This interactive presentation looks at different types of data and how they might be best suited to certain analysis techniques. It may to some extent seem an academic issue, however as you will hopefully see the type of data gathered dictates the type of analysis that is valid to carry out on it. To sensibly appraise a piece of research it is then important that you have some understanding of these issues, otherwise you would not be able to distinguish between a valid or invalid analysis. The buttons on the toolbar to the left are common to all the slides, the help button gives a full explanation. Other buttons on the slides will let you navigate around the main content. Some people will already be quite confident in their ability to interpret descriptive statistics used as research evidence, others less so. The interactive nature of the resource lets you learn at your own pace and check your learning.
Restart FAQs Examples Links/Reference Technical Test yourself Help Exit Data types Discrete and continuous data Paired and nonpaired data
The aim is to investigate the concept of different data types and how they might affect the analysis of data. Some of this might seem rather theoretical so don't worry if you feel your understanding remains rather superficial on this particular topic. Under the "Technical" button there are links to more active resources, these might be of interest for those who might use statistics in their own research later but not essential learning now.
Data types
More restricted in how they can be analysed
Click a link to see more detail about each type.
Nominal
Ordinal
These are data which give classes which have no real connection with numbers and cant be ordered meaningfully.
These are data which can be ordered meaningfully. They can be ranked but the distance between them has no real meaning, only their rank.
These are data which can be ordered meaningfully and the distance between them does have real meaning. However we can't say x is twice y. These are data which can be ordered meaningfully and the distance between them does have real meaning. We can also make statements like x is twice y.
Restart
Interval
Ratio
(Stevens, 1946)
Height is continuous. For example, a student could be 162.3cm tall or any number in a range, i.e. the range of human height. The number of matches in a box. The number of matches is discrete. It may be 1,2,3,4500,501 but not 2.54 or 56.79 The times taken for a person to run 100m. Time is continuous. For example, an athlete may run 100m in 10.4 seconds I may take slightly longer!
Discrete and Continuous Variables summary. A discrete variable is a variable whose possible values over the range of the data do not include all the values within that range e.g. how many children in a family. A continuous variable is a variable whose possible values include all the values within the range of the data e.g. the height of the eldest child.
Go on to Paired and non-paired data
Paired Data
Data are paired if the entries in each record are connected with each other.
Examples of Paired data: the ages and weights of a group of gymnasts the weights of a group of gymnasts before and after a training session Examples of Non-paired data : the weights of a group of gymnasts and a group of non-gymnasts the changes in weight of two groups of gymnasts given different kinds of training session
If you are not sure whether data are paired or not, think of them in two columns, consider whether rearranging the order of one of the columns would affect the data. If it would, they are paired. Paired data often occur in before and after situations. They are also known as related samples. Non-paired data can also be referred to as independent samples. Scatterplots (also called scattergrams) are only meaningful for paired data.
More on Paired data
These data are paired, just rearranging one column would make it meaningless!
A researcher is interested in the effect of exercise on weight. She weighs a group of 50 subjects then randomly ascribes half the subjects to a treatment group who get an exercise regime to follow and the other half who continue as normal. After one month she reweighs all the subjects again and calculates the difference in weight for every individual, she therefore ends up with one weight loss figure for each person and hopes to see more weight loss in the members of the treatment group. What kind of data does she have?
Summary
Although she has initially gathered paired data the data she is using for her analysis is not paired, she has the change in weight for people from two independent groups.
OK what next
Data might be paired or unpaired, if a client was measures three of four times this is still referred to as paired data and sometimes as repeated measures. Data might also be continuous or discrete. The level of measurement should also fall into the categories, Nominal, Ordinal, Interval and Ratio. This introduction has covered the basics on this topic. Use the buttons on the left to have a look at some examples then check your knowledge with the self test.
OK what next
Data might be paired or unpaired, if a client was measures three of four times this is still referred to as paired data and sometimes as repeated measures. Data might also be continuous or discrete. The level of measurement should also fall into the categories, Nominal, Ordinal, Interval and Ratio. This introduction has covered the basics on this topic. Use the buttons on the left to have a look at some examples then check your knowledge with the self test.
Examples
Example 1 do older people have a greater liking for classical music? an example of paired data.
Restart FAQs Examples Links/Reference Technical Test yourself Help Exit
Example 2 are women taller than men? an example of two groups of ratio data.
A Yes in some cases, for example we could take human heights which are ratio data and use them to place people in categories, small, medium and large. This new value would at best be ordinal data. Sometimes this is a sensible analysis technique, though always wonder why it is done and who picked the size and boundaries of the categories!
References.
Links
Definitions, Uses, Data Types, and Levels of Measurement http://www.andrews.edu/~calkins/math/webtexts/stat01.htm "StatPrimer" a useful in depth stats resource. http://www2.sjsu.edu/faculty/gerstman/StatPrimer/
Excel is not even vaguely concerned about what you do to the data you store on it! this is to some degree an advantage in flexibility of use but does allow us to undertake less appropriate analyses. It does make distinction between text, number and date/times.
SPSS is more concerned that you store data in a format that is relevant to it sensible analysis. The "Variable View" lets you set this if needed but usually the default automatic setting is fine. For some procedures SPSS might even ask to convert scale variables to categorical variables. The "Variable View" also lets you assign text labels to numerical categories (e.g. 1=Male, 2=Female), this makes output look much better. SPSS doesn't' quite use the variable types we have looked at here but the SPSS "scale" variable equates roughly to our Interval and Ratio types.
More important is the structure you choose for your data when coding it into SPSS. For more information on coding paired and non-paired data in SPSS click here.
Non-paired
A researcher measured students height and noted their gender (1=male, 10=female) to hopefully find out if males or females were generally taller. The data again can be stored in two columns BUT the Gender variable is categorical. This variable is used as a discriminatory variable by SPSS. The term "grouping variable" is often Gender Height (cm) used in preference to 1.00 193.00 1.00 185.00 "discriminatory 181.00 variable" these days, 1.00 1.00 184.00 it is the variable that 1.00 180.00 lets us discriminate 1.00 189.00 between the groups 2.00 167.00 we are interested in. 2.00 159.00 In our example here 2.00 184.00 2.00 165.00 there are just 2 but 2.00 174.00 there could be more.
2.00 2.00 2.00 2.00 2.00 172.00 177.00 166.00 169.00 167.00
Paired data.
Ordinal
Interval
Ratio
A researcher collects human body weight in kg, e.g. 67.23 kg. A researcher collects height in metres, e.g. 1.64m. A researcher collects the home town of her research subjects... Students are asked to score the taste of a new recipe of bun as; "Very good"/"Good"/"Poor"/"Very poor"
No No Yes No
No No No Yes
No No No No
Yes Yes No No
Discrete
Click or swipe the boxes to see the answer when you've had a think.
Continuous
Restart FAQs
No
The times taken for athletes to run 100m.
Yes Yes No No
Examples Links/Reference Technical Test yourself Help Exit The numbers of sprouts in 1000g bags of packed sprouts. Blood group (O, A, B, AB).
No Yes Yes
Paired data
Have a look at the examples below and see if you can decide whether the data are paired or not refer back if you want to help you fill in the table below...
Paired
Click or swipe the boxes to see the answer when you've had a think.
Not Paired
No
The times taken for each subject to run 100m and their body mass index. The number of cigarettes smoked by a subject and their forced expiatory volume (a measure of lung function).
Yes No No
Yes Yes
Help
What do the buttons do? Downloading the presentation to view at home.
Restart FAQs Examples Links/Reference Technical Test yourself Help Exit
If the link has changed simply type the words " PowerPoint viewer" into Google and follow the links to Microsoft's Office download site. The file is an executable program file that should download and saved to disk. If you are not on the Internet at home you can put it and the PowerPoint presentation files on a memory stick from a SHU PC to install and use at home. Full instructions for installing the viewer are on the Microsoft site.
Useful resources:
Key Skills Online: http://keyskills.shu.ac.uk/
There are Learning & IT Services Information and Support Desks on each campus where you can get help with your borrowing, information and IT needs. http://students.shu.ac.uk/lits/help/index.html
SHU also runs "Maths Help" sessions as part of the Education Guidance/Study Support service within the Student Services Centre.
Yes
No
References
Stevens, S.S. (1946). On the theory of scales of measurement. Science, 103, 677-680
Internet. (2001). [online]. In: The concise Oxford dictionary. (2001). 10th ed., edited by Judy Pearsall. Oxford University Press. From Oxford Reference Online. Last accessed on 7 July 2004 at: http://www.oxfordreference.com/