the class”.
 In addition, students were asked to
indicate which of several words the
lecturer uttered during the
previous lecture—an objective
measure of concentration and thus
engagement.
Methods used to collect  One complication is that
data and any methods disengaged students might not
used to enhance the answer the questions as
quality of measurements accurately. To assess this

Page 4 of 11
 possibility, participants also
completed the inattentive scale—a
measure that gauges the extent to
which individuals are concentrating
while completing the
questionnaire.
Information on validated  In past research, Cronbach’s  for
instruments such as the student engagement scale
psychometric and approximates 0.85 (Smith, 2005),
biometric properties implying high levels of internal
consistency (Nunnaly, 1975).
Furthermore, as evidence of
validity, research indicates that
student engagement, as measured
by this scale, strongly predicts
subsequent exam performance
even after controlling previous
grades (Brown, 2010).
7. Sample How sample size was  An a priori power analysis was
size determined and, when conducted using GPower. This
applicable, explanation of analysis indicated that 100
any interim analyses and participants would generate a
stopping rules power of 0.8, assuming a medium
effect size. Hence, the sample size
of 100 was deemed as suitable for
these circumstances.
8 Unit of assignment (the  Each individual was assigned to
Assignment unit being assigned to one of the two conditions.
method study condition, e.g.,
individual, group,
community)
Method used to assign  At the start of this semester,
units to study conditions, students received a timetable,
including details of any indicating whether they should
restriction (e.g., blocking, attend Tuesdays or Wednesdays.
stratification, minimization) An algorithm, designed to prevent
timetable clashes and equate the
number of students in each class,
governed this timetable.
 Thus, which day the student
attended was partly dependent
upon the other units in which they
were enrolled but partly random as
well.
Inclusion of aspects  The algorithm that allocated

Page 5 of 11
 employed to help minimize students to classes was designed
potential bias induced due to control grade point average.
to non-randomization (e.g., Thus, grade point average in the
matching) Tuesday and Wednesday classes
were roughly equivalent.
9. Blinding Whether or not  To diminish demand characteristics
participants, those and similar biases, the person
administering the administering the intervention—the
interventions, and those lecturer—was not informed of the
assessing the outcomes study hypotheses.
were blinded to study
condition assignment; if so,
statement regarding how
the blinding was
accomplished and how it
was assessed.
10 Unit of Description of the smallest  The engagement of each individual
analysis unit analysed to assess was included in the analyses
intervention effects (e.g.,
individual, group, or
community)
If the unit of analysis  NA
differs from the unit of
assignment, the analytical
method used to account
for this (e.g., adjusting the
standard error estimates
by the design effect or
using multilevel analysis)
11 Statistical Statistical methods used to  A mixed effects regression model
analysis compare study groups for was conducted to ascertain
primary methods whether engagement of each
outcome(s), including student at both times—as
complex methods of measured by both the
correlated data questionnaire and assessment of
concentration—differed between
the two conditions. In particular, a
structured covariance matrix was
utilized to…
Statistical methods used  Furthermore, to gauge the validity
for additional analyses, of these measures, Table 1
such as a subgroup presents the correlations between
analyses and adjusted the student engagement scale,
analysis concentration during the lecture,
and inattention during the survey

Page 6 of 11
 after both the first lecture and last
lecture.
Methods for imputing  The data from individuals who
missing data, if used completed the survey at only one
time were included in the analyses.
Hence, missing data were not
imputed.
Statistical software or  SPSS Version 22 was utilized to
programs used conduct these analyses.

Results
Item Description of the item Example
12 Enrolment: the numbers of  In this study, 60 students were
Participant participants screened for allocated to each of the two
flow [Often eligibility, found to be classes. However, only 55 and 50
include a eligible or not eligible, students attended the first Tuesday
diagram] declined to be enrolled, and first Wednesday class
and enrolled in the study respectively. In addition only 51
and 48 of these students
Assignment: the numbers completed the survey during the
of participants assigned to first Tuesday and first Wednesday
a study condition class respectively.
 Furthermore, 48 and 45 of these
Allocation and intervention students completed the survey
exposure: the number of during the last Tuesday and last
participants assigned to Wednesday class respectively.
each study condition and  Two students attended the last
the number of participants class, but not the first class; these
who received each individuals did not complete the
intervention survey, however.
 All the completed surveys were
included in the data analysis, even
if participants completed the
surveys during one time only.
Follow-up: the number of  NA
participants who
completed the follow- up or
did not complete the
follow-up (i.e., lost to
follow-up), by study
condition
Analysis: the number of 
participants included in or

Page 7 of 11
 excluded from the main
analysis, by study
condition
Description of protocol  Contrary to the protocol, one
deviations from study as participant attended both the
planned, along with Tuesday and Wednesday sessions.
reasons  The responses of this participant
were excluded from the analyses.
13 Dates defining the periods of  The first Tuesday and first
Recruitment recruitment and follow-up Wednesday classes were organized
in February 2018. Participants
received a plain language
statement during these classes.
The last Tuesday and first
Wednesday classes were organized
in May 2018
14 Baseline Baseline demographic and  Table 1a presents the mean age,
data clinical characteristics of the percentage of males and
participants in each study females, and percentage of
condition domestic and international
students in each condition during
the first Tuesday and Wednesday.
 Table 1b presents the same
demographics but during last
Tuesday and Wednesday.
Baseline characteristics for  Table 1c presents the same
each study condition demographic characteristics across
relevant to specific disease the entire university.
prevention research
Baseline comparisons of  NA
those lost to follow-up and
those retained, overall and
by study condition
Comparison between study  As these tables indicate, the age,
population at baseline and gender profile, and international
target population of status of students in this sample
interest and in the university overall are
comparable.
15 Baseline Data on study group  These tables indicate that mean
equivalence equivalence at baseline age and percentage of domestic
and statistical methods and international students does not
used to control for baseline differ significantly between the
differences Tuesday and Wednesday classes.
 However, the proportion of females
is higher in the Tuesday classes

Page 8 of 11
 relative to the Wednesday classes.
 Hence, gender was controlled in
the mixed effects regression model
16 Numbers Number of participants  The responses of 102 participants
analysed (denominator) included in were subjected to the first mixed
each analysis for each effects regression model…
study condition,
particularly when the
denominators change for
different outcomes;
statement of the results in
absolute numbers when
feasible
Indication of whether the  NA.
analysis strategy was
“intention to treat” or, if
not, description of how
non-compliers were
treated in the analyses
17 For each primary and  Table 2 presents the B coefficients,
Outcomes secondary outcome, a t values, semi-partial correlation,
and summary of results for and 95% confidence interval of the
estimation each estimation study B coefficients for each predictor as
condition, and the well as the R squared value.
estimated effect size and a
confidence interval to
indicate the precision
Inclusion of null and  NA
negative findings
Inclusion of results from  NA
testing pre-specified causal
pathways through which
the intervention was
intended to operate, if any
18 Ancillary Summary of other analyses  Table 3 presents the analyses after
analyses performed, including excluding participants who did not
subgroup or restricted complete the survey within 2 days.
analyses, indicating which As this table shows…
are pre-specified or
exploratory
19 Adverse Summary of all important  Interesting, subsequent interviews
events adverse events or indicated that some participants
unintended effects in each maintained the causal attire was
study condition (including unprofessional.
summary measures, effect

Page 9 of 11
 size estimates, and
confidence intervals)

Discussion
Item Description of the item Example
20 Interpretation of the  Consistent with the hypotheses,
Interpretatio results, taking into account both measures of engagement
n study hypotheses, sources were elevated whenever the
of potential bias, lecturer wore casual attire. This
imprecision of measures, effect of clothes, however, was not
multiplicative analyses, as pronounced after 12 weeks.
and other limitations or  Nevertheless, many other
weaknesses of the study differences between the two days,
such as the weather, could explain
this disparity between the Tuesday
and Wednesday classes.
 Other disparities between the
conditions could explain the
findings. Students allocated to the
Tuesdays tended to be studying
social work—indicating these
students might be more interested
in mental health, the topic of this
lecture series.
Discussion of results taking  The results are consistent with the
into account the notion that formal attire primes
mechanism by which the memories of conventional
intervention was intended circumstances and can diminish a
to work (causal pathways) sense of creativity and exploration
or alternative mechanisms  Alternatively, the students might
or explanations feel a greater sense of connection
with lecturers who wear casual
attire
Discussion of the success  NA
of and barriers to
implementing the
intervention, fidelity of
implementation
Discussion of research,  NA
programmatic, or policy
implications
21 Generalizability (external  In this study, all the participants

Page 10 of 11
 Generalizabil validity) of the trial were psychology students. In other
ity findings, taking into disciplines, students might not
account the study associate casual attire with
population, the creativity. For example, in
characteristics of the business, students might associate
intervention, length of casual attire with unprofessional
follow-up, incentives, behaviours, and thus such attire
compliance rates, specific might elicit a distinct sequence of
sites/settings involved in reactions.
the study, and other  In addition, many other features,
contextual issues besides whether the attire was
casual or formal, could have
differed between the conditions. To
illustrate, the formal attire covered
the lecturer and, therefore, could
have induced more sweat. The
sweat might have influenced the
responses of students.
22 Overall General interpretation of  These results vindicate the
evidence the results in the context proposition that informal clothing
of current evidence and might foster creativity, and that
current theory such creativity might promote
student engagement.
 Nevertheless, to substantiate this
account, future research should
assess whether creativity does
indeed mediate the association
between casual attire and student
engagement.

References

Des Jarlais, D. C., Lyles, C., Crepaz, N., & the Trend Group (2004). Improving the
reporting quality of nonrandomized evaluations of behavioral and public health
interventions: The TREND statement. American Journal of Public Health, 94, 361-
366. For more information, visit: http://www.cdc.gov/trendstatement/

Page 11 of 11