SAS Macro for Charting the Relative Costs of Reduced Factorial Designs

Downloads

(Available free to members. Register.)

SAS Macro:

Sample File 1:

Sample File 2:


	[X] Terms and Conditions of Use (c) 2012, The Pennsylvania State University. This program is free software; you can redistribute it and/or modify it under the terms described below. WARRANTY This software is provided "as is," and any express or implied warranties, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose, are expressly disclaimed. In no event shall Penn State University or its employees be liable for any direct, indirect, incidental, special, exemplary, or consequential damages (including, but not limited, the costs of procurement of substitute goods or services; loss of use, data or profits; or business interruption) however caused and on any theory of liability, whether in contract, strict liability, or tort (including negligence or otherwise) arising in any way out of the use of this software, even if advised in advance of the possibility of such damage. PUBLICATIONS Any peer-reviewed publication in which this software is used will include a citation of the software in the reference section. Sample citations in APA format will be available on The Methodology Center website for as long as the software is supported by the center. REDISTRIBUTION Redistribution and use of this software, with or without modification, are permitted, provided that the following conditions are met: 1. Any redistribution must include the above copyright notice and disclaimer and this list of conditions in any related documentation and, if feasible, in the redistributed software. 2. Any redistribution must include the acknowledgment, "This product includes software developed by Methodology Center, The Pennsylvania State University," in any related documentation and, if feasible, in the redistributed software. 3. The names "Penn State" and "Penn State University", and “The Pennsylvania State University” are trademarks of The Pennsylvania State University and may not be used in association with any goods or services, including products derived from these materials, without the prior express written agreement of The Pennsylvania State University. OTHER TERMS By downloading this software, you are agreeing to abide by the terms and conditions contained herein. By downloading, you are obtaining a non-exclusive license to use these materials in accordance with these terms. Nothing contained herein shall be deemed a waiver of any rights by The Pennsylvania State University in these materials.

Example 1 for Using the Macro

Example 2 for Using the Macro

4-minute video: How to run a SAS macro

The SAS macro RelativeCosts1 is provided as a supplement to Collins, Dziak, & Li (2009).

It automates the calculation and graphing of relative costs of implementing different designs for multiple-factor experimental designs for screening or testing the effects of several independent dichotomous factors. It is written for use with SAS 9.1 for Windows.

Read more information about factorial experimental designs and the multiphase optimization stragetgy (MOST).

Sample File 1 can be run to demonstrate its use in a situation in which per-condition overhead costs are expected to be greater than per-subject marginal costs, and Sample File 2 demonstrates its use in the opposite situation.

Before using the macro, you must tell SAS where you have saved the file, e.g.,

%INCLUDE("c:\RelativeCosts1.sas");

To use the macro, call it using the syntax below, replacing the “???”’s with the appropriate values.

%RelativeCosts1( number_of_factors=???,

desired_fract_resolution=???,

min_target_d_per_factor=???,

condition_costlier_than_subject=???,

max_cost_ratio=???,

num_graph_points=???,

max_graph_ratio=???) ;

The meaning of the arguments are listed below.

Number_of_factors should be an integer between 2 and 19, telling the number k of factors of interest.
Desired_fract_resolution should be an integer between 3 and 6, giving the desired resolution of the fractional factorial design to be considered.
Min_target_d_per_factor is the effect size for which a power of .80 to detect is desired. This effect size refers to the main effect for a full or fractional factorial design, or to the appropriate simple effect in a single-factor-many-levels or many-separate-experiments design.
Condition_costlier_than_subject should be 0 if the per-condition overhead cost is expected to be less than the per-subject cost (e.g., if the cost of a treatment level is $500 plus $5000 per subject) and 1 if it is expected to be greater (e.g., if the cost of a treatment level is $5000 plus $500 per subject).
Max_cost_ratio, num_graph_points, and max_graph ratio control aspects of the appearance of the graph. These arguments can often be omitted, in which case they are left at sensible default levels.

The macro prints information such as the following to the output window.

Doing separate experiments on each of the 9 factors requires at least:

7074 subjects total, i.e. 393 subjects in each of 18 cells.

A comparative setup with 9 groups plus a control group requires at least:

3930 subjects total, i.e. 393 subjects in each of 10 cells.

A 2^9 complete factorial experiment requires at least:

1024 subjects total, i.e. 2 subjects in each of 512 cells.

A resolution 4 fractional factorial experiment with 9 factors requires at least:

800 subjects total, i.e. 25 subjects in each of 32 cells

The macro also produces a graph showing the relative costs of each design as a function of the ratio of per-condition overhead cost to per-subject cost (if condition_costlier_than_subject = 1) or per-subject cost to per-condition overhead cost (if condition_costlier_than_subject = 0). The ratios considered are 0 to max_cost_ratio, where the latter has a default value of 100. This macro is free and is provided for use without any warranty, implied or otherwise.

Reference & Recommended Citation

Collins, L. M., Dziak J. J., & Li, R. (2009). Design of experiments with multiple independent variables: A resource management perspective on complete and reduced factorial designs. Psychological Methods, 14(3), 202-24. PMCID: PMC2796056

Need Help?

Please send questions or comments about the macro to MChelpdesk@psu.edu.

If your organization does not permit download and installation of files, please contact us with a shipping address and we will provide a CD containing all downloadable files.