The percentiles of this distribution range between a potential minimum value of 0 (if there are no observed frequencies at 0), and ALWAYS 100.
This distribution is what you find if you computed a standard frequency distribution for a set of scores in say SPSS, STATISTICA, or another stats package. Stanscore now computes the cumulative frequency distribution from the observed frequencies at each integer score, assuming an integer score is a discrete entity (no continuity assumption). I tried to make some sense of the competing definitions in one of my Technical Whitepapers entitled "Percentiles and Percentile Ranks - confused or what?" Technical whitepaper #3. It is factually incorrect to state the percentile for this score is the value below which n% of the norm-group score, because half the people scoring the maximum observed score are included in the cumulative proportion. So for the maximum possible score, only half the observed frequencies are used to express the final percentile. The "continuous" assumption calculation only uses half the frequencies at an observed score. There is no continuity there are just the numbers 0 to some maximum-possible score, with the intervals between scores as an indivisible single number of 1. The problem is this is just plain wrong for integer-scored data where percentiles need to be computed for each integer score of a psychological test. Another way of putting this is to say person X with percentile rank of say 80 scores higher than 80% of the group. The upshot of this is that the percentiles created using this formula are usually (but not always) interpreted as the values "BELOW WHICH" n% of individuals score. This used a formula that assumes a score is actually a mid-point between two integer scores. Stanscore version 3, in line with many authors, computed percentile ranks assuming continuous scores. StanScore calculates standardised scores in an on-screen, scrollable, detailed score distribution table (which can be can saved as an Excel spreadsheet).It also provides summary perecentile, stanine and sten look-up tables that are displayed on-screen, can be printed in tabular form, and can be saved to an Excel file for other uses or re-formatting for presentation-documentation purposes.
Missing Data are recognized for each variable separately by the program - recognizing a -9999 code, a null field (two commas together in a comma-separated "csv" or "txt" file), or the usual blank entries in SPSS, STATISTICA, Excel, or ACCESS. can be easily exported into say an Excel file, then read straight into Stanscore "as is" for processing. This means that data contained in such statistical packages as SPSS and STATISTICA etc.
#Spss code for scoring free#
mdb database tables, and standard comma-separated free text files. The program accepts data in several convenient formats such as from Microsoft Excel. StanScore is a simple but useful utility program that converts one or more sets of raw test scores into normalized-standardized z, stens, stanines, T-scores, and percentiles.