7.4.3 Factors Affecting the Choice of Statistical Test

Statistical Testing

Statistical testing provides a way of determining whether hypotheses should be accepted or rejected.

Significance – we cannot tell whether the difference in mean values found in a study is what psychologists refer to as significant. This difference may have occurred by chance, a statistical test must be conducted to find out.

Correlation Coefficients

A correlation coefficient (r) is a number between -1 and +1 that represents the direction and strength of a relationship between co-variables. The value of +1 represents a positive correlation and the value of -1 represents a negative correlation. The closer the coefficient is to -1 or +1, the stronger the relationship between the co-variables. The closer to 0, the weaker the relationship.

Descriptive statistics refer to graphs, tables, and summary statistics (such as measures of central tendency or dispersion). These are used to identify trends and analyse sets of data.

"Inferential Statistics" refers to the use of statistical tests which tell psychologists whether the differences or relationships they have found are statistically significant or not. This helps to decide which hypothesis to accept or reject.

Choosing a Statistical Test

Three factors are considered when choosing the appropriate statistical test for a set of results

1. Whether the researcher is looking for a difference or correlation – this should be obvious from the wording of the hypothesis.

2. In the case of a difference, what experimental design is being used: independent groups (unrelated design), repeated measures and/or matched pairs (both related designs) | | Non-parametric Test | Non-parametric Test | Parametric Test | |---|---|---|---| | Experimental Design | Nominal Data | Ordinal Data | Interval Data | | Repeated Measures and matched pair | Sign Test | Wilcoxon Test | Related T Test | | Independent Groups | Chi-Squared Test | Mann-Whitney U Test | Un-related T Test | | Correlations | | Spearman Rho Test | Pearson's Product Moment |

3. The level of measurement. Quantitative data can be divided into nominal, ordinal or interval data. Nominal data – data represented in the form of categories.

• Nominal data is discrete – one item can only appear in one of the categories.
• Ordinal data – data that is ordered in some way. It does not have equal intervals between each unit (it does not make sense to say someone who has rated something an 8 likes it twice as much as someone who rated it 4). It lacks precision because it is based on subjective opinion rather than objective measures and is therefore referred to as 'unsafe'.
• Interval data – data is based on numerical scales that include units of equal, precisely defined size. It is the most precise and sophisticated form of data in psychology and is a necessary criterion for the use of parametric tests (unrelated and related t-tests).