Brief description of the software

When you start the e2gStats Basic app, you are presented with the following tabbed interface:

The STATS tab allows you to choose any of the twenty supported analysis methods by selecting a group of methods from the OBJECTIVE spinner, then one of the objective's methods from the METHOD spinner.

The DATA tab provides editor access to data files stored on an SD card, allows downloading of data files over the Internet and lets you create data files from scratch. The DATA tab also allows addition or deletion of variables to existing data files and provides access to data subset and missing values definitions along with a capability to compute variable values from a formula.

The HELP tab accesses three types of support documentation: (1) help on running the software (this document), (2) a reference document defining data analysis terminology and methodology and (3) a mini-course that provides examples for using the supported data analysis methods with the sample data files loaded when e2gStats Basic was installed. Buttons provide access to each of the three capabilities on the HELP tab. A second button touch jumps to a location in the currently selected help file that is relevant to the current selection on the STATS or DATA tab.

Soft keyboards for numeric data entry

To facilitate numeric data entry on smartphones with small screens, beginning with v1.2 e2gStats Basic associates a telephone style keypad with numeric data entry fields:

Beginning with v1.3, touching the Next key moves to the next data entry field when there is a set of related input values. When the last field is reached, the key changes to Done. Pressing this key after entering the field's value closes the soft keyboard.

If you prefer to use the numeric shift of a full keyboard for numeric data entry, touch the menu key, then touch the popup menu to toggle the keyboard style:

This will shift to the following keyboard for numeric fields the next time a numeric field is selected (it won't change the keyboard for the currently selected data entry field):

To revert to the phone style keypad, again touch the menu button, then touch the option:

Alphanumeric fields always associate a full keyboard starting in lower case alpha shift with the data entry field. Because it is a free-format editor, the .csv data editor supports only the full alphanumeric keyboard.

Physical keyboards on Android devices are full keyboards that don't provide Next or Done keys, but pressing the Enter key will index through related data entry fields until the last related field is reached.

Saving data on exit

[ Index ]

Exiting from e2gStats Basic is accomplished by pressing the Back key while in any of the tabs. A confirmation dialog is displayed (v1.3) in case the key was touched accidentally:

Exiting from the HELP tab may require multiple presses of the Back key, since it serves as the browser back key. To exit quickly, change to another tab before trying to exit.

If a data file has been modified and not saved, the following warning dialog is provided:

The choices presented allow exit from the program with or without saving the data file, or cancelling the exit action. Pressing the home key does not check for an unsaved file since the program is only paused, not abandoned.

Data analysis methods: the Stats Tab

Selecting an analysis objective and method

[ Index ]

The e2gStats Basic software is designed to support studies that use multiple techniques to understand a set of data. Typically, such studies would involve a data gathering instrument such as a questionnaire. Once the questionnaire has been administered, the data will be captured in a data base then approached from multiple analysis directions.

The analysis objectives supported by the app include:

• Descriptive statistics: These methods include tabular, mathematical and graphical techniques that are used to summarize and communicate the essential meaning of data sets. They represent the usual starting point for unravelling the implications of a research study. Numerical estimates of the parameters defining probability distributions are often calculated and provide insight into the shape of the underlying distribution from which a data sample has been drawn.
• Hypothesis testing: Statistical methods may be used to decide whether a data sample subject to random variation provides sufficient evidence to conclude that an hypothesized effect has occurred. These methods allow evaluation of specific research questions: "Is there evidence that scientists are happier in their jobs than managers?" "Is there evidence that users of this product younger than 30 are at least 20% more likely to buy it again?"
• Correlation, regression, multivariate analysis: Correlation methods are used to examine relationships among variables to understand how they influence each other. Models based on correlation measures may be used to support prediction. The parameters of these correlation measures and models may be estimated from data and evaluated for statistical significance.
• Analysis of variance (ANOVA): When a variable is measured under two different conditions or "treatments" hypothesis tests comparing means may be used to evaluate the effect of the treatments. ANOVA techniques extend these hypothesis testing methods to the evaluation of differences among multiple treatments and treatment levels.
• Nonparametric statistics: When it is not reasonable to make assumptions about the underlying probability distribution, or the data to be evaluated are not measured on a scale suitable for parametric analysis, nonparametric techniques may be the last resort. The methods supported by e2gStats Basic are nonparametric analogs to parametric techniques available for the hypothesis testing, correlation and ANOVA objectives.

We'll run a descriptive statistics example using a e2gStats Basic data file automatically installed with the app to look at the steps required to perform an analysis.

The "Descriptive Statistics" objective is the default selection when the app loads. Touch the METHOD spinner to see a list of the available descriptive statistics methods:

Parameter entry interface

[ Index ]

Touch the Histograms entry to select this method, then touch the Parameters for this METHOD button and you will be presented with the histogram parameter entry dialog. To define a histogram you must select the data file from the file selection spinner, choose a data subset (if any have been defined from the data tab and you want to include only a subset of the data in the analysis) and select the variable to be represented in the histogram. You may optionally select the data value for the upper bound of the first class interval and a class width. Or, you may let e2gStats Basic pick these values.

Here are the example parameters for a first look at the distribution of values for the age variable:

Analysis results

[ Index ]

Touch the Start Analysis button and the histogram is displayed:

To back up to the parameter screen the Change Parameters button is used. You might want to produce the histogram with a different starting upper bound for the first class and different class interval. Touch the appropriate fields to input values for these parameters. If your Android device uses a soft or virtual keyboard, it will cover up part of the parameter screen. You can scroll the screen before or after the keyboard appears to view the parameter fields as you enter values. After touching the back or Done keys to close the keyboard, touch the Start Analysis button to generate the histogram based on the new parameters:

The exact format of the parameter entry dialogs is specific to each analysis method, and a brief description of inputs to each method is provided with the examples of each technique included in the Mini-Course document. Parameters are provided using touch rather than typing wherever possible. Each method requires selection of a data file from a spinner, optional selection of a subset (two subset selections required for the two sample means tests), specification of the variable(s) to be used in the analysis and (for some methods) keyboard entry of numeric parameter values.

Some methods allow entry of a series of variable values: correlation and regression methods, for example. This is done by touching the check boxes for all of the variables desired in the analysis (or making repeated selections from the variable spinner prior to version 1.5). Selecting a variable previously selected removes it from the list. In the following parameter screen for the multiple regression method, the dependent variable is X7 and the independent variables are X1, X2 and X4:

Sharing/printing results via e-mail

[ Index ]

You may send the results of any analysis as an e-mail attachment formatted as a Web page by touching the Share Results button on the results screen. Select an app from the list of apps that can send an .html file as an attachment (usually e-mail programs) that appears, then complete the requested information. The file attachment is always named output.html. This is the best way to obtain a printed copy of analysis results: e-mail them as an attachment to a location with print capability, load the attachment into a browser and print it from the browser. Most browsers have a "print preview" option that allows rescaling of the output before it is printed. To include analysis outputs in presentations (PowerPoint, for example) use a screen capture of a resized print preview of the output.

In order to display or print e2gStats Basic's graphic outputs (histograms, pie charts and scatter plots) the browser must be recent enough to support standard HTML 5 drawing commands: Firefox v1.5 or later and Microsoft Explorer v9 are examples of current browsers that provide appropriate support.

If you share results with another Android device user, the HTML quick viewer typically provided with e-mail apps to examine attached documents will display text output from e2gStats Basic but not graphics. To view graphics output these users will have to save the output to the SD card (it will be saved to the SD card root), load the browser from the home page then load the shared document (output.html) into the browser using the following Web address:

file:///sdcard/output.html

Note that there are THREE slashes following the file: protocol.

Data file creation and maintenance: the Data Tab

The DATA tab allows loading an existing data file from the SD card, downloading a file from the Web (which will automatically be stored on the SD card) and creating a new data file.

Current data file size limits: a maximum of 50 variables and 200 cases.

Data file format

[ Index ]

Data input to e2gStats Basic's analysis methods are defined by two text files. Variable labels and numeric data are represented in a text file in comma separated values (.csv) format. For example:

"AGE","INCOME-K"
30,25
42,45
60,100

This example defines two variables, with values for three individuals (three cases). The first row of values represents variable labels and each must be enclosed in quotation marks and will be truncated to eight characters when displayed in analysis outputs. All values must be separated by commas. The second and succeeding rows each represent one case. These values must be numeric, must be separated by commas and may (optionally) be enclosed in quotation marks. The data files are stored on the SD card in the sdcard/e2gStatsData subdirectory that is automatically created on your Android device the first time the app is run.

A second file, with a name that matches the data file, is created only by the e2gStats Basic app to store optional supplemental or "meta" data including a data file description, subset definitions (to extract subsets of the cases for analysis) and the numeric value to be treated as a "missing value" in the analysis procedures. A case with a missing value is skipped if that value would be an input to the analysis. These supplemental values are defined or modified only through the Subset dialog accessed from either of the data editors and are stored on the SD card in the sdcard/e2gStatsSubsets directory that is automatically created when the app is first run. If a data file is deleted, the corresponding subset file will also be deleted.

Any changes made to a data file (including the subset file) that is already selected in the STATS tab will require reloading the file upon returning to the parameter or method selection dialogs in the STATS tab. Adding a new file or deleting an existing file will require a similar reselection of parameters.

Creating data files outside of e2gStats Basic

[ Index ]

Many spreadsheet and database management programs (including some Android apps) can export their data in .csv format, and so may be used to create e2gStats Basic data. If you prepare data in this way, you may need to use a file management app to move the .csv file into the sdcard/e2gStatsData subdirectory. Many Android devices are delivered with a file manager pre-installed.

For large data files, you may want to prepare the data using a PC spreadsheet or database management program, exporting it in the .csv format. Several options are available for transferring a file prepared this way to your Android device's sdcard/e2gStatsData subdirectory. You could connect the Android device to the PC with a USB cable and do a direct file transfer. If you have a Web site, you could upload the file to the Web, then download it from e2gStats Basic's DATA tab. Finally, you could e-mail the file to your Android device as an attachment. The latter option might require the use of a file management app to move the .csv file into the sdcard/e2gStatsData subdirectory.

You must make sure data prepared for e2gStats Basic satisfies the following requirements:

• Labels define the first row of data: each label enclosed in double quotes, with commas between labels. Embedded quotation marks are NOT permitted in labels (using "" to embed a quotation mark is permitted in spreadsheets, but not e2gStats Basic).
• Successive rows contain only numeric values separated by commas. The number of values on each line must match the number of labels on the first line. It is acceptable to enclose numeric values in quotation marks: these will be stripped off by e2gStats Basic after any file modifications are accomplished.
• Each line ends with a carriage return and line feed (e2gStats Basic will accept a carriage return, line feed or carriage return followed by a line feed as input, and will reformat the lines to end with a carriage return followed by a line feed the next time the file is saved).

Selecting, importing or creating a data file

[ Index ]

The source toggle (showing SD in the screen image) allows selection among the input options: SD card, Web and creating a new file. When the SD option is selected, touching the spinner provides a list of all data files in the /sdcard/e2gStatsData subdirectory that have a .csv extension. When one of these files has been selected, it may be loaded or deleted by touching the appropriate button. A progress dialog and Android "toast" message bracket the loading activity and provide an explanation if the loading is unsuccessful. Once the file is loaded, buttons providing options to View/Edit the file with the grid or csv editor are displayed.

If the source toggle is touched once, it displays Web and a Web address may be typed to download a data file over the Internet. The address may be typed with or without including the http:// protocol:

Touching the source toggle again allows a new file to be created. The file name is entered with (optionally) a .csv extension. If no file extension is provided .csv will be added. If any other extension is entered it will be stripped off and replaced with .csv.

If the data file is successfully loaded or created, access to a choice of editors is provided:

The grid editor: data entry, adding and deleting variables and cases

[ Index ]

If the View/Edit Grid button is touched after the Load Data File action, the data file is displayed along with options for modifying the data. The display format is a grid similar to a spreadsheet with columns representing variables and rows representing cases:

Data files loaded into the grid editor are error checked to make sure all values other than the variable labels are numeric and there are an equal number of values for each case. If errors are found, a dialog box detailing the error is displayed and the grid editor is not opened. You will need to repair the data file with the .csv editor or an external editing program. The .csv data files used by e2gStats Basic are text files, so many external editors are available. By default variable labels are blank and data values are set to the missing value defined with the subset editor: the default value is zero.

The active (highlighted) cell in the grid is selected by touch or by using cursor keys/d-pad/touchpad controls available on a specific Android device to move the highlighted cell. Touch the screen and drag it to reach cells outside the displayed range. Three grid resolutions are available, accessed by using the pinch-zoom gesture if supported: touch the screen with two fingers and move them apart to get larger cells or together for smaller cells. Each pinch-zoom gesture increments or decrements the resolution by one level.

[ Index ]

The content of the active cell is displayed at the top of the screen. Touching the text entry area brings up a virtual keyboard if there is no physical keyboard. Modifying the displayed cell value, then pressing the enter key on the virtual or physical keyboard, replaces the value in the active cell. If quotation marks are input in a variable label, they will be changed to single quotes. If the value entered for the variable in a case is not a numeric value, the input will be ignored.

The grid editor provides automatic cell indexing. When the active cell is in the top (variable label) row, pressing the enter key moves the active cell to the next cell on the right. When the active cell is in any other row, pressing the enter key moves the active cell one variable to the right until the last variable in the data file (defined by the last non-blank label) is reached. It then moves to the leftmost variable in the next row. To enter data in a new data file efficiently, enter all of the variable labels first, move to the first entry for case 1, then begin entering the numeric data. At the end of each case, the active cell will move to the start of the next case.

Occasionally, while indexing through the cells to enter data, display of the value being entered will stop. This appears to be a result of a documented, but as yet unresolved, problem in the Android OS. WORKAROUND: If this problem occurs while you are entering data, switch to another tab then back to the DATA tab and continue entering data.

Cells within the current range of variables and cases are displayed with a white background. Out of range cells are displayed with a gray background.

The Select X to delete/add spinner allows a column of data (representing the values of a single variable) to be deleted or added. Touch the spinner, then touch the variable that is to be deleted, or that precedes the location of the variable to be added, and the following dialog appears:

Adding a variable assigns the missing value defined by the subset editor to all cases. Deleting or adding a variable causes all following variables to be renumbered (X5 becomes X4 if X4 is deleted). A progress dialog is displayed while the data file reformatting is accomplished.

[ Index ]

Variables may also be added to the data file by entering a new variable label to the right of the current highest numbered variable. Variables should only be deleted using the Select X to delete/add spinner.

Touching one of the displayed case numbers opens the following dialog that allows a case to be deleted from the data file or new case to be inserted:

Cases may be also be added to the end of the data file by entering values for a new case below the current last case that contains data. Cases may be deleted by touching the case number and following the dialog instructions, or by changing every value in the case to the missing value -- using the dialog is preferred.

Attempts to add variables more that one column to the right of the current highest numbered variable (last column displayed with a white background) or more than one row below the current highest numbered case (last row displayed with a white background) will be ignored.

Three additional options allow searching for a value and jumping directly to a specific case or column. To use any of these options the selected cell must NOT be on the variable label line. To search in a forward direction for a specific numeric value in a cell, select a cell in the column representing the search variable, type the @ sign (or * if using the phone style keypad) followed (with no spaces) by the search value and touch the enter key:

The search will begin at the next case beyond the currently selected case, and the highlight will move to the requested value if it is found, scrolling the grid if necessary. If the value is not found, the currently selected cell remains selected.

To move to a specific case, type the # sign followed by the case number and touch the enter key:

The highlight will move to the requested case number if it is within the range of cases currently entered in the data file, scrolling the grid if necessary. If the requested case is out of range, the selected cell moves to the highest numbered in-range case.

To move to a specific column, type an X (or / if using the phone style keypad) followed by the desired column number and touch the enter key:

The highlight will move to the requested column in the currently selected case if it is within the range of variables currently entered in the data file, scrolling the grid if necessary. If the requested column is out of range, the selected cell moves to the rightmost in-range column.

It is very important with any of these three options to move the highlighted cell out of the label row before entering the @, *, #, X or / options. These are legal characters to begin a label, and that's what they will do if entered when the selected cell is a label cell !

Buttons at the bottom of the grid editor screen provide access to the subset editor and variable calculation capabilities. Both are discussed following the introduction to the .CSV editor that follows.

The .CSV editor: free format data entry, adding and deleting variables

[ Index ]

If the View/Edit CSV button is touched after the Load Data File action, the data file is displayed along with options for modifying the data. The display format is the .csv text format.

In most cases, touching the screen area that contains the data (or is blank if this is a new data file) displays a virtual keyboard that may cover part of the data. The data entry area may be dragged vertically and horizontally to allow access to all of the data.

As with the grid editor, the Select X to delete/add spinner allows a column of data (representing the values of a single variable) to be deleted or added. Touch the spinner, then touch the variable that is to be deleted, or that precedes the location of the variable to be added, and the following dialog appears:

Adding a variable assigns the missing value defined on the Subset screen to all cases. Deleting or adding a variable causes all following variables to be renumbered (X5 becomes X4 if X4 is deleted). A progress dialog is displayed while the data file reformatting is accomplished. This reformatting includes removing quotation marks if any were included around the numeric data and removing extraneous blanks.

Supplemental data: defining subsets and missing values

[ Index ]

Touching the Subset button from either editor accesses the following dialog. This dialog accepts inputs that are stored in a file separate from the data file (in the sdcard/e2gStatsSubsets directory on the SD card) with the same name as the data file. This file is automatically created by e2gStats Basic for new or imported data files, and is deleted from the SD card if its matching data file is deleted:

Three kinds of information are accessed from the dialog: a data file description that will appear in the heading of e2gStats Basic outputs ("Smartphone focus group" in the example), definition of up to four data file subsets and the numeric value to be treated as a missing value (bad data) when encountered in the data file. If the missing value is changed, returning to the edit screen by touching the View/Edit or Select File buttons results in a dialog offering the choice to replace existing missing values in the data file with the newly defined value. To save changes made using the subset dialog, you must leave the dialog by touching either the View/Edit or Select File buttons. This dialog is vertically scrollable to access input elements that may be covered by a virtual keyboard.

Modifying data with the computational facility

[ Index ]

Touching the Calc button on either editor screen accesses the variable calculation dialog:

Parameters defining the calculation of new values for a variable include designation of the target variable (always required), a constant value "C" (if the formula to be used includes C), two input variables to the calculation (designated Var1 and Var2 in the formulas requiring one or two variables), the calculation formula itself (always required) and a subset selection if new values are to be computed for only part of the data file.

Touching the Calc formula button presents a list of available computations. The result of evaluating the formula on a case-by-case basis is stored in the target variable. It is possible to use the same variable as the target and Var1 or Var2, allowing the summing of variables through successive calculations.

Here is a complete list of available calculation formulas:

• C
• Var1 + Var2
• Var1 - Var2
• Var1 * Var2
• Var1 / Var2
• Var1 + C
• C - Var1
• Var1 * C
• Var1 ^ C
• C ^ Var1
• Ln(Var1)
• Int(Var1)

The C option sets the value of the target variable to C, and no Var1 or Var2 selection is required. Ln(Var1) calculates the natural (base e) logarithm of the value in Var1 and Int(Var1) rounds the value to an integer. The ^ symbol denotes exponentiation. If any of the calculations results in an error (division by zero, for example) the target variable's value is set to the missing value defined on the subset screen.

When the Calc button is touched to start the calculations, a check is first made to make sure selection of a target variable, calculation formula and Var1/Var2 (as required for the chosen calculation formula) have been accomplished. If not, an error dialog appears, allowing selection of the remaining components of the calculation.

Using the e2gStats Basic help system: the Help Tab

e2gStats help

When the app starts, e2gStats help (this document) is the default help file loaded into the help tab. A Context button appears at the bottom of the help tab that, if touched, moves the help file to a location relevant to the current activity selected on the STATS or DATA tabs. Although the help files load offline (wi-fi not required), they are large files with many graphics, so take some seconds to load completely -- and the links within the files won't work correctly until they are loaded.

Analysis help

[ Index ]

Touching the Analysis button switches the help display to a refrence document that defines data analysis terminology and briefly describes the various analysis techniques supported by e2gStats Basic. A Context button provides quick access to information relevant to the current activity selected on the STATS tab. For example, if the multiple regression parameter entry screen is currently active in the STATS tab, the Analysis button on the HELP tab is touched, then the Context button is touched, the following information will be presented:

Mini-course

[ Index ]

The Mini-course button accesses training scenarious that provide hypothetical questionnaires and responses to illustrate the analysis methods through examples. When the Mini-course button is touched, an Example button replaces it. Touching this button moves the display to an example of the analysis method currently active on the STATS tab.

An index of all of the examples is also included in the Mini-Course.

e2gStats Basic Errors

Analysis errors

[ Index ]

Two types of analysis errors are recognized by e2gStats Basic: errors in the parameters input to an analysis method, and errors that result from a problem when the analysis computations are performed. Both types of errors are reported in a dialog box with an "OK" button that allows return to the current parameter entry dialog to correct the error. A common mistake is failing to select values for all of the parameters required by a specific method:

Another type of error is entering conflicting parameter values, for example requesting the same variable as both the dependent and an independent variable in a regression analysis:

Computational errors are generally specific to an analysis method. For example, trying to enter a variable in a stepwise regression that is a near linear combination of variables already in the equation would cause an error in the calculations. Therefore, entry of the variable is prohibited:

Data file maintenance errors

[ Index ]

All data (other than variable labels) in e2gStats Basic data files must be numeric. The data editor accessed from the DATA tab allows free format editing of the .csv formated data file. When data are loaded for analysis purposes from a method selected in the STATS tab, for data file restructuring (adding or deleting variables, for example) in the DATA tab or for editing with the grid editor, the data are examined to be sure each value is numeric. If not, data loading terminates with the following message:

After touching "OK" the .csv editor or an external editor should be used to correct the problem: in this case changing the value "6$" in the second record to a numeric value. The grid editor can't be used to fix the error because of its requirement that data be loaded into a numeric matrix.

When a data file is loaded for analysis or restructuring, the variable labels in the first row in the .csv file are loaded and counted. All successive data records must include the same number of numeric values. If there are too many or too few, a warning message like the following is provided:

Again, the appropriate response is to use the .csv editor or an external editor to correct the problem.

If the number of variables or number of cases exceeds the program limits of 50 variables and 200 cases when data are loaded from a .csv file for analysis or file restructuring, a similar error dialog is displayed. Variables or cases will have to be removed from the data file before it is usable in e2gStats Basic.