Technology Tradeoffs John H. Krantz Hanover College Workshop on Web Research Centre for Affective Sciences, May 2008.

Technology Tradeoffs

John H. KrantzHanover College

Workshop on Web ResearchCentre for Affective Sciences, May 2008

Outline

• Terms• General Perspective• Main Comparison• Server-Side Solutions• Client-Side Solutions

Some General Terms

• Server vs. Client• Server is machine/software the study resides• Client is the machine/software in use by the

participant

Some General Terms

• Push vs. Pull• Client-Pull: information is drawn from server

at request of client browser– More polite

• Server-Push: information is driven by the server– What is presented is now no longer under the

control of the participant

Some General Terms

• IP address – Machine Address on the network

• IVp4: xxx.xxx.xxx.xxx• IVp6 –coming down the pike• In many cases dynamically assigned• For an server, try to get fixed IP address

• Domain Name– We don’t like numbers– Names to make machine location more memorable– Tiny URL

Some General Terms

• Protocols and URL’s– A protocol is the means of communicating– Sort of like a language– E.g. hypertext transfer protocol (http) vs. file transfer

protocol (ftp) vs. hypertext transfer protocol secured (https)

• Uniform Resource Located– Common format for getting formation:– Protocol://server.domain/(path/filename)– http://psych.hanover.edu/

Some General Terms

• Internet Connectivity– People connect in different ways– Modem– Cable modem– Wireless– LAN

• These difference impact time to download and, thus, willingness to continue

Some General Terms

• Technical Variance– Data variation resulting from the use of

technology

• Denial of Participation:– Prevention or reduced likelihood of participation

based upon the technical requirements of study– e.g., visually impaired

General Perspective

• Web is a constantly changing environment• Commercial issues are not the same as ours• Lowest Common Denominator (Possible)– The more complex the technology the more

participants you lose– Loss is not random– Complexity also often, but not always adds delay,

and people are impatient on the web• Know the reach of your technology

General Issues Continued

• Population is changing– Early studies: mostly male (Krantz & Dalal, 2000)– More Recent studies: mostly female (e.g., Draper

& Elmer, 2008)– Feminization of Psychology– Still topic and probably technology dependent

(e.g., Braun & Drew, 2008)– May well interact with technology

Server-Side vs. Client-Side

Server-Side• The experiment is run on

the server• Complexity is on the server• The client needs only a

browser with minimal capabilities– Forms, Dynamically Created

Web pages

• All actions require return to the server

Client-Side• The experiment is run on

the client’s machine• Complexity is on the client?• Often requires plugin

– JavaScript, Flash Player, Java interpreter, etc.

– Some will not be pre-installed

• No need to worry about network traffic once experiment begins

Division is extreme

• Every experiment uses both server-side and client-side operations

• At the minimum:– on the server-side, experiment must send page(s)

and collect data– One the client-side, participant must have browser

to display html and forms

Server-Side Solutions

• Minimal requirements on the client side– HTML• The basic language of web pages

– Forms• Goes back to the early versions of Mosaic (the first

media capable browser)

CGI

• Common Gateway Interface– Interface between server and other program or

script– Perl and PHP– Often used to handle the data in experiments– Can do anything any program can do

Dynamic Webpage Generation• Use some CGI Mechanism to generate web pages

on fly• Often from some database as well as from

information from client • PHP – a common open source solution can run on many

server platforms• ASP– Microsoft’s version and requires their server platform

• Requires browser that can handle these pages, most can these days

Server-Side Summary

Advantages• Low requirements for Client• Single known platform for

most program execution– Excepting HTML and Form

interpretation

• Great flexibility is possible

Disadvantages• Limited control of display

– Only what is possible in HTML, XML, PHP …

• Timing is limited– Can get some timing

differences (see Ulf’s Method)– But will have a large constant

error

• Network traffic during study

Server-Side Software

• Perl: http://www.perl.com./• PHP: http://www.php.net/– Generic PHP Form Processor:

http://www.goeritz.net/brmic/– MySQL: http://www.mysql.com/

Client-Side Solutions

• Program runs on participant’s computer• Allows direct control of participant’s computer• Usually requires some plugin or interpreter to

be present on the participant’s computer

JavaScript

• A scripting language interpreted in client browser– Requires a scripting interpreter in the browser– Is an object-orienting scripting language– Viewers of the page can get the program– Is not related to Java though some similarity in

commands– Installed on most browsers but some people turn

off or limit

JavaScript Editors

• JavaScript Editor Lite: http://www.yaldex.com/JSFactory.htm

• JSEclipse (plugin for Eclipse): http://www.interaktonline.com/Products/Eclipse/JSEclipse/Overview/

Java

• Object-Oriented Program Language– Partially compiled/Partially interpreted– Allows for more platform independence– Runtime engines are written for each environment• But it does change

– Comes with most browsers now and lot of other software• But several versions exist

– Microsoft’s version is disappearing (YEAH!)

Some Java Limitations

• Java is limited, partly to make web safe• Cannot sync drawing to beginning of frame– An example

• Timing is problematic (low priority and difficult to steal priority from other machine operations)– Eichstadt (2001) gives way to handle timing issues

An Example of Java Timing

Writing in Java

• Eclipse: http://www.eclipse.org/ – An open source Integrated Development

Environment

• Java not used much due to complexity• Need an experiment library

Adobe Plugins (formerly Macromedia)

• Flash/Authorware (http://www.adobe.com)– These are really high level languages with specially

designed IDE’s– They require special plugins/players to run– Flash usually preinstalled on most browsers– Others less likely– Support for Shockwave/Authorware on wane– Flash is being pushed by publishers

Adobe Software

• Flash: http://www.adobe.com/products/flash/• Authorware:

http://www.adobe.com/products/authorware/– Authorware used at the PsychExps site – Largely basis of experiments at Online Psychology

Laboratory (OPL) by American Psychological Association (http://opl.apa.org)

Client-Side Summary

Advantages• Greater control over

participant’s computer• Can increase interactive

aspects of experiment• Less need to communicate

with server during experiment

Disadvantages• Requires plugin which

participant may not have• Greater impact of

participant’s setup• Participants may turnoff

ability to run plugin• May require large download

to run• Not all timing issues have

been solved

Learning Curve vs. PowerLearning Curve

Shallow Medium Steep

Power Low HTMLForms

Medium PerlPHPJavaScript

High FlashAuthorwareJava

Some Additional Issues to Consider

• Who can and will use the media?• Who cannot or will not use the media?– Issues of external vs. internal validity

• Will this impact my results?– Will usage depend upon condition?– If download time or complexity varies depending

upon condition you have a confound

Conclusion

• Test! Test! Test!• Browser bugs!• Consider is the web the best choice: not the

easiest choice