18

Dec 10

Stats Helper v0.6 Released

Problem:

Doing statistical analysis isn’t always as simple as it should be. While products like JMP and Minitab are extremely powerful and extremely useful, they are also more complex than is often needed. JMP specifically has a hard time dealing with summarized data, while Minitab can be a maze of menus and options. Quite often, I find myself just needing to run a quick hypothesis test, or needing to quickly determine a confidence interval for a proportion. While many people end up using the more complex tests that a full stats package offers eventually, many people never need more than the most basic features, or worse yet misuse the advanced features, thinking they are doing something simple. I once helped someone who, having only a very mild stats background, nevertheless wanted to run a hypothesis test, and had mistakenly chosen to run a Kruskal-Wallis test in Minitab. He knew he wanted a p-value, and Kruskal-Wallis gave a p-value. Luckily in this case, the answer ended up being nearly identical to a 2 Sample t-Test, but things could have gotten ugly.

Another issue is the cost. Most people won’t want to justify spending over $1,000 on stats software that they won’t fully use.

Basically, it seems like many people who could benefit from simple statistical analysis aren’t doing so, because the software available is either difficult to use, very expensive, some combination of both. Business professionals end up making decisions based on hunches because they don’t have the time to learn how to use powerful software, and making decisions based on hunches is a scary proposition.

Solve:

There’s got to be a simpler way. What many people need is just a simple tool to run a few quick tests. This tool should be require minimal knowledge of stats to run, give understandable results, and be extremely inexpensive. Excel is a viable platform, since most people, both students and business professionals, have access to it. Being compatible with OpenOffice would be a boon, as well.

There are numerous Excel stats packages, but many of them still cost hundreds of dollars, and approach the same level of complexity or specialization as Minitab or JMP. Excel itself comes with the Analysis Tookpak, but many people shy away from using it because it can give wrong answers. Even if it does give the right answers, interpreting the results is cumbersome and involves a steep learning curve.

Create:

As part of learning Six Sigma on the Master Black Belt level, I made a spreadsheet to do all the cool cool analysis I was learning. I found myself using it often at work and school, and a few people mentioned that I should sell it. Realizing that I shouldn’t argue with people who say I have something they want to buy, I decided to clean up the tool and offer it for sale. So I cleaned up the interface, added a few features, pulled out most of the advanced ones that would require an instruction manual, and figured out a way to sell it online via Paypal. For now, I’m calling it Stats Helper.

Here’s what it does, for now, divided into 4 sections:

Samples:

All you have to enter are the mean, standard deviation, and sample size for 2 populations, as well as the desired confidence level. The software then does some calculations and draws some graphs:

  • Each distribution entered in is automatically graphed, with dynamically updating scales.
  • Hypothesis test: Stats Helper automatically switches between using a Z- or t-Distribution, depending on sample size, and calculates the P-Value for a 2 tailed hypothesis test. It then offers a Plain English interpretation.
  • Confidence Intervals: Stats Helper automatically calculates the confidence intervals for the distribution(s) entered, again automatically selected the Z- or t-Distribution, based on the number of samples. These confidence intervals are also graphed automatically, so that you can quickly tell where the mean may lie in the population.
  • Confidence Interval for the Difference between Means: Stats helper also calculates the interval for the difference between the means of the 2 distributions entered, and offers a Plain English interpretation.

Proportions:

Again keeping it simple, all you have to enter are the number of trials (sample size) and the number of events, as well as the desired confidence level.

  • As with the samples section, each distribution entered in is automatically graphed, with dynamically updating scales.
  • Hypothesis test: Stats Helper automatically calculates the P-Value for a 2 tailed hypothesis test. It then offers a Plain English interpretation of the test.
  • Confidence Intervals: Stats Helper automatically calculates and graphs the confidence intervals of the means for the distribution(s) entered.
  • Binomial Probabilities: Stats Helper calculates a translated confidence interval of how many events would be expected within the confidence level specified.
  • Confidence Interval for the Difference between Means: Stats helper also calculates the interval for the difference between the means of the 2 distributions entered, and offers a Plain English interpretation.

Process Capability:

This section offers both a way to calculate Cpk and a Capability Translation calculator.

  • Capability Calculations: By inputing a few variables, as well as indicating whether you are entering short-term or long-term data, Stats Helper calculates Cp, Cpk, Pp, and Ppk, as well as calculates a confidence interval for the Cpk.
  • Capability Translations: By entering one known value, Stats Helper displays all the rest, be it Cpk, Defects per Million Opportunities (DPMO), Yields, or a Z score.

Power and Sample Size:

This section offers the ability to enter a few variables, and Stats helper calculates the rest.

  • Samples: Enter 3 of the following (confidence level, number of samples, standard deviation, difference to detect), and Stats Helper calculates the missing value.
  • Proportions: Enter a proportion and 2 of the following (confidence level, number of samples, difference to detect), and Stats Helper calculates the missing value.

Share:

Visit the software page to purchase and download your copy of Stats Helper.

6 Comments

  1. pharmacist says:

    Terrific work! This is the type of information that should be shared around the web. Shame on the search engines for not positioning this post higher!

  2. candex says:

    This is good work but rather simplistic.
    Wondering if you could point the way for a rather complex task of predicting pseudo random numbers as per list given below:
    40 99528
    39 305472
    38 797878
    37 651086
    36 860502
    35 695977
    34 115557
    33 900774
    32 422601
    31 869860
    30 700780
    29 495417
    28 155006
    27 865922
    26 279684
    25 143421
    24 607963
    23 879479
    22 606670
    21 705328
    20 692795
    19 210706
    18 530290
    17 628188
    16 552176
    15 327378
    14 443946
    13 323589
    12 595531
    11 308266
    10 759282
    9 101483
    8 73183
    7 491067
    6 390241
    5 203769
    4 559700
    3 271731
    2 452093
    1 614211
    What will be for 41,42,43,44,45….

    1. Jed Campbell says:

      This looks interesting. Give me some time to think of an approach. Are you certain #40 is correct? At first glance, it appears out of place.

  3. J Wilson says:

    I purchased your stat helper program on 8-10-11, and downloaded the zip file. When I enter the serial number, the file closes and disappears. Any suggestions. I downloaded the file twice. Same problem. Running XP and Excel 2000 with data analysis.

    1. Jed Campbell says:

      I’ll email you all the files that are in the archive. This problem has happened once before, so I’ll look into trying to recreate it in order to fix what’s causing the problem. Sorry for the wait…

  4. I have been reading your blogs problem.solve.create.share » Stats Helper v0.6 Released and they are pretty useful for me to know something new and special. And You can also visit our web also all hyips monitor.

Leave a Reply