Research methods - Data types, graphs, descriptive statistics

Quantitative data

Numerical data (eg. quiz scores, times, categories). Easy to analyse using descriptive and inferential statistics, therefore conclusions are objective and unbiased. Lack detail and oversimplifies reality so conclusions lack external validity and meaning

Qualitative data

Non-numerical/descriptive data (eg. interviews or observations). Detailed and meaningful insight so have a have a high external validity. Difficult to analyse and draw conclusions. Rely on subjective interpretations which are subject to bias (especially if there are expectations)

Primary data 

New data - data collected by the researcher for a study first hand that is specifically related to the aims +/or hypothesis for the study. Did not exist before current study was conducted. Controlled by researcher to make sure it fits the aims and hypothesis of researcher’s current study. Expensive and lengthy process

Secondary data

Old data - data collected for a purpose/study that is not the one it is being used for, it come from before the study (eg. previous study, collected by other researchers or institutions). Used by meta-analysis studies. Quicker and cheaper to access. Research may not fit the aims and hypotheses of the study as the researcher has no control over the data. 

Meta-analysis

Uses secondary data. A review of existing research. Identifies studies with the same aim and compiles data to calculate an overall effect size of difference or correlation. Increased external validity of conclusion as multiple studies means a large sample of participants is used. Prone to investigator effects when a researcher may not reference all relevant studies (leave out negative results) - incorrect conclusions would be made!

Measures of central tendency

Descriptive statistic used for quantitative data - provides researchers with an average. Consists of three measures: Mean, median and mode

Mean

The average for all participants considering the entire data set. Total data/no participants. Most sensitive measure of central tendency as it represents data as a whole by using all values in the data set. Easily distorted by extreme values

Median

Middle piece of data for all recorded data when in order. 1+num pts /2 = participant with median. Not distorted by extreme values. Less sensitive as does not include all values in data set in calculation.

Mode

Most popular/frequently occurring result from all pts. Not distorted by extreme values. Not useful when describing data if all values of data in a set are different.

Measures of Dispersion

Provide researchers with an indication of how spread out the data set is. Consists of two measures: range and standard deviation

Range

The spread of data in a data set. Largest piece of data - smallest piece of data = range. Easy to calculate. Distorted by extreme values.

Standard Deviation

Spread of data in a data set around the mean. The average distance from the mean. The lower this is the closer together the data is to the mean. Most sensitive measure of dispersion as includes all values of data set in the calculation. Distorted by extreme values

Graphical Representation

Used to represent quantitative data and acts as a visual aid to help observe patterns in data. All graphs need: a title, labels on axis, to operationalise variables clearly in the above and an appropriate scale.

Bar Graphs

They present discrete data and are divided into categories with no particular order. The space between each bar represents a lack of continuity. The IV must be operationalised on the X-axis and the DV must be operationalised on the Y-axis. 

What does a bar graph look like?

Discrete data

Data that cannot be subdivided more than it already is (eg. between sound and silence)

Continuous data

Can be meaningfully split into smaller groups (eg. Age or score on a quiz.

Histograms

The present continuous data which is divided into categories in a specific order. The bars touch to represent the continuity of the data. The IV and DV must be operationalised on the X-axis and Y-axis respectively.

What does a histogram look like?

Scattergrams

They present a correlation, one co-variable is presented on each axis. Points are plotted where the two variables meet and the pattern of points reveals the type of correlation (positive, negative or zero).

What does a scattergram look like?