## Measures, Correlations, and Explained Variances

Rosenthal and Rubin (1979) deliberately challenge conventional statistical wisdom by means of a paradoxical interpretation of a simple data set. They suppose that 100 patients in a medical study are randomly assigned to new treatment N and 100 to standard treatment S. At the end of one year, 70% of the patients under treatment N are still alive, but only 30% under treatment S. Here is their data matrix:

```Treatment            Survival after
New            30     70
Standard       70     30
```

How much better is Treatment N than Treatment S?
1) The Pearson product-moment correlation, r, between surviving and receiving treatment N is 0.4. Not an impressive value.

2) The variance in "survival explained" by assignment to treatment is r^2 = 0.4^2 = 0.16, i.e., only 16%. This is less than 1/6th of the total variance. Obviously a small and, therefore, obviously an "unimportant" amount.

But, when considering one's own medical treatment, Treatment N is clearly preferable to Treatment S. Rosenthal and Rubin (1979) stop here, having made their point that the relationship between statistics and substance is not always obvious. Let us continue.

3) The standard error of r for testing whether r=0 is approximately 1/sqrt(200) = 0.07. t = 0.4/0.07 = 5.7. Thus the probability that there is no correlation between treatment and survival is <.0001. This is statistically very significant, but it enables us only to declare that N and S are "not the same". It does not say how much better N is than S.

4) The logit distance between the effects of treatments N and S is the difference between the log-odds for survival with N and the log-odds for survival with S = loge(70/30) - loge (30/70) = 1.7 logits, with a standard error of sqrt(200/(30*70)) = 0.3 logits. Testing the hypothesis that the distance is zero, t = 1.7/0.3 = 5.7 (as expected).

5) Let us take this extremely simple logit analysis a step further. The analysis provides a useful 95% confidence interval for the survival advantage of the New treatment over the Standard:

(New - Standard) = 1.7 +- (2 * 0.3) = 1.7 +- 0.6 logits

Now we can conclude, with 95% confidence, that the survival advantage of New over Standard is:

1.1 logits < (New - Standard) < 2.3 logits

6) Finally, what are these advantages of 1.1, 1.7 and 2.3 logits on New treatment benefit worth to you, a prospective patient? The advantage to you depends on your Standard probability of survival.

```Your               Your New survival probability
Survival           At least    Average     At most
Probability         1.1         1.7         2.3
.10                 .25         .38         .50
.20                 .43         .58         .71
.30 (in example)    .56         .70         .81
.40                 .67         .78         .87
.50                 .75         .85         .91
```

Despite the low correlation of 0.4 and low "explained variance" of only 16%, are you in any doubt as to which treatment is best for you? These data, when understood from the Rasch measurement point of view, are in no way paradoxical. They provide an entirely decisive answer to your important question.

By choosing to think with measures rather than correlations and variances, the conflict between the substantive and the statistical findings disappears.

Rosenthal R, Rubin DR (1979) A note on percent variance explained as a measure of the importance of effects. Journal of Applied Social Psychology 9:5, 395-396.

Measures, correlations and explained variances. Rosentahl R, Rubin DR, Linacre Jm, Wright BD. … Rasch Measurement Transactions, 1995, 9:2 p.435

Rasch Publications
Rasch Measurement Transactions (free, online) Rasch Measurement research papers (free, online) Probabilistic Models for Some Intelligence and Attainment Tests, Georg Rasch Applying the Rasch Model 3rd. Ed., Bond & Fox Best Test Design, Wright & Stone
Rating Scale Analysis, Wright & Masters Introduction to Rasch Measurement, E. Smith & R. Smith Introduction to Many-Facet Rasch Measurement, Thomas Eckes Invariant Measurement: Using Rasch Models in the Social, Behavioral, and Health Sciences, George Engelhard, Jr. Statistical Analyses for Language Testers, Rita Green
Rasch Models: Foundations, Recent Developments, and Applications, Fischer & Molenaar Journal of Applied Measurement Rasch models for measurement, David Andrich Constructing Measures, Mark Wilson Rasch Analysis in the Human Sciences, Boone, Stave, Yale
in Spanish: Análisis de Rasch para todos, Agustín Tristán Mediciones, Posicionamientos y Diagnósticos Competitivos, Juan Ramón Oreja Rodríguez

 Forum Rasch Measurement Forum to discuss any Rasch-related topic

Go to Top of Page
Go to index of all Rasch Measurement Transactions
AERA members: Join the Rasch Measurement SIG and receive the printed version of RMT
Some back issues of RMT are available as bound volumes
Subscribe to Journal of Applied Measurement

Go to Institute for Objective Measurement Home Page. The Rasch Measurement SIG (AERA) thanks the Institute for Objective Measurement for inviting the publication of Rasch Measurement Transactions on the Institute's website, www.rasch.org.

Coming Rasch-related Events
June 30 - July 29, 2017, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Further Topics (E. Smith, Winsteps), www.statistics.com
July 31 - Aug. 3, 2017, Mon.-Thurs. Joint IMEKO TC1-TC7-TC13 Symposium 2017: Measurement Science challenges in Natural and Social Sciences, Rio de Janeiro, Brazil, imeko-tc7-rio.org.br
Aug. 7-9, 2017, Mon-Wed. In-person workshop and research coloquium: Effect size of family and school indexes in writing competence using TERCE data (C. Pardo, A. Atorressi, Winsteps), Bariloche Argentina. Carlos Pardo, Universidad Catòlica de Colombia
Aug. 7-9, 2017, Mon-Wed. PROMS 2017: Pacific Rim Objective Measurement Symposium, Sabah, Borneo, Malaysia, proms.promsociety.org/2017/
Aug. 10, 2017, Thurs. In-person Winsteps Training Workshop (M. Linacre, Winsteps), Sydney, Australia. www.winsteps.com/sydneyws.htm
Aug. 11 - Sept. 8, 2017, Fri.-Fri. On-line workshop: Many-Facet Rasch Measurement (E. Smith, Facets), www.statistics.com
Aug. 18-21, 2017, Fri.-Mon. IACAT 2017: International Association for Computerized Adaptive Testing, Niigata, Japan, iacat.org
Sept. 15-16, 2017, Fri.-Sat. IOMC 2017: International Outcome Measurement Conference, Chicago, jampress.org/iomc2017.htm
Oct. 13 - Nov. 10, 2017, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com
Jan. 5 - Feb. 2, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com
Jan. 10-16, 2018, Wed.-Tues. In-person workshop: Advanced Course in Rasch Measurement Theory and the application of RUMM2030, Perth, Australia (D. Andrich), Announcement
Jan. 17-19, 2018, Wed.-Fri. Rasch Conference: Seventh International Conference on Probabilistic Models for Measurement, Matilda Bay Club, Perth, Australia, Website
April 13-17, 2018, Fri.-Tues. AERA, New York, NY, www.aera.net
May 25 - June 22, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com
June 29 - July 27, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Further Topics (E. Smith, Winsteps), www.statistics.com
Aug. 10 - Sept. 7, 2018, Fri.-Fri. On-line workshop: Many-Facet Rasch Measurement (E. Smith, Facets), www.statistics.com
Oct. 12 - Nov. 9, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps), www.statistics.com