ChartDog Chart Generator: On-Line Manual

Close Window
Last Updated: 25 January 05

Section 1: An Introduction to ChartDog 2.0

ChartDog is a web-based application that allows you to create progress-monitoring graphs. The application follows common conventions for graphing single-subject research data (Hayes, 1981; Kazdin, 1982). With ChartDog, you can:

enter up to two sets of data to be graphed.
group data series into discrete phases on the time-series graph.
attach custom labels to each graph phase.
create customized labels for the X- and Y-axes (as well as a customized title for the chart).
compute trend (regression) lines, means, and percentage of non-overlapping data-points for each data phase.

ChartDog is a useful tool for charting and analyzing outcome data to judge whether a student has made significant improvements in academic or behavioral goals in response to classroom interventions.

Section 2: Getting Started With ChartDog 2.0

When you first open ChartDog 2.0, a 'starter' page wil be displayed (see figure below). The page directs you to select from a drop-down list the number of data points that you want to enter into the application and to click the 'Open ChartDog!' button. Immediately, the page reloads in a format that allows you to enter data to be charted.

ChartDog 2.0: Starter Page

Don't worry if you are not sure how many observations that you will ultimately chart: ChartDog allows you to add additional lines for data entry at any time

Section 3: Customizing ChartDog 2.0 Graphs

ChartDog gives you the ability to customize your data graph to suit your own needs. You can change the following chart settings:

Chart Title. You can type in a custom title for your graph (Figure 1A). Chart titles can be up to 50 characters in length.

Figure 1: Enter Graph, X-Axis, and Y-Axis Names

Figure 2: Select Line-Plot Options

Y-axis Title. When creating a label for the Y (vertical) axis of your chart (Figure 1B), you can choose from a preformatted drop-list of Curriculum-Based Measurement choices (e.g., 'CBM Correct Digits/Per 2 Mins'). Or you can enter your own custom label (up to 50 characters). If you should type in a custom label and select a choice from the drop-list, ChartDog will accept your custom label to display on the graph.
X-axis Title. The X (horizontal) axis displays on the chart the dates when time-series data were collected. When creating a label for the X axis (Figure 1C), you can choose from a preformatted drop-list of choices (e.g., 'Calendar Days'). Or you can enter your own custom label (up to 50 characters). If you should type in a custom label and select a choice from the drop-list, ChartDog will accept your custom label to display on the graph.
Line-Plot Options. ChartDog allows you to choose the specific appearance of the line-plots. For each data-series, you can select from drop-lists to determine the shape (e.g., circle, diamond, triangle) and color (e.g., red, blue, green) for line-plot elements. Figure 2A shows the drop-lists that appear on the ChartDog web form. Figure 2B displays the line-plots that appear on the graph, as dictated by the user's selection.

Figure 3: Show Numeric Values of Data-Points

Figure 4: Select Optional 'Short Names' For Data-Series

Show Data Values. For each line-plot, you can choose to show or hide the numeric value of the data-points. (Displaying values as numbers on the graph can help users to better grasp the data.) Figure 3A illustrates the button on the ChartDog form that allows you to display numeric data values. Figure 3B provides a visual example of a line-plot that includes data values.
Short Names for Data-Series. If you want to assign unique labels to your data series, you can type those labels into the ChartDog web form (Figure 4A). Each label can be up to 25 characters in length. Labels that you enter will appear in the graph legend (see example in Figure 4B). If you do not name your data series, ChartDog will assign generic names to them (e.g., "Data Series 1").
Display Data Table in Report? You have the option to have a table of data values appear in your generated chart report, to be displayed beneath the time-series chart. (Figure 5A shows how the user selects or deselects the 'Display Data Table' option, while Figure 5B presents a sample Data Table as it would appear in your chart report.)

Figure 5A: Display Data Table in Report

Figure 5B: Example of Chart Data Table

Type in Comments for Report. If you comments of your own into the box 'Type in Any Comments to Be Added to Report' (See Figure 6), that text will also be displayed in the final chart report.

Figure 6: Type Comments to Be Added to Report

Section 4: Entering Dates & Data into ChartDog 2.0

For each observation date, you can enter up to two separate data values into ChartDog and select the date on which the observation(s) were collected. ChartDog also allows you to mark multiple phase changes on the data chart and to add custom phase labels.

Entering Dates of Observations. For each data point that you enter into ChartDog, you must also enter a corresponding date on which that data value was collected. The ChartDog web form contains drop-down lists with month, day, and year (see Figures 7A, B, & C, respectively). Use these lists to select the date of the observation. Entering Dates: Time-Saving Tip!! ChartDog has a great time-saving feature for entering date information. When you select a month or year (for example, on Observation #1), ChartDog 'remembers' that each of these date-values has been selected. As you enter additional observations, if the month remains unchanged from previous data points, you do not need to select the month again. The year-value, too, does not need to be reselected for later data values if it remains unchanged. Only a new day-value must be selected for each observation.

Figure 7: Enter Dates & Data Values For Each Observation

Figure 8: Example of Labeled Graph

Entering Numeric Data. Type each data observation into the appropriate box (Figure 7D). ChartDog will accept only numeric values as valid data. The application can handle integer and decimal values, as well as negative numbers. If you are entering values for two data series, type the number value for the first data series into the box 'Obsv 1' and type the number value for the second data series into the box 'Obsv 2'. (See Figure 7D for an example of how to format the entry of two values). When you enter observations for two data series, ChartDog will plot these series as separate line-plots (as shown on the sample graph in Figure 8).

Marking Phase Changes . In ChartDog, you can mark significant 'phase changes' on the graph by clicking the 'Phase Change?' box. (See Figure 7E for an illustration of the 'Phase Change?' checkbox.) Phase changes are locations in a data series in which a significant modification has taken place in the conditions of the program or treatment being measured--e.g., a major alteration of a student's academic program. When you select a phase change for a specific observation date, ChartDog inserts a break into the charted line-plot(s) and places a vertical dividing line on the graph to mark the phase change. (Refer to Figure 8B for an example of a charted phase change).
Inserting Phase Labels . You have the option to label phase changes on your graph. Just type the content of the phase-label into the 'Phase Label?" box (Figure 7F), which will accept up to 40 characters. Your phase labels will appear as tabs on top of the time-series chart (see phase-label example in Figure 8A).

Create Chart. Once you have entered all of your data and chart settings, you create a chart simply by clicking on the 'Create Chart!' button that appears at both the start and end of Section 3: Enter Data Observations (Figure 9A). The progress-monitoring chart will appear in a separate window. See Figure 12 below for an example of a progress-monitoring chart created by ChartDog.
Erase Chart Data & Settings. If you wish to erase all of your data and chart settings in order to begin a fresh chart, you simply click on the 'Erase Chart Data & Settings' button' that appears at both the start and end of Section 3: Enter Data Observations (Figure 9A). A prompt will appear asking: Are you sure that you want to erase your chart settings and all data from this form? If you click 'OK' to the prompt, all of your data will be erased and a new ChartDog screen will appear

Figure 9A: Create Chart/Erase Chart Data & Settings

Adding More Lines for Data Entry. If you have run out of lines in which to enter observation dates and data, you can add up to 10 additional 'lines'at a time. (See Figure 9B.) Using the drop-down list that appears at the bottom of the ChartDog web page, select anywhere from 1 to 10 lines to add to the form. Then click the button 'Do it!'. The ChartDog page will reload with the specified number of additional lines for data entry appearing at the bottom of the page--and will retain your chart settings and any data that you had previously entered.

Figure 9B: Add Data Lines

Section 5: Saving Data & Chart Settings in ChartDog 2.0

ChartDog 2.0 now allows you to save your chart settings and any data that you have entered into the program. You must complete two steps to permanently save data in ChartDog:

STEP 1: SAVE DATA TO THE WEB PAGE: If no data have yet been saved to the page, you will see a red checked border displayed, along with the message 'No Data Has Been Saved to This Page'(Figure 10A).

To save data, you enter the data you wish to have appear in the chart and customize any chart settings that you wish to save. Then you click the 'Save Data & Chart Settings to This Page' button. The page will reload. A green checked border will be displayed, along with the message 'Your Data Has Been TEMPORARILY Saved to This Web Page' (Figure 10B).

Figure 10A: 'No Data Saved to Page' Message

Figure 10B: 'Data Temporarily Saved to Page' Message

STEP 2: SAVE DATA TO YOUR HARD DRIVE. Please note that your data is not PERMANENTLY saved until you save the page to your hard drive. To save a Web page, click File in the browser's Menu bar. Netscape users, click Save Page As, choose a folder to save the Web page to, and then click Save. IE users, click File and then Save As. Choose a folder and then click Save.

Once you have successfully saved a ChartDog web page with data to your hard drive, you can open that page weeks or months later and all of the data will appear on the page just as you left it. If your computer is connected to the Internet when you open the saved ChartDog page, you can immediately create a chart or add more data and resave the page to the hard drive.

Section 6: ChartDog 2.0 Data Analysis Options

You are most likely to use ChartDog to measure the impact of an intervention, or 'treatment', on a single subject (e.g., a student) over time. If that treatment results in immediate improvements, you may be able to tell just by looking at the pattern of data points on your graph that your intervention has proven highly effective. However, on many time-series graphs, it is typical for the data to vary considerably from day to day. When faced with so much 'scatter' or variability in the data points on your chart, you may find it difficult to pick out any strong underlying pattern or trend through visual inspection alone. ChartDog gives you several computational and statistical tools that you can use to help you to analyze and draw conclusions from your charted data.

Figure 11: Choose One or More Methods of Data Analysis for Each Data-Series

By selecting data-analysis tools from the ChartDog drop-list (Figure 11), you can:

compute a least-squares regression (trend) line for a data series (Figure 11A)
calculate the mean value for each phase on your data chart (Figure 11B)
compile the percentage of non-overlapping data points by phase (Figure 11C)
compute an effect size for any treatment phase when compared to a baseline phase (Figure 11D).

Figure 12: Example of a ChartDog Time-Series Chart Report

You can choose to use any or all of these tools to analyze data from a particular graph. The results of each analysis appear on the chart report, below the time-series graph (see sample chart above in Figure 12). Each data-analysis tool has its own strengths and limitations--as described below:

'Compute Trend (Regression) Lines for Data Series...' When drawn on a graph, trend lines summarize the direction and rate of change in a data series (Franklin, Gorman, Beasley, & Alison, 1996). ChartDog calculates trend by using ordinary least-squares (OLS) regression (Moore & McCabe, 1989), a statistical formula that is widely used in time-series graphs (Franklin, Gorman, Beasley, & Alison, 1996), including those used for curriculum-based measurement monitoring (Shinn, Good, & Stein, 1989). As shown in Figure 12, the computation of OLS regression results in slope and intercept values. Using these values, ChartDog draws a dotted-line through the data series that shows the best 'linear fit' of all values (see example in Figure 14A).

Figure 13: Formula for Computing Least-Squares Regression

Figure 14: Chart Examples of Regression Analysis (A) and Mean Value (B)

You can view the trend line to get a visual estimate of the direction and rate of improvement for the individual whose data are being graphed. Slope and intercept values also appear in the 'Graph Notes' window.

Cautions and Limitations in Interpreting OLS Regression:

While OLS regression is a useful tool for data analysis, you should place less confidence in its results if your data are highly variable (Moore & McCabe, 1989). When graphing Curriculum-Based Measurement data, for example, educators often discover a great deal of variability in a student's data over time. In such cases, OLS regression loses some of its predictive power. If faced with this situation, educators should consider supplementing the use of a trend line with visual inspection or other techniques to help them to judge the rate of student improvement (Dunn & Eckert, 2002).
Keep in mind that the accuracy of the least-squares regression formula diminishes with fewer data points. For this reason, ChartDog will not compute a trend-line for any data series containing fewer than 5 data points.

'Compute Means for Data Series...' For each phase in your graph, ChartDog can compute the mean value of data points present within that phase. When you select this option, you see that mean values are represented on your graph as horizontal lines within each phase (see Figure 14B for an example). Means for each data series are drawn in the same color that you select for that data series (e.g., green, red, blue). Mean values also appear in the 'Graph Notes' window.

Cautions and Limitations in Interpreting Phase Means:

The mean values of data series are useful, but do have limitations. Specifically, they provide no information about either the variability of data in the series or the trend (rate of change) occurring within that data series.
Because means for each data series represent the average of all data points, they can be skewed (distorted) by extreme (high or low) outlier values in that series

'Compute Percentage of Non-Overlapping Data Points (PNDs) for Data Series...' One method for quantifying the impact of a treatment or intervention phase in a data series is to calculate the percentage of non-overlapping data points (PNDs) in the treatment phase when compared to the baseline phase (Faith, Allison, & Gorman, 1996); Scruggs, Mastropieri, & Casto, 1987). PNDs are computed in four steps:

Decide whether the target behavior that you are measuring is intended to increase or to decrease during the treatment phase.
Count up the number of data points in the treatment phase that do not overlap the data points in the baseline phase.
- If you intend for your target behavior to INCREASE, you count only those treatment data points that are HIGHER than the highest point in the baseline data series.
- If you intend for your target behavior to DECREASE, you count only those treatment data points that fall BELOW the lowest point in the baseline data series.
Divide the number of non-overlapping data points in the treatment series by the TOTAL number of data points in the treatment series.
Multiply the figure from step 3 by 100.

A brief example will illustrate the calculation of PNDs. Figure 15 shows a series of 4 data points (treatment phase) being compared to a series of 3 data points (baseline phase). Higher data points represent the desired direction for improvement in this data series (Step 1). There are 4 data points in the treatment phase that do not overlap baseline data (Step 2).

Figure 15: Example of PND Calculation

We find that there are also a total of four data points in the entire treatment series. When we divide 4 non-overlapping data points by the total of 4 values in the series, we get a quotient of 1.0 (Step 3). We then multiply this quotient by 100 to find that the PND for the treatment phase is 100%. ChartDog displays PND values in the 'Graph Notes' window.

Cautions and Limitations in Interpreting PNDs:

A high degree of variability in baseline data may reduce the PND value of a treatment phase that follows. Extreme outlier values in baseline data may overlap a large percentage of the treatment phase data points, even if overall the treatment phase appears highly successful (Faith, Allison, & Gorman, 1996).
If you are running a single-subject reversal design, you are likely to find an 'extinction effect' when you move from a treatment phase to the second baseline phase. It is expected under these conditions that the subject's behaviors will change from treatment levels to baseline levels quite quickly with the withdrawal of the treatment. Because this accelerating or decelerating 'extinction pattern' is likely to span a large range of data values before it returns to baseline levels, any new treatment phase that immediately follows is unlikely to have a high PND value--even if the treatment appears upon visual inspection to have been successful (Scruggs, Mastropieri, & Casto, 1987).

'Compute Effect Sizes for Data Series...' Advanced users of ChartDog may wish to calculate a standardized 'effect size' for treatment data collected on an individual subject. By calculating effect sizes for each subject in a multi-participant study, a researcher can then report the mean of those values as the 'average' effect size, or outcome, of the study. Or ChartDog users may want to convert the findings of a single-subject study to an effect-size so that they can compare their results to published studies of a similar nature.

While a number of formulas exist for calculating effect sizes in single-subject designs, ChartDog employs a commonly used 'standardized difference approach' (Faith, Allison, & Gorman, 1996; Shernoff, Kratochwill, & Stoiber, 2002). Using this formula (Figure 16):

the mean of the baseline phase is subtracted from the mean of the treatment phase
the difference is then divided by the standard deviation of data values in the baseline phase.

Figure 16: Formula for Computing Effect Size Between Treatment and Baseline Phases (Faith, Allison, & Gorman, 1996)

When you prompt ChartDog to calculate effect sizes, the program begins with the second phase and cycles through all phases appearing in the chart. Treating the current phase as the 'treatment' phase and the phase preceding the current phase as the 'baseline' phase, ChartDog successively calculates an effect size for each treatment phase. Effect-size values appear in decimal format and are displayed in the 'Graph Notes' window.

Cautions and Limitations in Interpreting Effect Sizes (Faith, Allison, & Gorman, 1996):

Effect sizes do not provide information about slope (or rate of change) in the treatment data series.
Single-subject time-series data may contain positive or negative autocorrelation--which can bias effect-size values.

Section 7: About ChartDog 2.0

ChartDog is a web-based program that constructs customized progress-monitoring time-series graphs from data-values entered by the user. It is programmed in PHP, an Internet [computer] scripting language. ChartDog is built upon the foundation of JPGraph, a freeware suite of PHP code that creates dynamic charts for the Internet. JPGraph was created by Johan Persson, a Swedish programmer who has demonstrated great generosity in making his software available at no cost to web developers. Learn more about JPGraph at: http://www.aditus.nu/jpgraph.

The dog icons used throughout the ChartDog web form and manual came from Nory Island, a Japanese web site with neat graphics.

I (Jim Wright) would also like to express my appreciation to some folks who helped me to create or improve ChartDog, including:

Dr. Joe Sliker, a school psychologist with Milwaukee Public Schools. Joe was kind enough to give me a tutorial in using JPGraph when I was there in February 2003 presenting workshops to Milwaukee Public Schools staff. Without Joe's strategic assistance, ChartDog would never have been created!
Kari Sassu. As a doctoral school psychology student at the University of Connecticut, Kari diligently emailed me at several points in 2004 to alert me to bugs in the original version of ChartDog--which I was then able to fix.
School psychologist Dr. Seth Aldrich, who gave me detailed feedback on an earlier version of ChartDog and was invaluable in helping me to select features to include in this upgrade. Seth is a friend and colleague (and my neighbor!) who offered lots of valuable recommendations,including the suggestions that ChartDog allow the user to plot trendlines for every phase of data appearing in a time-series chart and that the user be able to permanently save data in ChartDog. Both of those recommendations were incorporated into ChartDog 2.0.

References
Dunn, E.K. & Eckert, T.L. (2002). Curriculum-based measurement in reading: A comparison of similar versus challenging material. School Psychology Quarterly, 17, 24-46.
Faith,M.S., Allison, D.B., & Gorman, B.S.(1996). Meta-analysis of single-case research. In R.D.Franklin, D.B.Allison, & B.S.Gorman (Eds.), Design and analysis of single-case research. (pp.245-277). Mahwah, NJ: Lawrence Erlbaum Associates.
Franklin,R.D., Gorman, B.S., Beasley, T.M., & Allison, D.B. (1996). Graphical display and visual analysis. In R.D.Franklin, D.B.Allison, & B.S.Gorman (Eds.), Design and analysis of single-case research. (pp.119-158). Mahwah, NJ: Lawrence Erlbaum Associates
Hayes, S.C. (1981). Single case experimental design and empirical clinical practice. Journal of Consulting and Clinical Psychology, 49, 193-211.
Kazdin, A.E. (1982). Single-case research designs: Methods for clinical and applied settings. New York: Oxford Press.
Moore, D.S., & McCabe, G.P. (1989). Introduction to the practice of statistics. New York: W.H. Freeman and Company.
Scruggs, T.E., Mastropieri, M.A., & Casto, G. (1987). The quantitative synthesis of single-subject research: Methodology and validation. Remedial and Special Education, 8(2), 24-33.
Shernoff, E.S., Kratochwill, T.R., & Stoiber, K.C. (2002). Evidence-based interventions in school psychology: An illustration of task force coding criteria using single-participant research design. School Psychology Quarterly, 17, 390-422.
Shinn, M.R., Good, R.H., & Stein, S. (1989). Summarizing trends in student achievement: A comparison of methods. School Psychology Review, 18, 356-370.

Close Window