Rep IV

<pageby nominor="false" comments="false"/>

Definition

Rep IV is the name for a series of tools for repertory grid technique (RepGrid) and visual syntactic structures used for representing construct networks (RepNet).

As of Feb 2009 there are three versions according to the official Rep IV (RepGrid, SocioGrids, WebGrid & RepNet) page:

Personal Version 1.11

This “provides simplified versions of RepGrid and RepNet that are freely available for personal (non-commercial) use. It can elicit and analyze single grids with up to 15 elements and 15 constructs, and nets having a syntax from the set provided.”

Research Version 1.12

This provides the full versions of RepGrid, SocioGrids and RepNet for professional use (commercial)

Web Version 1.12

This extends the Research version with web server capabilities supporting WebGrid-style grid elicitation and analysis through web browsers.

A functional test version is available on the Knowledge Science Group Server, University of Calgary. Daniel K. Schneider used both WebGrid III and WebGrid IV to generate the examples discussed in repertory grid analysis techniques

An introduction to Rep IV/RepGrid personal edition

This entry does by no means replace the well done official one (available in the download) and which contains much more details. Also, read repertory grid technique if you are not familiar with the concepts !

I (Daniel K. Schneider) will use screen captures and other "data" from a quickly done elicitation regarding on-line environments for teaching and learning that I use myself (Moodle, Pageflakes, Edutechwiki, TecfaSeed, ELGG, Blackboard) or that are used by coworkers (Tiki, Dokeos, Acolad) and with which I am familiar. Some elements are missing, e.g. other wiki variants, web pages + forums or Ning. Also, constructs are not complete. This example was just used to discover this software and to gain some extra insight into repertory grid technique.

Elicitating a Grid

Step 1 - Create a new Grid

• Click on "New" and fill in the form

Screen capture of Rep Grid IV Personal edition

Tips

• Name should be the name of the participant (E.g. Daniel). It will show in the elicitation dialog.
• The same principle is true for "Purpose". E.g. the participant will see "Daniel is considering TECFA's on-line environments for ..."
• You must choose a scale, e.g. 5 points (negative numbers are not allowed, max. is 9).
• It's a good idea to change terminology, so that users can understand what you mean, e.g. to study representations people have of our (TECFA's) online environments that we currently used, we used:
  • system for element
  • aspect for construct

.... The rest should be obvious

Step 2 - Elicit elements and constructs

You also could administer an already made grid or enter a with the simpler Edit Grid script.

Click on the Scripts tab and click Elicit Grid. The system now will elicit elements and constructs with the classical triadic method until it suggest that sufficient constructs have been entered to discriminate between all elements (no 2 elements should be the same) and that there are sufficient elements to discriminate constructs (no 2 constructs should be same).

Break construct match

An interesting feature of this system is "break construct match" warning and "dialog options". E.g. at some point in the elicitation process, you will get this kind of warning:

Break construct match

The system is telling the participant that two constructs (systems in our case) are nearly the same, i.e. that they they discriminate very little among elements (aspects in our case). It then prompts use to look for a system that has different combination on the scale, e.g in our case a system made for pedagogy and that also favors and open share space. We could for instance add a system like Knowledge Forum (but we don't use if for real at TECFA).

On a side note, Daniel K. Schneider simply would like to add that indeed pedagogical platforms are deficient in more than one aspect and systems made for other purposes share many aspect since they are often made to support a community.

Break element march

In the same way, if two elements are too close, RepGrid IV will ask for another construct that might separate them. E.g. in our case it found Dokeos and Blackboard to be very similar systems:

Break element match

The feature I found was simple: One is commercial and the other is free and open source.

At any point you also can add manually an element or a construct by clicking on the "Elements" or the "Constructs" panel (see below "fine tuning")

Step 3 - Fine tuning

Since in our case it was difficult to enter absolute values for given constructs (aspects), we suggest that in similar cases, one has to go through each construct and adjust values for elements. Best method is to use the elements panel, then select constructs from the pull-down menu.

In the Elements panel you can select the construct on which you want to work with the pull-down menu. You then can change the values by typing or by using the right-click menu:

Constructs editing in the Elements panel

In the Constructs panel you can select the element for which you wish to enter values.

• You can change text text by clicking in a cell
• You can change value by editing the cell (or with a right-click menu)

Step 4 - Analysis

RepGrid IV personal edition will produce three graphics:

• The repertory grid table (raw data)
• A two-way cluster analysis (Focus button). It will highlight "positive" poles", reorder columns and rows in order to attach the dendogram.
• A principal component analysis of constructs showing elements (PrinGrid)

Almost no statistical data are available in the free personal version. The only information concerns explained variance of the two extracted factors.

In our testing example we got the following repertory grid, Focus and clusters, and principal component analysis:

RepGrid IV personal edition - results

• There are three big clusters of elements: (a) LMS systems under foreign control + our own Moodle, (b) systems other than LMSs, and (3) a web 2.0 webtop.

• Constructs that go together are: (a) commercial, under foreign control and hip (mhh..), (b) web 2.0, co-construction, not popular, made for other purposes, multi-purposes, etc. (c) Many tools, (d) Difficult interface.

• The main factor (58%) could be named Isolated spaces made for teaching vs. Open shared multi-purpose systems. Factor two (26%) is Ugly, under control and free vs. hip and commercial/not controlled. I personally favor the systems in the upper right quadrant as you may guess ...

Disclaimer. This "test" analysis shows how Daniel K. Schneider perceived his use of systems within his research and teaching unit. Although some constructs rather refer to generic properties of these systems. That should be made more clear in a publication, for example. Nevertheless I find these pictures quite interesting, but will refrain from further analysis for now :)

Creating a Grid with "Edit Grid "

Here is a short summary of the procedure.

• Create new grid

• Choose the rating scale (e.g. 5)

• Directly select the "Scripts" Tab.

• Right-click on top-right button "Elicit Grid" and select "Enter Grid". The button now should show "Enter Grid". Finally, click on this button.

• Answer all the prompts. The system will first ask for the list of elements and then ask for the constructs. For each construct it will directly ask to judge the existing elements.

The rest (tuning) is like above. I.e. if you exit from the dialog by clicking on another tab, you will have to complete either through elements, constructs panel. If you hit "Enter Grid" again, you will be thrown in the typical knowledge elicitation dialog, i.e. the system will try to complete values of the constructs and then check break constructs and elements.

Reusing a Grid, elements or constructs

There are three possibilities for reusing a grid. In the main panel:

• Click on Exchange to reuse elements and constructs. Then select an existing RepGrid (*.grid file) from your system
• Click on Elements to reuse just the elements.
• Click on Constructs to reuse just the constructs.

You then can either add/delete elements and constructs or edit the existing ones through any input panel (elements, constructs, scripts, i.e. the Elicit Grid Script' or the Edit Grid Script).

In an empirical situation, it is probably best to use the Elicit Grid script. It will do the "right thing" and provide explanations to the participant. E.g. in "exchange mode", ipt will prompt for each element construct to rate constructs. Users can click on the panel and a popup window with the values will appear.

Additional tips

Deleting items

To delete an element or a construct: select the line in the "Elements" or "Constructs" tag then hit BACKSPACE (the DEL key will not work not the "Delete" item in the "Edit menu".

Editing cells with text

To edit a cell with text in either the elements or the constructs panel: Click on it, then right-click or double-click (this is fairly counter intuitive).

Watch out for menus

Watch out for contextual menus (if the cursor shows a menu, then try right-click). Therefore, also watch the shape of the cursor when you move it over a cell or some text.

Use graphics output in your publications

You may cut and paste (ctrl-C/ctrl-V) figures to the clipboard (i.e. into programs like word). You also can drag and drop selected lines to other applications. In other words, publishing little studies with this software should be really easy. Typical small grids easily fit on a portrait A4 page and they area readable.

In addition, you also can move around items in the biplot, i.e. drag elements of constructs with the mouse. This is also a really nice feature, in particular when several constructs are highly correlated!

Exporting repertory grids, element and construct descriptors

Exporting is not available in the personal edition, but there are some workarounds. E.g. open the .rgrid file with Excel or a programming editor. Then copy/paste or write a filtering script.

Here is how to import manually a grid to the free Idiogrid software:

• Open the rgrid file in Excel (hit next, next, finish). Data are TAB separated.
• Select "Enter/Edit constructs or elements" from the Edit menu of Idiogrid
• Then, paste elements, as well as left and right poles of constructs from the Excel file into the three columns.
  • Copy the element names first (they go to the right most column)
  • Excel column I pole items go to the left columns (the emergent pole)
  • Excel column H items goe to the right column (the implicit pole)
  • Click ok, then open it again and remove some extra whitespaces at the end of each item.
• Now, copying the data is more complicated since columns and lines are reversed in rgrid files.
  • Select the data area (easy to identify) and copy (ctrl-C) all the cells.
  • Then in an empty spot in the Excel worksheet copy/transpose. e.g. in Excel 2007: On the Home tab, in the Clipboard group, click the arrow below Paste, and then click Transpose.
  • Then copy this new Excel area.
  • Finally, in Idiogrid's "Grid Data" panel, click the "Paste values to grid" button in the menu bar (top right) and fix grid type (e.g. if you used a seven point scale, enter 0,6.

... I am soo proud of my Excel knowledge. More seriously, this is something you should consider if you need to do more sophisticated analysis. Unfortunately, graphics output of Idogrid's factor analysis is not as nice and it doesn't do cluster analysis (unless I didn't find it), but the program can export to SPPS that can...

So, below is output from Idiogrid for factors 1 and 2 of a three-factor "Slater analysis" (probably the same technique). Not as pretty, but as I said, you get real statistics and plus other possibilities for analysis.

Screendump from Idiogrid 2.4 with the same data

You can see that results are similar, but probably the parameters used were not exactly the same. Idiogrid users are expected to understand what they do and parametrize a lot, whereas RepGrid IV personal edition implements just a single typical (and well done) way of doing it ...

Finally, read the bloody manual too !

Discussion and background information

The free personal edition of Grid IV has some missing features that a researcher might need, but also some interesting capabilities that are not found in other systems. Overally speaking, after a single test, Daniel K. Schneider believes that this software would be useful for smaller projects or for teaching.

Ergonomics

At least on Win Vista it is not easy to see the cursor and where exactly one should click. This may be a problem with some users. However, the manual explains the GUI.

Most important missing features

Of course there is the "pro" version, but Daniel K. Schneider believes that a free version could support a few extra features

• Poles of constructs can't be changed (without editing . This is too bad, since for a casual user "high" values are associated with "positive" poles and "high values" are highlighted in the graphic.
• There should be some more statistics, e.g. the typical factor analysis data
• One should be able to export the (e.g. as command or tab separated ascii file). A workaround is to extract these with an editor from the *.rgrid file or copy/paste via Excel.

The Elicit Grid script

WebGrid has an interpreter that can read scripts (and you can find these in the installation directory)

Daniel K. Schneider found the Elicit Grid script well done. Since it is principled that is no surprise. It emulates the behavior of Shaw’s (1980) interactive repertory grid elicitation program, PEGASUS. According to the manual (available from the downloads page, it is programmed to:

1. Request missing fields in the “Status” window such as the user’s name and the purpose of the elicitation

2. To ask the user to enter six or more elements

3. To elicit constructs from triads of elements until there are four

4. To check element and construct matches and offer the user the opportunity to enter more constructs or elements to break the matches

5. To offer the user the option to elicit more constructs from triads, edit and enter elements or constructs, or to finish the elicitation

6. To ask the user to rate the elements on any given constructs when the elicitation process is finished 6-8

7. To modify the elicitation process appropriately to elicit ratings for “exchange” grids in which the elements and constructs are given but the ratings are open, “elements” grids in which the “elements” are given, and “constructs” grids in which the constructs are given.

Once steps 1 through 3 are complete the scripts loop between steps 4 and 5. At any time during the elicitation the user can click on the “Analysis” menu or the “Display,” “Focus,” or

“PrinGrid” buttons to display or analyze the grid, and then continue the elicitation.

(Grid IV manual: 6-7)

The script can be run at any time, e.g. after manually changing the data in the elements or constructs panel. If steps 1-3 are complete, it will loop again between steps 4 and 5.

The Enter Grid script

works in a similar way, except that

• it provides less explanations
• it does not use triadic elicitation for the constructs

Factor analysis

“The grid has been treated as if the elements were points plotted in an n-dimensional space defined by the constructs as axes centered on the means of the elements. The data has then been rotated through principal components analysis to spread the elements out as much as possible in a 2-dimensional plot. Slater (1976; 1977) is the definitive work on such analysis of grids. Joliffe (1986) is a good general source on principal components analysis. Gower (1966) is the definitive work on the spatial model involved, and Gower and Hand (1995) on the presentation of the analysis as a biplot.” (

Focus sorting and clustering

The manual (Gains & Shaw, 2005: 7-2) refers to Shaw, 1980; Shaw, 1981; Shaw and McKnight, 1981; Denicolo and Pope, 2001; Jankowicz, 2003).

Data format and files

RepGrid files are saved as simple ascii files with the *.rgrid extension. Elements are separated by lines and TABs. The format is not really human readable, but it's nevertheless quite understandable. E.g. you can open it with a spreadsheet program like Excel. But to do "surgery" I suggest to use a real programming editor like Emacs.

Download and manual

• Rep IV Home Page Download for Win & MacOSX, manual

WebGrid IV

The experimental web grid system is a much more powerful system than RepGrid IV personal edition. In addition to the same features you get:

• Elicitation through the web (obviously)
• More statistics
• More analysis types
• Much larger parametrization (of all features)

Todo: I wonder if one could import RepGrid *.rgrid data files in webgrid IV.

Today (Daniel K. Schneider 17:32, 14 February 2009 (UTC)), it is down, you can see some result screen captures in the repertory grid technique article in this wiki.

Bibliography

• Denicolo, P. and Pope, M.L. (2001). Transformative Professional Practice: Personal Construct Approaches to Education and Research. London, Whurr.

• Gains, Brian R. and Shaw Mildred L.G. (2005), Rep IV' Manual for Personal Version 1.10., Centre for Person-Computer Studies, [available from http://repgrid.com/RepIV/index.html Rep IV Home Page], retrieved 16:16, 14 February 2009 (UTC).

• Gower, J. (1966). Some distance properties of latent root and vector methods used in multivariate analysis. Biometrika 53 325-338.

• Gower, J.C. and Hand, D.J. (1995). Biplots. London, Chapman & Hall.

• Jankowicz, D. (2003). The Easy Guide to Repertory Grids. Chichester, UK, Wiley.

• Jolliffe, I.T. (1986). Principal Component Analysis. New York, Springer.

• Shaw, M.L.G. (1980). On Becoming A Personal Scientist: Interactive Computer Elicitation of Personal Models Of The World. London, Academic Press (now only available from http://repgrid.com or http://www.gallowglassbooks.com).

• Shaw, M.L.G., Ed. (1981). Recent Advances in Personal Construct Technology. London, Academic Press (now only available from http://repgrid.com or http://www.gallowglassbooks.com).

• Shaw, M.L.G. and McKnight, C. (1981). Think Again: Personal Problem-Solving and Decision-Making. Cobble Hill, BC, CPCS (now only available from http://repgrid.com or http://www.gallowglassbooks.com).

• Slater, P., Ed. (1976). Dimensions of Intrapersonal Space: Volume 1. London, John Wiley.

• Slater, P., Ed. (1977). Dimensions of Intrapersonal Space: Volume 2. London, John Wiley.