Section 1 Topic 1 Levels of Measurement Categorical Data S1T1 1 Statistics Descriptive Purpose to organise, display and summarise the data that have been collected Inferential Purpose is to make generalisations, estimates, predictions or decisions about some measure of a population from a sample. S1T1 2 Descriptive Statistics 1. 2. Begin by examining each variable by itself. Then move on to study the relationships among the variables. Begin with a graph. Then add numerical summaries for specific aspects of the data. S1T1 3 Section 1 Topic 1 Displaying and summarising categorical data What are the four levels of measurement? Why do we bother with levels of measurement? How do we display categorical data? How do we summarise categorical data? S1T1 4 Variables Life Expectancy 1. Country 2. Sex 3. Year 4. Life Expectancy 15 countries S1T1 5 Notes p.18 Variables and Values Variables Quantities about which we record information Eg: Sex, country, Income level Values How is the data recorded or coded? Sex could be coded Male, female M, F 0, 1 S1T1 6 Level of Measurement Numbers mean different things in different situations. Q: “What number did you wear in the race?” A: “5” Q: “What place did you finish in?” A: “5” Q: “How many minutes did it take you?” A: “5” S1T1 7 Level of Measurement nominal scale ordinal scale interval scale ratio scale S1T1 8 Notes p.20 The Nominal Scale Lowest level of measurement Numbers used to name or nominate and numbers can be interchanged, or changed Eg: 1= “female”, 2= “male” or 1=“male”, 2= ‘female” or 0 = “male”, 1 = “female” S1T1 9 The Nominal Scale For example, we might have the variable Location of home, with: 1 = “northern suburbs” 2 = “southern suburbs” 3 = “western suburbs” 4 = “eastern suburbs” S1T1 10 Ordinal Data Numbers are used to both label and order Example: Participants asked to rate a painting 1 2 3 4 5 least appealing less appealing unsure more appealing most appealing S1T1 11 *Exercise 3: Ordinal or Nominal? religion nominal (1 = Protestant, 2 = Roman Catholic, 3 = Other, 4 = None) ordinal year of course (1 = year 1, 2 = year 2, 3 = year 3) suburb nominal (1 = eastern, 2 = southern, 3 = central, 4 = western, 5 = northern) ordinal family income (1 = low, 2 = medium, 3 = high) S1T1 12 Notes p.21 The Interval Scale •Has properties of ordinal scale plus •Intervals between the numbers are equal •Has no true zero point S1T1 13 Notes p.21 Interval: Celsius Scale Intervals on the scale shown represent equal differences of 5oC in temperature. 0°C does not mean complete absence of heat. cannot say “a day of 40°C is twice as hot as a day of 20°C”. S1T1 14 Interval Scale Example IQ Scale Person A: 112 1. 2. 3. 4. Person B: 113 Person C: 114 They have different IQ's (nominal property of the scale) Person C scored higher on the test than person B who scored higher than person A (ordinal property of the scale) There is the same difference in intelligence (in theory at least) between person A and B as there is between B and C. We cannot say is that a person who scores 0 on an IQ test has no intelligence, nor that someone with an IQ of 150 is twice as smart as someone with an IQ of 75. S1T1 15 Ratio Scale Examples: Height Weight measured in metres, centimetres … measured in kilograms, grams… Reaction time Measure in seconds, minutes … S1T1 16 Notes p.22 Ratio Scale All properties of interval scale But “zero” means absence of the quantity Consequently ratio statements such as Alice (150cm) is “twice as tall” as Ruby (75cm) S1T1 17 *Exercise 4: Identify the level of measurement nominal political party preference (1 = Labor, 2 = Liberal, 3 = National, 4 = Other) ratio interval time taken to solve a mental puzzle in seconds self-esteem as measured on a standardised Psychological test S1T1 18 Notes p.22 *Exercise 4: Identify the level of measurement ordinal health rating (1 = excellent, 2 = good, 3 = satisfactory, 4 = poor, 5 = very poor) ratio ratio ordinal number of children weight in kilograms weight (1 = below average, 2 = average, 3 = above average) S1T1 19 Categorical and Metric Data Level of Measurement Metric Interval Categorical Ratio nominal S1T1 ordinal 20 Notes p.23 SPSS Levels of Measurement Nominal Ordinal Scale – (Interval/ratio) S1T1 21 Notes p.23