1. P-value calculations using JMP
2. Quiz

P-value calculations in JMP

First lets show how to find the area under a normal or a Chi-squared frequency curve.

1. start up JMP.
2. In the first column, enter values for standardized scores you are interested in.
3. To create a column of the probability of a value less than Z, as in Table 8.1, double click on the area for column 2; this should bring up the "column info" window.
4. Select "Formula", and then click on OK. This will bring up the calculator.
5. Go to the scrollbar menu in the middle and select "probability". To get the probability of a value less than Z, select "Normal Distribution" at the top of the scroll bar menu on the right. Finally, select Column1 from the left window to use as the argument of the normDist function. In the formula window at the bottom it should say, normDist(Column 1). Click "evaluate", then close the window. If all went well, you will have values for the probability of a value less than your standardize score.
6. If you change the values in column one, then you will need to click in the row in column 2 (or the area at the top of the column) to have the value updated. This can be used to fill in for values not in Table 8.1.
7. To get the area for the Chi-squared distribution, select Chi-Square Distribution in stead of normal in step 5. The formula will have two boxes for arguments. The first should the column with chi-squared statistics (i.e. enter in column 3). The second value is a value for the degrees of freedom. For a two by two table, this is 1. Click on the second box, type the number 1, then click on "constant" in the calculator pad. The formula window should show "chi-square Dist(Column 3, 1 DF, centered at 0). Click on evaluate, then close the calculator window. (Try this for a chi-squared value of 3.84 - did you get 0.95? This returns the value for the probability of getting a value less than the chisquared value. the p-value is one minus this. <\ol>

   For a standardized score or test statistic that has a normal frequency distribution, use the following steps to get the p-value. Remember the values that are "extreme" are those that are unlikely under the null hypothesis, but in the direction of the alternative hypothesis.

   1. If the alternative hypothesis is that the mean is not equal to zero, find the probability that a normal score is less than -|Z|, and then multiply by 2 to obtain the p-value. Z = (observed value - mean under the null hypothesis)/standard error.
   2. If the alternative hypothesis is that the mean is greater than 0, then the p-value is the probability that a normal score is greater than Z. (this is 1 minue the values returned by the calculator)
   3. if the alternative hypothesis is that the mean is less than 0, then the p-value is the probability that a normal score is less than Z. This is exactly what the calculator returns.