Search results

  1. K

    Determining temperature variance and sampling rate

    Can we make things a little clearer please. What would the answer to question 1 look like? It seems that you want something like "after sampling 10 (say) blast sites of 200 holes, we can be 95% sure that 95% of temperatures will lie between 86 and 94 degrees." Is this what you are after? If...
  2. K

    Generalisation of 'overlapping confidence intervals'

    There is a derivation of the 83% here. https://chris-said.io/2014/12/01/independent-t-tests-and-the-83-confidence-interval-a-useful-trick-for-eyeballing-your-data/ Informally, significance starts at about half an error bar overlap. I've used the attached spreadsheet at workshops to illustrate.
  3. K

    Using a probabilistic approach to determine missing information

    I can't see it. You want to know how often Rotarians and video gamers interact within their own groups when all you know is how often the two groups interact.
  4. K

    What kind of distribution to model new business success

    Here is something you can experiment with. The graph shows the BLS data you mentioned in your first post - the first lot from 1994. The x axis is years from 1994 and the black line is 1/(1+0.2*years). kat
  5. K

    What kind of distribution to model new business success

    The data looks somewhat like exponential decay, but if you want a distribution for say Monte Carlo risk analysis, you can always make a piece-wise function using the data you have.
  6. K

    Likert item analysis

    The common p < 0.05 cutoff for significance acknowledges the possibility that there is no real difference anywhere but just by bad luck your sample looks as if there really is a difference. The tests and cutoffs are designed so that you will claim there is a real difference when there isn't one...
  7. K

    Which T-Test tails/types do I use in Excel?

    The t test is used to compare two sets of numbers that could be expected to be the same if there was nothing special going on. There is no way you could expect the average of the hours to be the same as the average of the marks, so a t test is completely the wrong approach. Google correlation...
  8. K

    Which T-Test tails/types do I use in Excel?

    It sounds like you probably need correlations rather than t tests. Correlation answers questions like "Do subjects with higher (or lower) motivation scores tend to have higher (or lower) LMS use scores?" Is this the sort of thing you want?
  9. K

    Which T-Test tails/types do I use in Excel?

    What is the question you are trying to answer? The t tests are seeing if there is a difference in the mean scores of two sets of marks. It looks as if the AssessMark column is not comparable with the other columns.
  10. K

    Rare event probability

    This may be useful. There is a statistical rule of thumb called the rule of three which says that if you have N successes without a failure, you can be 95% sure that the error rate is less than 1 in N/3. So, if you observe say 600 cars without a miss, you can be 95% sure that the error rate is...
  11. K

    How many times do we have to flip a coin to have it being as close to its true mean of 50%?

    The actual formula is MoE = z*sqrt(p(1-p)/n) (just google ci for proportion). For 95% z = 1.96 or near enough to 2. p is about 0.5 so MoE = 2*sqrt(.5(1-.5)/n) = 2*sqrt(.5(1-.5))/sqrt(n) = 1/sqrt(n) as a simple rule of thumb.
  12. K

    Likert item analysis

    Let's just look at one typical test you are planning? Say Experience and Factor1. It looks like you have split the subjects into several Experience groups. The question is "do at least two of the various groups score noticeably differently on Factor1?" One approach is certainly Kruskal-Wallis...
  13. K

    How many times do we have to flip a coin to have it being as close to its true mean of 50%?

    You can get a rough idea of the accuracy using MoE = 1/sqrt(n) for midrange proportions. So, in round numbers, a MoE of 1% for a coin needs about 10000 flips. The MoE often quoted on political polls is 3% from 1000 people. 1/sqrt(1000) is about 3%.
  14. K

    Finding P-Values For a Null Hypothesis with Neutral Trend

    You can also load the standard Excel Analysis ToolPak and do the regression from it. The p value is sitting there either in the anova table or the summary table. Plenty of examples on the net.
  15. K

    Confidence interval for fuction of random variables

    So long as the samples are not to small, margin of error MOE(A+B+C)/3) = (sqrt(MOE(A)^2+MOE(B)^2+MOE(B)^2))/3 and you can get the MOEs from the CIs.
  16. K

    Nonparametrics Help

    I think we are all interested in helping. Perhaps it would help our understanding if you made up a mock data set and posted it here.. We could try various things.
  17. K

    Unequal Sample Size

    You could find a confidence interval for the employee mean and see if the CEO lies outside it - effectively a one sample t test. Wjth 42 employees there is enough data for a one sample test. However, with 8 companies and 3 styles there are 24 tests so even if there was no real difference...
  18. K

    I am having trouble getting a normal bell curve out.... - Any help is appreciated. Will pay

    It depends on what the fitted normal was going to be used for. If you wanted a confidence interval for the mean, for instance, you could do a bootstrap CI using the actual data. You could compare one area or period with another using a permutation test. If you wanted to do a risk analysis which...
  19. K

    I am having trouble getting a normal bell curve out.... - Any help is appreciated. Will pay

    Just thinking. Poisson isn't the best because with a mean of around 1 or 2, you should not see anything above about 6. In the meantime, you can use the data itself as its own probability distribution and use a Monte Carlo technique. You could also consider zero-truncated negative binomial but...