ANOVA Analysis in SPSS PDF

Preview

Summary

Run through of ANOVA analysis in SPSS. Including how to interpret the output and writing this up. There is also an example experiment included, step by step screenshots and detailed information for those struggling with this particular analysis....

Description

SPSS and Statistics Guides

Guide to Conducting ANOVA in SPSS

Rebecca Jane Randles BSc AFHEA MBPsS

Contents Experiment Example ........................................................................... 3 What is the Design? ......................................................................... 4 Assumptions of ANOVA .................................................................... 4 Running ANOVA .................................................................................. 5 Creating Graphs ............................................................................... 7 Writing up Analyses ............................................................................ 8 Discussing the Graphs ...................................................................... 8 The Interaction Effect ..................................................................... 10 Linear and Quadratic ...................................................................... 10 Effect Analyses ............................................................................... 11

Exp erim ent Ex am pl e Experim eriment Exam ampl ple Mindfulness is a meditation-like practice that encourages attentional focus and reduces mindwandering (Williams, 2014). While there is evidence to support the efficacy of mindfulness techniques, the underlying cognitive and neural mechanisms by which mindfulness affects behavioural outcomes is unclear. Recently, the practice of mindfulness has been applied to vulnerable populations suffering from trait vulnerability to anxiety and depression with the aim to enhance attentional control and reduce anxiety and depressive-linked attentional biases for threatening material (e.g., Derakshan et al., in preparation). A major characteristic of anxious and depressed individuals is an impaired ability to disengage from the processing of threatening material and ruminative thoughts due to impairments in attentional control (see Berggren & Derakshan, 2013, for a review). Here, mindfulness was applied to see if it would result in a reduction in attentional biases for threat in anxious and depressed samples.

There were two groups of Anxious (not depressed) and Depressed (not anxious) participants: Half of whom underwent a 20 minute single session mindfulness training, with the other half a control training that involved reading a passage from a book. Assessment of attentional biases for threat was made using a visual search task where participants searched for a ‘neutral’ target face amongst a crowd of angry faces, and attentional dwell time (in ms) on each crowd face, prior to target face was recorded using eye-movements (participants completed other conditions but for the purpose of the analysis here only this condition will be discussed). Thus, longer dwell times indicate greater capture by threat and an impaired ability to disengage from the processing of the face. Participants completed a version of this task before training, a similar version immediately after training, and a week after training to see if any change in attentional bias as a result of training was sustained.

When looking at an experiment it is best to break it down so that you can understand it better. Highlight the main points of the experiment that you need to know and understand. For example, you will need to know the independent variable, dependent variable and the levels of this. The first paragraph includes a small amount of background information to help you to understand the context of the study. However, all the important information about the research is contained in the second paragraph: There were two groups of Anxious (not depressed) and Depressed (not anxious) participants: Half of whom underwent a 20 minute single session mindfulness training, with the other half a control training that involved reading a passage from a book. Assessment of attentional biases for threat was made using a visual search task where participants searched for a ‘neutral’ target face amongst a crowd of angry faces, and attentional dwell time (in ms) on each crowd face, prior to target face was recorded using eye-movements (participants completed other conditions but for the purpose of the analysis here only this condition will be discussed). Thus, longer dwell times indicate greater capture by threat and an impaired ability to disengage from the processing of the face. Participants completed a version of this task before training, a similar version immediately after training, and a week after training to see if any change in attentional bias as a result of training was sustained. This breaks the experiment down into the following:

Two groups:  

Anxious (not depressed) Depressed (not anxious)

In each of these groups half of the participants:  

Completed mindfulness training Read a book

All participants completed a visual search task at three points in time:   

Before the task Immediately after the task One week after the task

By breaking down the experiment, it seems more manageable and makes more sense to understand. You can also take information from the background itself as to what it is you are looking for. For example, in this case, the information states that the longer the participant dwells, the greater the threat therefore it is hoped that the number will reduce.

What is the Design? There are two independent variables (IV): The group (Anxious/Depressed) and the Training (mindfulness/control). “Levels” refers to the number of different categories within that IV/DV. For example, the group has two levels, as there is anxious and depressed. The dependent variable is time (before/after/one week after); this has three levels, as there are three different parts. The same participants are completing each of the different measures; therefore, this is a repeated measure design. As the participants are repeated in each of the conditions (before, after, one week after). Hypothesis can be either one tailed or two tailed. A one tailed a hypothesis indicated what direction you expect a significant result to go, where as a two tailed hypothesis indicates that you just believe there will be a significant difference but you’re unaware what this may be. You can make predictions based on the literature. For example, the background section of the example suggests that there will be a significant reduction in attentional bias when the participant undertakes mindfulness training (one tailed hypothesis). You may just want to indicate that there will be a significant result rather than saying that it will reduce, this turns it into a two tailed hypothesis. It is up to you what you want to predict based on the literature.

Assumptions of ANOVA ANOVA stands for analysis of variance and is used to determine whether there are any statistically significant differences between the means of two or more groups. Before conducting ANOVA, it is important that you are aware of the assumptions it has. This means that this statistical test assumes that your data has certain characteristics. 1. The dependent variable is continuous: This essentially means that the variable has an infinite number of possible values such as height, weight etc. 2. The independent variable has at least two categorical groups: There should at least be two groups, though usually there are three, which are categorical. In this case we have the group and we have training. These are both categorical variables.

3. No significant outliers 4. Data is normally distributed 5. Variances are equal: In this case, for ANOVA, variance is shown by sphericity. In order for this assumption to be met, the sphericity table should have a non-significant result. I will show this later.

Run ning ANOV A Running ANOVA Running ANOVA in SPSS depends upon the type of ANOVA you are running. In this case we are running a repeated measures. ANOVA is a General Linear Model. Univariate is when there is one dependent variable. Multivariate is when there are several dependant variables. Repeated measures is of course when it is repeated measures design i.e. same participants repeated.

Analyze > General Linear Model > Repeated Measures. Here is where you type in the name of your DV, name it something that you will be able to recognise easily. As discussed earlier, it has three levels: before, after and one week after. Once you have done this, click Add.

Here is where you type in the name of the measure, so for example in this case its attentional bias, I’m just going to type AB.

You then need to define these, click define.

Then click add.

Now you need to define the variables and tell the ANOVA where the levels of this measure is.

Transfer each of the time times into the time points of 1, 2 and 3.

Put the independent variables into the between-subjects factors box.

Creating Graphs Then click on the plots button to the right hand side. This is where you can add the graphs that you want to create.

In this case we are plotting separate graphs for anxiety and depression (hence groups is in the separate plots section). We want to see the difference between mindfulness and control so we want them to be on separate lines. Finally, time periods will be on the horizontal axis so we can see how the measure changes over the time points. You then need to click Add. You can also create any other plots that you wish to view that you may find useful. Once you have added all the graphs that you want, press continue.

Click the options button. Here you want to put time into “display means for”. You also want to make sure compare main effects is ticked and the “Confidence interval adjustment” is set to Bonferroni. You can also set further information; I would suggest descriptive statistics, effect sizes and homogeneity tests to be ticked. All of this information can be very useful for writing up the results. Press continue and then OK.

ritingg u up Analyse alysess Writin p An alyse Upon running the ANOVA, it is important to understand the best way to write up the analysis. This is often done in full sentences with the support of the numbers within brackets. When writing up information in a report, you generally start with descriptive statistics. This would include producing a table that including the Means and SDs, with a paragraph or two below describing what they indicate. You would then discuss the inferentials and statistical analysis.

Discussing the Graphs Below is the two graphs that have come out of the ANOVA analysis. One graph indicates the anxious group, the other graph indicates the depression group.

The two lines on each of the graph indicate the two different types of training; mindfulness and control. But what do the graphs indicate? You want to have in your mind the thing which is being measured and what it is that we want. In this case, we want a reduction in the numbers as this would mean less attentional bias. Therefore, the lower the number the better. Firstly, looking at the anxious graph; straight away you can see that the mindfulness group has a really large reduction from before to after the training. There seems to also be a slight reduction for the control group, however you can already see that the mindfulness group has a much larger reduction. To write up this information you may write something like the following, after labelling the graphs figure one for the anxious group and figure two for the depressed group:

Figure one shows the differences between the mindfulness training and the control for the anxious group of participants. The lines on this graph are indicative that the mindfulness training was potentially more effective than the control group as there is a larger drop from the before measurement of attentional bias to the after measurement. Although there is a slight increase after one week, the mindfulness group still appears to have reduced the bias a lot more than the control group.

You could then discuss the graph for the depressed group and compare how they differ. For example, you can see that after one week the depressed group appeared to go back to their original state.

The Interaction Effect In an ANOVA analysis, there are two types of effect that you can look at and comment on: 

Main effect: This is simply how one of the independent variables effects the dependant variable. For example in this case a main effect may be how the training effects the attentional bias over time. Interaction effect: This refers to how the independent variables interact with eachother in relation to the dependent variable.

Here shows the interaction effect between the two IVs and the DV. In this case you can see P= .012. This is therefore significant as P.05) You would report this as:

There was evidence for a quadratic trend for the main effect of time (F (1, 28) = 2.13, P...