SPSS

(Statistical package for Social Sciences)

The first version of the program was released in 1967. SPSS quickly became one of the most widely used statistical programs for market research, survey analysis, medicine, and the social sciences. From the very beginning, the authors focused on academic users and tried to keep it easy to use and adapted to users who are not computer literate. IBM purchased SPSS in 2009.

Historically, SPSS has enjoyed immense popularity in regional academic circles. It is intended for the social sciences. Many functions are available via a graphical user interface and a menu system, while advanced functions must be programmed in syntax and are not available via menus. The menus also generate syntax that can be displayed in the output. Programs can be run interactively or via the Production Job Facility. A macro language can be used to write language subroutines, and there is a Python extension that allows access to the data dictionary and dynamically create syntax commands. The Python extension also allows the execution of any R function.

Depending on the type of license, it costs between $1000 and $12000. A rich set of statistical techniques is implemented. The basic license includes basic statistical techniques and methods used in the social sciences, such as ANOVA, cluster analysis, discriminant analysis, factor analysis, regression and correlation, various types of tests, multidimensional scaling, etc.

By purchasing additional licenses, additional modules are also obtained, such as:

  • time series analysis.
  • questionnaire processing, etc.

SPSS is often used in biology and social sciences because it allows you to very easily perform a standard set of statistical analyses required in many academic disciplines. On the other hand, physicists, computer scientists, and statisticians tend to want more highly non-standard algorithms that don't fit easily into these traditional methods, so SPSS is less commonly used in these areas.

SPSS places restrictions on file structure, data types, processing methods, and corresponding files, which greatly simplifies programming. Data is always in the form of a two-dimensional table in which rows represent cases (individuals) and columns represent measurements (variables). There are only two types of data: numeric and text.

The graphical user interface has two tabs. One displays the data matrix, and the other displays metadata, i.e. variable properties (name, label, variable type, etc.). It has excellent capabilities for loading and saving data in various formats, including access to relational databases. In addition to the original (proprietary) format, data can be saved in Microsoft Word, PDF, Excel and other formats, and graphic output can be saved in various standard graphic formats.

The following features speak in favor of using SPSS:

  • it is very easy to learn,
  • it is more powerful than Minitab,
  • historically, it was the dominant package in academia, though modern industry and advanced research have largely shifted toward Python and R,
  • it has a menu and a command line,
  • it is one of the most powerful packages, and at the same time it is easy to use.

On the other hand, it is very expensive and is not adequate for modeling and boundary statistical analysis.

SAS

(Statistical Analysis System)

It has a very complicated pricing model. Licensed annually, with a first-year cost of $8,500.

SAS was founded in 1976. The company has a record of growing profits that is unprecedented in the software industry. New versions are released very frequently, making SAS considered the most innovative statistical software. SAS is also known as one of the best companies to work for. The software runs on a wide variety of platforms.

Commands are run via the command line, although there is also a graphical user interface and a menu system that usually only automates the generation of SAS programs. Functionality can be accessed through an application interface in the form of commands and procedures. Compared to general-purpose programming languages, this structure allows the programmer to focus less on technical details and how to store data, and more on the information contained in the data. This blurs the line between programmer and user.

SAS enforces highly secure, centralized data management architectures, and its analytical engine provides robust, industry-standard implementations for ANOVA, mixed models, multivariate analysis, and logistic regression.

SAS has the advantages of being widely accepted as a leader in statistical analysis and modeling, widespread use in industry and academia, flexibility, and power. On the other hand, SAS is very expensive, does not have a user-friendly interface, is slow to learn, and has relatively poor graphics capabilities.

EViews

It is oriented toward econometric analyses. It can be used for general statistical analyses as well as for specific econometric analyses such as cross -sectional analysis, panel data analysis, and time series analysis.

Minitab

It was created in 1980 at Pennsylvania State University, and the price of one license is $1395. It is often used in basic statistics courses because it is easy to learn. Minitab can be used to do descriptive statistics, graphical display, sample size calculation, random number generation, regression analysis, etc., but it is surprisingly limited in the number of implemented functionalities compared to SPSS or SAS.

Because it is easy to learn, it is commonly used in schools in basic statistics courses. It is also widely used in engineering for process improvement. However, it is relatively expensive compared to other licensed statistical packages and is not suitable for very complicated calculations and analyses, so it is not usually used in academic research.

Matlab

It has elegant matrix support and allows good visualization. On the other hand, it is expensive and has a limited set of statistical techniques compared to other packages. Its proprietary programming language is highly optimized for matrix computations, offering robust support for vectorization and algorithm development. It is very good for developing new mathematical algorithms and is very popular in the field of machine learning.

It is typically used by engineers because it is more oriented towards mathematics than statistics.