- Sequential probability ratio test
The

**sequential probability ratio test**(SPRT) is a specific sequential hypothesis test, developed byAbraham Wald . [*cite journal*] Neyman and Pearson's 1933 result inspired Wald to reformulate it as a sequential analysis problem. The Neyman-Pearson lemma, by contrast, offers a

first=Abraham

last=Wald

title=Sequential Tests of Statistical Hypotheses

journal=Annals of Mathematical Statistics

volume=16

issue=2

date=June, 1945

pages=117–186

url=http://links.jstor.org/sici?sici=0003-4851%28194506%2916%3A2%3C117%3ASTOSH%3E2.0.CO%3B2-7

doi=10.1214/aoms/1177731118rule of thumb for when the all the data is collected (and its likelihood ratio known).While originally developed for use in

quality control studies in the realm of manufacturing, SPRT has been formulated for use in the computerized testing of human examinees as a termination criterion. [*Ferguson, Richard L. (1969). [*] [*http://eric.ed.gov/ERICWebPortal/custom/portlets/recordDetails/detailmini.jsp?_nfpb=true&_&ERICExtSearch_SearchValue_0=ED034406&ERICExtSearch_SearchType_0=no&accno=ED034406 The development, implementation, and evaluation of a computer-assisted branched test for a program of individually prescribed instruction*] . Unpublished doctoral dissertation, University of Pittsburgh.*Reckase, M. D. (1983). A procedure for decision making using tailored testing. In D. J. Weiss (Ed.), New horizons in testing: Latent trait theory and computerized adaptive testing (pp. 237-254). New York: Academic Press.*] cite journal

author = Eggen, T. J. H. M.

year = 1999

title = Item Selection in Adaptive Testing with the Sequential Probability Ratio Test

journal = Applied Psychological Measurement

volume = 23

issue = 3

pages = 249-261

doi = 10.1177/01466219922031365]**Theory**As in classical

hypothesis testing , SPRT starts with a pair of hypotheses, say $H\_0$ and $H\_1$ for thenull hypothesis andalternative hypothesis respectively. They must be specified as follows::$H\_0:\; p=p\_0$:$H\_1:\; p=p\_1$

The next step is calculate the cumulative sum of the log-

likelihood ratio , $Lambda\_i$, as new data arrives::$S\_i=S\_\{i-1\}+\; log\; Lambda\_i$

The

stopping rule is a simple thresholding scheme:* $a\; <\; S\_i\; <\; b$: continue monitoring ("critical inequality")

* $S\_i\; geq\; b$: Accept $H\_1$

* $S\_i\; leq\; a$: Accept $H\_0$where a and b ($0math>)\; depend\; on\; the\; desiredtype\; I\; and\; type\; II\; errors,$ alpha$and$ eta$.\; They\; may\; be\; chosen\; as\; follows:$

$a\; approx\; log\; frac\{\; eta\; \}\{1-alpha\}$ and $b\; approx\; log\; frac\{1-eta\}\{alpha\}$

In other words, $alpha$ and $eta$ must be decided beforehand in order to set the thresholds appropriately. The numerical value will depend on the application. The reason for using approximation signs is that, in the discrete case, the signal may cross the threshold between samples. Thus, depending on the penalty of making an error and the

sampling frequency , one might set the thresholds more aggressively. Of course, the exact bounds may be used in the continuous case.**Example**A textbook example is

parameter estimation of aprobability distribution function . Let us consider theexponential distribution ::$f\_\; heta(x)=\; heta^\{-1\}expleft(-x/\; heta\; ight),\; x,\; heta>0$

The hypotheses are simply $H\_0:\; heta=\; heta\_0$ and $H\_1:\; heta=\; heta\_1$, with $heta\_1>\; heta\_0$. Then the log-likelihood function (LLF) for one sample is

:$egin\{align\}log\; Lambda(x)=log\; left\; [\; frac\{\; heta\_1^\{-1\}expleft(-x/\; heta\_1\; ight)\}\{\; heta\_0^\{-1\}expleft(-x/\; heta\_0\; ight)\}\; ight]\; \backslash =log\; left\; [\; frac\{\; heta\_0\}\{\; heta\_1\}\; exp\; left(x/\; heta\_0\; -\; x/\; heta\_1\; ight)\; ight]\; \backslash =frac\{\; heta\_1-\; heta\_0\}\{\; heta\_0\; heta\_1\}\; x\; -\; log\; frac\{\; heta\_1\}\{\; heta\_0\}end\{align\}$

The cumulative sum of the LLFs for all x is

:$S\_n=sum\_\{i=1\}^n\; log\; Lambda(x\_i)=frac\{\; heta\_1-\; heta\_0\}\{\; heta\_0\; heta\_1\}\; sum\_\{i=1\}^n\; x\_i\; -\; n\; log\; frac\{\; heta\_1\}\{\; heta\_0\}$

Accordingly, the

stopping rule is:$b\{\; heta\_1-\; heta\_0\}\{\; heta\_0\; heta\_1\}\; sum\_\{i="1\}^n"\; x\_i\; -\; n\; log\; frac\{\; heta\_1\}\{\; heta\_0\}math>$After re-arranging we finally find:$b+n\; log\; frac\{\; heta\_1\}\{\; heta\_0\}\; <\; sum\_\{i=1\}^n\; x\_i\; <\; a+n\; log\; frac\{\; heta\_1\}\{\; heta\_0\}$

The thresholds are simply two

parallel lines withslope $log\; (\; heta\_1/\; heta\_0\; )$. Sampling should stop when the sum of the samples makes an excursion outside the "continue-sampling region".**Applications****Manufacturing**The test is done on the proportion metric, and tests that a variable "p" is equal to one of two desired points, "p

_{1}" or "p_{2}". The region between these two points is known as the "indifference region" (IR). For example, suppose you are performing a quality control study on a factory lot of widgets. Management would like the lot to have 3% or less defective widgets, but 1% or less is the ideal lot that would pass with flying colors. In this example, "p_{1}= 0.01" and "p_{2}= 0.03" and the region between them is the IR because management considers these lots to be marginal and is OK with them being classified either way. Widgets would be sampled one at a time from the lot (sequential analysis) until the test determines, within an acceptable error level, that the lot is ideal or should be rejected.**Testing of human examinees**The SPRT is currently the predominant method of classifying examinees in a variable-length

computerized classification test (CCT). The two parameters are "p_{1}" and "p_{2}" are specified by determining a cutscore (threshold) for examinees on the proportion correct metric, and selecting a point above and below that cutscore. For instance, suppose the cutscore is set at 70% for a test. We could select "p_{1}= 0.65" and "p_{2}= 0.75" . The test then evaluates the likelihood that an examinee's true score on that metric is equal to one of those two points. If the examinee is determined to be at 75%, they pass, and they fail if they are determined to be at 65%.These points are not specified completely arbitrarily. A cutscore should always be set with a legally defensible method, such as a modified Angoff procedure. Again, the indifference region represents the region of scores that the test designer is OK with going either way (pass or fail). The upper parameter "p

_{2}" is conceptually the highest level that the test designer is willing to accept for a Fail (because everyone below it has a good chance of failing), and the lower parameter "p_{1}" is the lowest level that the test designer is willing to accept for a pass (because everyone above it has a decent chance of passing). While this definition may seem to be a relatively small burden, consider the high-stakes case of a licensing test for medical doctors: at just what point should we consider somebody to be at one of these two levels?While the SPRT was first applied to testing in the days of

classical test theory , as is applied in the previous paragraph , Reckase (1983) suggested thatitem response theory be used to determine the "p_{1}" and "p_{2}" parameters. The cutscore and indifference region are defined on the latent ability (theta) metric, and translated onto the proportion metric for computation. Research on CCT since then has applied this methodology for several reasons:#Large item banks tend to be calibrated with IRT

#This allows more accurate specification of the parameters

#By using the item response function for each item, the parameters are easily allowed to vary between items.**ee also***

CUSUM

*Computerized classification test

*Wald test

*Likelihood-ratio test **References**

*Wikimedia Foundation.
2010.*