Re: Results Scaling
- To: mathgroup at smc.vnet.net
- Subject: [mg19617] Re: [mg19580] Results Scaling
- From: BobHanlon at aol.com
- Date: Sun, 5 Sep 1999 16:57:40 -0400
- Sender: owner-wri-mathgroup at wolfram.com
Anthony,
One approach is to use T-Scores which are normalized to a mean of 50 and a
standard deviation of 10. A T-Score readily gives the deviation from the
mean in standard deviations. For example, a T-Score of 65 is 1.5 standard
deviations above the mean.
Bob Hanlon
____________________
Needs["Statistics`ContinuousDistributions`"]
Needs["Statistics`DataManipulation`"];
Needs["Graphics`Graphics`"];
Needs["Utilities`FilterOptions`"];
Options[plotContDistData] = {nbrBins -> 12};
plotContDistData::usage =
"plotContDistData[dist, data] overlays the PDF for the specified \
continuous distribution over a bar chart of the data list.";
plotContDistData[dist_, data_List, opts___?OptionQ] :=
Module[{mu, sigma, xmin, xmax, pltPDF, x, nbrVal = Length[data], step, k,
pltData, nBins, pltOpts},
nBins = (nbrBins /. Flatten[{opts}]) /. Options[plotContDistData];
pltOpts = FilterOptions[Plot, opts];
mu = N[Mean[dist]];
sigma = N[StandardDeviation[dist]];
xmin = Max[mu - 3sigma, Domain[dist][[1, 1]]];
xmax = Min[mu + 3sigma, Domain[dist][[1, 2]]];
pltPDF =
Plot[PDF[dist, x], {x, xmin, xmax},
PlotStyle -> AbsoluteThickness[2], DisplayFunction -> Identity,
Evaluate[pltOpts]];
step = (xmax - xmin)/nBins;
pltData =
GeneralizedBarChart[
Transpose[{Table[xmin + step(k - 1/2), {k, nBins}],
BinCounts[data, {xmin, xmax, step}]/(step*nbrVal),
Table[step, {nBins}]}], DisplayFunction -> Identity,
Evaluate[pltOpts]];
Show[{pltData, pltPDF}, DisplayFunction -> $DisplayFunction]];
dist = NormalDistribution[70, 5];
scores = RandomArray[dist, 200];
{mu = Mean[scores], sigma = StandardDeviation[scores]}
{70.1659, 4.61972}
plotContDistData[dist, scores];
The T-Scores are normalized to a mean of 50 and a standard deviation of 10.
normalizedScores = 50 + 10(scores - mu)/sigma;
{Mean[normalizedScores], StandardDeviation[normalizedScores]}
{50., 10.}
plotContDistData[NormalDistribution[50, 10], normalizedScores];
In a message dated 9/4/99 4:45:09 AM, antonyip at ihug.co.nz writes:
>I am currently developing a student results entry application.
>One of features of this application is to allow user to statistical scale
>students results.
>I am just wondering if any body here have similar experience before.
>I really want to know the statistical algorithm that can be used to scale
>student results.
>