Observed/Expected Ratios and Exact Confidence Intervals [sg29: STB-23] ------------------------------------------------------- ^smr^ obsvar expvar [^if^ exp] [^in^ range] [^,^ ^icd(^varname^)^ ^l^evel^(^#^)^ ^r^owlab^(^varname^)^ ^s^umonly ^t^otal ] -- or -- ^smri^ #o #e [^,^ ^l^evel^(^#^)^ ] Description ----------- For each observation, ^smr^ computes the ratio of the observed to the expected count (^obsvar^/^expvar^) together with a confidence interval based on assuming that ^obsvar^ is a Poisson count and the expected value is known without error. A one-sided confidence-interval is provided if the observed count is zero. Ratios that are significant are marked with one asterisk for p-values less than .05, two asterisks for p-values less than .01, and three asterisks for p-values less then .001. None of the p-values and confidence intervals are adjusted for the implicit multiple comparisons. P-values are twice the one- sided Poisson probabilities of observing a more extreme count. ^smri^ is the immediate form of ^smr^; see [4] immediate. Options ------- ^level^ specifies the significance level in percent for the confidence interval. ^icd(^varname^)^ is a synonym for ^rowlab()^. ^rowlab(^varname^)^ is used to label the rows of the table. ^varname^ contains row labels for the observed and expected counts. It can be a numeric or a string variable. If ^varname^ contains floating point numbers, it it is advisable to read and store the variable as a ^double^ rather than a ^float^ to avoid unattractive display. ^rowlab^ is ignored if ^icd^ is specified. ^sumonly^ specifies that only the totals should be calculated and displayed. ^total^ specifies that the total observed count together with its expectation, the ratio O/E, and its confidence interval should be calculated and displayed in addition to the usual output. Author ------ Peter Sasieni Imperial Cancer Research Fund, London FAX 011-44-171-269-3429 Also see -------- STB: sg29 (STB-23)