Visualising likert scale data collected at two time points.

Let’s say you have collected likert scale data at pre and post some sort of intervention, and you want to see if the intervention has changed people’s opinions.

Sample of 30 participants, and their pre and post intervention opinions.

Notes:

1 – To properly test the effects of the intervention, it is best practice to have a control, for a before-after-control-impact designed study. This is not always possible, and as such we will limit discussion to before-after data.

2 – Likert scale data is really ordinal data. Unless your sample size is large (say greater than 100), you are probably going to want to use a non-parametric method to test for significant changes in opinion. But with a smaller sample size, there are more options for visualising the data.

For likert scale data, I like to use divergent bar charts. The following graph was produced using the R library, likert.

Divergent bar charts showing what proportion of participants agree/disagree pre and post the intervention.

Looking at the divergent bar charts, it appears that the post intervention responses are slightly more positive than the pre intervention data. However this graph does not take into account the paired nature of the data. Given the small sample size, it is possible to plot individual changes in opinions.

Graph showing how individual participants opinions changed from pre to post intervention,using arrows.

This shows that, for example, all individuals who originally strongly disagreed, tempered their opinions post intervention. The graph may get a little busy with more participants. Another way of presenting is through a heat map representation of a transition matrix.

Proportion of participants that have a particular opinion post intervention, for each pre intervention opinion.

This graph shows that most participants retained the same opinion, or had a more positive opinion post intervention. The same data could also be represented using bar charts.

Proportion of participants that have a particular opinion post intervention, for each pre intervention opinion.

Since we essentially have a transition matrix, we could represent this as a network. However this does suggest ongoing transitions, while the data is based on one transition (from pre to post).

Network representation of transition matrix, generated using the function plotmat in the R library, diagram.

Finally, to test for significant change in opinion from pre to post intervention, I would suggest a non-parametric equivalent to the paired sample T-test, such as a Wilcoxon signed rank test, a permutation test or a bootstrap test.

For this data, the wilcoxon signed rank test showed a significant change in opinion from pre to post intervention (V=21, p-value=0.018).

Workshops in 2023

Wed 05 April10.00am – 12.00pmStats Sessions: Introduction to SPSS 
Wed 10 May*10.00am – 12.00pmStats Sessions: ANOVA, Linear Regression and Logistic Regression in SPSS
Wed 07 June10.00am – 12.00pmStats Sessions: Factor Analysis Using SPSS
Wed 05 July10.00am – 12.00pmStats Sessions: Mediation / Moderations / SEMs
Wed 02 Aug10.00am – 12.00pmStats Sessions: Multivariate Analysis and Repeated Measures ANOVA
Wed 06 Sept10.00am – 12.00pmStats Sessions: Introduction to R
Wed 04 Oct10.00am – 12.00pmStats Sessions: GLMs and Mixed Effects Models in R
Wed 01 Nov10.00am – 12.00pmStats Sessions: Meta-analysis Using R

Click this link for registration: HDR workshops at WSU

Workshops this year

I just finished giving my first online stats session. Here is a full list of what’s available:

Wed 16 Feb10.00am – 12.00pmStats Sessions: Introduction to Statistics
Wed 09 March10.00am – 12.00pmStats Sessions: Experimental Design, Power Analysis, Sample Size Estimation
Wed 06 April10.00am – 12.00pmStats Sessions: Introduction to SPSS
Wed 04 May10.00am – 12.00pmStats Sessions: ANOVA, Linear Regression and Logistic Regression in SPSS
Wed 01 June10.00am – 12.00pmStats Sessions: Factor Analysis Using SPSS
Wed 06 July10.00am – 12.00pmStats Sessions: Mediation / Moderations / SEMs
Wed 03 Aug10.00am – 12.00pmStats Sessions: Multivariate Analysis and Repeated Measures ANOVA
Wed 07 Sept10.00am – 12.00pmStats Sessions: Introduction to R
Wed 05 Oct10.00am – 12.00pmStats Sessions: GLMs and Mixed Effects Models in R
Wed 02 Nov10.00am – 12.00pmStats Sessions: Meta-analysis Using R
Wed 07 Dec10.00am – 12.00pmStats Sessoins: Introduction to STATA Workshop