tuesday ets air symposium data analysis

Upload: shyam-gande

Post on 08-Apr-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    1/44

    An Overview of Data Analysis

    Jonathan SteinbergSenior Research Data Analyst, Data Analysis Research

    Bruce A. Kaplan

    Director, Data Analysis Research

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    2/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Outline

    Key Considerations in Data Analysis

    Components of a Data Analysis Plan

    Distinguishing Data Types Distinguishing Different Types of Analyses

    Overview of Different Statistical Software

    Case Study: ACESTM

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    3/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Key Considerations in

    Data Analysis

    Identify the purpose of the analysis or project

    Understand the sample(s) under study

    Understand the instruments being used to

    collect data

    Be cognizant of data layouts and formats

    Establish a unique identifier if matching ormerging is necessary

    Plan your work and work your plan!

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    4/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Components of a

    Data Analysis Plan

    Statement of research questions

    Methods used to answer research questions

    Timeline Budget

    File restructuring procedures (syntax creation, adding new

    variables as needed)

    Algorithms for scoring, equating, etc. Data cleaning procedures (e.g. removing outliers)

    Quality control procedures at every step in the project

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    5/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Examples of Analyses

    Frequency Distributions and Cross -Tabulations

    Descriptive Statistics (Means, Std. Deviations, Correlations)

    T-tests and Analysis of Variance (ANOVA)

    Regression

    Principal Components/Factor Analysis (Data Reduction)

    Cluster and Discriminant Analyses (Segmentation)

    Latent Class Analysis (Classification)

    Hierarchical Linear Modeling (HLM)

    Differential Item Functioning (DIF)

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    6/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    More Advanced the Analysis,

    Greater the Amount of Preparation

    Most analyses can be executed straight from a

    working data file Some analyses may require transformations of

    the raw data, subsets, or specific input data to

    comply with statistical software

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    7/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Data Types & Representation

    Variables may require special coding for

    different data representation Numeric

    String

    Date & time

    Monetary

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    8/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ASCII Text Files

    Usually rectangular in structure One record per observation

    Each data variable in same position on each record

    Each record may have multiple instances of data Arrays

    Repeating blocks (sets of variables)

    File may have multiple records per observation Number of records per observation can be variable

    Most government data files come in this format at aminimum

    Every software package can handle this file type

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    9/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    CSV Files

    (Comma-Separated Values)

    Individual data elements separated by commas

    Usually rectangular structure One record (line) per observation

    Fixed number of elements on each record

    Problems if data elements contain delimiter or blank

    spaces (i.e. text strings) Missing data must be represented by nulls

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    10/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    System Files

    (SAS

    , SPSS

    )

    Data stored in binary (machine) format

    Issues of portability across platforms

    Structured as rectangular tables

    SAS files can be indexed for direct access

    Self-contained documentation

    Data variable labels & formats

    Data value labels

    Most analysis packages provide facility for reading (but not writing!)system files from other packages (SPSS more than SAS)

    Using default data formats can yield system files that are much largerthan source ASCII files

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    11/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Relational Databases

    Accessible using Structured Query Language

    (SQL)

    Optimal for filtering & sorting

    One or more rectangular tables contain data of

    interest

    This format is often at odds with statistical analysis

    needs

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    12/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Relational Database

    Example

    Index Student School Gender Ethnicity

    1 29101 291 F 3

    2 29102 291 M 2

    3 29201 292 F 6

    Student

    SAT

    Index Student TestDate Measure Score

    1 29101 10/13/2006 Verbal 650

    2 29101 10/13/2006 Math 640

    3 29102 5/21/1995 Verbal 370

    4 29102 5/21/1995 Math 400

    5 29201 2/3/2000 Verbal 520

    6 29201 2/3/2000 Math 550

    Index Student TestDate Test Measure Score

    1 29101 5/6/2006 General Verbal 760

    2 29101 5/6/2006 General Quantitative 540

    3 29102 5/3/2006 General Verbal 340

    4 29102 5/3/2006 General Quantitative 420

    5 29102 5/3/2006 General Analytical 0

    6 29102 Subject Subject1 540

    7 29201 5/6/2006 General Quantitative 500

    8 29201 5/6/2006 General Verbal 530

    9 29201 Subject Subject1 490

    10 29201 Subject Subject2 530

    GRE

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    13/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Hierarchy of

    Statistical Software

    Excel

    General Stats

    Psychometric

    Advanced Modeling

    SAS, SPSS

    IRT: PARSCALE, BILOG

    Factor Analysis: LISREL, EQS

    Nested Models: HLM

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    14/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Overview

    SAS and SPSS are most commonly used and tend to focus onthe classic statistical routines: Descriptive statistics and non-parametric (distribution-free) tests

    ANOVA / Regression Factor analysis

    However, many psychometric procedures (e.g. IRT) and newerstatistical models are not as well supported by these programs Very specialized programs are used

    Designed to do a specific task or validate a theory

    Specialized programs may have issues Interface not very user-friendly

    Additional data types or files required

    Expense

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    15/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    What is Excel?

    Data are organized by worksheets, rows and columns

    Worksheet limits are 256 columns and 65,536 total cells

    C

    ells contain data or formulas with relative or absolute references toother cells

    Direct manipulation of data and flexibility to move data around (e.g.sorting, replacing, merging)

    Opens many file types

    Quite useful in prepping files for use in SPSS, SAS or other programs Conditional formatting

    Also features macro capabilities, replicating user actions, allowingsimple automation of regular tasks

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    16/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Data Presentation

    Options in Excel

    Tables and graphs can be exported to a wide variety of software packages

    Can tweak and perfect example graph or table and then replicate by

    replacing only the data being used

    Main advantage is ability to combine data from multiple sources not just

    what is found in the data file

    Two-for-One deal - table creation usually puts data into a format that leads

    to easy graph creation

    User has control over virtually all aspects of a graph - size, colors, fonts, titles,

    legends, labels, etc. Can combine graphs with tables and use cell layout to produce more complex

    presentations

    Final graphs can be of publication quality

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    17/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Data Presentation

    Options - Example

    0%

    10%

    20%

    0%

    0%

    0%

    60%

    0%

    80%

    90%

    100%

    00 and

    belo

    10 to

    800

    810 to

    900

    910 to

    1000

    1010 to

    1100

    1110 to

    1200

    1210 to

    1 00

    1 10 to

    1

    00

    1

    10 to

    1 00

    1 10 to

    1600

    C

    mbi

    (V

    rb

    l + M

    t

    )

    r

    i trib ti f Fr m G b C mbi r

    A(

    .90 to

    .2!

    )

    A-(

    .60 to

    .89)

    " +(

    .

    0 to

    .!

    9)

    " (2.90 to

    .29)

    "

    -

    (2.60 to 2.89)

    C+

    (2.

    0 to 2.!

    9)

    C(1.90 to 2.29)

    C-

    (1.60 to 1.89)D

    (0.60 to 1.!

    9)

    #

    ("

    elo$

    0.!

    9)

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    18/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    What is SAS?

    A general purpose statistical package with a basicprogramming capability utilizing scores of statistical andmathematical functions in numerous modules

    Can readily access data from a wide variety of sources,perform data management, and present findings in avariety of report and graph formats

    Provides powerful tools for both specialized andenterprise-wide analytical needs

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    19/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    SAS - Strengths

    Versatile data input and output formats

    SAS provides both SQL and DATA steps tomanipulate data: SQL provides a way of carrying out relational algebra

    on tables and views SAS data sets can be indexed for direct access or

    processed sequentially, without reading all records

    into memory, which is sometimes much more efficient

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    20/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    SAS - Weaknesses

    Steep learning curve; volume of functions, options anddocumentation can be overwhelming for the novice

    Inconsistent syntax across different procedures ormodules

    Not a good choice for applications that interact withexternal systems such as hardware devices orsoftware programs because of its inconvenient

    interface

    Difficult interaction with other programming languages

    Expensive

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    21/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    What is SPSS?

    A commercially produced statistical software packagethat is widely used in the fields of Education andPsychology

    Program functionality is broken into over a dozendifferent modules which are sold individually

    Most commonly used are Base, Regression Models, andAdvanced Models

    Other modules can be installed to run more complexanalyses

    SPSS data files include both the data and also variableinformation (variable and value labels, formats andmissing values)

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    22/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    SPSS - Strengths

    Easily opens data from other programs such as Excel and SAS

    Variable view screen allows for quick overview of file contents andallows for easy modifications of names, formats, labels, and variable

    order Having all data information in a single file allows sharing files on a

    project to be very easy

    Point-and-click menus do not require memorizing syntax for majorityof procedures

    Many procedures can be expanded beyond the menu options in syntax

    Split-file command allows all output to be replicated for variousgroups through a single command

    Journal file tracks all commands used for life of program, with goodresources to find code accidentally deleted

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    23/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    SPSS - Weaknesses

    Ease of doing data manipulation can sometimes lead tomistakes as the program does not preclude inappropriatemodifications to the data

    Matching feature requires exact match Duplicate records generate warnings but can be marked in file

    Error logs are hard to interpret at times

    Incompleteness of menus means some options are onlyavailable via syntax

    While the majority of output is saved as pivot tables allowinggreat flexibility in modifying tables

    Output tables and graphs generally not done as well as Exceland are harder to manipulate

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    24/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    LISREL

    Ideal for discrete data types

    Test data, Likert scale item data

    Data can be imported in various types

    ASCII, Access, Excel, SAS, SPSS, etc.

    Variable names have length restrictions

    Data files then stored as system files for later use

    Basic statistics (e.g. means and correlations) are generated in an

    underlying program called PRELIS LISREL itself is used to confirm the structural validity of a

    measurement model for any assessment

    Requires syntax and input matrices

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    25/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    EQS

    Ideal for continuous data types (test subscales)

    Data can be imported in various types ASCII,

    Access, Excel, SAS, SPSS, etc. but variable names

    have length restrictions

    Data files then stored as system files for later use

    EQS itself is used to confirm the structural validity ofa measurement model for any assessment

    Some model syntax can be built through the menus

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    26/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    HLM

    Hierarchical Linear Modeling (HLM) is becoming a more

    popular type of analysis, namely in cohort trend modeling

    Also allows you to look at variance component estimatesand regression models given a nested sample ofrespondents

    Students within countries within global regions on personality

    variables

    More tedious to set up analysis with fewer available filetypes

    Also requires more upfront work as multiple data files are

    needed

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    27/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    What Program

    Should I Use?

    Microsoft Excel is the most basic and accessible spreadsheet programavailable today

    It is most ideal for general data exploration, histograms, scatter plots, etc.

    Appearance of tables can be customized to meet APA standards Allows for easy transition to other programs to complete analyses and

    write reports

    However, its heritage is not as a statistical analysis program

    Certain statistical programs are designed for specific analytic tasks

    Balance the results and what will being presented Choose wisely in the interests of efficiency and accuracy of results

    Some output is good forlookingat the data through basic exploration andto generate basic tables, but not to present the data

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    28/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Summary

    Be very clear about the analysis objectives

    Be very familiar with all aspects of what defines your data

    Develop and stay true to your data analysis plans and

    research questions

    Be cognizant of which statistical software programs can best

    answer your research questions and present your results

    Be thorough in your analyses, express openness to additionalinvestigations, yet be mindful of limitations given the data and

    the programs you are using

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    29/44

    Overview of the Admitted Class

    Evaluation Service (AC

    ES)Bruce A. Kaplan

    Director, Data Analysis Research

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    30/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    What is ACES?

    ACES is a validity study service offered by theCollege Board that is operated and maintained

    by ETS.

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    31/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Types of ACES Studies

    There are two types of ACES studies:

    Admission

    Evaluate the validity of tests in predicting how well

    students will do in an institution of higher education

    (used for admitting students into an institution)

    Placement Evaluate the validity of tests in placing students into

    or out of a class

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    32/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Types of Data

    Institutions have many variables to choose from

    when requesting their models The required variables are:

    First-Year Grade Point Average (FGPA), supplied by the

    institution

    A measure of high school success (either from the SATQuestionnaire or supplied by the institution)

    A choice from a list of SAT Test scores

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    33/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Types of Data (continued)

    The optional variables are:

    Course data (name of course, grade received, and

    number of credits awarded) for each student,supplied by the institution

    A choice from a list of SAT Subject Test scores.

    Additional predictors, either supplied by the

    institution or chosen from a list of ACES predictors Additional subgroups to be analyzed, either

    supplied by the institution or chosen from a list of

    ACES subgroups

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    34/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Data Merging, Matching, -

    and Cleaning

    ACES data poses many challenges

    Merging

    Institutions have choices of format type, and sending two files orone

    ACES merges all this information in a more or less standardized

    format

    Matching

    Institution data is matched against a file of college seniors to pull

    SAT scores, SAT subject scores and SAT QuestionnaireResponses

    Cleaning

    Data is checked for outliers, out of range, and implausible values

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    35/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Types of statistics used

    Basic Statistics

    Means and Standard Deviations for Total Group andby Gender

    Correlations for various predictor variables with:

    First-Year Grade Point Average (FGPA) for Total group andsubgroups

    Adjusted First-Year Grade Point Average (AGPA)

    Adjusted for course difficulty

    Both are corrected for Restriction of Range

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    36/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Types of statistics used (continued)

    Various Regressions with up to six classes of

    predictor variables

    SAT (SAT-r) variables only

    SAT-r and high school measure (either HSGPA or HS rank)

    SAT-r, HS measure and school requested variables

    SAT-r and SAT Subject Tests (SAT-s)

    SAT-r, SAT-s, and HS measure

    SAT-r, SAT-s, HS measure and school requested variables

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    37/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Types of statistics used (continued)

    At Risk Students are students whose Actual FGPA

    fell 1.5 or more standard deviations below their

    Predicted Grade Point Average (PGPA) PGPA for a student is computed from regression

    model with the most data available for that student

    Return to institutions the data files used so they can

    conduct further research. This includes any data sentto ACES along with the matched SAT variables, and

    calculated values like PGPA and AGPA

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    38/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Restriction of Range

    What is Restriction of Range?

    We want to measure the correlation between FGPAand SAT scores for the applicant pool for a given

    institution

    We only have the attending student pool for a giveninstitution

    What does that do to our correlations?

    Demonstration of Restriction of Range: http://cnx.org/content/m11196/latest/

    Our solution: use the SAT senior cohort to helpestimate the correlations

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    39/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Admission -

    Course Adjustment

    Are all FGPAs at an institution really equal?

    Adjusted FGPA (AGPA) using institution-supplied coursegrades

    For the institution, calculate each students predicted FGPA using alinear regression with SAT CR, SAT M, SAT W (if used), and HSvariable as independent variables and FGPA as a dependentvariable

    Calculate an average residual per course

    C

    alculate an adjustment factor for each student, using the averageresiduals for the courses they took, weighted by the number ofcredits for that course

    Apply the adjustment factor to the FGPA to obtain theAGPA

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    40/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Placement -

    Statistics Used

    Distributions of Predictor variables (Accuplacer

    scores, CLEP scores, SAT Scores, or school

    supplied variables) by Different levels ofCourse

    Grade

    Logistic regression calculated for two dependent

    variables

    Pass = B or higher, else Fail

    Pass = C or higher, else Fail

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    41/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    ACES Placement -

    Statistics Used (continued)

    Models provided for single predictors, and multiple

    predictors

    For each Model

    A measure of association between the dependent and

    predictor variables is computed

    Percent correctly placed

    Cut Scores Associated with Predicted Probability of

    Success

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    42/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    You can submit a study

    for your Institution

    Examples of sample reports:

    Admissions Report

    http://www.collegeboard.com/prod_downloads/highered/apr/aces/sample_admiss.pdf

    Placement Report

    http://www.collegeboard.com/prod_downloads/highered/apr/aces/sample_place.pdf

    ACES site to request a study

    http://professionals.collegeboard.com/higher-

    ed/validity

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    43/44

    Copyright 2008 by Educational Tes ting Service. All rights reserved. ETS, the ETS logo and LISTENING. LEARNING. LEADING. are registered trademarks of

    Educational Testing Servic e (ETS). The AIR logo is a trademark of the Associati on for Institutional Res earch. 9116

    Contact Information

    Jonathan Steinberg:[email protected]

    Bruce Kaplan: [email protected]

    For additional ACES help and information:

    [email protected] Or call (609) 921-9000 and ask for help with ACES

  • 8/7/2019 Tuesday ETS AIR Symposium Data Analysis

    44/44

    Questions orComments