From Yahoo Answers

**Question:**I need to calculate the p-value significance of a table of data and don't understand what I need to do/or how the df is involved? Can you help? I can't use an online calculator, plus I need to work out the initial sum/formula to get to the value bit. HELP! I have an exam tomorrow and need to be able to do this! I don't know how to work out the 'x' or the 'df'!

**Answers:**use microsoft excel use the chidist function it is listed under the category of statistical functions enter x and the degrees of freedom it will compute the p=value

**Question:**This would have to deal with research methods in the social science.

**Answers:**The null hypothesis for a chi square tests is like the null hypothesis for parametric methods, i.e., that the difference in he means is zero. However, rather than continuous numerical values that have normal distributions like IQ or shoe size, you're dealing with categorical variables that in the simplest sense have yes/no answers like male/female or gay/straight. So for a chi square you're talking about the means of frequencies (or percentages) rather than the means of values. Instead of a1-a2=0 (or a1=a2) as with e.g., a t-test, the null hypothesis would be a1b1=a1b2=a2b1=a2b2=25%. For example, if you asked your male (a1) and female (a2) subjects if they were gay (b1) or straight (b2), the null hypothesis would predict that half the subjects would be male and half of males and females would be straight, i.e., 25% in each box. Of course it doesn't have to be a "square"; you can use a 2 x 3 design and ask male and female subjects if they're gay, straight, or bisexual. In this case the null hypothesis would be that 16.667% of your subjects would fall into each of the six categories. Of course if you draw your subjects from the population at large, you will find that there are significantly more straights than gays or bi's, among both men and women, and you will reject the null hypothesis. Depending on your sampling methods, you may also end up rejecting the null hypothesis if you have a greater proportion of males than females or vice versa. Of course you're not limited to a 2 x 3 design either. With today's powerful stats software, there is no practical limit on the number of categories you can use (although there may be a real limit).

**Question:**P2= 2PQ= Q2= P= .756 + (.242) = .877 Q= (.242) + .002= .123 P2=(.877)2= .769 Q2=(.123)2= .051 2PQ= 2(.877 x .123) = .216 = .769 + .216 + .051=1.04 .887+.123= 1.01 X2=(o-e/e)2 X2=[(.756-.769)2 / .769 ] + [(.242-.216)2 / .216] + [(.002 -.051)2/ .051] = X2=[(-.013)2/.769] + [(.026)2 / . 216] + [(.-.049)2/ . 051]= X2=2.198x10-4 + .0031 + .0471= .0504198= .0504

**Answers:**You find your chi-square value and, using your degrees of freedom, read off the chi-square chart for the corresponding probability. If your probability is lower than your pre-determined alpha level, then you reject your null. If not, then you accept it.

From Youtube