06.04.2016 Views

Do Randomized Controlled Trials Meet the “Gold Standard”?

Do-randomized-controlled-trials-meet-the-gold-standard

Do-randomized-controlled-trials-meet-the-gold-standard

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

DO RANDOMIZED CONTROLLED TRIALS MEET THE “GOLD STANDARD”?<br />

ALAN GINSBURG AND MARSHALL S. SMITH<br />

adjustments to <strong>the</strong> analyses where ei<strong>the</strong>r baseline differences<br />

or attrition appears excessive. The WWC also<br />

rigorously requires adjustments of statistical tests for clustering<br />

of students within schools and classrooms. 21<br />

There are 18 different curricula. Some curricula<br />

were involved in more than one RCT. Table 1 names<br />

<strong>the</strong> curricula and <strong>the</strong> number of RCT studies in two<br />

different categories: 19 studies have RCTs that meet all<br />

WWC standards for quality of randomization and statistical<br />

procedures without reservations, and 8 studies<br />

do not fully meet all standards but are deemed by <strong>the</strong><br />

WWC as usable with reservations. 22<br />

Methodology<br />

We ga<strong>the</strong>red all <strong>the</strong> available material from <strong>the</strong> WWC<br />

website on each RCT study, as well as online original<br />

reports or relevant articles. 23 In a few cases, we emailed<br />

<strong>the</strong> authors to ask <strong>the</strong>m to clarify something related to<br />

<strong>the</strong>ir study.<br />

We specifically examined each RCT study against<br />

certain criteria to gauge whe<strong>the</strong>r it had threats to <strong>the</strong><br />

standard of credible, reliable, and useful evidence. 24 For<br />

all <strong>the</strong> RCTs, we were able to get a reliable and arguably<br />

accurate reading on all or many of <strong>the</strong> criteria.<br />

Table 2 is a sketch of our conceptualization of <strong>the</strong><br />

desired criteria for a useful RCT. We have eight general<br />

categories where threats may occur: strong <strong>the</strong>ory,<br />

independence of association with curriculum developer,<br />

curriculum well-implemented, identified comparison<br />

group, appropriate grade coverage, objective measurement<br />

of outcomes, curriculum not out of date, and<br />

replication.<br />

This report does not explore whe<strong>the</strong>r <strong>the</strong> 27 RCT<br />

studies met <strong>the</strong> six crosshatched criteria in Table 2. We<br />

assume that <strong>the</strong> WWC staff in <strong>the</strong> IES fully reviewed<br />

and accepted three traditional statistical criteria: unbiased<br />

sampling procedure, sample attrition, and appropriate<br />

statistical analyses.<br />

In addition, we did not explore whe<strong>the</strong>r <strong>the</strong> intervention<br />

was based on a reasonable <strong>the</strong>ory, if <strong>the</strong> implementation<br />

had been replicated, or whe<strong>the</strong>r it used a<br />

single or double blind structure, even though <strong>the</strong>se criteria<br />

are important. 25 None of <strong>the</strong> WWC summaries<br />

provided adequate information about any of <strong>the</strong>se three<br />

potential threats. A few of <strong>the</strong> research articles discussed<br />

<strong>the</strong> <strong>the</strong>ory of <strong>the</strong> intervention, but this was atypical and<br />

did not provide enough detail to review.<br />

This leaves us with six of <strong>the</strong> eight categories and 12<br />

potentially threatened criteria. They are:<br />

Study independent of association with curriculum<br />

developer<br />

1. Evaluators are independent of association with<br />

curriculum developer.<br />

Curriculum well-implemented<br />

2. Curriculum is not studied in first year of<br />

implementation.<br />

3. Available evidence of implementation shows<br />

fidelity.<br />

Comparison identified<br />

4. Outcomes are reported by each identified<br />

curriculum.<br />

5. Comparison and intervention have equivalent<br />

instructional time.<br />

Appropriate grade coverage<br />

6. Study is longitudinal, if appropriate.<br />

7. There is broad coverage of a multigrade<br />

curriculum.<br />

8. WWC reports outcomes by grade, where available<br />

in <strong>the</strong> original study.<br />

Outcomes objectively measured, correctly analyzed,<br />

and fully reported<br />

9. Aligned assessments do not favor treatment or<br />

comparison curricula.<br />

10. Student interactions with outcomes are assessed<br />

and reported, where available.<br />

11. Teacher interactions with outcomes are assessed<br />

and reported, where available.<br />

Curriculum is not out of date<br />

12. Curriculum is current: it is still published, was<br />

initially released after 2000, and has not been<br />

replaced by a career- and college-ready state<br />

standards version (such as <strong>the</strong> Common Core).<br />

4

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!