Probability Calculator

Here is a program with a very limited audience (researchers, statistics students). It replaces some of the more common statistical tables. It will compute the p value (probability of a more extreme result) for the Z, t, F, and chi square distributions. Or, for a given alpha level, it will calculate the corresponding critical value of the statistic. For the Z and t distributions, you have your choice of one tailed or two tailed tests. For the t and F distributions it will also report the corresponding value of Eta. Eta is a measure of effect size - the nonlinear correlation between the independent and dependent variables. Eta squared can be interpreted as the proportion of explained variance.

For me, it is quite nice to have these functions available on my touch phone (Samsung Omnia). Authors frequently neglect to report effect size but I can just whip out my phone, plug in the t or F value, and see for myself. Useful for improving accracy in interpreting research results.

I now return you to the real world . . .

Some image file are missing.

Sorry about that. I did not include the graphics because they are embedded in the .exe file. I should have realized that they would be useful if someone wanted to run the source code. I have attached a new zip file that should have everything.

By the way, I just noticed the same .exe file runs on both my desktop and on my windows mobile devices. That is really nice!

Thanks.
As long as your program doesn't use any device specific keywords (liks SIP) or libraries, the executable will run on the desktop as well.

A small clarification for those pedants amongst us

Erel said:

Thanks.
As long as your program doesn't use any device specific keywords (liks SIP) or device specific libraries, the executable will run on the desktop as well.

Click to expand...

Interesting... So you say that the phrase 'device specific' in a sentence like 'device specific keywords or libraries' doesn't apply to the second term?

Erel said:

doesn't apply to the second term?

Click to expand...

Not necessarily, it is an ambiguous construction that can mean either as there is no explicit way in English of grouping the second term with the first. I've consulted my partner, she's more of a grammarian than I, and she suggests the less (but still) ambiguous phrase "keywords or libraries specific to the device".

Interestingly "libraries or keywords specific to the device" is more ambiguous than "keywords or libraries specific to the device" as "libraries" in the first form could imply "any libraries" but "keywords" in the second form is unlikely to mean "any keywords" and so will be grouped in the readers mind with "specific ...".

I don't know about other languages but lawyers and legislators use specific forms of English, usually without punctuation, in order to try to avoid the ambiguities that abound in the English language in order to specify absolutely precise meaning.

She went for a bath, had a think while relaxing and came up with "keywords or libraries, either of which are specific to the device"

:sign0013::sign0006:

"keywords or libraries, either of which are specific to the device"

Click to expand...

It is too long. I tried to avoid "device specific keywords or device specific libraries", which is shorter.

I'll settle on:

Erel said:

I'll settle on:

B4X:

If device_specific AND [B]([/B]keywords OR libraries[B])[/B] Then
   will also work on desktop
End If

Click to expand...

Shouldn't that be


Optimized compilation doesn't like spaces in variable names :sign0100:

How can this probability calculator be used specifically? What are the real world applications and who would use them exactly? Is it just for students?

marathon332 said:

How can this probability calculator be used specifically? What are the real world applications and who would use them exactly? Is it just for students?

Click to expand...

To be totally truthful, I've not even down loaded it.
I have now, and will respond tomorrow evening :sign0148:

ProbCalc deals with four common statistical distributions: Z, t, F, and chi square. These distributions are used for statistical inference, usually within the context of experimental or applied research. ProbCalc can be used in lieu of tables printed in the back of statistical text books.

In null hypothesis testing, an experiment is conducted usually with at least one experimental group and one control group. For example, one group of people may be given a "brain booster" drug, and another group given a placebo. Then an IQ test is administered to see if the drug had any effect. Rarely will two groups get exactly the same results even when the experimental treatment had no effect. Statistics can be computed to determine the liklihood of the observed outcome under the presumption that the experimental treatment had no effect. ProbCalc computes this probability.

A probability calculator has the advantage (over tables) of being able to calculate precise p (pobability) values for any given experimental (i.e., statistical) result. There are more comprehensive probability calculators available for Windows based computers but this is the first (I believe) for a PPC.

ProbCalc also computes the degree of relationship between the independent and dependent variables in the experiment (for t and F tests). For the example, this would be the relationship between the drug and the outcome on the IQ test. When squared, Eta (the relationship statistic computed by ProbCalc) can be interpreted as the percentage of explained variance. So if, for example, Eta = .40 for the current example, you could concluded that 16% of the observed variance in IQ scores could be explained by the drug.

The advantage of having a portable version available is that one can use it in the library when reading the research literature (yeah, I know the library is not the real world . . .). Many authors do not report effect size (relationships) and only report whether the experiment was statistically significant (P<.05, for example). I have seen MANY published research reports with very strong conclusions that were based on research with a very small Eta. In other words, an experiment can be statistically significant without being practically significant. ProbCalc is an extremely useful tool for putting published research into a real word context -- and it takes just seconds.

Hope I didn't bore you too much.

Thanks for that detailed explanation....