https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514610#M2766 <P>Hi&nbsp;<a href="https://communities.sas.com/t5/user/viewprofilepage/user-id/130109">@chrissowden</a>,</P> <P>&nbsp;</P> <P>The risks in the two groups (row 1 and row 2 in the PROC FREQ cross tabulation)&nbsp;seem to be fairly similar: Assuming the true risks were equal to the average empirical risk in your sample (0.08 for column 1, see "Total" row in the&nbsp;upper table), one would expect about 0.08*538≈43 and 0.08*675=54 "cases" in row 1 and row 2, respectively. (I hope I guessed the numbers 538 and 675 correctly.) The slightly different observed numbers, 39 (i.e. 4 less) and 58 (4 more), can be due to chance. This is what I would conclude from the fact that zero is contained in the confidence interval for the risk difference: [-0.0439, 0.0170]. An interval around zero has necessarily a negative lower limit (and a positive upper limit). So, even though the&nbsp;observed relative frequency of "cases" in row 2 was slightly (by 1.34 percentage points) higher than in row 1, it cannot be ruled out that the <EM>true</EM> (column 1) risk in row 1 is higher. Your data don't provide enough evidence to decide this.</P> Mon, 19 Nov 2018 22:18:20 GMT FreelanceReinh 2018-11-19T22:18:20Z https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514565#M2761 <P><span class="lia-inline-image-display-wrapper lia-image-align-inline" image-alt="Capture.PNG" style="width: 600px;"><img src="https://communities.sas.com/t5/image/serverpage/image-id/25051i5C419E2A764F4E91/image-size/large?v=v2&amp;px=999" role="button" title="Capture.PNG" alt="Capture.PNG" /></span></P><P>I am trying to understand&nbsp; what a&nbsp; negative number in the Asym 95% in the&nbsp; Difference row.</P> Mon, 19 Nov 2018 20:25:04 GMT https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514565#M2761 chrissowden 2018-11-19T20:25:04Z https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514610#M2766 <P>Hi&nbsp;<a href="https://communities.sas.com/t5/user/viewprofilepage/user-id/130109">@chrissowden</a>,</P> <P>&nbsp;</P> <P>The risks in the two groups (row 1 and row 2 in the PROC FREQ cross tabulation)&nbsp;seem to be fairly similar: Assuming the true risks were equal to the average empirical risk in your sample (0.08 for column 1, see "Total" row in the&nbsp;upper table), one would expect about 0.08*538≈43 and 0.08*675=54 "cases" in row 1 and row 2, respectively. (I hope I guessed the numbers 538 and 675 correctly.) The slightly different observed numbers, 39 (i.e. 4 less) and 58 (4 more), can be due to chance. This is what I would conclude from the fact that zero is contained in the confidence interval for the risk difference: [-0.0439, 0.0170]. An interval around zero has necessarily a negative lower limit (and a positive upper limit). So, even though the&nbsp;observed relative frequency of "cases" in row 2 was slightly (by 1.34 percentage points) higher than in row 1, it cannot be ruled out that the <EM>true</EM> (column 1) risk in row 1 is higher. Your data don't provide enough evidence to decide this.</P> Mon, 19 Nov 2018 22:18:20 GMT https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514610#M2766 FreelanceReinh 2018-11-19T22:18:20Z https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514726#M2792 <P>That is Confidence Interval for diff of Risk, which contains ZERO , i.e. You can't reject H0: Risk in Row1=Risk in Row2</P> Tue, 20 Nov 2018 12:40:51 GMT https://communities.sas.com/t5/New-SAS-User/Negative-Difference-in-Risk-Tables/m-p/514726#M2792 Ksharp 2018-11-20T12:40:51Z