In this example of use of a Time variable, a Derived Single Response variable is used to group Q1b, “What time did you arrive?” into suitable ranges for analysis on a table or chart.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable details:
   • Name: Time
   • Label: Time of arrival
   • Type: Derived (the variable will derive its data from Q1b, the existing time of arrival question).
   • Response: Single (each respondent will fall into only one of the new codes as there is only one time of arrival per case).
4. Specify the Code Details. Any recognisable time format can be used in the Value specification but if using the 12 hour clock the am and pm suffixes must be applied

.

5. Click to save the variable. This derived variable can now be used in tables and charts in the usual way.

Categorising time responses using a time function

You can also categorise time responses using time functions. This example uses the hour function in a Derived variable to group responses to Q1b, “What time did you arrive?” into suitable ranges for analysis.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable details:
   • Name: Arrivalhour
   • Label: Hour of arrival
   • Type: Derived (the variable will derive its data from other existing variables).
   • Response: Single (each respondent will fall into only one of the new codes, as there is only one time of arrival per case).
4. Set up the code values to categorise by hour.

5. Click to save the variable. You can now use this variable to analyse by respondents’ hour of arrival.

The post Categorising time responses appeared first on SnapSurveys.

Data will have already been entered into the questionnaires; including answers to the time variables (the variables must have a response type of Time). For this example, Q1b is the time of arrival and Q1c is the time of departure.

1. Click to display the Variables window.
2. Click to add a new variable.
3. To calculate the duration of waiting for each respondent, specify the Variable Details as:
   • Name: Wait
   • Label: Duration of time
   • Type: Derived (the variable will derive its data from other existing variables).
   • Response: Quantity (the response will be a numeric value for each case).
4. Specify the Code Details to calculate the amount of time in seconds. The Not Asked value is specified to ensure that the calculation is only performed if data exists for the time question.

5. Click to save the variable. The variable can then be used to produce a statistics table of the duration of wait.

The post Calculating time intervals appeared first on SnapSurveys.

• Times can be categorised as falling into particular periods in the day, such as ‘rush hour’; ‘mid-morning peak’ or ‘lunchtime’.
• The difference between two times (for example between a time of arrival and a time of departure) can be calculated and the totals grouped into appropriate ranges, such as ’15 minutes waiting time’; ’30 minutes waiting time’ and ‘1 hour waiting time’.

Time formats

A wide range of ways of entering time data is supported.

• When time data is entered, use a consistent time format. If you use the twelve hour clock remember to add the am or pm suffix otherwise all times will be read as am.
• Any calculations of time between two periods that fall either side of midnight (e.g. 11.58 pm to 3.30 am) produces a positive number in calculations such as ‘Duration of time’ from time of departure minus time of arrival.

Time functions

When defining times, there are a number of functions that can be used to perform calculations and tests on the times.

The post Analysing time responses appeared first on SnapSurveys.

This example shows the use of the date function weekday. You define a Derived Single Response variable to work out the day of the week for a particular date.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable Details:
   • Name: V6
   • Label: Day of the week restaurant visited
   • Type: Derived (the variable will derive its data from other existing variables).
   • Response: Single (each respondent will fall into only one of the new codes, as there is only one date of visit per case).
4. Specify the Code Details:

5. Click to save the variable. The variable can then be used in tables and charts in the usual way.

Calculating age from a date of birth

These instructions tell you how to create variables to calculate the respondent’s age from their date of birth and then put the respondents into age bands, based on the difference between a date of birth variable and the date function today. The instructions also apply if you have another date question in the questionnaire, which you want to use instead of today.

For this example, the date of birth variable is Q1a. It assumes a four-digit year.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable Details as:
   • Name: Age
   • Label: Calculated age in years
   • Type: Derived (the variable will derive its data from other existing variables).
   • Response: Quantity (the response will be a numeric value for each case).

4. Specify the Code Values to calculate the number of years. Note the use of \ to extract the whole number of years.

5. Click to save the variable. You can then use this to analyse respondents’ ages.
6. The Not Asked value is specified to ensure that the calculation is only performed if data exists for the date question.
7. To band the respondents into age groups, create a new variable:
   • Name: Agegroup
   • Label: Age groups
   • Type: Derived
   • Response: Single (each respondent will only fall into one of the new codes).
8. Specify the Code Details:

9. The codes are assigned in order, so all responses that have not been sorted into previous bands appear as 65 plus.
10. Click to save the variable.

The post Date function examples appeared first on SnapSurveys.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable details:
   • Name: Date
   • Label: Quarterly visits
   • Type: Derived (the variable will derive its data from the existing date of visit question).
   • Response: Single (each respondent will fall into only one of the new codes as there is only one date of visit per case).
4. Specify the Code Details as shown in the example below. Any recognisable date format can be used in the Value specification.

5. Click to save the variable. The variable can then be used in tables and charts in the usual way.

The post Categorising date responses appeared first on SnapSurveys.

• Dates are categorised as falling in particular years, months or weeks of the year, or on particular days of the week.
• The difference between two dates can be calculated, such as calculating a respondent’s age. These can be categorised or grouped into appropriate ranges or used in the production of descriptive statistics.

Date formats

The software supports a wide range of date formats. All dates are based on the Gregorian calendar.

Day

The day number must appear in the range 1 to 31, with optional leading zeros. For example, 01, 02 are acceptable. Checks are made on the validity of a day number for the given month. For example, 31 September will be invalid. In the case of the month of February, the calculations also check whether the year is a leap year.

The day number can appear before or after the month, e.g. 1 Jan or Jan 1.

You cannot use the day suffixes such st, nd, rd, th, unless you apply a suitable date pattern to the variable.

The system settings determine the local rules for dates, such as the US standard of Month/Day/Year. Consequently, 01/05/97 will be calculated as January 5th 1997 in the US and May 1st 1997 in the UK (the computer date settings control these formats).

The characters, full stop (.), a hyphen (-) or a slash (/) separate the month, day and year. A space is used when the month is a word.

In the special case where the month name is given before the day number, a comma can be used to separate the day and year, e.g. January 5,1997

Month

The month can appear in full, e.g. January, or abbreviated to the first 3 characters, e.g. Jan, or as a number, e.g. 01.

The only exceptions to the 3 character abbreviation rule are July which can be Jul or Jly, and September which can be Sep or Sept.

The characters, full stop (.), a hyphen (-) or a slash (/) separate the month, day and year. The space character separates the month from the rest of the date if it is in word format.

Year

The year number has two formats available, either the last two digits, e.g. 97, or the full year, e.g. 1997. The characters, full stop (.), a hyphen (-) or a slash (/) separate the month, day and year.

When a year has only two digits, the default sets the year between 1930 and 2029 inclusive. In this way, 97 represents the year 1997, 00 represents 2000, and 29 represents 2029. Update the default year range by opening the Data Entry Tailoring dialog, using the Tailoring | Data Entry menu item. The Dates section contains the Date format and 2 Digit Years fields, which sets the default year range.

Date functions

When defining dates, there are a number of functions that can perform calculations and tests on the dates. The function names are not case sensitive and can be written in uppercase or mixed case.

Examples

Date functions are useful in filter expressions to return a group of data responses based on a date, for example all responses completed in a particular month or year. Filters are available in analyses, reports and when looking at responses in the data entry window.

Here are some examples of date filters using the date functions:

• ID.Date year=2023 gives responses for everyone who responded in 2023
• ID.Date year=Today year gives responses completed this year
• VisitDate Month = Today Month gives responses where respondents visited this month.
• VisitDate Month = Today Month AND VisitDate Year = Today Year returns responses where respondents visited this month in the current year, which you can use if your survey runs over a number of years
• VisitDate Month = 1 gives responses where respondents visited in January
• VisitDate Month name = “January” also gives responses where respondents visited in January
• VisitDate day = 1 gives responses where respondents visited on the first day of the month
• VisitDate weekday = 1 gives responses where respondents visited on Monday
• VisitDate weekday name = “Monday” is an alternative filter that also gives responses where respondents visited on Monday
• ID.Date weekday = (1 TO 5) gives responses completed on a weekday from Monday to Friday
• ID.Date weekday = (6 TO 7) gives responses completed on a weekend from Saturday to Sunday
• ID.Date weekday > 6 is an alternative filter which also gives responses completed on a weekend

The post Analysing date responses appeared first on SnapSurveys.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable Details:
   • Name: V5
   • Label: Amount spent
   • Type: Derived (the variable will derive its data from other existing variables).
   • Response: Single (each respondent will only fall into one of the new codes).

4. Specify the Code Details:

5. Click to save the variable. The variable can then be used in tables and charts in the usual way.

The post Categorising quantity responses with a derived variable appeared first on SnapSurveys.

You can display indexed counts or sums to give a relative measure of individual cell values.

Applying the chi-square test

The Chi-square test can be applied to single-response variables. It compares observed (actual) and expected (theoretical) values in order to establish whether there is a significant relationship between two variables in a table.

To calculate the Chi-square Statistic and related statistics for a cross-tabulation or grid table,

• select the Chi square option in the Analysis Definition dialog
• click on the table analysis toolbar

The Chi-square and associated statistics will be displayed above the table

You can display the expected values by selecting the Expected Counts option in the Analysis Definition dialog .These are the values that would be expected in each cell of the table if the row variable were not influenced by the column variable or vice-versa (the Null Hypothesis). They are calculated as:

(Row Total) x (Column Total) / Table Total (Base).

Since you cannot always compare Chi-squared values directly, Snap generates a statement about the result.

The interpretation of the Chi-square test result follows a seven point scale and categorises the certainty of a relationship (or non-relationship) between the variables in the cross-tabulation as follows:

The information given in the Chi-square report is as follows:

Detail	Meaning
Chi-square value	where O is the observed (actual) value, and E the expected value for each cell.
Degrees of freedom	Relates to the number of choices that can be made in fixing the values of the expected frequencies. It is calculated as where r is the number of rows in the table and c is the number of columns. The above is a 5 by 5 table, and therefore has 16 (= 4 4) degrees of freedom.
Cramérs V	Used to compare tables of different dimensions and different sample sizes. It is calculated as where n is the number of cases and k the smaller of the number of rows and columns. Its value always lies between 0 and 1.
Phi Coefficient	Calculated as where n is the number of cases. For 2 by 2 tables Phi always lies between 0 and 1. If one dimension is greater than 2, Phi can be greater than 1. Phi can be used to compare tables of the same dimension but different sample sizes.
Contingency Coefficient	Calculated as It will always have a value between 0 and 1. Tables of different dimensions cannot be compared.
Evidence of Relationship	A statement indicating the strength of the relationship (or non-relationship) between the row and column variables. This is used because comparison of chi-square values from tables of different dimensions or different sample sizes is meaningless.
Warning	This appears when a large number of cells in the table have small Expected Values. The reliability of the Chi-square test reduces as the proportion of cells with small expected values increases. A warning appears if more than 20% of the cells of a table have an expected value of less than 5.

Indexed Values

Indexed Values are widely used in product research and other projects to give a relative measure of individual cell values. Indexed Values can be shown on the table by checking the appropriate box (Indexed Counts or Indexed Sums) option in Definition tab of the Analysis Definition dialog.

In general, for any individual cell in the table, the corresponding index value can be calculated from the formula:

The figure of 100 represents a norm or expected value. Thus index values less than 100 would be considered lower than expected and index values of greater than 100 would be considered higher than expected.

The post Statistical calculations on tables appeared first on SnapSurveys.

• A list of Quantity Responses can be produced either for all respondents or for respondents selected on the basis of a filter specification. This is done in the same way as a list of Literal Responses
• Statistics of the quantity values can be calculated either for all respondents or for respondents selected on the basis of a filter specification.

A Derived Variable can be set up to allocate the quantity data into coded ranges. These are used as an analysis or break variable on a table.

Overview of descriptive statistics

You can produce a table of descriptive statistics which shows, for each variable in the specification, the mean, mode, median, quartiles, sum, min, max, range, standard deviation, and variance, standard error of the mean, skewness and kurtosis. Any of the statistics can be excluded using the Descriptive Statistics tab on the Analysis Definition dialog.

Descriptive statistics can be calculated for all cases or for those selected on the basis of a Filter Specification. Missing values such as No Reply will be excluded from statistical calculations. The statistics can also be weighted, perhaps to correct some imbalance in the sampling, by providing a suitable weight specification.

To describe each of the statistics that may be calculated assume that a survey of households was conducted. As part of that survey the number of persons living in the household was requested. The following replies were received from the eleven households interviewed:

1, 2, 3, 4, No Reply, 3, 4, 5, 4, 6, 2

Statistic	Description
Mean	This is often called the average, and is defined as the sum of the items divided by the number of items. Mean = (1 + 2 + 3 + 4 + 3 + 4 + 5 + 4 + 6 + 2) = 34 10 = 3.4 Since one respondent gave no reply, the mean and all other statistics are calculated using a sample of 10.
Mode	The mode of a distribution is the most frequent or most popular item. If two values tie for the mode, Snap will choose the lower. Mode = 4, since 4 is the most frequently occurring value (three occurrences).
Quartile 1	25% through a range of values
Median	The midpoint or 50% through a range of values. To calculate the median, the items of the distribution are arranged in order of magnitude starting with either the smallest or the largest, then: if the number of items is odd, the median is the value of the middle item. if the number of items is even, the median is the mean of the two middle items. 1, 2, 2, 3, 3, 4, 4, 4, 5, 6 Median = (3 + 4) ÷ 2 = 3.5
Quartile 3	75% through a range of values.
Sum	The sum is calculated by adding all the values of a distribution. Sum = 1 + 2 + 3 + 4 + 3 + 4 + 5 + 4 + 6 + 2 = 34
Minimum	The minimum is the smallest value of the distribution. Minimum = 1
Maximum	The maximum is the largest value of the distribution. Maximum = 6
Range	The range shows the spread of the distribution and is calculated by subtracting the smallest value (minimum) from the largest value (maximum). Range = 6 – 1 = 5
Standard Deviation	The standard deviation is a measure of dispersion of values in a distribution. It gives an indication of how much the values deviate from the mean. Thus, a distribution with a large range would have a larger standard deviation than one with a small range. The standard deviation is calculated as: where xi is each value in the distribution, is the mean of the values and n is the number of cases. For the sample in question: Standard Deviation = 1.428286
Variance	The variance is another measure of dispersion of values in a distribution and is calculated as the square of the standard deviation: Variance = 2.04 Snap calculates the standard deviation and variance by assuming the data represents a sample rather than an entire population.
Standard Error of the Mean	The standard error of the mean is calculated by dividing the standard deviation by the square root of the number of items in the sample. It is defined as the standard deviation of the distribution of the sample mean and gives an indication of how far individual scores deviate from the mean score shown. The larger the sample, and/or the closer the individual scores are to the mean score, the smaller the standard error. Standard Error of the Mean = 1.428286 10 = 0.451664
Skewness	A distribution that is not symmetrical but has more cases toward one end of the distribution than the other is called skewed. The measures of central tendency (mean, mode and median) can vary considerably. If the mean is larger than the mid point of the range (the median) and the most frequently occurring value (the mode), the sample is said to be positively skewed. If the mean is smaller than the mid point of the range (the median) and the most frequently occurring value (the mode), the sample is said to be negatively skewed. A small skewness value (close to 0) indicates that the data is evenly distributed about the mean. With this type of distribution it would be expected that the values for mean, mode and median be similar. The skewness of the example is 0.098843 indicating a small positive skewness.
Kurtosis	Kurtosis also gives an indication of the shape of a distribution in the form of the extent to which, for a given standard deviation, the data clusters around a central point. A positive value for kurtosis indicates a distribution that is more peaked than usual. A distribution of this type would typically have most of the values clustered around a central point. A negative value for kurtosis indicates a flatter or more widely dispersed distribution. The kurtosis for the example is -0.75202

The post Analysing quantity responses appeared first on SnapSurveys.

The example below shows coding for the free text for Other food item.

1. Click to display the Variables window.
2. Click to add a new variable.
3. Specify the Variable details:
   • Name: Q4b
   • Label: Other items ordered
   • Type: Question
   • Response: Multiple (each respondent could fall into more than one of the new codes if they ordered more than one item).
4. Specify the Code Details:

5. Click to save the variable. This variable can then be used instead of the original Literal Response variable (Q4a) for data entry.

The post Pre-coding answers appeared first on SnapSurveys.

Click to display the Variables window.

Click to add a new variable.

5. Ensure that the variable from which the derived one will gather its data has a relevant pattern applied, such as lower case.
6. Click to save the variable. The variable can then be used in tables and charts in the usual way.

Unless a pattern has been applied to the source variable, the specification of the text string must be precise. The pattern match process searches for an exact match (i.e. it is case sensitive) and looks for a capital M at the beginning of the word Milkshake and ignore lower case m. Q4a=”ilkshake” would find both.

The post Combining literal responses with a coded question appeared first on SnapSurveys.

Click to display the Variables window.

Select the variable which will act as a source to the derived variable. Double click the variable or click to open the Variable Details dialog.

Click to display the Pattern drop-down. If it is not already selected choose lower case as the pattern.

Click to add a new variable.

8. Click to save the variable. The variable can then be used in tables and charts in the usual way.

Snap is case sensitive so an appropriate pattern must be applied to the source variable to ensure that all cases with relevant comments are captured.

Categorising postcodes

The best way of analysing postcodes is to create a Derived Single Response variable and to apply a pattern to the variable.

The following instructions show how to categorise the postcodes by area.

1. Select Tailor | Patterns to open the Patterns dialog.
2. Select postcode uk from the list and click Clone.
3. Give the pattern a new name, such as UK Postcode area.
4. Clear the Result box if required. Hover over the Result box and right click.
5. Highlight Component and a list will appear containing the components of the pattern.
6. Choose area. This will now appear in the Result box.
7. Click OK to return to the Patterns dialog
8. This pattern will be used as part of an ‘as’ expression in the derived variable.

9. From the Variables Window click to add a new variable.
10. Specify the Variable Details:
    • Name: V3
    • Label: Area
    • Type: Derived
    • Response: Single (each respondent will only fall into one of the new codes)
11. Specify the Code Details:

12. Click to save the variable. The variable can then be used in tables and charts in the usual way.

If you wish to check your results you may do so by using the filter button in the Data Entry window. Using the example above, you could enter a filter value of V3=1 to ensure you have specified the value correctly for the ‘Bristol’ area.

The post Categorising literal responses appeared first on SnapSurveys.

1. Click to display the Data Entry window for the current survey.
2. Select the menu option File | Export to display the Data Export dialog box.
3. Set the Format to the required format.
4. Set the Destination for the data report as either File or Clipboard. The clipboard provides the most straightforward method of output and will be suitable for most surveys. However if there is a very large number of cases the computer may not have enough memory to hold all the required data on the clipboard in which case the file option should be used.
5. In the Filter field type the specification to select the cases to be reported.
6. For example to report only those cases where a response has been given to Q4a specify Q4a OK. Leaving the Content field blank will include all data in for each case that meets the filter specification in raw data format.
7. Specify the Content as the name of the variable or variables to be printed:
   • Individual questions (e.g. Q4a)
   • List of questions (e.g. Q1,Q4)
   • Range of questions (e.g. Q1 TO Q4)
   • Include the CASE variable to associate the data with a case number

8. Click OK and the data report will be output to the clipboard or the file specified.
9. Switch to the word processing package and paste the report from the clipboard into a document. Alternatively load the file if the File destination has been used.

The post Exporting literal responses appeared first on SnapSurveys.

Click to display the Data Entry window for the current survey.

Click to display the Print Data dialog box.

3. Specify a Title for the report if required.
4. In the Filter field type the specification to select the cases to be printed:
5. For example, to print only those cases where a response has been given to Q4a specify Q4a OK.
6. Specify the Content as the name of the variable or variables to be printed:
   • Individual questions (e.g. Q4a)
   • List of questions (e.g. Q1,Q4)
   • Range of questions (e.g. Q1 TO Q4)
7. Click Print to print the report.

The post Printing literal responses appeared first on SnapSurveys.

1. Click Analysis List on the Snap XMP Desktop toolbar to create a list definition.
2. Enter the literal variable definition, such as Q9 in the Analysis field.
3. Press Apply to display the list.

Using a derived literal response

You can use derived literals to enter other text in your list. You can also convert dates, times and quantity variables to text in derived literals.

For example, in the Crocodile Rock Cafe survey supplied with Snap XMP Desktop, there is a date variable Q1.a and a literal variable Q8 for “Any other comments”. To create a report with the information in a readable format, you can create a derived literal that writes the date a comment was received together with the comment, and then create a list analysis of the data.

1. Click to open the Variables window.
2. Click to add a new variable.
3. Set the type to Derived and the Response to Literal.
4. Enter the text “Comment'{Q9}’ was received on:{Q1.a}.”

Note that the variable names are enclosed in {} within the text. The quotation marks included in the text (‘) are different to those that mark the beginning and end of the text in the derived literal (“). (If you need to have double quotes (“) in your text, you must use single quotes (‘) to mark the beginning and end of the text.

5. You only want to list if there are comments, so in the Not Asked field, type unless Q9 ok. This means the your derived literal will be left empty if there is no comment

6. Click to save the variable.
7. Create a list as before, but use your new derived variable, V1 in the Analysis field.

The post Creating a list of literal responses appeared first on SnapSurveys.