INDEX ​

Power Analysis

Statistics for Data Analysis 28 includes an enhancement to the Power Analysis procedures to create a graphical representation of results. Power analysis plays a pivotal role in study plan, design, and conduction. The calculation of power is usually before any sample data have been collected, except possibly from a small pilot study. The precise estimation of the power may tell investigators how likely it is that a statistically significant difference will be detected based on a finite sample size under a true alternative hypothesis.

All of the Power Analysis procedures now allow the user to enter a series of power values rather than a single one. These can take the form of separate (point) values, a series of values incremented by a specific amount, or a combination of both. This enhancement now results in a table of power values and, optionally, an accompanying power curve graph.

​

Ratio Statistics

The Ratio Statistics procedure in Statistics for Data Analysishas long been a convenient option for obtaining descriptive statistics on ratios of two variables, as well as the go-to procedure for property assessment professionals, whose work involves working with ratios of property assessment values to sales prices. In the time since the original procedure was implemented, changes in thinking about the use of some statistics in property valuation have occurred. The Ratio Statistics procedure has been enhanced to make it more consistent with current approaches, as described in the Standard on Ratio Studies document published by the International Association of Assessing Officers.

The most important change to the procedure is the addition of the coefficient of price-related bias (PRB) and an associated confidence interval. The PRB is an index of whether assessment to price ratios are systematically higher or lower for higher-priced properties. It is calculated by regressing percentage differences in assessment ratios from the median ratio on the base 2 logarithms of value proxy measures calculated as the averages of sales prices and the ratios of assessed values to the median ratio, and gives the percentage change in assessment ratios for a 100 percent change in value. This newly available measure is selected by default in the Statistics dialog when using the graphical user interface to the procedure, which is accessed via Analyze > Descriptive Statistics > Ratio.

Other changes include removing the coefficient of variation based on the median ratio from the Statistics dialog options, relabeling the remaining coefficient of variation option based on the mean ratio as COV, as is common usage, and adding the option in the main Ratio Statistics dialog to include the number of cases or properties on which statistics are based in the output table, instead of only offering that in a saved output file.

​

Other procedural enhancements

1. A new graphical user interface control was designed for GLM and UNIANOVA. Specifically, an additional Compare simple main effects checkbox was added to the EM Means… sub-dialog to support the specification of comparisons among the main effects nested within the levels of other factors.

2. Adjustments were made to Change Contrast in the LOGISTIC REGRESSION and COXREG dialogs:

​​

a. The default reference category specification for contrast types that use reference category specifications (Indicator, Simple, and Deviation) was changed from Last to First.

​

b. The positions of the Last and First buttons were switched.

 

c.The listing of variables in the Categorical Covariates target list now includes (last).

​

  3. In prior versions users were able to run a two-sided T-Test, but additional calculations were necessary to get output for a one-sided
      T-Test. Users often were left to reference an IBM Tech Note to get the results they required for a one-sided T-Test. Starting with V28
      Beta, users will be able to run a one-sided T-Test through the dialogs.

​