A2 - Psychology - More on inferential statistics

Using statistical tests

Drawing conclusions

Inferential statistical tests help us to draw inferences (conclusions) about populations from the samples of data tested. The main aim of any research study is to make some sense of the findings and use these to help explain human behaviour. This is what we mean when we 'draw conclusions'. What does the data show us about human behaviour in general (the population rather than the sample)? In your AS course you studied deascriptive statistics. These can be useful when it comes to drawing conclusions because they provide a summary of the data. They help us detect general patterns and trends. However, you connot truly draw conclusions from such statistics because you cannot go beyond the particular sample to draw inferences about people in general (the 'population). For example, you can conclude from that people who studied an organised word list were able to remember more words compared with people who studied a random word list. However, you cannot conclude that people remember organised word lists better (i.e. organisation aids memory) because then you are assuming that people in the population will behave the same as the participants in your study. In order to draw such inferences, we need inferenitial statistical tests.

Different statistical tests

Different statistical tests are used for different research designs and for data with different levels of measurement. For example, if a study involves looking at the correlation between two variables then the test used ti determine whether or not there is a significant correlation is a correlation test such as Spearman's Correlation Test. There are five statistical tests thay you are required to study: Spearman's Rank Oder Correlation Coefficient (Spearman's Correlation Test), the Chi-squared Test, the Sign Test, the Wilcoxon Matched Pairs Signed Ranks Test (Wilcoxon T Test) and the Mann-Whitney U Test.

Observed values

Each statistical test involves taking the data collected in a study and doing some calculations which produce a signle number called the test statistic. In the case of Spearman's Rank Order Correlation Coefficient that tests statistic is called rho whereas for Mann-Whittney Test it is U. The rho value calculated for any set of data is called the observed value (because it is based on the observations made). This is sometimes also called the calculated value because it is the value…