Skip to content

Page 1

DDPP

DISCRETE DATA PROGRAM PACKAGE

ND 210142

[Photo: Office scene with people at desks and computers, plant and wall decoration, ND office environment]

ND
Norsk Data

Page 2

DDPP

DISCRETE DATA PROGRAM PACKAGE

Introduction

DDPP is a general program system for file handling, data manipulation, and statistical analysis of different types of numeric and alphanumeric data. The system is easy to use in both interactive and batch processing.

DDPP provides the most commonly used descriptive statistics, various standard analysis techniques, such as one-way variance, multiple regression, factor, and discriminant analysis. Also provided are Multiple Classification Analysis (MCA), Automatic Interaction Detection (AID), non-parametric causal analysis, and sociometric analysis.

Moreover, the system offers a wide range of functions and operations for data manipulation, such as checking and correction, recoding, and construction/analysis of complex data structures.

DDPP can be compared to the better-known system SPSS, although its statistical repertoire is somewhat limited.

Features

  • Easy to use; requires very little knowledge of computers and data processing.
  • Can handle very large data masses, mixing several kinds of data types.
  • Offers flexible routines for data input from and output to different storage devices.
  • A general, natural command language for interactive use as well as batch processing.
  • Easy to update and enlarge.
  • Data documentation facilities by an associated codebook, which supports the output result with textual information.
  • Well-suited file administration functions, allowing operation on several files in one task, where intermediate file handling is performed without user involvement.
  • Flexible data import/export facilities accept files in various formats and codes residing on all available storage devices.
  • Support for a broad range of data handling/manipulation functions.
  • Good documentation/description tools both for describing the initial data structure and for recording the data manipulation history.

Product Description

The user communicates with DDPP through a simple user-orientated language. This language consists of commands specifying the functions to be performed and a set of appropriate parameter options with each command. Most commands lines are interpreted one at a time. Hence, in interactive processing, the user may immediately correct erroneous commands and continue processing.

DDPP is file-oriented: a DDPP file consists of a data matrix and an associated codebook with data description. Non-DDPP files are easily converted to DDPP format within the system.

The codebook contains a description of the data structures, and it includes variable names, their type and length, value labels, category names, and additional general information.

The data part (file) comprises a matrix of variables. Fairly complex data structures (e.g., hierarchy or network) may be represented and analyzed by several programs which allow operation on several DDPP files at the same time.

Data entries can be accessed in two ways: a. Directly, so that each DDPP position (item) can be addressed separately. Varying variable fields may be defined by the user each time. This applies to all data-handling programs, which then accept alphanumeric information as well.

b. Indirectly, by first accessing the codebook to retrieve information about variable type and length, and then automatically converting the character strings comprising a variable to an integer/real number.

This solution offers the user maximum flexibility, and allows mixing of different data types, such as discrete, continuous, and textual variables.

Hence, almost all operations may be performed on all variable types. DDPP files may be sorted according to an alphanumeric key, and rearranging of variables can be done. Different files can be merged; new variables and/or cases are easily added to an existing DDPP file.

Interactive Facilities

When running DDPP in conversational mode, the user commands may be read either from the terminal or from a separate command file. A special interactive program is available to help the novice user in setting up the file-handling specifications.

The command file can be created by an ordinary editor (PED or TED) or by a simple line editor in DDPP.

If a fatal error occurs, e.g., during file accessing, DDPP will help the user to continue execution with a minimum loss of information.

Page 3

DDPP

DISCRETE DATA PROGRAM PACKAGE

Statistics

DDPP covers the most common statistics such as

  • Simple frequency straight-run tables and cross-runs tables comprising up to 10 variables (items) with various non-parametric statistics.
  • Median, skewness, kurtosis, Gini-index, percentiles.
  • Mean, min-max values and standard deviation with one-way variance analysis including F and T tests.
  • Correlation analysis: PEARSON, KENDALL, SPEARMAN and Tetrachoric coefficients.
  • Stepwise REGRESSION analysis and simple FACTOR analysis.
  • Non-metric causal analysis based upon parent differences.
  • DISCRIMINANT analysis and MANN-WHITNEY test.
  • Automatic Interaction Detection (AID).
  • Multiple Classification Analysis (MCA).
  • Sociometric Analysis based on one-way or mutual case-to-case relationship.

Statistics may be excluded according to various criteria: Illegal (missing) values, subset selections, user-defined sampling algorithms and so forth. The statistics can be automatically repeated for different subsets of data, and weight variables can be specified.

Data Manipulation

Several programs are available for various types of data modification and transformation:

  • Error checking, listing, editing and recording, printing in user specified format.
  • Generation of new variables from existing ones, transformation from textual to numeric values, arithmetic and logical functions within or between subjects.
  • Manipulation of textual data: Comparison, frequency counting, editing, sorting etc.
  • Complete rearranging of the codebook or data part.
  • Reading, converting, counting of binary coded data (e.g. multiple choice response), with operations for consistency checking and correcting.

Programs

Some programs should be particularly mentioned:

  • GROUPING – for data aggregation and handling of hierarchical file structures, e.g. for the accumulation of data on the level of units to some higher level as groups or school classes etc.
  • MIXING – to allow the reverse process of combining higher level data with data of units or subjects.
  • CALCULATION – for writing user-defined algorithms. A simple "programming language" resembling BASIC is offered, which includes assignment statements, arithmetic and Boolean functions, IF and GOTO, and separate INPUT-OUTPUT statements.

Technical Aspects, Limitations and Requirements

DDPP consists of about 35 subprograms, which are almost strictly written in standard FORTRAN. Only 10 assembly language subroutines are supported. The programs are organized in a 3-level overlay structure with a common monitor, data (blocks) and subroutine library providing I/O, character-handling file management and interprogram communication. This allows easy switching between different functions in a simple manner.

The limitations upon data size is 2048 DDPP variables, while number of cases (subjects) on 16 bits integer FORTRAN is 32767. (When using 32 bits integer only restriction will be available mass storage). 40 - 60Kbyte memory is required.

Documentation

Per H. Jacobsen. Statistics and computing in Social Science. The EDP Center, University of Oslo, 1978.