Flashcards
SCC 202 7B - Measuring the User experience 2
Cards in this deck (20)
- 1Define descriptive statistics?Essential for any interval and ratio-level data.
Describing the data, without saying anything about the larger population. Inferential statistics let you draw some conclusions about the population beyond your participants. - 2List the common types of data analysis?Measures of central tendency
Measures of variance
Confidence interval - 3How can we measure central tendency and describe these ways?Mean- Most common in UX analysis (“average”)
Median- Useful if outliers skew the distribution
Mode- Can be useful for limited scales with few attributes - 4How can we measure variability and describe these ways?Range - Useful for identifying outliers Variance- How spread out the data is from the mean
Std. Dev - Square root of variance, most common - 5How can we measure confidence or alpha level and describe these ways?Interval around the mean where we have certain confidence that it contains the true value Report routinely for any means!
- 6What is the excel formula to work out confidence levels?EXCEL: = CONFIDENCE(alpha, standard deviation, sample size)
- 7What does confidence level show you?CONFIDENCE extrapolates to the population of which your participants are representatives.
- Estimate the range in which the true value would lie for 95% of tests with anybody in the target population. - 8What does standard deviation show you?STDDEV tells you about variability in the data.
Puts the mean into perspective; perhaps for some participants the other design was better?
BUT still only representing the participants you had. - 9Confidence intervals can be shown as error bars. Describe all the different types of results and what this shows us (overlaps)?No overlap: you can assume significant difference.
Wide overlap: no significant difference. Slight overlap: not sure. Run t test to determine significance. - 10What does the type of statistics you use depend on?The kind of statistics you can use depend on the type of data
- 11Best statistics to represent interval and ratio data? When should you use different t tests? What should use use to compare two samples?- Comparing two samples: t test gives a probability for the measured difference being due to chance.
- Different t tests for independent samples (e.g. different user groups), and for paired samples (e.g. same group of users using different designs).
- Comparing more than two samples: use ANOVA to determine significant differences across more than two groups. - 12List 5 basic data graphs?Column or bar charts.
Line charts.
Scatterplots
Pie or donut charts.
Stacked bar or column charts. - 13When should we use bar graphs? What must the axis start at?Column or bar graphs are appropriate for presenting values of ratio data (e.g. times) for discrete items (e.g. tasks, users)
Axis must start at zero and not extend beyond theoretical maximum (because length of the bar is important: conveying ratios at a glance) - 14When should we use line graphs? What do the lines only help us see? What should you set the start vertical axis to?Line graphs show trends: appropriate for presenting values of one continuous variable as function of another continuous variable.
Show your data points: the lines aren’t the data, they only help see trends. Include a legend if there is more than one line.
Normally start vertical axis at zero but not as important here - 15When should we use scatter plots? What is this good for showing? What do the variables have to be?Scatterplots show pairs of values: good for exploring relationships between two variables.
Variables have to be continuous (interval/ratio) - 16When should we use pie charts? How can we improve pie charts and what should the parts of a pie chart add up to?Pie or donut charts are appropriate only to illustrate the parts or percentages of a whole.
- Parts must add up to 100%
- Minimise number of segments
- Include percentage labels - 17What is pie chart data encoded as? The effect of this to our visual system?Data encoded as:
Area of slice.
Angle at centre.
Length of perimeter.
Our visual system is not good at decoding any of these.
Only good for rough estimates. - 18When should we use Stacked bar or column chart? What are they appropriate for presentingStacked bar charts are like multiple pie charts shown in bar form
Appropriate for presenting a series of data sets (categorical data) each showing parts of a whole.
More than three segments per bar are hard to interpret! - 19List four design principles to acknowledge when using charts/graphs?Proximity
Similarity
Multiples / ratios
Colour discrimination - 20State some pre-attentive features?Seeing things without thinking
- Size (-> Bar charts better than pie chart)
- Direction (-> Line chart)
- Colour (-> have selected items stands out)
Information
- Description
- Flashcards on SCC 202 7B - Measuring the User experience 2, created by Ayo Olutobi on 28/01/2017.
- Created by
- Ayo Olutobi
- Views
- 0
- Rating
More on Descriptive Statistics and Data Presentation