Frequently Asked Questions

Adult Reading Test, Second Edition (ART-2) - Frequently Asked Questions

 Use of Table 1 or Tables A10 and A1:

We would recommend the use of table 1 to interpret the results from ART2. 

However,  if confidence intervals are needed, then base the size of the confidence interval on the size of the interval in tables A.10 and A.11.  
Please note that these tables differ in terms of standard scores because the IQ-based standard scores (ie, those based on a mean of 100 and standard deviation of 15), which are presented in the tables at the end of the manual (tables A.10 and A11), have been determined via a formula that assumes a normal distribution.

The ‘Standard scores equivalents’ in the table within the body of the manual (table 1) are those associated with the centile scores in the table – and these centile values do not assume a normal distribution. If the data actually do produce a normal distribution, then the two methods will give the same results.

Unfortunately, a normal distribution rarely happens with real data produced by individuals under normal educational testing conditions. A normal distribution can be produced by clipping the ends of the data and manipulating the data in various ways; however, if the ends of the distribution are clipped then the values in table 1 will no longer represent the actual data collected. 

Tables A10 and A11 have been included in the manual by the authors as they may be useful for assessors if they wanted to focus more on IQ-based standard scores as part of the assessment, and if they wanted to see how to convert parts of the test or sections of the results into IQ-based standard scores, and, especially, if they wished to quote associated confidence intervals. 

Given that the aim of the test is to support assessment of learning needs, within the context of using other measures of ability, then we have included within the body of the manual table 1 that represents the data collected from the full sample of participants – to best represent students from the full range of FE and HE backgrounds included in the work. These data do not follow a normal distribution in all cases – and they have not been clipped or manipulated in any way to better represent a normal distribution. Therefore, the centile scores are the best way of representing the data and should be used in most cases, although it is recognised that standard scores are required, especially, by many examination bodies.

What do you think?