1. Click to open the Analysis Definition window to build a table.
2. Specify the Analysis variables for the table or chart as Q6a TO Q6e (“How do you rate the following?”).
3. Specify the Break as Q12 (gender).
4. Select Means & Differences. Specify the Calculate variable/weight as Score5 (or another appropriate weight for the rating scale).

5. Click OK to build the table.

Each cell in the table shows the mean ratings of speed of service, cleanliness, etc., as calculated according to the values specified in Score5, both for the survey sample as a whole and for males and females separately. The various Analysis Difference, Break Difference and Base Difference options can all still be ticked.

The post Showing mean values in a grid cross-tabulation appeared first on SnapSurveys.

The only response types that can be used in means and differences tables/charts are Single Response and Quantity Response.

1. Click to open the Analysis Definition window to build a table.
2. Specify the Analysis variable for the table or chart as Q11 (age categories).
3. Specify the Break variable for the table or chart as Q12 (gender).
4. Select Means & Differences as the Calculate option.
5. Specify the Calculate variable as Q5 (amount spent).
6. The Show options on the Analysis Definition change to difference options.
   • Choose Analysis Difference if you want to see a comparison of the means and difference of men and women of each age group.
   • Choose Break Difference if you want to see a comparison of the means and difference of men in each age category and a separate comparison of the means and difference of women in each of the age groups.
   • Choose Base Difference if you want to see a comparison of the means of each cell with the base mean.

7. Click OK to build the table.

This example displays Break Difference so you can see the mean amount spent by men of each age category compared with the overall mean of amount spent for men. We can also see the mean amount spent by women of each age category compared with the overall mean of amount spent for women.

Converting to a chart

The difference values can usefully be displayed in the form of a 2D bar chart:

1. Click to re-define the table and change the form to have the Bar Counts style.
2. Uncheck the Means box and click OK to build the chart.

The post Tables and charts showing means and differences appeared first on SnapSurveys.

The only response types that can be used in means and significances tables/charts are Single Response and Quantity Response.

For example, in the Crocodile Rock Cafe survey it might be required to see whether there is a significant difference in spending in a comparison of men and women who are in the same age bracket or, alternatively, it might be necessary to look at whether there is a significant difference between the amount spent by men in different age categories and the amount spent by women in different age brackets. In other words, the first example will compare men and women of the same age, whereas the second example will compare men in age categories and there will be a separate comparison between women in age categories.

1. Click to open the Analysis Definition window to build a table.
2. Specify the Analysis variable for the table or chart as Q11 (Age categories).
3. Set the Break variable for the table or chart as Q12 (Gender).
4. Set Calculate to Means & Significances.
5. Specify the Calculate variable as Q5 (Amount spent).
6. The Show options in the Analysis Definition now show significance options.
   • Choose Analysis Significance if you want to see a comparison of means between men and women.
   • Choose Break Significance if you want to see a comparison between men of different age groups and a separate comparison of women of different age groups.

7. Click OK to build the table

Each cell in the table shows the average amount spent and the significance of that value compared with every other mean in either the same column or the same row, depending on whether Analysis Significance was selected, or Break Significance.

The table displays Break Significances. You can see that there are highly significant differences between the means of females aged under 18 and those who are 55+. Amongst men, there are no significant differences. The table also displays any significant differences between the base means (all respondents) for each age category.

Converting to a chart

On a chart you can show either the mean or the significance but not both.

1. To convert this table to a chart format, click .
2. Uncheck the Break Significance option.
3. Change the form from Table to Chart and from the drop-down menu select the style Bar Counts.
4. Click OK to build the chart.

The post Tables and charts showing means and significances appeared first on SnapSurveys.

The example describes how to display a table using the data in the Crocodile Rock Cafe survey showing the average amount spent (Q5) broken down by frequency of visit (Q2) and gender of the respondent (Q12).

1. Click to open the Analysis Definition dialog to build a table.
2. Specify the Analysis variable for the table or chart as Q2 (frequency of visit).
3. Specify the Break variable for the table or chart as Q12 (gender).
4. Set Calculate to either Means & Differences or Means & Significances. Only the means are required for this table; differences or significances will not be shown, as only the Means option is ticked.
5. Specify the Calculate variable as Q5 (amount spent).

6. Click OK to build the table.

Each cell in the table shows the average amount spent. For example, the average amount spent by males who visit the restaurant daily is 15.17.

To convert this table to a chart format

1. Click on in the table window.
2. Change the form from Table to Chart.
3. Click Browse by the Style field, and select the style Bar Counts from the drop-down list.
4. Click OK to build the chart.

The post Analyses of mean values appeared first on SnapSurveys.

If you want these figures to be the default values for the survey, you must set them in the Analysis tailoring dialog. This also has a Summary Statistics tab. You open the Analysis Tailoring dialog by selecting Analysis from the Tailor menu.

1. Open the appropriate dialog (Analysis Definition or Analysis tailoring)
2. Click the Summary Statistics tab
3. Select the Significance (t-test) in either of the columns. The possible options appear below.

• Comparison enables you to select the base used when comparing the mean of base to the mean of each category on your table. Either use:
  • Base: the mean for all respondents
  • Base less current: the mean for respondents that are not included in the category being compared.
• Score is the score applied for all calculations
• Decimal places give the number of decimal places the significance (t-test) value is calculated to. (It is usual to keep the significance (t-test) decimal places to 0 and the mean decimal places to 2.)

The post Setting the options for Significance (t-test) appeared first on SnapSurveys.

For example, in the Crocodile Rock Cafe survey it might be required to see whether there is a significant difference in perception of speed of service between different age groups.

1. Click to open the analysis Definition dialog to build a table.
2. Specify the Analysis variable for the table as Q6c (Rating of Parking).
3. Specify the Break variable for the table as Q11 (Age range).
4. Click the Summary Statistics tab.
5. Select Significance (t-test) in the Available column and click on the > button to move the option into the Used column.
6. Select Mean and click on > to move it into the Used column.
7. Specify Score5 as the score. (Score5 is the name given to a Weight created in the Crocodile Rock Cafe survey.)
8. Click OK to build the table.

The table shows the counts and percentages of each age group against the rating of speed of service. It also shows the mean rating of speed of service as specified by each age group, and the t-test (significance of the mean) in each category compared with the t-test (significance of the mean) of the base. The significance is shown as a percentage, so if we use the 95% and 99% significance levels as our markers we can see that the under 18s and 35-44 age groups have highly significant results, which means these results are likely to be repeatable in other surveys. The other age categories are not showing significant results and would not necessarily be repeated if the survey was conducted again with a similar group.

The post Using Significance (t-test) in tables appeared first on SnapSurveys.

If you want these figures to be the default values for the survey, you must set them in the Analysis tailoring dialog. This also has a Summary Statistics tab. You open the Analysis Tailoring dialog by selecting Analysis from the Tailor menu.

1. Open the appropriate dialog (Analysis Definition or Analysis tailoring)
2. Click the Summary Statistics tab
3. Select the t-test or U-test in either of the columns. The possible options appear below.

The post Setting the options for t-tests and U-tests appeared first on SnapSurveys.

The U test can be used to identify differences between the ratings for the different age groups.

1. Click to open the Analysis Definition dialog to build a table.
2. Specify the Analysis variable for the table as Q6a.
3. Specify the Break variable for the table as Q11 (Age).
4. Select the Summary Statistics tab.
5. Select U-Test in the Available column and click on the > button to move the U-Test option into the Used column.

6. Click OK to build the table.

Each cell in the table shows the number of respondents who ticked each particular age group and speed of service rating. The U-Test result displays 6 characters in each age group column representing the results of comparing a column with those of each other column. It indicates whether there is a significant difference or whether it is just a chance result.

This implies that there is a significant difference between those under 18, and those between 45 and 54. There may be other significant differences which are concealed because your sample size is too small or somehow unrepresentative.

The post Using U-tests in tables appeared first on SnapSurveys.

A table is set up with mean scores applied. The level of confidence between the mean scores for two different groupings can then be displayed by using a t-test.

For example, in the Crocodile Rock Cafe survey it might be required to see ratings for the overall performance of the restaurant question by age groups and to show the mean score for each of these groupings. The t-test can then be used to establish which mean scores are significantly different. For example, under 18s’ rating of overall performance is high; therefore, it would be useful to find out if this is a significant result and likely to be repeated in further surveys or just a chance thing.

1. Click to open the Analysis Definition dialog to build a table.
2. Specify the Analysis variable for the table as Attitude (the derived variable we set up to assess Overall rating of performance).
3. Specify the Break variable for the table as Q11 (Age).
4. Click the Summary Statistics tab
5. Double-click Mean in the Available column to move it to the Used column.
6. Enter score10 in the Score field. Instructions on creating the weight score10 are given in Calculating mean scores after banding a quantity variable

7. Double-click t-test in the Available column to move it into the Used column. Leave the t-test settings as they are.
8. Click on the Base/Labels tab. Because the analysis variable has been scored, change the analysis labels to make this obvious on the table.
   • Click Insert by the Analysis Label and select {label}. This inserts the appropriate code label of the analysis variable into the table.
   • Type “scored as” into the Analysis Label field
   • Click Insert by the Analysis Label and select {score}. This inserts the score for that code into the table.
9. Click OK to build the table.

Each cell in the table shows the number of respondents who ticked each particular age group and overall performance rating. The mean displays seven scores at the bottom of the table, which indicate the mean response for the base and for each age group. The mean score was set up so that good scores are positive and bad scores are negatives. (Excellent scores 5 through to Very Poor scoring -5). The Under 18 age group, showing a mean score of 1.98, indicates that they scored overall performance of the restaurant as good. The 55+ age group has the lowest mean score of -2.5 which represents an average score of poor.

The t-test result displays 6 characters in each age group column representing the results of comparing the mean score value of one column with those of each other column. It indicates whether there is a significant result between the two means or whether it is just a chance result. For example, is it just chance that the under 18s scored highly compared to the 55+ or is this highly significant and will be a repeatable trend?

These levels of confidence can be changed on the Summary Statistics tab of the Analysis Definition dialog. Highlight t-test and alter the numbers in the Upper Level and Lower Level dialog boxes. A 99% level of confidence means that 99 times out of 100 this group of people will always score more highly. The 100 represents 100 surveys. With this level of confidence it is highly likely that the result is repeatable and not a random result.

In the example table the “under 18” t-test result displays “-bCDEF”.

• The first character place refers to itself (under 18s) and is represented with a hyphen; no significant difference is always recorded when comparing with itself.
• The second character place displays a lower case ‘b’ indicating that the result is significant at 95% confidence between the under 18s and the 18-24s.
• Character three place has a ‘C’ recorded which represents 25-34s and indicates a significant result. This result shows 99% confidence that the Under 18 age group will be more satisfied with the overall performance than the 25 to 34 age group. This result can also be seen with the next two age groups as capitals are displayed for D, E and F.

Although the t-test is a reliable statistical method that allows you to analyse any major categories of respondents and see if the way they are responding is significantly different, when comparing more than two categories of data it can produce some unexpected results. Therefore when doing analysis across more than two categories of data it is advisable to apply Tukey’s Correction, which is available in the Summary statistics tab of the Analysis Definition dialog.

The post Using t-tests in tables appeared first on SnapSurveys.

1. Click to display the Analysis Definition dialog box.
2. Specify the Analysis as Q6e, Choice of food.
3. Specify the Break as Q11, Age range.
4. Click on the Summary Statistics tab.
5. Highlight Median from the ‘Available’ list and click > to add it to the Used column. You can position the Median scores above the body of the table by using the Move Up button.
6. Click OK to display the table.

Remember that calculating the median means finding the mid-point. In this example 65 respondents are 25-34; therefore the mid-point is 32. The 32^nd respondent in that age group responded ‘Good’ so the median is ‘Good’. 72 respondents are 35-44 and the mid-point is 36. The 36^th respondent comes within the responses of ‘OK’ and the median is therefore ‘OK’.

You can add a score to the table by checking the Scored box. The median will be calculated using the score displayed.

The post Calculating median scores appeared first on SnapSurveys.

You can calculate the mean scores of the banded variable by scoring it, so that the different codes in the derived variables are matched back to numeric values. Open the Crocodile Rock Cafe survey.

1. Click to display the Weights Window.
2. Click to add a new weight (score10) and specify the weight matrix as:

3. You can use other scoring values if you wish. For example 100, 90, 80, etc.
4. Click to display the Variables window.
5. Click to add a new variable.
6. Specify the Variable Details for derived variable which bands Q8 into ten codes. The example shown labels the bands as 1 to 10, but you could label them Poor to Excellent.

7. Click to display the Analysis Definition dialog box.
8. Enter your new derived variable in the Analysis.
9. Specify the Break as Q11, age range.
10. Click the Summary Statistics tab.
11. Double-click Mean in the left-hand column to add it to the list of statistics.
12. A box appears for you to score the mean. Specify the Score as your new score (score10).

13. Click on the Base/Labels tab. Because the analysis variable has been scored, change the Analysis labels to make this obvious on the table.
14. Click Insert by the Analysis label and select {label}. This inserts the appropriate code label of the analysis variable into the table.
15. Type “scored as” into the Analysis label field.
16. Click Insert by the Analysis label and select {score}. This inserts the score for that code into the table.

17. Click OK to create the table.

The post Calculating mean scores after banding a quantity variable appeared first on SnapSurveys.